FINANCIAL LIBERALISATION AND INTERNATIONAL
REMITTANCES IN SUB-SAHARAN AFRICA:
A PANEL DATA ANALYSIS
by
Deodat Emilson Adenutsi
Dissertation presented for the degree of
Doctor of Philosophy in Business Management and Administration at
Stellenbosch University
Promoter: Professor Matthew K. Ocran
Co-Promoter: Professor Meshach J. Aziakpono
December 2014
Stellenbosch University http://scholar.sun.ac.za
DECLARATION
By submitting this dissertation, I, Deodat Emilson Adenutsi, declare that the entirety of the work
contained therein is my own original work, that I am the authorship owner thereof (unless to the
extent explicitly otherwise stated), and that I have not previously in its entirety or in part
submitted it for obtaining any qualification.
December 2014
Deodat Emilson Adenutsi
Copyright © 2014 Stellenbosch University
All rights reserved
ii
Stellenbosch University http://scholar.sun.ac.za
ABSTRACT
This study analyses the implications of financial liberalisation programme for international
remittance inflows with regard to the macroeconomic determinants and also the implications of
remittances for economic growth and development in sub-Saharan Africa (SSA) between 1980
and 2009. The methodological approach to the analytical framework of this study is based on
the hypothesis that financial liberalisation causes higher inflows of international migrant
remittances through official channels to augment the scarce domestic financial resources, and
to stimulate economic growth for sustainable development in capital-constrained SSA.
Prior to the macroeconometric analyses, the study addressed definitional and measurement
issues on international remittances and financial liberalisation, and provided an overview of the
macroeconomic policy environment in post-independent SSA, as well as the magnitude and the
trends in remittances received by SSA relative to other developing economies. First, the
system Generalised Method of Moment (GMM) for dynamic panel-data estimation was used to
determine the macroeconomic factors responsible for the changing trends in remittance
inflows. Then an inquiry into the impact and causal effects of financial liberalisation on
international remittance inflows in SSA following the static panel-data modelling and panel
Granger non-causality estimation procedures was undertaken. Following this, the system GMM
was further employed to examine the impact of remittances on long-run economic growth, and
the effects of remittance inflows on economic development in SSA. Essentially, the economic
development indicators considered in this study are poverty, income inequality, labour market
outcomes, human capital development, and financial development.
It is revealed in this study that the most appropriate measure of international migrant
remittances is the sum of “workers‟ remittances” and “compensation of employees” excluding
“migrant transfers”. Using remittances per capita, which the study found to be the best proxy for
remittances per migrant rather than the commonly used remittances as a percentage of GDP, it
is shown that SSA is the least recipient of official migrant remittances in the world, with no SSA
country receiving remittances worth US$1 per day. This study further establishes that the
macroeconomic factors that influence remittance inflows in SSA have varying rather than static
impact in response to changing macroeconomic policy environment. Also, macroeconomic
factors have different influences on attracting remittances from abroad in relation to migrant
duration status – permanent or temporary. Although financial liberalisation Granger-causes
iii
Stellenbosch University http://scholar.sun.ac.za
international remittances, not sufficient evidence exists that a significant proportion of the
official remittances received in SSA passes through the banking system. Besides, the extent to
which financial liberalisation can Granger-cause and/or positively impact on international
remittance inflows in SSA is directly and ultimately conditional to the macroeconomic
fundamentals of the remittance-receiving SSA country.
It was also found out that generally, international migrant remittances propel higher economic
growth in SSA, with greater impact on SSA countries with relatively higher growth rates.
International remittance inflows have significant positive developmental impact, with no
sufficient evidence of moral hazard effects. Overall, international remittances contribute to
reducing poverty and unemployment but not necessarily income inequality and, at worse,
remittances have no significant impact on labour productivity and participation in SSA. Higher
remittance inflows promote human welfare, educational attainment, life expectancy, and
financial development in SSA. With the exception of educational attainment, the developmental
effects of remittances vary across countries, depending upon the level of economic
development.
KEYWORDS:
Financial Liberalisation, Financial Development, International Remittances, Economic Growth,
Economic Development, Migrants, Panel Data Analysis, Developing Countries, System GMM,
Panel Fixed Effects, Panel Random Effects, sub-Saharan Africa
iv
Stellenbosch University http://scholar.sun.ac.za
DEDICATION
I dedicate this work to the following:
The memory of my mom, Mrs Salome Ama-Dapaah Adenutsi (1942-2002)
for giving me everything I need to become who Jehovah wants me to become.
Mrs Gifty Dzifa Adenutsi, for her unlimited support, faithfulness, love and care.
Christian Fafa Adenutsi, Sally Ama-Dapaah Adenutsi, Portia Esenam Adenutsi, and Cyril
Yohannes Mawuyram Adenutsi, for being my “royals”, and motivation for success.
v
Stellenbosch University http://scholar.sun.ac.za
ACKNOWLEDGEMENTS
I owe the beginning and the successful completion of this dissertation to my promoter,
Professor Matthew K. Ocran of Nelson Mandela Metropolitan University. I am grateful for his
inspiring role, critical but generous and motivational review comments, an unimaginably high
level of tolerance and the trust and confidence he had in me prior to my enrolment, and
throughout the pursuit of my programme. It is his excellent leadership, guidance and sacrifice
that urged me on to complete this programme at the University of Stellenbosch.
May l also express my sincere appreciation to Professor Meshach J. Aziakpono, who cosupervised this dissertation, for what I call „a luxury of time supervision‟, as measured in his
detailed critical reviews and suggestions. Besides the thorough academic guidance, on a
number of occasions, Professor Aziakpono went the extra mile to offer priceless life counselling
to me, especially during the early stages of writing this dissertation. In all these, Professor
Aziakpono also exhibited a lot of patience and tolerance, worthy of admiration.
No amount of words can be sufficient to express my heartfelt appreciation to my wife, Mrs. Gifty
Dzifa Adenutsi, for her love, motivation, spiritual and financial support, and other sacrifices
throughout the period of my writing this dissertation. It is her unlimited sacrifices that gave me
the peace of mind l needed to undertake this study and complete the programme in good time.
And to my pride, pleasure and “divine royals”, Fafa, Ama-Dapaah, Esenam, and Mawuyram, I
owe a lifetime debt of gratitude for their wonderful spiritual support and outstanding motivation.
I am grateful to the University of Stellenbosch for granting me its lucrative bursary for
Outstanding PhD Research Student to enable me study full-time at its prestigious Graduate
School of Business throughout the entire duration of the programme. I admit that, without this
bursary, it would not have been possible for me to pursue this programme on a full-time basis.
My thanks also go my employers, Central University College, Accra, Ghana for granting me
study leave to enable me to undertake this programme on a full-time basis.
I am very thankful to Professor Frikkie J. Herbst (Graduate School of Business, University of
Stellenbosch) and Professor Evan Gilbert (Department of Economics, University of
Stellenbosch), not only for the confidence they had in me prior to my enrolment into the
vi
Stellenbosch University http://scholar.sun.ac.za
programme, but also for motivating me, virtually on daily basis, whilst writing this dissertation.
Professor Nicolas Biekpe of Africagrowth Institute, Cape Town, and the University of Cape
Town, who initiated my admission process whilst he was with the University of Stellenbosch, is
also hereby duly acknowledged.
To my father, Mr. Seth Ahia Adenutsi, I shall for ever remain grateful for his disciplinary
upbringing and the spirit of honesty, diligence and determination he instilled in me, which have
enabled me to surmount very difficult moments in the course of undertaking this study. Indeed,
the inspiration and unflinching support of my father went a long way, not only to help me
commence and complete this programme on schedule, but also in convincing me to complete
my PhD programme at a highly reputable academic institution such as the Graduate School of
Business, University of Stellenbosch.
To my very best friend and spiritual brother, Christian R.K. Ahortor (West African Monetary
Institute, Headquarters, Accra; and University of Cape Coast, Ghana) who supported me and
my family, materially and spiritually from the very beginning to the very end of this programme,
I am extremely grateful.
The completion of this programme would have been a mirage without the contribution of certain
important personalities at the Graduate School of Business, University of Stellenbosch (USB).
Deserving of particular mention are Professor John Powell (Director), Professor Eon Smit
(immediate past Director), Professor Sylvanus Ikhide (Head of PhD Programme), Professor
Charles D.K. Adjasi (Graduate School of Business), Mrs Marietjie van Zyl (Senior
Administrator, PhD Programme), Mrs Norma Saayman (Assistant to Prof. Ikhedi), the USB IT
staff, and the USB library staff.
I extend my hand of appreciation to Abdul Abiad, Martin Schindler, Enrica Detragiache, Rabah
Arezki, (all of the International Monetary Fund (IMF), Washington, D.C., USA) and Richard H.
Adams jr. (World Bank, Washington, D.C., USA) for supplying me with very important data and
reading materials without which the timely completion of this dissertation would have been
impossible.
I further take pleasure in acknowledging the African Economic Research Consortium (Nairobi,
Kenya) for granting me the PhD Research Award which contributed in no small measure
vii
Stellenbosch University http://scholar.sun.ac.za
toward the successful completion of this study.
My special appreciation goes to a host of other individuals and institutions that assisted me in
various ways, ranging from personal motivation, prompt response to my requests for data and
literature, and reviewing comments on portions of the dissertation presented at conferences,
seminars, and the regular USB PhD colloquia. Even though it is not possible for me to provide
an exhaustive list of this supportive group of individuals and institutions, the following cannot go
unmentioned: IMF (Washington, D.C., USA), United Nations Economic Commission for Africa
(Addis Ababa, Ethiopia), West African Monetary Institute (Accra, Ghana), and the World Bank
(Washington, D.C., USA). I would also like to mention country-desk officials at the African
Department of the IMF and the World Bank (Washington, D.C., USA), the West African Central
Bank (Dakar, Senegal), the Central (or the Federal) Banks of Ethiopia, Kenya, Madagascar,
Mozambique, South Africa, Tanzania, and Uganda. Others who have contributed in various
important ways include Alina Carare (Deputy Division Chief, African Division, IMF Institute,
Washington, D.C.); Dalia S. Hakura (Deputy Division Chief, IMF Institute, Washington, D.C.);
and Charles A. Yartey (Economist, IMF, Washington, D.C.).
I sincerely appreciate the professional editorial support services of Mrs Melanie Bailey (Cape
Town, South Africa), and Mr. David Doade (Accra, Ghana) for their thorough editorial review
which has in no small way improved the quality of this work.
My gratitude also goes to all others who have contributed in one way or the other to the
successful completion of this dissertation, but whose names I have not specifically mentioned.
Finally, and most importantly, to Jehovah, the Omnipotent God, be the glory and praise.
Nonetheless, I should be held solely responsible for any error or omission that remains in this
dissertation.
ADENUTSI, D.E.
Graduate School of Business
University of Stellenbosch
Republic of South Africa
September 14, 2013.
viii
Stellenbosch University http://scholar.sun.ac.za
ACRONYMS AND ABBREVIATIONS
A-B test
Arellano-Bond test statistic
AAF-SAP
African Alternative Framework to Structural Adjustment Programme for
Socioeconomic Recovery and Transformation
ADI
African Development Indicators
AfDB
African Development Bank
AFRODAD
African Forum and Network on Debt and Development
AIC
Akaike Information Criterion
AID
Foreign Aid
APPER
Africa‟s Priority Programme for Economic Recovery
AR
Autoregressive
AU
African Union (formerly known as Organisation for African Unity)
B-P stat
Breusch-Pagan test statistic
BKS
Banking Supervision
BoP(S)
Balance of Payments (Statistics)
BT
Breitung t-statistic
CfA
Commission for Africa
COMP(PC)
Compensation of Employees (per capita)
Cor_
Correlation Index
CPI
Consumer Price Index
DCRR
Directed Credit, Reserve Requirement and Aggregate Credit Ceilings
EAP
East Asia and the Pacific
EBC
Entry Barriers or pro-Competition
ECA
Europe and Central Asia
e.g.
exempli gratia (= for example)
EG2S
Engle-Granger 2-Step
ERP
Economic Recovery Programme
et al.
et alii (= and other people)
etc.
et cetera (= and other similar things)
FDI
Foreign Direct Investment
FDV
Financial Development
FE
Fixed Effects
FLB(I)
Financial Liberalisation (Index)
ix
Stellenbosch University http://scholar.sun.ac.za
GDF
Global Development Finance
GDP
Gross Domestic Product
G(F)CF
Gross (Fixed) Capital Formation
GGFCE
General Government Final Consumption Expenditure
GLS
Generalised Least Squares
GMM
Generalised Method of Moment
GXP
Government Expenditure
H
Hypothesis
HCA
Human Capital Accumulation
HDR
Human Development Report
HFCE
Household Final Consumption Expenditure
HHC
Hadri Heteroskedasticity Conditional z-statistic
HIPC
Heavily Indebted Poor Countries
HIV/AIDS
Human Immune-Deficiency Virus / Acquired Immune Deficiency Syndrome
ibid
ibidem (= in the same place)
IBRD
International Bank for Reconstruction and Development
ICF
International Capital Flows
ICRG
International Country Risk Guide
IDA
International Development Association
i.e.
id est (= that is)
IFS
International Financial Statistics
IMF
International Monetary Fund
INF
Inflation
INS
Institutional Quality
INV
Investment
IOM
International Organisation for Migration
IPS
Im, Pesaran and Shin
IRC
Interest Rate Control
IV
Instrumental Variable
JUCR
Johansen Unrestricted Cointegration Rank
KPSS
Kwiatkowski-Phillips-Schmidt-Shin
LAC
Latin America and Caribbean
LDCs
Less Developed Countries
LIBOR
London Interbank Offered Rate
x
Stellenbosch University http://scholar.sun.ac.za
LLC
Levin, Lin and Chu
LM
Lagrange Multiplier
ln
Natural Logarithm
Log
Logarithm
LSDV
Least Squares Dummy Variable
M
Imports
M2
Money plus Quasi Money (or Broad Money)
M2/GDP
Broad Money-GDP ratio
MDG
Millennium Development Goal
MDV
Medium-Dummy Variable
MNA
Middle East and North Africa
MoF
Ministry of Finance
MRem
Migrant Remittances
MRF-2011
Migration and Remittances Factbook 2011
MRPC
Migrant Remittances per capita (also represented by REMPC)
MRPM
Migrant Remittances per Migrant
MT
Migrant Transfers
MTOs
Money Transfer Operators
NELM
New Economics of Labour Migration
NEPAD
New Partnership for Africa‟s Development
NGO(s)
Non-Governmental Organisation(s)
NPISHs
Non-Profit Institutions Serving Households
OAU
Organisation for African Unity (now called the African Union)
Obs
Observations
ODA
Official Development Assistance
OECD
Organisation for Economic Co-operation and Development
OLS
Ordinary Least Squares
op. cit.
opposite citation
OPN
Openness to Trade
P-P
Phillips-Perron
PCA
Principal Component Analysis
PGARCH
Panel Generalised Autoregressive Conditional Heteroskedasticity
PPP
Purchasing Power Parity
PRGF
Poverty Reduction and Growth Facility
xi
Stellenbosch University http://scholar.sun.ac.za
PRS
Poverty Reduction Strategy
PSC
Bank Credit to Private Sector
PVZ
Privatisation
R&D
Research and Development
RE
Random Effects
REER
Real Effective Exchange Rate
REMGDP
Migrant Remittances as ratio to GDP
REMPC
Migrant Remittances per Capita (same as MRPC)
RIR
Real Deposit Interest Rate
RXR
Real Exchange Rate
SADC
Southern African Development Community
SALs
Structural Adjustment Loans
SAP
Structural Adjustment Programme
SAS
South Asia
SIC
Schwarz Information Criterion
SMEs
Small and Medium Scale Enterprises
SMK
Stock Market Development
SOEs
State-Owned Enterprises
SSA
Sub-Saharan Africa
Sys-GMM
System GMM
2SLS
Two-Stage Least Squares
TFP
Total Factor Productivity
ToT
Terms of Trade
UK
United Kingdom
UN
United Nations
US(A)
United States of America
US$
US Dollars
VAT
Valued Added Tax
viz.
dated (= namely)
WB
The World Bank
WDI
World Development Indicators
WEO
World Economic Outlook
WREM(PC)
Workers‟ Remittances (per capita)
X
Exports
xii
Stellenbosch University http://scholar.sun.ac.za
TABLE OF CONTENTS
Declaration ................................................................................................................................................... ii
Abstract ....................................................................................................................................................... iii
Dedication..................................................................................................................................................... v
Acknowledgements ..................................................................................................................................... vi
Acronyms And Abbreviations ...................................................................................................................... ix
CHAPTER ONE ...........................................................................................................................................1
GENERAL INTRODUCTION........................................................................................................................1
1.0 Introduction .........................................................................................................................................1
1.1 Background ........................................................................................................................................1
1.2 The Research Problem ......................................................................................................................5
1.3 The Research Questions....................................................................................................................7
1.4 The Research Objectives ...................................................................................................................7
1.5 Motivation for the Study......................................................................................................................8
1.6 Specific Motivations and the Research Hypotheses ..........................................................................9
1.7 Scope ...............................................................................................................................................11
1.8 Structure of the Dissertation .............................................................................................................11
1.9 Chapter Summary and Conclusions ................................................................................................13
Appendix 1..............................................................................................................................................14
CHAPTER TWO .........................................................................................................................................15
CONCEPTUAL FRAMEWORK AND MEASUREMENT ISSUES ..............................................................15
2.0 Introduction .......................................................................................................................................15
2.1 International Remittances.................................................................................................................15
2.1.1 Concept Definition .....................................................................................................................15
2.1.2 Measurement of International Remittances ..............................................................................17
xiii
Stellenbosch University http://scholar.sun.ac.za
2.2 Financial Liberalisation .....................................................................................................................24
2.2.1 Concept Definition .....................................................................................................................24
2.2.2 Measurement of Financial Liberalisation...................................................................................26
2.3 Chapter Summary and Conclusions ................................................................................................31
Appendix 2..............................................................................................................................................33
CHAPTER THREE .....................................................................................................................................38
MACROECONOMIC ENVIRONMENT AND EXTERNAL CAPITAL FLOWS TO SUB-SAHARAN AFRICA
(1960-2009) ................................................................................................................................................38
3.0 Introduction .......................................................................................................................................38
3.1 Background ......................................................................................................................................38
3.2 A Contextual Analysis of Policy Environment and Macroeconomic Performance of SSA (19602009) ......................................................................................................................................................40
3.2.1 The Pre-Reforms Era (1960-1979) ...........................................................................................41
3.2.2 The Reforms Era (1980-1989) ..................................................................................................42
3.2.3 The Post-Reforms Era (1990-2009) ..........................................................................................46
3.2.4 Macroeconomic Performance and Policy Environment in SSA ................................................47
3.3 External Capital Flows to SSA (1960-2009).....................................................................................50
3.3.1 Composition and Trends in External Capital Flows to SSA: A Global Outlook.........................50
3.3.2 The Dynamics of Remittances and the Macroeconomic Environment in SSA .........................57
3.4 The Stylised Facts of Migrant Remittance Flows to SSA.................................................................66
3.5 Remittances and Macroeconomic Policy Imperatives in SSA .........................................................68
3.6 Chapter Summary and Conclusions ................................................................................................69
Appendix 3..............................................................................................................................................72
xiv
Stellenbosch University http://scholar.sun.ac.za
CHAPTER FOUR .......................................................................................................................................83
MACROECONOMIC DETERMINANTS OF INTERNATIONAL REMITTANCES IN SUB-SAHARAN
,
AFRICA ......................................................................................................................................................83
4.0 Introduction .......................................................................................................................................83
4.1 Background ......................................................................................................................................83
4.2 Selected Stylised Facts on Remittance Flows to SSA .....................................................................87
4.2.1: The Cyclical Behaviour of Remittance Flows to SSA, 1980-2009 ...........................................88
4.2.2 The Composition of Migrant Remittances Received in SSA, 1980-2009 .................................90
4.2.3 Migratory Patterns in SSA: Main Destinations and Sources of Remittances............................92
4.3 Literature Review .............................................................................................................................95
4.3.1 The Microeconomic Foundation and Theoretical Underpinnings of Remittances ....................95
4.3.2 Theoretical Review of Macroeconomic Determinants of Remittances ...................................106
4.3.3 Empirical Review of Macroeconomic Determinants of Remittances ......................................112
4.4 Theoretical Framework...................................................................................................................115
4.5 Empirical Model, Methodological Approach and Data Issues ........................................................120
4.5.1 The Empirical Model................................................................................................................120
4.5.2 The Methodological Approach ................................................................................................122
4.5.3 Data Measurement, Sources and Expected Impact on Remittances .....................................132
4.6 Empirical Results and Discussions ................................................................................................137
4.6.1 Results of Robustness Models and Diagnostic Tests .............................................................137
4.6.2 Macroeconomic Determinants of Migrant Remittances ..........................................................140
4.6.3 Macroeconomic Determinants of Workers‟ Remittances ........................................................149
4.6.4 Macroeconomic Determinants of Compensation of Employees .............................................157
4.7 Conclusions, Policy Implications and Recommendations ..............................................................163
Appendix 4............................................................................................................................................168
xv
Stellenbosch University http://scholar.sun.ac.za
CHAPTER FIVE .......................................................................................................................................183
THE IMPACT AND CAUSAL EFFECTS OF FINANCIAL LIBERALISATION ON INTERNATIONAL
REMITTANCE INFLOWS IN SUB-SAHARAN AFRICA ..........................................................................183
5.0 Introduction .....................................................................................................................................183
5.1 Background ....................................................................................................................................183
5.2 Selected Stylised Facts ..................................................................................................................189
5.3 Literature Review ...........................................................................................................................192
5.3.1 Theoretical Literature ..............................................................................................................192
5.3.2 Related Empirical Literature ....................................................................................................198
5.4 Empirical Model, Methodological Approach and Data Issues ........................................................200
5.4.1 Empirical Panel Granger Non-Causality Model and Analytical Approach ..............................201
5.4.2 Empirical Static Panel Model and Methodological Approach..................................................203
5.4.3 Data Type, Description and Sources ......................................................................................208
5.5 Empirical Results and Discussions ................................................................................................208
5.5.1 The Causal Effects of Financial Liberalisation on Remittance Inflows in SSA .......................208
5.5.2 Empirical Results on the Impact of FLB on International Remittances in SSA .......................210
5.6 Conclusions and Policy Recommendations ...................................................................................219
Appendix 5............................................................................................................................................222
CHAPTER SIX .........................................................................................................................................229
,
REMITTANCES AND ECONOMIC GROWTH IN SUB-SAHARAN AFRICA ..........................................229
6.0 Introduction .....................................................................................................................................229
6.1 Background ....................................................................................................................................229
6.2 Selected Stylised Facts ..................................................................................................................234
6.3 Theoretical Framework and Literature Review ..............................................................................236
6.3.1 Theoretical Framework............................................................................................................236
6.3.2 Empirical Literature on International Remittance Inflows and Economic Growth ...................248
xvi
Stellenbosch University http://scholar.sun.ac.za
6.3.3 A Brief Literature Review on Other Potential Determinants of Economic Growth ..................252
6.4 Empirical Model, Methodology and Data Issues ............................................................................263
6.4.1 The Empirical Model and Methodology ...................................................................................263
6.4.2 Data Sources and Description ................................................................................................267
6.5 Empirical Results and Discussions ................................................................................................268
6.6 Conclusions and Policy Recommendations ...................................................................................278
Appendix 6............................................................................................................................................282
CHAPTER SEVEN ...................................................................................................................................298
THE DEVELOPMENTAL-IMPACT OF REMITTANCES IN SUB-SAHARAN AFRICA ............................298
7.0 Introduction .....................................................................................................................................298
7.1 Background ....................................................................................................................................298
7.2 The Literature on Remittance Inflows and Economic Development ..............................................302
7.2.1 Theories of the Developmental-Impact of International Migrant Remittances ........................302
7.2.2 Literature Review on Effects of Remittances and Developmental Outcomes ........................306
7.3 Analytical Framework, Empirical Model and Data Issues ..............................................................315
7.3.1 Analytical Framework and Empirical Model ............................................................................315
7.3.2 Data Issues .............................................................................................................................318
7.4 Empirical Results and Discussions ................................................................................................319
7.4.1: The Impact of Remittances on Poverty and Income Inequality in SSA .................................319
7.4.2 The Impact of Remittances on Labour Market Outcomes in SSA ..........................................322
7.4.3 The Impact of Remittances on Human Welfare and Development in SSA .............................325
7.4.4: The Impact of Remittances on Financial Development in SSA .............................................327
7.5 Conclusions and Policy Recommendations ...................................................................................337
Appendix 7............................................................................................................................................341
xvii
Stellenbosch University http://scholar.sun.ac.za
CHAPTER EIGHT ....................................................................................................................................353
SUMMARY, CONCLUSIONS, POLICY IMPLICATIONS AND RECOMMENDATIONS .........................353
8.0 Introduction .....................................................................................................................................353
8.1 Summary ........................................................................................................................................353
8.2 Conclusions ....................................................................................................................................359
8.3 Policy Implications and Recommendations....................................................................................362
8.4 Contributions to Knowledge ...........................................................................................................365
8.5 Limitations and Directions of Future Research ..............................................................................367
REFERENCES .........................................................................................................................................370
xviii
Stellenbosch University http://scholar.sun.ac.za
LISTS OF BOXES
Box A2.1: Summary of New Measurement and Definition of Remittances ...............................................33
Box A2.2: Coding Rules for the Financial Liberalisation Index (FLB) ........................................................34
Box A3.1: The Millennium Development Goals (MDGs) ...........................................................................77
Box A4.1: Matrices Corresponding to the Instruments Used in the Estimation ......................................174
xix
Stellenbosch University http://scholar.sun.ac.za
LISTS OF FIGURES
Figure 3.1: Trends in Selected Macroeconomic Indicators in SSA, 1960-2009 ........................................46
Figure 3.2: Trends in External Capital Flows to SSA, 1970-2009..............................................................50
Figure 3.3: Remittances Received per Migrant (US$) in Developing Economies, 1970-2009 ..................52
Figure 3.4: Correlation between Migrant Remittances and Financial Development Indicators, 1990-2009
............................................................................................................................................................57
Figure 3.5: Migrant Remittances Received per capita by SSA Countries (in US$), 1980-2009 ................62
Figure A3.1: Trends in External Capital Flows to Developing Economies, 1970-2009 .............................78
Figure A3.2: Migrant Remittance Flows to Developing Economies, 1970-2009 (actual, per capita & % of
GDP) ...................................................................................................................................................79
Figure A3.3: Remittances per capita vs per Migrant in Developing Economies, 1970-2009 .....................80
Figure A3.4: Migrant Remittance-Recipient Countries in SSA (average, based on % of GDP), 1980-2009
............................................................................................................................................................81
Figure A3.5: Migrant Remittances Received in SSA Countries (period average in US$‟m), 1980-2009 ..82
Figure 4.1: Trends in Migrant Remittances, Household Consumption and Income in SSA, 1980-2009 ...88
Figure 4.2: Trends in Components of Migrant Remittances and GDP per capita in SSA, 1980-2009 ......89
Figure 4.3: Composition of Migrant Remittances Received by SSA Countries, 1980-2009 ......................91
Figure 5.1: Remittances Received and Financial Liberalisation in SSA, 1980-2009 ...............................190
Figure 5.2: Correlation between Remittances Received and Financial Liberalisation in SSA, 1980-2009
..........................................................................................................................................................192
Figure 6.1: Total Inflows and Outflows of Remittances in Developing Economies, 1980-2009 ..............234
Figure 6.2: Correlation between Remittances and Key Macroeconomic Indicators in SSA, 1980-2009.
........................................................................................................................................................2345
Figure A6.1: Global Outlook of Migrant Remittances Received and Paid, 1980-2009 ............................282
xx
Stellenbosch University http://scholar.sun.ac.za
LIST OF TABLES
Table A1.1: Estimates of Additional Critical Financial Resources for SSA ................................................14
Table A1.2: Target Population and the Sample .........................................................................................14
Table 2.1: Examples of Measures of Remittances in Recent Empirical Studies .......................................20
Table 2.2: Measures of Financial Repression / Liberalisation ...................................................................27
Table 2.3: Components and Coverage of Existing Indices of Financial Liberalisation ..............................28
Table 3.1: Macroeconomic Performance and Policy Environment in SSA, 1960-2009 .............................48
Table 3.2: International Trade Performance and Policy Environment in SSA, 1960-2009 ........................48
Table 3.3: Financial Sector Performance and Monetary Policy Environment in SSA, 1960-2009 ............49
Table 3.4: Correlation between Remittances and Selected Macroeconomic Indicators, 1990-2009 ........55
Table 3.5: Comparative Analysis of Top-10 and Bottom-10 Migrant Remittance per capita in SSA .........59
Table A3.1: HIPC Status and Date of Political Independence of SSA Countries ......................................72
Table A3.2: Summary of Major Economic Policies Pursued in SSA since Post-Independence, 1960-2009
............................................................................................................................................................73
Table A3.3: Remittances Received by Sampled SSA Countries, 1980-2009 (period averages) ..............76
Table 4.1: Host Countries of SSA Migrants Resident outside SSA ...........................................................93
Table 4.2: Estimated Results of Migrant Remittances (REMPC) Flows to SSA, 1980-2009 ..................141
Table 4.2.1: Results of Decade-Based Parameter Evolution and Instability Tests for Migrant Remittances
..........................................................................................................................................................148
Table 4.3: Estimated Results of Workers‟ Remittances (WREMPC) Flows to SSA, 1980-2009 .............151
Table 4.3.1: Results of Decade-Based Parameter Evolution and Instability Tests for Workers‟
Remittances......................................................................................................................................155
Table 4.4: Results on Compensation of Employees (COMPPC) Flows to SSA, 1980-2009 ...................158
Table 4.4.1: Results of Decade-Based Parameter Evolution and Instability Tests for Compensation of
Employees ........................................................................................................................................162
Table A4.1: Summary of Empirical Studies on Macroeconomic Determinants of Remittances ..............168
Table A4.2: Data Description, Measurement and Sources ......................................................................176
xxi
Stellenbosch University http://scholar.sun.ac.za
Table A4.3: Host Countries of SSA Migrants ...........................................................................................177
Table A4.4 Robustness Test Results of International Migrant Remittance Flows to SSA, 1980-2009 ..178
Table A4.5: Results of Panel Unit Root Tests ..........................................................................................179
Table A4.6: Static Panel-Data Modelling of Migrant Remittance Inflows in SSA, 1980-2009 .................180
Table A4.7: Static Panel-Data Modelling of Workers‟ Remittance Inflows in SSA, 1980-2009 ...............181
Table A4.8: Static Panel-Data Modelling of Compensation of Employees to SSA, 1980-2009 ..............182
Table 5.1: Implementation of Financial Liberalisation in SSA ..................................................................189
Table 5.2: Expected Impact of Financial Liberalisation Policies on International Remittance Inflows....200
Table 5.3: Financial Liberalisation-Remittances Bivariate Panel Granger Non-Causality Results in SSA,
1980-2009 ........................................................................................................................................209
Table 5.4: Results of the Impact of Financial Liberalisation on International Remittance Inflows in SSA,
1980-09 ............................................................................................................................................212
Table 5.4.1: Financial Liberalisation-Remittance Impact by Rank of Economic Significance in SSA 198009 ......................................................................................................................................................213
Table 5.4.2: Parameter Evolution and Instability Test Results in Frontier and Emerging SSA Financial
Markets .............................................................................................................................................214
Table 5.4.3: Parameter Evolution and Instability Test Results in Underdeveloped SSA Financial Markets
..........................................................................................................................................................216
Table 5.4.4: Financial Liberalisation-Remittance Parameter Evolution and Instability Test Results in SSA
..........................................................................................................................................................218
Table A5.1: Panel Unit Root Test Results ................................................................................................222
Table A5.2: Results of Panel Co-integration Tests ..................................................................................222
Table A5.3: Financial Liberalisation-Remittances Bivariate Panel Granger Non-Causality Results in SSA,
1990-2009 ........................................................................................................................................223
Table A5.4: Empirical Modelling Robustness Test for Impact of Financial Liberalisation on International
Remittances in SSA, 1980-2009 ......................................................................................................224
Table A5.5: Pairwise Correlation Coefficients of Financial Liberalisation Indicators and Remittances in
SSA, 1980-2009 ...............................................................................................................................225
Table A5.6: Descriptive Statistics of Financial Liberalisation Indicators and Remittances Data .............226
Table A5.7: Degree of Financial Liberalisation in Contemporary SSA, 2005-2009 .................................227
Table A5.8: Data Description, Measurement and Sources ......................................................................228
xxii
Stellenbosch University http://scholar.sun.ac.za
Table 6.1: Estimated Impact of Remittances on Economic Growth in SSA, 1980-2009 .........................270
Table 6.2: Results of Decade-Based Parameter Evolution and Instability Tests for Impact of Migrant
Remittances on Growth in SSA ........................................................................................................274
Table 6.3: Estimated Results of Remittance-Growth Size-Effect on SSA, 1980-2009 ............................275
Table A6.1: Global Inflows of Migrant Remittances and Major Forms of External Capital (as of 2009) ..282
Table A6.2: Data Description, Measurement and Sources ......................................................................283
Table A6.3: Descriptive Statistics of Dataset ...........................................................................................284
Table A6.4: Bivariate Correlation of Variables .........................................................................................284
Table A6.5: Results of Panel Unit Root Tests ..........................................................................................285
Table A6.6: Estimated Impact of Median-Dummy Variable (MDV) on Growth in SSA, 1980-2009 ........286
Table A6.7: Median-Dummy Variable-Remittances Interactive Effect on Economic Growth in SSA, 19802009 ..................................................................................................................................................287
Table A6.8.1: The Contemporaneous Impact of Remittances on Growth in SSA, 1980-2009 ................288
Table A6.8.2: Contemporaneous Size-Effect of Remittances on Growth in SSA, 1980-2009 ................289
Table A6.9: Robustness Test Results of Contemporaneous Investment and Remittances on Growth in
SSA ..................................................................................................................................................290
Table A6.10: Static Panel-Data Modelling of Remittances on Economic Growth in SSA, 1980-2009 ....291
Table A6.11: Summary of Empirical Studies on the Impact of Remittances on Economic Growth .........292
Table 7.1: Functions of Financial System and Financial Sector Development Indicators .......................313
Table 7.2: Impact of Remittances on Poverty and Inequality in SSA, 1980-2009 ...................................320
Table 7.2.1: Comparative Analysis of Remittance Effects on Poverty and Inequality in SSA .................321
Table 7.3: Impact of Remittances on Labour Market Outcomes in SSA, 1980-2009 ..............................323
Table 7.3.1: Comparative Analysis of Remittance Effects on Labour Market Outcomes ........................324
Table 7.4: Human Development and Welfare Impact of Remittances in SSA, 1980-2009 .....................325
Table 7.4.1: Comparative Analysis of Remittance Effects on Human Development and Welfare ..........327
Table 7.5.1: Impact of Remittances on Private Sector Bank Credit in SSA, 1980-2009 .........................328
Table 7.5.1.1: Results of Parameter Evolution and Instability Tests for Impact of Migrant Remittances on
Private Sector Credit in SSA ............................................................................................................330
xxiii
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.1.2: Comparative Analysis of Remittance Effects on Private Sector Credit in SSA ................331
Table 7.5.2: Impact of Remittances on Broad Money-GDP Ratio in SSA, 1980-2009 ............................333
Table 7.5.2.1: Comparative Analysis of Remittance Effects on Broad Money-GDP Ratio in SSA ..........334
Table 7.5.2.2: Results of Parameter Evolution and Instability Tests for Impact of Migrant Remittances on
Broad Money Supply in SSA ............................................................................................................335
Table A7.1: Summary of Empirical Studies on the Impact of Remittances on Economic Development .341
Table A7.2: Set of Control Variables in the Empirical Models .................................................................349
Table A7.3: Median of Endogenous Variables and Specification of Median Dummy Variables ..............349
Table A7.4 Static Panel-Data Modelling of Remittances on Private Sector Bank Credit in SSA, 19802009 ..................................................................................................................................................350
Table A7.5: Static Panel-Data Modelling of Remittances on M2/GDP in SSA, 1980-2009 .....................351
Table A7.6: Data Description, Measurement and Sources ......................................................................352
xxiv
Stellenbosch University http://scholar.sun.ac.za
CHAPTER ONE
GENERAL INTRODUCTION
1.0 INTRODUCTION
This chapter provides a broad insight into the nucleus and the outline of this dissertation. In
particular, the background of the study, the research problem, the research questions, the
motivation for the study and the research objectives are discussed. The chapter also presents
the research hypotheses, the scope of the study, as well as the structure of the dissertation.
1.1 BACKGROUND
The search for rapid growth and sustainable development for the underdeveloped economies,
particularly sub-Saharan Africa (SSA), has been continuing for a long time. This has led to the
adoption of economic reform policies such as the liberalisation of the financial sector in an
apparent recognition of the widely held view that the financial sector can play a crucial role in
accelerated economic growth and sustainable development. For instance, as far back as the
1870s, Bagehot (1873) recognised and consequently emphasised the critical role of the
financial sector in resource mobilisation to finance economic growth and development. Later, a
new generation of prominent economists, notably Schumpeter (1912), Cameron (1967),
McKinnon (1973) and Shaw (1973), re-emphasised the relevance of the financial sector in
propelling economic growth and development.
These policy prescriptions, notwithstanding, many governments in developing countries, have
until recently, at one time or another, intervened in the smooth development process of their
respective domestic financial markets through the imposition of various forms of restrictions
and control measures that limited the scope, pace and operations of financial institutions.
These actions subsequently crowded-out private sector initiatives and investment as financial
institutions under state control directed credit in favour of government projects and public sector
institutions.
Meanwhile, Cameron (1967), McKinnon (1973) and Shaw (1973) maintain that the benefits
accruing from a well-functioning and properly developed financial system can be enormous.
First, through an efficient financial intermediation process, lenders and borrowers are easily
1
Stellenbosch University http://scholar.sun.ac.za
brought together, which reduces transaction and search costs. Second, financial institutions
provide relevant services to their clients and thereby helping reduce information costs, provide
risk management services and reduce risks involved in financial transactions in general. Third,
financial institutions create liquidity in an economy by converting short-term borrowings into
medium- and long-term financial assets by way of lending and other forms of business finance.
Fourth, the intermediaries bring the benefits of asset diversification to the economy. Fifth,
financial institutions mobilise savings from atomised individuals for investment, thereby solving
the problem of indivisibility in financial transactions. Above all, through a well-functioning
financial system, mobilised savings are invested in the most productive projects. This
investment creates opportunities for full employment of factors of production to propel rapid
economic growth and development.
Essentially, the numerous merits of financial intermediation can translate into economy-wide
benefits (Levine, 1993; 1997), which influence governments to adopt financial liberalisation
programmes in economies where the financial sector is considered underdeveloped. These
programmes which comprise a series of policy reforms are designed mainly to increase the
process of financial resource mobilisation from domestic and foreign sources channelled
through the formal financial sector; improve the efficiency of financial intermediation; and
enhance the effectiveness of monetary policy.
Based on these expectations, many developing countries, including those in SSA, embarked
upon the implementation of policies of financial liberalisation as a component of the Structural
Adjustment Programme (SAP) under varying financial structures and different macroeconomic
fundamentals. For instance, at the commencement of the reforms within the West African subregion, Nigeria already had relatively more advanced financial institutions and assets than
Ghana, Sierra Leone and the Gambia. Generally, however, the financial reform programmes
were initiated in these countries as a response to macroeconomic imbalance and financial
distress.
Through the removal of the elements of financial repression, particularly controlled interest
rates, financial sector reform is expected to lead to higher nominal and real interest rates, which
are, in turn, expected to serve as incentives for financial resource mobilisation and efficient
credit allocation. This is the supposition of the liberalist hypothesis (McKinnon, 1973; Shaw,
1973). A higher real deposit rate encourages economic agents to substitute consumption for
2
Stellenbosch University http://scholar.sun.ac.za
savings (the substitution effect). In addition, higher interest-income on savings enables savers
to achieve their saving targets with lower stock of savings (the wealth or income effect). The
two effects operate in opposite directions and the net outcome depends on which one
dominates the other. The underlying reasoning of the McKinnon-Shaw doctrine is that the
substitution effect outweighs the wealth effect. Accordingly, financial savings will further be
boosted by a shift in the savers‟ wealth portfolios from non-financial assets to financial assets
(asset substitution effect).
Contrary to the McKinnon-Shaw premise, the increased real interest rate may not necessarily
lead to improved domestic financial resource mobilisation. In very low-income countries like
those in SSA, for instance, the level of income could be so low that households spend a very
high proportion of their earnings on basic needs1. Under this circumstance, even with high real
deposit rates, very little or no proportion of income can be saved. It must also be emphasised
that in Less Developed Countries (LDCs), subsistence economic activities are vibrant and quite
pre-dominant in rural communities. These rural economies which form the largest sector in
LDCs have the highest population of illiterate peasant farmers and petty traders who still
engage in barter trading since household incomes are more in kind than in cash. This implies
that the McKinnon-Shaw proposition is probably not entirely relevant to developing economies.
A study of this proposition by Ogaki et al. (1996) shows that a 100 per cent rise in real deposit
rate leads to a 66.7 per cent rise in savings in high-income countries, but to only 10 per cent
rise in very low-income countries in the long run. This “basic needs” explanation and even the
tendency of dissaving in LDCs and, for that matter SSA, could be the likely explanation for the
insensitivity of financial savings to real deposit interest rates in many African countries2.
In this era of globalisation, macroeconomic policies and programmes for all countries, including
those in SSA, have, since the 1980s, invariably and as a matter of necessity, become more
liberal and market oriented. This has enhanced the global mobility of factors of production in
general and capital in particular. For instance, remittances have become topical in international
finance and development economics as the rate and volume of cross-border asset transfers
have been increasing exponentially since the 1980s. In 1995, migrant remittances to
developing countries totalled US$57.8 billion and this soared up to US$96.5 billion in 2001
(World Bank, 2006a). In 2005, the World Bank estimated that migrant remittances to
1
2
When households‟ incomes are at subsistence level, their marginal propensity to consume is equal to one.
See Oshikoya (1992) for the case of Kenya.
3
Stellenbosch University http://scholar.sun.ac.za
developing countries totalled more than US$167 billion, but the actual amount could be 50 per
cent higher or more, while others put the figure at US$298 billion. In 2006, the World Bank
(WB) reported that official remittances increased to US$206 billion. Thus, the growth of
remittances has now exceeded private capital flows or foreign direct investment (FDI) and
official development assistance (ODA) to developing countries. Besides, remittances are a
reliable source of foreign capital and the least volatile source of foreign exchange since the
1990s and now account for a third of global finance (World Bank, 2006a).
In spite of the consistent growing trend in international remittances, the implications of
remittances for an underdeveloped economy appear rather ambiguous. Thus, while it is true
that increased remittances to developing countries could lead to rapid economic growth,
macroeconomic stability, and improved livelihoods, it is also possible that continuous colossal
remittance inflows could result in increasing brain drain, dollarisation, inflation, over-reliance
and abandonment of the pursuit of pro-growth economic policies, and moral hazards where
recipients heavily depend on these transfers, thereby reducing supply of labour3. Increased
remittance inflows to developing countries could also lead to real exchange rate appreciation
and less international competitiveness, culminating in what has been referred to as the „Dutch
Disease‟. Altogether, these costs of high international remittance inflows could possibly retard
the economic growth and economic development process of underdeveloped economies.
The reasons for the adverse effects of remittance inflows are not far-fetched. Among the
prominent features of underdeveloped economies are high population growth rates resulting in
excess labour supply, high unemployment rates, low per capita incomes, widespread poverty
and rural-urban migration (Lewis, 1954; Todaro and Smith, 2002). According to Lewis (1954),
rural economies are subsistent in nature with low productivity and low industrialisation, and a
high desire among the active population to move to industrialised economies where it is
presumed that there are ready jobs with relatively higher incomes. Therefore, it is conceivable
that in a globalised world, once migrants abroad continue to remit home consistently, those at
home who are earning relatively abysmal incomes will yearn to join the exodus wagon leading
to brain drain in underdeveloped economies. Besides, since developing countries have less
developed financial markets which are not strongly integrated into the global financial system,
there is a higher tendency among migrants from the developing world to remit home through
3
Some recipients of regular remittances may become over-dependent and choose to be voluntarily unemployed or
underemployed especially in developing SSA countries where working conditions are poor and real wages are
unattractively low.
4
Stellenbosch University http://scholar.sun.ac.za
unofficial routes4. As foreign currency denominated financial remittances continue to flow into
developing countries which generally have difficulty in stabilising their national currencies, the
existing desire for holding foreign currencies is likely to rise, culminating in de facto
dollarisation. In fact, all over the world, there appears to be a correlation between remittance
inflows and financial dollarisation as is evident in Latin America and the Caribbean (LAC)
countries as well as in East Asia and the Pacific (EAP) countries which are the leading
recipients of migrant remittances (see Adenutsi and Yartey, 2007). Unofficial dollarisation is a
recipe for inflation as monetary authorities will find it difficult to determine the actual volume of
total money supply in the economy correctly5. Arguably, if nationals of developing countries
continue to seek and secure more lucrative jobs abroad and remit home to support family
members left behind, the pressure on governments of underdeveloped economies to create
jobs and even to industrialise will reduce considerably. This is a more likely event in developing
countries where governments receive significant revenue during episodes of higher inflation in
the form of seigniorage (Adenutsi, 2008).
Thus, though the role of international remittances in an economy has remained theoretically
controversial, in recent times, some development economists, including Stahl and Arnold
(1986), and Massey et al. (1998), seem to agree that generally, at least, there are good
reasons to believe that remittances can play a critical role in economic growth and the
development process by aiding beneficiary developing countries in poverty alleviation and
minimising balance of payments problems. It is also widely acknowledged that remittances
constitute an invaluable resource for consumption and employment creation through business
finance in many developing countries (Taylor, 1992; Brown, 1994; Adams, 1998).
1.2 THE RESEARCH PROBLEM
An estimated 175 million people worldwide, implying one in every 35 and approximately three
per cent of the total world population, had settled in countries other than their native countries
by the beginning of the 21st century (United Nations, 2002). With the advent of globalisation and
the increasing development gap between the industrialised world and developing countries, the
number of international migrants is estimated to increase by roughly 2.5 per cent per annum
(IOM, 2010). Without doubt, international migration has offered an opportunity for developing
4
World Bank (2005) estimates that the recorded remittances received by developing countries are just about 50 per
cent of the actual volume received.
5
Adenutsi (2008) found that from official sources alone, foreign currencies form more than a third of total monetary
aggregates in developing countries whilst economic openness causes dollarisation and inflation in Ghana.
5
Stellenbosch University http://scholar.sun.ac.za
countries to consider ways of benefiting from their nationals living and working abroad through
the implementation of monetary policies and the adoption of pro-growth strategies to attract
optimal remittances to finance their development projects rather than over-emphasizing the
perceived negative effects. In today‟s world of globalisation and international competition, the
significant role of remittances in propelling the development agenda of SSA6 and other
developing economies has become even more crucial and the need to offer incentives to
attract such transfers into local savings and investment funds has become more inevitable.
Currently, SSA receives not more than seven per cent of global remittances. This is by far the
smallest share to developing economies and less than half of the amount received by India
alone, whilst the EAP receives nearly 30 per cent with South Asia (SAS) receiving 24 per cent
of global remittances in 20097. Similarly, the remaining developing economies comprising
Europe and Central Asia (ECA), LAC, and the Middle East and North Africa (MNA) received
about 42 per cent (or an average of 14 per cent) global remittances in 2009. Even across
Africa, SSA significantly lags behind. The questions that arise then are: Why is SSA alone
lagging behind in attracting international remittances to augment its scanty domestic
resources? In what ways can SSA enhance international remittance inflows and thereby
maximise these remittances from the large pool of their citizens living abroad that could serve
as a compensation for losing their skills to the advanced countries? How do remittance inflows
impact on the economic growth and development in SSA?
The problem is that, notwithstanding the emerging interest and extensive work on both
remittances and economic growth and development in underdeveloped economies8, the links
between the role of financial sector policies in mobilising and managing international
remittances for economic growth and development in SSA as a sub-region remains
underexplored. Hence, in the case of SSA, as a sub-region, as at now, very little is known
about the underlying factors of remittance inflows, the linkages between remittances and
financial liberalisation, and the implications of remittance inflows for economic growth and
development in a liberalised financial environment. Thus, the fact remains that countries within
SSA are generally poor but they remain a major „net exporter of labour‟ into the industrialised
countries, yet SSA has been the least recipient of remittances over the years. Can this be
6
SSA in particular is still in dire need of colossal resources to finance its development agenda. See estimates of the
sub-region‟s critical resource gap in Table A1.1 in the Appendix.
7
Author based on World Bank (2011a; 2011b). See Figure A3.2 in Chapter Three for evidence.
8
See Chami et al. (2005), Giuliano and Ruiz-Arranz (2009) and Adenutsi (2010).
6
Stellenbosch University http://scholar.sun.ac.za
attributed to the underdevelopment and non-competitive nature of the financial market? Or is it
a case that SSA does not, in fact, receive the least international migrant remittances if
remittances were appropriately defined? And, has the adoption of financial liberalisation
programme three decades ago led to higher inflows of international remittances through official
channels to be transmitted through the banking system? In order to provide some useful
information relevant to the understanding of the linkages among financial sector reform
programmes, international remittances, and economic growth and development in developing
countries, this study explores the linkages between financial liberalisation and international
remittance inflows, and the implications of remittances for economic growth and development
for developing countries with special reference to SSA.
1.3 THE RESEARCH QUESTIONS
Therefore, the broad and pertinent questions explored in this study with reference to SSA
include:
i.
Does the SSA economy broadly demonstrate any significant improvement in economic
development since the adoption of economic reform programmes in the 1980s? And
what has been the trend in international remittance inflows since the pursuit of financial
liberalisation in the 1980s?
ii. What are the macroeconomic determinants of international remittance inflows to SSA
under liberalised financial regime?
iii. Does the implementation of financial liberalisation have any impact or causal effect on
international remittance inflows? If so, which specific policies under financial
liberalisation programme have been the most important in this regard?
iv. Do international remittances impact on long-run growth under liberalised financial
regime? And has this impact changed over time in response to the cyclical behaviour of
remittance inflows?
v. To what extent do international remittance inflows promote economic development?
1.4 THE RESEARCH OBJECTIVES
In response to the above research questions, the objectives are to explore the macroeconomic
factors that explain the changing levels of remittance flows to SSA and to examine the
implications of international remittances for the financial liberalisation and economic growth and
development in SSA empirically. More specifically, on the one hand, this study seeks to find the
7
Stellenbosch University http://scholar.sun.ac.za
empirical nexus between international remittance inflows and financial liberalisation, as well as
the impact of remittances on growth and development in SSA. On another hand, this study
seeks to propose the best measure for international remittances, and to explore how the
changing macroeconomic policy environment affects remittance inflows into SSA.
The specific objectives of this study include the following:
i.
to verify whether or not, the macroeconomic environment of SSA has changed
significantly with the pursuit of financial liberalisation programmes, and if it does,
whether this change has any correlation with international migrant remittance inflows;
ii. to identify the macroeconomic factors that explain variations in international remittance
flows to SSA under liberalised financial regime;
iii. to trace the causal effect and examine the impact of financial liberalisation on
international remittance inflows in SSA;
iv. to evaluate the impact of international remittance inflows on economic growth in SSA;
and,
v. to determine the developmental-impact of international remittance inflows in SSA.
1.5 MOTIVATION FOR THE STUDY
There is a need to examine the implications of international remittances for policies and
developments within the financial sector, economic growth and economic development in SSA
empirically. This is essential because, currently, there is no apparent reason to expect a
paradigm shift in economic policy design in favour of an inward-looking approach imbedded in
a socialist doctrine. This is the result of the collapse of communist states. Also, there appears
to be no reversibility from globalisation of economies, given the vast merits of economic
openness over the states in autarky equilibrium positions. Clearly, if these expectations are
upheld, then, given the wide development gap between the North and the South, in the interim,
governments in SSA, and indeed, their counterparts in other less developed regions of the
world, can do very little to prevent their active population from migrating to industrialised
economies where higher remuneration and better conditions of work are envisaged. Evidently,
remittance flows to developing countries, including SSA, in the form of migrant transfers have
been rising consistently in recent years. The steady and appreciable increases in remittances
are likely to have a strong positive correlation with the exodus of both skilled (professionals)
and unskilled labour from developing countries to the industrialised world.
8
Stellenbosch University http://scholar.sun.ac.za
As the volume of remittance inflows continues to be high and growing exponentially, broad
money supply will continue to rise in SSA due to the likely increases in foreign currencies in
circulation. The role of the financial sector in enhancing the mobilisation of remittances through
official transfer channels has become even more crucial. Growth in money supply has obvious
consequences for inflation, interest rates, and exchange rates as demand for foreign goods
increases in import-dependent countries like those of SSA. Also, the international reserve
component of the balance sheets of Central Banks will be enriched with the upsurge of official
remittance inflows. This notwithstanding, the SSA sub-region has been traditionally known for
its deficiency in formulating and implementing effective pro-growth macroeconomic policies
over the years, which has resulted in a somewhat vicious cycle of perpetual economic
instability, stagnation and underdevelopment. Therefore, as a result of these imperatives, it is
important to investigate the causes, macroeconomic determinants, and the implications of
increasing inflows of remittances for growth and development under the liberalised financial
environment in SSA. Broadly speaking, there is motivation to explore the causal effects of
financial liberalisation in attracting international remittances through the banking system of SSA
as well as to examine the determinants and implications of remittances for economic growth
and development in SSA.
1.6 SPECIFIC MOTIVATIONS AND THE RESEARCH HYPOTHESES
Consistent with the afore-stated objectives, the set of hypotheses (H) that are fundamental to
guiding the focus of this study, with reference to SSA, includes the following:
1.6.1 Macroeconomic Determinants of International Remittances in SSA
Various empirical studies (see Table A4.1 in Chapter Four) have shown that macroeconomic
factors in native (or home) countries and resident (or host) countries of migrants play crucial
roles in determining international remittances. To verify this, within the context of SSA, the
following central hypotheses were tested:
H1: Macroeconomic factors are not determinants of international remittance inflows.
H2: Macroeconomic determinants do not have the same influence on attracting
remittances from permanent migrants (workers‟ remittances) and remittances from
temporary migrants (compensation of employees).
9
Stellenbosch University http://scholar.sun.ac.za
1.6.2 Financial Liberalisation and International Remittance Inflows
Essentially, financial liberalisation is implemented to deepen and widen the financial market
and also to make the financial market more open and competitive towards accelerated financial
development for economic growth. With these developments, it is expected that the domestic
financial market should become attractive to the private sector as far as resource mobilisation
is concerned since improved efficiency and reduced operational costs lead to cheaper financial
services. Also, under competitive financial market environment, banks are expected to become
more innovative in designing products and services for different segments of their target
customers including international migrants. This study, therefore, examines the impact and
causal effects of financial liberalisation on international remittance inflows in SSA as specified
in H3 and H4.
H3: There is no causal relationship between financial liberalisation and international
remittance inflows.
H4: Financial liberalisation does not impact on international remittance inflows.
1.6.3 The Long-Run Growth and Developmental-Impact of International Remittances in SSA
Both theoretically and empirically, the controversy over the developmental-impact of
international migrant remittances has remained unresolved as evident in the conclusions drawn
by various scholars (see Tables A6.6 and A7.1 in Chapters Six and Seven respectively). To
contribute to this debate, hypotheses H5-H14 were tested with respect to SSA:
H5: International remittance inflows do not affect economic growth.
H6: International remittance inflows do not impact on poverty.
H7: International remittance inflows do not influence income inequality.
H8: International remittance inflows have no impact on unemployment.
H9: International remittance inflows do not affect labour participation.
H10: International remittance inflows do not influence labour productivity.
H11: International remittance inflows have no effect on human welfare.
H12: International remittance inflows have no impact on educational attainment.
H13: International remittance inflows do not impact on life expectancy.
H14: International remittance inflows have no relationship with financial development.
10
Stellenbosch University http://scholar.sun.ac.za
1.7 SCOPE
The study period of this dissertation is restricted to 1980-2009. This is because financial
liberalisation programmes in SSA were essentially initiated in the 1980s. Although officially
reported data on some of the relevant variables are available up to 2011 for 27 of the 36
sampled countries at the time of this study, the researcher was more interested in fairly
representing the sub-region with a higher number of countries as much as possible. This is one
of the primary underlying motivations for restricting the study to 2009 for 36 sampled countries.
Another justification for restricting the upper limit time coverage of this study to the year 2009 is
not only to provide for consistent decade-by-decade analysis but also to allow for a consistent
econometric approach for testing the stability of the varying estimated coefficients across the
three decades. Furthermore, because some of the variables used as indicators of economic
development, notably measures of poverty and income inequality are reported in a five-year
interval by the World Bank, stretching the study period beyond 2009 to say 2011 will imply
using different study periods in the various chapters of this study. Finally, extending the study
period beyond the year 2009 will require collecting new survey data on at least seven
components of financial liberalisation identified by Abiad et al. (2010). Financial constraint and
the slow response rate from the various central banks and stock exchanges of the sampled
SSA countries will affect the timely completion of this study, hence the decision to restrict the
upper study period to the year 2009.
Thus, based strictly on availability of balanced panel data (see Table A1.2), this study is
generally limited to only 36 SSA countries. Countries included in the broad panel are Benin,
Botswana, Burkina Faso, Cameroon, Cape Verde, Comoros, Congo Republic, Côte d‟Ivoire,
Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea Bissau, Kenya, Lesotho, Madagascar,
Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia, Niger, Nigeria, Rwanda, São Tomé
and Príncipe, Senegal, Seychelles, Sierra Leone, South Africa, Sudan, Swaziland, Tanzania,
Togo and Uganda.
1.8 STRUCTURE OF THE DISSERTATION
This dissertation comprises eight chapters. The outline of presentation of the remaining seven
chapters is as follows:
Chapter Two: Conceptual Framework and Measurement Issues
This chapter was undertaken to achieve the specific objective (i) and in response to research
question (i). In particular, the concepts of financial liberalisation and international remittances
11
Stellenbosch University http://scholar.sun.ac.za
were explained with measurement issues discussed and justified.
Chapter Three: Macroeconomic Environment and External Capital Flows to SSA (1960-2009)
To begin with, an attempt was made to provide a detailed insight into the trends in
macroeconomic performance of SSA and the flow of external capital to the sub-region since
1960. This was meant to provide a visual impression on trends in key macroeconomic
performance indicators under the three main policy environments, viz. the pre-reforms era, the
reforms era, and the post-reforms era in post-independence SSA. This chapter addresses
research question (ii) and specific objective (ii).
Chapter Four: Macroeconomic Determinants of International Remittance Flows to SSA
In order to address research question (iii), achieve specific objective (iii), and evaluate H 1 and
H2, the system Generalised Method of Moment (sys-GMM) procedure for estimating dynamic
panel-data models was employed to determine the macroeconomic factors that affect
international remittances at the aggregated and the disaggregated levels.
Chapter Five: The Impact and Causal Effects of International Remittances on Financial
Liberalisation in SSA
In line with specific objective (iv), research question (iv), and hypotheses H3 and H4, following
the Granger panel analytical framework, the empirical causal relationship between financial
liberalisation and international remittances was investigated.
The static panel estimation
approach for single equations was further employed to evaluate the impact of financial
liberalisation on international remittance inflows in SSA.
Chapter Six: Impact of International Remittances on Economic Growth in SSA
To respond to question (v), achieve specific objective (v) and evaluate H5, the system GMM
estimation procedure was followed to examine the long-run impact of international remittance
inflows on economic growth in SSA from 1980 to 2009.
Chapter Seven: The Development-Impact of International Remittances on SSA
The dynamic panel model, following system GMM estimation technique, was followed to
examine the hypotheses H6–H14 and in response to research question (vi) and specific
objective (vi). In effect, the impact of international remittance inflows on indicators of poverty,
income inequality, labour market outcomes, human development, and financial development
12
Stellenbosch University http://scholar.sun.ac.za
were explored in this chapter.
Chapter Eight: Summary, Conclusions and Recommendations
As the final chapter concluding the dissertation as a whole, a summary of the major findings,
the conclusions drawn in connection with the research questions, objectives and hypotheses,
policy recommendations, and the suggestions for relevant areas for future research are
outlined in this chapter.
1.9 CHAPTER SUMMARY AND CONCLUSIONS
This chapter provided the general introduction to the study by presenting a wide-ranging
outlook and the motivation for this study. In particular, in the background of the study, issues
concerning the pattern of macroeconomic policies and management during the postindependence era, and the circumstances leading to the adoption of financial liberalisation
across SSA in the 1980s were discussed. It also provided information that helps to explain
what the picture looks like with regard to the changing trends in international capital flows, the
possible causes and the likely reasons behind this new development. Following the
background information, the research problem was formulated and the relevant research
questions identified were raised. The central motivation for this study is the need for an
empirical understanding of why although SSA has been a consistent leading „net exporter of
labour‟ over the years, it has steadily remained the region receiving the least international
remittances which are non-debt external funds critically required to address the numerous
socioeconomic problems confronting the sub-region since post-independence. Based on the
research problem, the research questions and the motivation for the study, the research
objectives were specified. The general objective, from which this study takes its stimulus, was
to identify the macroeconomic factors that explain variations in migrant remittances to SSA and
to examine, empirically, the linkages between international remittances and financial
liberalisation; and the determinants and impact of remittances on the economic growth and
development in SSA.
Other essential subjects related to the specific motivation behind each aspect of the research
problem tackled, the hypotheses guiding the research, the scope of the study, as well as the
structure of the dissertation were also addressed in this chapter. The stage has now been set
for the study to proceed to Chapter Two, which is devoted to addressing issues related to the
definition and measurement of international remittances and financial liberalisation.
13
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 1
Table A1.1: Estimates of Additional Critical Financial Resources for SSA
Author(s)
Zedillo et al. (2001)
Devarajan et al. (2002)
Estimated Amount
Required Yearly
US$50 billion
US$40-60 billion
Funke and Nsouli (2003)
CfA (2005)
US$64 billion
US$37.5 billion
Gupta, Powell and Yang
(2006)
US$38-46 billion
Comments/Method of Estimation/Justification
Developing countries in general
Used two estimation methods. One approach estimated the
MDGs resource needs by computing the required economic
growth rates of countries, and then the investment required to
achieve these. The other method separately estimated the
costs of achieving individual goals. Both estimates exclude
certain costs, notably those of the complementary
infrastructure needed to support the required rates of growth
and investment.
Equivalent to 12 per cent of Africa‟s GDP.
Required to finance public expenditure until 2010. A third of
this amount is expected to come from domestic sources whilst
the remaining two-thirds US$25 billion comes from foreign aid.
US$14-18 billion required for 2006-2008 whilst US$24-28
billion is required by 2015 to finance infrastructural
improvement and human development.
Source: Author‟s compilation
Table A1.2: Target Population and the Sample
Country
Sampled? If NO, why not?
Country
Sampled? If NO, why not?
Angola
NO; data not available except for 1996, 2008
Madagascar
YES
Benin
YES
Malawi
YES
Botswana
YES
Mali
YES
Burkina Faso
YES
Mauritania
YES
Burundi
NO; data not available prior to 2004
Mauritius
YES
Cameroon
YES
Mayotte
NO; data not available for 1994-2009
Cape Verde
YES
Mozambique YES
CAR
NO; data not available except for 1990 -1993
Namibia
YES
Chad
NO; data not available for 1994-2009
Niger
YES
Comoros
YES
Nigeria
YES
Congo, DR
NO; data not reported for any of the years
Rwanda
YES
Congo Rep
YES
ST&P
YES
Côte d'Ivoire
YES
Senegal
YES
Eq. Guinea
NO; data not available except for 1992 & 1997
Seychelles
YES
Eritrea
NO; data not available prior to 1998 & 2001 9
Sierra Leone YES
Ethiopia
YES
Somalia
NO; data not available for 1985-2009
Gabon
YES
South Africa
YES
Gambia
YES
Sudan
YES
Ghana
YES
Swaziland
YES
Guinea
YES
Tanzania
YES
G-Bissau
YES
Togo
YES
Kenya
YES
Uganda
YES
Lesotho
YES
Zambia
NO; data not available prior to 2003
Liberia
NO; data not available prior to 2004
Zimbabwe
NO; data not available for 1995-2009
Source: Author. Notes: CAR, Eq. Guinea, G-Bissau and ST&P represent Central African Republic, Equatorial
Guinea, Guinea-Bissau, and São Tomé & Príncipe respectively. Data availability here is restricted to remittances.
14
Stellenbosch University http://scholar.sun.ac.za
CHAPTER TWO
CONCEPTUAL FRAMEWORK AND MEASUREMENT ISSUES
2.0 INTRODUCTION
This chapter discusses the main concepts used in this study. The concepts of international
remittances and financial liberalisation have, thus, been elucidated and rationalised from a
theoretical and practical viewpoint within the purview of this study. Therefore, the general
motivation for this chapter is to define and review existing alternative measures as well as to
justify and outline the procedure used in computing the selected indicator of international
remittances and financial liberalisation in sub-Saharan Africa (SSA). This is imperative because
what is incorrectly defined can only be correctly measured by coincidence. The chapter also
presents the procedure for computing the selected index of financial liberalisation and justifies
why this indicator was selected among the existing alternatives. The remaining part of this
chapter is structured as follows: Section 2.1 presents the definition and issues on the empirical
measurement of international remittances. This is followed by the working definition and
measurement of the concept of financial liberalisation in Section 2.2. Then, the summary and
conclusions of this chapter are presented in Section 2.3.
2.1 INTERNATIONAL REMITTANCES
2.1.1 Concept Definition
There are different definitions of international remittances but the following definition makes the
relevant points: migrant-related assets transferred across international borders from the
migrant‟s country of residence, usually to his/her native or adopted country of citizenship. Some
definitions of international remittances which have been advanced in contemporary studies are:
i.
Kapur (2004: 1) defines remittances broadly as financial resource flows arising from the
cross-border movement of nationals of a country. In the narrowest sense, remittances
as “unrequited transfers refer primarily to money sent by migrants to family and friends
on whom there are no claims by the sender unlike other financial flows such as debt or
equity flows”.
ii.
“Remittances are person-to-person flows (from migrants to their friends and families),
well targeted to the needs of the recipients, who are often poor. Such remittances do
not typically suffer from the governance problems that may be associated with official
15
Stellenbosch University http://scholar.sun.ac.za
aid flows” (Ratha and Mohapatra, 2007: 1).
From the definitions provided above, it is understandable that international remittances can be
defined from a narrow or a broad viewpoint9. From a narrow perspective, international
remittances are typically monetary transfers from international migrants to their countries of
origin or nationality where their families or other relatives are residing. The key features of the
narrow definition of international remittances are that: (i) remittances are generally monetary
transfers; (ii) remittances are from a migrant (the relatively wealthier resident outside his/her
home country) to a non-migrant (the relatively poorer relative/associate resident in his/her
native country); (iii) remittances are aimed at solving or managing a specific known problem;
and (iv) remittances are often in small amounts but with some exhibits of regularity and stability
in flow. It is this narrow definition that is referred to as “international migrant remittances” in this
study because the concept is directly related to individual migrants and the remittances are
expected to be highly dependent upon the personal earnings of migrants relative to the
average earnings of the target recipient. International migrant remittances, in this context, also
include non-monetary transfers of small but valuable goods from a migrant to his/her family,
friends or other relatives in his/her home country.
With regard to the broader definition, international remittances are financial flows mainly
occasioned by migration, from a person (the migrant) or an international benevolent
organisation (such as the migrant association of a particular ethnic group) to persons and/or
social institutions (such as orphanages, refugees, or the physically challenged) in poorer
countries10. Thus, international remittances should be seen to include the narrow definition plus
other non-debt transfers in the form of money or materials sent by migrants (either as
individuals or as a group) and organisations (often specialised non-governmental organisations
(NGOs) serving households notably migrant associations) to individuals or charitable social
institutions in poorer nations. The distinction between Official Development Assistance (ODA)
and the type of international remittances sent by humanitarian NGOs is that, unlike the former,
the latter is essentially unofficial, relatively small in value, more regular and stable in flow,
directed specifically at the target beneficiary, and does not require any technical and
9
The current standardised definition of remittances in the Appendix 5 of IMF‟s Balance of Payments and
th
International Investment Position Manual, 6 Edition, is household income from foreign economies arising mainly
from temporary or permanent movement of people to those economies. Measurement in accordance with this new
definition is yet to be formally reported in the IMF‟s BoPS or WB‟s WDI.
10
International remittances may also flow to countries that may not be necessarily poor but hit by civil war or natural
disasters.
16
Stellenbosch University http://scholar.sun.ac.za
managerial costs or „conditionalities‟. As Ghosh (2006: 14) puts it, “…. their characteristics,
modus operandi, and role sharply differ, defying facile comparison. Remittances are private
transfers, ODA flows, which are transactions between governments, are not”.
2.1.2 Measurement of International Remittances
The measurement of international remittances has consistently posed a great challenge to
economists since the recognition of its importance as an essential source of external finance
about two and a half decades ago (Adams and Page, 2005; Ghosh, 1997; 2006). The main
reasons for the difficulty of measuring international remittances are that:
i.
remittance inflows to developing countries are underreported, as a colossal amount is
believed to be channelled through the informal sector;11,12
ii. international remittances are undercounted because in many developing countries it is
not mandatory to report “small” remittances (Gupta, Pattillo and Wagh, 2009: 3);
iii. illegal migration is high among natives of developing countries whilst data on migration
in general suffer from worse problems than even data on remittances;
iv. even if data on migration were adequately and correctly reported, accurate data on
skills of migrants, type of employment and skills-related migrant employment, and the
changing residential status of migrants are non-existent across countries and over
regular time intervals;
v. some countries report all forms of remittances as workers‟ remittances, thus ignoring
the standard reporting system requiring categorisation according to migrant status;
vi. migrants from poor countries are not likely to find remitting home through the formal
channel convenient and appealing due to high illiteracy rates and illegal status;
vii. migrants find it less costly and probably more appropriate to use the informal
transmission channels than the officially approved routes;13
11
For instance, the World Bank (2006a) estimates that, globally, at least 50 per cent of remittances are transferred
through informal channels. Freund and Spatafora (2005) estimate that remittances through informal channels are
relatively higher in SSA, ranging between 45 and 65 per cent of formal flows as against the range of 5-20 per cent in
Latin America.
12
It is important to state that this problem of data omission or underestimation in developing countries is not peculiar
to migrant remittances alone. In fact, Schneider and Enste (2000) report that GDP values for developing countries
are underestimated by between 25-75 per cent due to the neglect of the rather large informal sector dominated by a
series of interrelated unreported subsistent activities. This exactly equates the magnitude of underestimation of
officially reported migrant remittances in developing countries as noted by Fruend and Spatafora (2005).
13
This is especially due to the underdevelopment of the financial system in SSA and other developing economies.
17
Stellenbosch University http://scholar.sun.ac.za
viii. some countries only collate data from banks and overlook remittances through money
transfer operators (MTOs), post offices, and credit unions (de Luna Martinez, 2005);
and,
ix. as noted by the World Bank (2006a), statisticians in some countries adopt a fuzzy
approach for the reporting system whereby the estimated number of emigrants is
multiplied by the average remittance inflows.
The World Bank, IMF and the UN, like many others, including Shelburne and Palacin (2007: 6),
confirm that obtaining data on remittances is extremely difficult because many migrants are
illegal, and with many others being “poorly educated and unskilled …with limited knowledge of
the local language or customs,” and/or for the avoidance of tax, patronise unofficial financial
channels which are not reported. Apart from this, the definition of the various components of
remittances is either misunderstood by national institutions responsible for compiling the data
or they are ambiguous and lack specificity14. The main issue is that the concepts of “residency”
and “migration status” upon which the definitions and the distinctions are generally based are
difficult if not impossible to measure in various countries. For instance, if a resident household
member leaves the country where his/her household is residing and returns to his/her
household after a limited time period of less than one year (i.e. not exceeding 12 months), the
individual continues to be a resident of his/her home country even if he/she undertakes
frequent journeys outside his/her „native‟ economic territory. Similarly, a person who leaves
his/her „native‟ country with the intention of living in another country for a year or more ceases
to be a resident of his/her native country and is considered a resident of the new economy (with
a few exceptions, notably students, medical patients, diplomats, military personnel, and
international volunteers). Moreover, IMF‟s Balance of Payments Guide (1993) does not specify
any explicit definition for international migrants.
Furthermore, transfers are recorded in the BoP as contra-entries to the provision of a resource
such as grants and gifts, in cash or in kind, without a quid pro quo. Depending on the nature of
the intended use of the „transferred resource‟, transfers are recorded as current transfers in the
14
To address this problem the World Bank, IMF and the UN have been organising a series of collaborative technical
meetings on measuring remittances in recent years since the Heads of G-8 in 2004 emphasised the need for
measuring remittances accurately. The latest is the International Technical Meeting on Measuring Remittances held
in Washington, DC on June 11-12, 2009. The World Bank and the UN had previously organised international
meetings, seminars and working group discussions on the issue of measurement and statistics on remittances in
2005, 2006, 2007 and 2008. See Box A2.1 in the Appendix for the summary of consensus reached so far on
improved measurement of remittances.
18
Stellenbosch University http://scholar.sun.ac.za
current account or as capital transfers in the capital account section of the capital and financial
account. As noted by Reinke (2007), “workers‟ remittances” and “migrants‟ transfers” are
transfers, whereas “compensation of employees” records remuneration for labour. “Workers‟
remittances” involve a current transfer between residents of different countries, whilst “migrants‟
transfers” relate to the capital account changes caused by the change of residence of a
household at the time this relocation/resettlement takes place. “Depending on their specific
interest, data users can decide which of these components best represents their notion on
remittances” (Reinke, 2007: 3).
As a consequence of this state of no uniformity, international remittances have been measured
in many empirical studies from broad and narrow perspectives in various dimensions by various
scholars (see Table 2.1 below). These measurements of remittances are: (i) workers‟
remittances; (ii) the sum of compensation of employees, workers‟ remittances and migrants‟
transfers; (iii) the sum of compensation of employees and migrants‟ transfers; (iv) the total of
migrants‟ transfers plus an additional category in the BoPS, namely „other current transfers‟;
and (v) compensation of employees, workers‟ remittances, migrants‟ transfers, and other
current transfers. It is essential to stress that the IMF reports remittances under four different
sections in its BoPS (see, for example, IMF, 2011a). IMF defines compensations of employees
as the gross earnings of workers residing abroad for less than 12 months, including the value
of in-kind benefits (under the current account subcategory, “income”). Workers‟ remittances are
the value of monetary transfers sent home from workers residing abroad for more than one
year (under the current account subcategory, “current transfers”). Migrants‟ transfers represent
the net wealth of migrants who move from their country of employment to another, often their
native country (under the capital account subcategory, “capital transfers”). More technically,
migrants‟ transfers are contra-entries to the flow of goods and changes in financial items that
arise from the migration of individuals from one economy to another. „Other current transfers‟ is
the component that covers transfers in cash or in kind between individuals, between non-official
organisations such as migrant associations and between an individual and a non-official
organisation. Such transfers include gifts, inheritances, alimony and other support remittances,
non-contractual pensions from NGOs, compensation for damages, and so on recorded under
„other private transfers‟. This component also includes non-contractual pensions from foreign
governments recorded under „other official unrequited transfers‟. Official fund transfers from
foreign governments and enterprises, in the form of donations, aid, sponsorships for education
and cultural exchange programmes including scholarships, which directly or indirectly benefit
19
Stellenbosch University http://scholar.sun.ac.za
households are types of the „other unrequited transfers‟ recorded in the BoPS (IMF, 1993;
1995).
Notwithstanding the identification of international remittances noted above, most researchers
do not take the „other current transfers‟ component into account in empirical studies, probably
because they consider it to be less migrant-related. It is for this reason that many researchers
including Harrison (2003), and Adams and Page (2005), and subsequent debates treated
remittances as certain transactions initiated by persons living or working outside their native
countries and directly related to economic participation of a migrant. It, thus, follows that the
main components of the remittances that have received attention in empirical studies are
compensation of employees, workers‟ remittances and migrants‟ transfers as shown in Table
2.1.
Table 2.1: Examples of Measures of Remittances in Recent Empirical Studies
Author(s)
Kapur (2004)
Abdel-Rahman
(2006)
Niimi
and
Özden (2006)
Vargas-Silva
and
Huang
(2006)
Shelburne and
Palacin (2007)
Adenutsi
and
Ahortor (2008)
Ahortor
and
Adenutsi (2009)
Giuliano
and
Ruiz-Arranz
(2009)
Gupta, Pattillo &
Wagh (2009)
Research Problem Investigated
The economic and political effects of
remittances
Determinants of Foreign Worker
Remittances in the Kingdom of Saudi
Arabia
Migration and Remittances: Causes
and Linkages
Macroeconomic
Determinants
of
Workers‟ Remittances: Host vs. Home
Country‟s Economic Conditions
Remittances in the CIS: Their
Economic Implications and a New
Estimation Procedure
Remittances, Exchange Rate and
Monetary Policy in Ghana
The Impact of Remittances on
Economic Growth in Small-Open
Developing Economies
Remittances, Financial Development
and Growth
Impact of Remittances on Poverty
and Financial Development in SSA
Long-Run Macroeconomic Impact of
Adenutsi
International Migrant Remittances on
(2010a)
Human Development in Low-Income
Countries: Evidence from SSA
Barajas et al. The impact of the global economic
(2010)
crisis on African GDP via the
remittance channel during 2009-2010.
Singh et al. Determinants and Macroeconomic
(2010)
Impact of Remittances in subSaharan Africa
Source: Author‟s compilation
20
Measurement of Remittances
Migrant transfers plus compensation of employees
Workers‟ remittances
Workers‟ remittances, compensation of employees
plus migrants‟ remittances
Private remittances plus other current transfers
Sum
of
workers‟
compensation,
remittances, and migrants‟ transfers.
workers‟
Workers‟ remittances plus compensation of
employees.
Sum of workers‟ remittances, compensation of
employees, migrants‟ transfers plus other current
transfers.
Workers‟ remittances, compensation of employees,
and migrants‟ transfers
Sum of workers‟ remittances, compensation of
employees plus migrants‟ transfers.
Workers‟ remittances and migrants‟ transfers.
Workers‟ remittances.
Workers‟ remittances, compensation of employees
plus migrants‟ transfers or only other current
transfers.
Stellenbosch University http://scholar.sun.ac.za
Given the above, this study recognises that international remittances are measurable from two
main perspectives - the narrow and the broad. Of these two, this study used the narrow
measure as a proxy for international remittances. In the narrow sense, international remittances
emphasise remittances from migrants only, which are measured in this study as the sum of
compensation of employees and workers‟ remittances. This is because these are the two most
consistently reported components of remittances across countries. Also, based on IMF‟s BoPS,
the World Bank concentrates on reporting only these two components in its World Development
Indicators (WDI), Migration and Remittances Factbook, and Global Development Finance
(GDF). Furthermore, Shelburne and Palacin (2007) like many other scholars, point out that
generally these two components form at least 75 per cent of total remittances received by every
country. Shelburne and Palacin (2007:5) observe that “workers‟ remittances are by far the
largest component accounting for well over one-half of the total remittances; compensation of
employees accounts for approximately another third while migrants‟ transfers are relatively
small”.
Additionally, given that these two components (compensation of employees and workers‟
remittances) directly relate to migrants‟ current earnings and are reported in the current account
section of the BoPS under income and current transfers categories respectively, unlike the
remaining two components (migrants‟ transfers and other current transfers) there is a
justification for homogeneity in the measurement of remittances in this context. Again, by
definition, compensation of employees and workers‟ remittances are more regular in flow than
the remaining two components (migrants‟ transfers and other current transfers). For example,
migrants‟ transfers are earned and, hence, recorded only when a migrant changes his/her
country of residence. It is possible migrant resettlement in another foreign country may not
occur frequently, and even in many cases, relocation may never occur in the life of a migrant
who has attained permanent residence status, and if the migrant decides never to return home.
Remittances sent by international benevolent organisations recorded under „other current
transfers‟ may be occasioned by famine, wars, natural disasters, and when a migrant
association decides to support a particular project at home. Usually, these are events that are
not permanent and do not induce regularity in the flow of remittances from these benevolent
institutions. Finally, if the perception that many migrants from poor countries are illegal is
anything to go by, then the issue of permanent residency should be considered as immaterial in
21
Stellenbosch University http://scholar.sun.ac.za
categorising remittance flows to developing countries, particularly those in SSA. Thus, the sum
of workers‟ remittances and compensation of employees should be seen as a more direct
measure of migrant remittance flows to a typical developing economy in SSA. This definition is
the most representative of the newly adopted component in the measurement of international
remittances called „personal remittances‟ in the 2009 edition of the IMF‟s International
Transactions in Remittances: Guide for Compilers and Users as well as the 6th edition of IMF‟s
Balance of Payments and International Investment Position Manual (refer to Box A2.1 in the
Appendix for details).
The broad measure of international remittances is the sum of all the four components viz.
compensation of employees, workers‟ remittances, migrant‟s transfers plus other current
transfers, thus taking into account the total cross-border capital inflows, directly or indirectly,
linked to international migration. In connection with the implications of remittance inflows for
financial dollarisation and price fluctuation, total remittances should be seen as more relevant
and appropriate for policy design if data is available. The reason being that, unlike migrant
remittances (the narrow measure), total remittance inflows are more representative of the
actual addition to currency in circulation if they are immediately spent on domestically produced
goods and services, culminating in a rise in money supply in the remittance-receiving country.
Likewise, if international remittances are spent on imported consumables as found in many
survey studies (for example, Tongamoa, 1987; Dennis, 2003), the threat to exchange rate
stability and monetary policy effectiveness is obvious in an import-dependent region like SSA,
which also has an unfavourable history of high inflation (see Chapter 3). As international
remittances are denominated in foreign currencies, continuous inflows of these funds into a
region with „softer‟ domestic currencies could trigger dollarisation, when households prefer to
hold the „harder‟ foreign currencies. This is likely to be more pervasive in SSA where, due to
the high rates of inflation, real deposit interest rates have either been low or negative.
It is true that measuring remittances in its broadest sense will help reduce the magnitude of
underestimating errors associated with recorded migrant remittances received in SSA, as it is
widely believed that migrants from these poor countries often use unapproved/unofficial
channels to remit due to illiteracy or underdevelopment and low integration of the financial
markets of their native countries into the international financial system. Besides, it is likely that
due to their status as illegal migrants, most migrants from developing countries and, indeed
SSA, cannot conveniently remit home via the officially approved money transfer routes and
22
Stellenbosch University http://scholar.sun.ac.za
hence could be inclined to remit under disguise through native associations abroad, religious or
other benevolent organisations, or by sending easily tradable small-value gift items such as cell
phones, laptop computers, wristwatches, healthcare products, and clothing. Meanwhile, it is
difficult to track and quantify these informal remittances as the informal routes of remitting
which include hand-carried cash and portable valuables are complex whilst the process is
evolutionary. The broad definition is the most comprehensive and representative of the
definitions of the newly adopted component in the measurement of total international
remittances designated „total remittances and transfers to non-profit institutions serving
households (NPISHs)‟ in the 2009 edition of the IMF‟s International Transactions in
Remittances: Guide for Compilers and Users as well as the 6th edition of IMF‟s Balance of
Payments and International Investment Position Manual15 (refer to Box A2.1 in the Appendix for
details).
Despite the expected merits of measuring international remittances in the broadest sense, the
main challenge that confronts analysts when using „total international remittances‟ rather than
the narrow definition (i.e. international migrant remittances) is the absence of consistently
reported data on „migrants‟ transfers‟ and „other current transfers‟ in most developing countries
due to a number of reasons, some of which were mentioned earlier under this very sub-section.
Indeed, reported data on migrants‟ transfers and other current transfers is relatively scarce in
comparison with data on workers‟ remittances and compensation of employees. For example,
GDF, WDI, Migration and Remittance Factbook, and the e-database of the UN report
remittance data on workers‟ remittances and compensation of employees only, apparently
because, generally, workers‟ remittances and compensation of employees constitute about
three-quarters of the total remittances received in migrant-home countries. Another drawback
of using total international remittances is that, by their very nature, the two additional
components – migrant transfers and other current transfers – are less regular in flow compared
to workers‟ remittances and compensation of employees, which are invariably earnings
remitted by permanent and temporary migrants. Therefore, in analysing the motives behind the
flow of remittances involving interpersonal transfers from permanent and temporary
international migrants, it is the narrow definition that should be seen as the more appropriate. In
fact, it is this narrow definition of measuring international remittances that is used throughout
the empirical analysis of this study.
15
For full references, refer to IMF (2009a) and IMF (2009b) respectively.
23
Stellenbosch University http://scholar.sun.ac.za
2.2 FINANCIAL LIBERALISATION
2.2.1 Concept Definition
In separate works, McKinnon (1973) and Shaw (1973) provide the prominent theoretical
framework in support of financial liberalisation16. The McKinnon-Shaw financial liberalisation
theory suggests that the level of financial intermediation should be closely related to the
prevailing level of real interest rates, because the level of real interest rate when held below the
normal competitive levels, indicates the level of financial repression (De Gregorio and Guidotti,
1995). Financial repression, especially under inflationary conditions, stimulates demand for
physical wealth and encourages capital flight (Fry, 1995). According to the McKinnon-Shaw
hypothesis, a positive real interest rate stimulates financial savings and intermediation, thereby
increasing the supply of credit to the private sector, which in turn, stimulates investment and
economic growth (Fry, 1995). Thus, positive real interest rates that are consistent with the
equilibrium interest rates make the allocation of investible funds more efficient, thereby
providing positive implications for economic growth. Financial liberalisation is, therefore, a
possible policy response involving a package of measures intended to remove and/or reform
any undesirable state-imposed constraints on the free mechanism of the financial markets
(Arestis, 2005).
The financial liberalisation thesis was criticised by Neostructuralists notably Taylor (1983), van
Wijnbergen (1982; 1983a; 1983b), Bufie (1984), and Kohsaka (1984) on the grounds that the
absolute operation of a market-based financial system is most unlikely to propel growth in Less
Developed Countries (LDCs). In the opinion of the Neostructuralists, the role of the informal
financial institutions is crucial in financial resource mobilisation and allocation in LDCs. First
and foremost, this is as a result of the low level of incomes and fragmentation of the financial
system. Secondly, information asymmetry is a common feature of the financial sector in
developing countries and, thirdly, because in the formal financial sector, where the banking
sector dominates, the bulk of mandatory reserve requirements forms an important leakage in
the circular flow of funds during the process of financial intermediation. These reasons
underscore the importance of curb financial markets in the mobilisation of household saving
and credit extension (Fry, 1995). The Neostructuralists conclude that, practically, for developing
countries in particular, the pursuit of financial liberalisation is likely to inhibit the rate of
economic growth due to credit constraints because higher real interest rates increase costs of
16
Earlier to this, Bagehot (1873), Schumpeter (1912), Gurley and Shaw (1955; 1960), and Goldsmith (1969)
questioned the wisdom behind the pursuit of repressive financial policies.
24
Stellenbosch University http://scholar.sun.ac.za
production, reduce real wages, and cause stagflation (van Wijnbergen, 1982; Taylor, 1983). In
fact, Stiglitz (1994) was very emphatic in suggesting that government intervention in the
financial market will not only make financial markets function better but will also improve the
performance of an economy.
The Neostructuralists‟ scepticism of financial liberalisation notwithstanding, it is quite clear that
the controversy between the contrasting schools of thought is not much up to this point other
than, perhaps, with regard to the degree and pace with which liberalisation should be pursued.
For instance, as pointed out by the advoates for financial liberalisation, the process still
recommends a gradual approach (Edwards, 1989; McKinnon, 1991) and makes room for
“adequate banking supervision, aiming to ensure that banks have a well-diversified loan
portfolio, macroeconomic stability, which refers to low and stable inflation and a sustainable
fiscal deficit, and the sequencing of financial reforms” (Arestis, 2005: 7). Indeed, as evident in
the composite de jure indices constructed by various authors17, the implementation of financial
reform policies has been gradual although the sequencing of the process differs across
countries. The concept of financial liberalisation as used in this study is, thus, based on the
McKinnon-Shaw theory.
Accordingly, within the framework of the McKinnon-Shaw financial liberalisation theory, this
study defines financial liberalisation as the process of eliminating repressive elements of
financial regulations and policies to rational limits, which should enable financial institutions to
operate more efficiently based on market signals at home and abroad thereby enhancing the
free flow of financial resources. This definition seems adequate to capture the multidimensional
nature of the financial liberalisation process, which Kaminsky and Schmukler (2003) suggest
consists of the deregulation of the domestic financial sector, the foreign sector capital account,
and the stock market18. Financial repression originates from government policies of direct
imposition and indirect interventions in the financial market which ultimately results in restrictive
tendencies and unfair practices commonly associated with imperfect competition within the
financial sector (Fry, 1995). Beim and Calomiris (2001) observe that over the years,
researchers have identified six main ways through which governments often repress their
17
See for example Kaminsky and Schmukler (2003), Abiad and Mody (2005), Shrestha and Chowdhury (2006),
McDonald and Schumacher (2007) and Abiad et al. (2008, 2010).
18
Similarly, Beim and Calomiris (2001: 119) define financial liberalisation as “some combination of the following six
kinds of constraint relaxation: elimination of interest controls, lowering of bank reserve requirements, reduction of
government interference in banks‟ lending decisions, privatisation of nationalised banks, introduction of foreign bank
competition, and facilitation and encouragement of capital inflows”.
25
Stellenbosch University http://scholar.sun.ac.za
financial system. These are: (i) by imposing ceilings on interest rates paid by banks for
deposits; (ii) by imposing high reserves requirements on banks; (iii) by lending to industry
and/or directing bank credit; (iv) by owning and/or micromanaging banks, leaving them with
little autonomy; (v) by restricting entry into the financial market, especially by foreigners; and
(vi) by restricting capital inflows and outflows across international borders (Beim and Calomiris,
2001: 47).
This clearly suggests that financial repression is the opposite of financial liberalisation. In other
words, if an economy is proven to have a high level of financial repression, this directly implies
the economy in question has a lowly-liberalised financial system, and vice versa. In effect, the
indicators of financial repression invariably measure financial liberalisation on the same scale
but at opposite ends of the scale.
2.2.2 Measurement of Financial Liberalisation
Beim and Calomiris (2001) identify six measures of financial liberalisation. The definition and
measurement of each of these indicators are summarised in Table 2.2. Each of these
measures falls into the category of what Gemech and Struthers (2003) describe as policy
outcome measures of financial liberalisation, because they are not capable of capturing the
liberalisation process itself. In early empirical studies, these indicators (see Table 2.2), were
used to proxy the degree of financial liberalisation (Gemech and Struthers, 2003).
Over the last two decades, however, these outcome-based measures (also known as de facto
or ex post measures) have been used to evaluate the developments rather than the actual
process of the liberalisation of the financial sector in many empirical studies including, inter alia,
those by Greenwood and Jovanovic (1990), King and Levine (1993a; 1993b), Hermes (1994),
Demetriades and Hussein (1996), Lynch (1996), Arestis and Demetriades (1997), Levine
(1997), Kar and Pentecost (2000), Levine et al. (2000), and Ang and McKibbin (2007). Unless
these indicators are integrated into one composite index, each of the indicators is used to
measure a specific aspect of financial sector reforms (Williamson and Mahar, 1998; Bandiera et
al., 2000; Laeven, 2003; Abiad and Mody, 2005; Shrestha and Chowdhury, 2006). For
example, earlier works prior to these recent attempts at developing a comprehensive index of
financial sector reforms which merely used ex post measures19 are those of Abe et al. (1977),
19
Some of the financial reforms policy outcome indicators which were widely used in the past include nominal
interest rates, real interest rates, interest rate spread, proportion of bank credit to the private sector, and liquidity
ratio (i.e. M2/GDP).
26
Stellenbosch University http://scholar.sun.ac.za
Cho (1986; 1988), Gupta (1984), Snowden (1987), Ahmed (1988), Morisset (1991), Arestis and
Demetriades (1992; 1993), Roubini and Sala-i-Martin (1992), Bayoumi (1993), AfDB (1994),
and Loayza et al. (2000).
Table 2.2: Measures of Financial Repression / Liberalisation
Indicator
Reserve ratio
Real rates
Liquidity
Description of Indicator
Bank reserves as a ratio of money
plus quasi-money (M2) less
currency held outside of banks.
Nominal annual interest on bank
deposits (i) adjusted for the
realised annual rate of inflation
( ) : r (1 i ) /(1 ) 1 .
Short-term liquid liabilities (M3 if
available, else M2) as a ratio of
GDP.
Application of Indicator in measuring FLB
Higher reserve ratio limits banks ability to create money
and hence an indication of financial repression.
Negative real interest rates means the financial sector is
repressed due to interest rate controls, whilst positive real
interest rates suggests liberalised financial sector via
deregulation of interest rates.
Higher liquidity or monetisation shows the real size of the
financial sector and the extent to which money functions
as a means of payment for essential services and a store
of value. The higher the liquidity ratio the less repressed
the financial sector.
Private
Claims on private sector as a ratio Higher private sector credit allocation indicates financial
borrowing
of total domestic credit.
liberalisation since credit to government and state-owned
enterprises are often directed and, hence uncompetitive.
Bank lending
Deposit bank assets as a ratio of A financial sector is less repressive if this ratio is high
bank assets plus central bank because commercial banks, for mobilising more private
assets.
sector savings than the Central Bank, are expected to
extend more credit than the Central Bank in financing
profitable private sector projects.
Market value
Aggregate
stock
market Higher market capitalisation ratio symbolises improved
financial liberalisation via vibrant equity market, improved
capitalisation as a ratio of GDP.
investors‟ access to correct market information, and
superior management of investment funds.
Source: Author based on Beim and Calomiris (2001: 66)
In more recent years, various attempts have been made to develop a comprehensive index for
measuring the actual dimensions and pace of financial liberalisation processes, such that,
invariably, the ex post measures aforementioned are used essentially to measure the degree of
financial development. The need for a multi-dimensional financial liberalisation index became
increasingly imperative because financial liberalisation is a process involving a wide range of
policy initiatives alongside institutional and structural reforms. These policy actions and reforms
cover the licensing and restructuring of institutions; the development of the appropriate legal,
information and liquidity infrastructure; the operational arrangement for markets; and the design
of instruments (World Bank, 2005). Following this, Williamson and Mahar (1998), Gelbard and
Leite (1999), Bandiera et al. (2000), Kaminsky and Schmukler (2003), Abiad and Mody (2005),
Bekaert et al. (2005), Shrestha and Chowdhury (2006), McDonald and Schumacher (2007) and
Abiad et al. (2008; 2010), made rigorous attempts towards the development of a more
representative and realistic financial liberalisation index. Of these, the indices developed by
27
Stellenbosch University http://scholar.sun.ac.za
Bandiera et al. (2000), Abiad and Mody (2005) and Abiad et al. (2008; 2010) appear to have
enjoyed the most patronage from researchers in recent years as shown in Shrestha (2005),
Shrestha and Chwodhury (2006), Ang and Warwick (2007), Ang (2010), and Ang and Sen
(2011). Table 2.3 presents a summary of some of the popular existing indices of financial
liberalisation developed in recent years20.
Table 2.3: Components and Coverage of Existing Indices of Financial Liberalisation
Author(s), Year
Williamson
and
Mahar (1998)
Bandiera et al.
(2000)
Components / Dimensions of Index, (Coverage)
Credit controls; interest rate controls; entry barriers; regulation of bank operations; bank
privatisation; restrictions on international capital flows (6 variables; 34 countries)
Interest regulation, reserve requirements, direct credit, bank ownership (moves toward
privatisation), liberalisation of securities markets, prudential regulation, and international
financial liberalisation (6 variables; 8 countries)
Kaminsky
and Capital account liberalisation (capital mobility), domestic financial sector liberalisation
Schmukler (2003) (regulations on interest rates, credit allocation and foreign currency deposits), stock market
liberalisation (evolution of regulations on acquisition of shares in the domestic stock market
by foreigners, repatriation of capital, interests and dividends) (3 variables; 28 countries)
Laeven (2003)
Credit controls, interest rate controls, entry barriers, operational restrictions, bank
privatisation, capital account restrictions (6 variables; 13 countries)
Abiad and Mody Credit controls, interest rate controls, entry barriers, operational restrictions, bank
privatisation, restrictions on international financial transactions (6 variables; 35 countries)
(2005)
Shrestha
and Interest rate deregulation, removal of entry barriers, reduction in reserve requirements,
Chowdhury
easing in credit controls, stock market reform, privatisation of state-owned banks, external
account liberalisation (8 variables, 1 country; Nepal)
(2006)
McDonald
and Interest rate liberalisation, number of years real lending and real deposit rates have been
positive, the existence of a significant informal sector and directed credit allocation (4
Schumacher
variables; 37 countries)
(2007)
Abiad
et
al. Credit controls, aggregate credit ceilings and reserve requirements; interest rate controls;
(2010)
entry barriers in the banking sector; state ownership of the banking sector; financial account
restrictions; prudential regulations and supervision of the banking sector; securities market
policy (7 variables; 91 countries)
Source: Author‟s compilation
However, Abiad et al. (2008) criticise the preceding financial liberalisation indices developed
prior to the launch of their index in 2008. According to Abiad et al. (2008), these pre-2008
existing indices are basically de jure measures based on subjective coding and poorly captured
the intensity of the factual liberalisation process. Therefore, drawing lessons from Williamson
and Mahar (1998), Kaminsky and Schmukler (2003), and Abiad and Mody (2005), Abiad et al.
(2008; 2010) developed the most recent index which captures seven dimensions21 of the
financial sector reforms (Table 2.3).22 This new index is likely to enjoy popularity in its
application in empirical studies because, apart from being the most recently developed index
20
Only indices constructed from more than one dimension of financial reforms are reported. Accordingly, Behaert et
al. (2005) is excluded.
21
It is, however, important to emphasise that Abiad et al. (2008) which is the earlier version of Abiad et al. (2010)
attempted the construction of an eight dimensional index by separating directed credit and reserve requirements
from aggregate credit ceilings.
22
For details on how this index was constructed see Section 2.2.2.1.
28
Stellenbosch University http://scholar.sun.ac.za
for financial liberalisation, the new index provides an adequate coverage of the essential
dynamics in the financial policy environment (Ang, 2010). Besides, the policy variables used in
the computation of this new index are more explicit, easy to measure and quite common in
most countries.
With the exception of Kaminsky and Schmukler (2003), Abiad and Mody (2005), and Abiad et
al. (2010)23, all known efforts at computing a unified index for financial liberalisation, for
example, the attempts by Williamson and Mahar (1998), Bandiera et al. (2000), Laeven (2003),
Shrestha and Chowdhury (2006) and McDonald and Schumacher (2007) assigned arbitrary
scores ranging between zero (for fully repressed) and one (for fully liberalised); such that 0.25
or 0.33, 0.50, and 0.66 or 0.75 were assigned to capture partial and gradual deregulation of
each dimension of the financial sector policy reforms. Using Principal Components Analysis
(PCA), the weight of each of the components of the policy reform variables was derived. The
principal component selected is that which accounts for the highest percentage of the total
variance of the system variables. Mathematically, the computation of financial liberalisation
index (FLBI) as determined by these pre-2008 authors can generally be expressed as:
FLBIt w1FLP1,t w2 FLP2,t w3 FLP3,t ........ wn FLPn,t
(2.1)
where FLBI t is the index of financial liberalisation at time t, w is the weight of a specific
component of financial liberalisation policy (FLP) given by the respective eigenvector of the
selected
principal
component,
with
denoting
n
the
number
of
the
reforms
components/dimensions included in the computation of the index. The index for each individual
reforms policy component is obtained when the arbitrary scores initially assigned to
FLP1 , FLP2 ,...., FLPn are multiplied with their respective weights ( wi ) obtained from the PCA.
For each year (t ) , FLBI is derived through a horizontal summation of the calculated values for
the number of the policy-specific reforms components (n) covered by the specific author.
The approach adopted by Kaminsky and Schmukler (2003), Abiad and Mody (2005), and Abiad
et al. (2010) is comparable to the above, except that these authors, rather than assigning
23
All these authors computed FLB using Principal Components Analysis (PCA), but did not assign equal arbitrary
scores to the components. Kaminsky and Schmukler (2003) assigned more weight to capital account liberalisation
whilst Abiad and Mody (2005) emphasize reforms in the domestic financial sector, and apart from that PCA also
used simple sum, sum of squares and sum of square roots. Abiad and Mody (2005) found that the correlations
among the various series obtained from different methods were highly comparable, mostly above 95 per cent with
none below 90 per cent.
29
Stellenbosch University http://scholar.sun.ac.za
arbitrary scores used discrete scores along the different dimensions of the financial policy
reforms identified. The scores for each category of the financial policy reform process are then
combined in a graded index that is normalised between zero and one (Abiad and Mody, 2005;
Abiad et al. 2010). Of these indices, Abiad et al. (2010) is the most comprehensive having
recognised the improved multifaceted nature of the financial reform process and hence
recorded financial policy changes along seven key dimensions instead of the four dimensions
used by Kaminsky and Schmukler (2003) and the six by Abiad and Mody (2005).
2.2.2.1 Computation of the Empirical Financial Liberalisation Index (FLBI)
This study relied on the index constructed by Abiad et al. (2010) on financial sector reforms not
only because it is the most comprehensive among the three approaches that followed a
discrete coding system, but also because an improved set of coding rules were used to
minimise the degree of discretion in assigning scores (see Box A2.2 in the Appendix).
Additionally, Abiad et al. (2010: 286) following Abiad and Mody (2005) allow room for
reversibility in the financial reforms process such that “the imposition of capital controls or
interest rate controls are recorded as shifts from a higher to lower score”, which contributes to
making this index “a much more precise determination of the magnitude and timing of various
events in the financial liberalisation process”.
As suggested by Abiad et al. (2010), a raw score was first assigned to each of the seven
dimensions on the specific scales outlined in Box A2.2. Next, each raw score was normalised
between zero and three according to the rule specified in Box A2.2. Along each of the seven
dimensions identified, therefore, a country was assigned a final score on a graded scale from
zero (for full repression) to three (for full liberalisation). Since the maximum score a country can
obtain for a particular year, which represents full liberalisation in all seven dimensions, is 21
(which is obtained when the optimal normalised score for fully liberalised (3) is multiplied by
the number of dimensions covered, (7)), the actual normalised score for any particular year is
expressed as a ratio of 21. Accordingly, the index of financial liberalisation (FLBI) was
computed using Equation 2.1 which adequately represents the approach adopted by Abiad et
al. (2010).
n
FLBI t
d
i 1
n
i ,t
D
i 1
i
(2.1)
30
Stellenbosch University http://scholar.sun.ac.za
where n is the number of measurement dimensions (in this case n 7 ); d i is the partial
normalised score assigned to each specific component of financial reforms in year t such that
n
0 di ,t 21 , whilst Di is the full normalisation score upon which d i was assigned for each
i 1
n
of the components of financial reforms, so that
D
i 1
i
21 since D 3 . For example, if for a
particular year (say 2000) a country (say X) scored 1.5, 3, 2, 0, 1, 3, and 3 respectively (after
normalisation)24 on each of the seven dimensions/components of financial reforms, then FLBI t
for this country in year 2000 is determined as follows:
X
FLBI 2000
13.5 21 0.6428571.
(2.2)
This result shows that the financial market of country X is partially liberalised as at year 2000
since the computed index exceeds the standard average score of 0.50 and the maximum FLBI
score any country can obtain in a particular year to represent full liberalisation of its financial
market is one.
2.3 CHAPTER SUMMARY AND CONCLUSIONS
This chapter defined the concepts of international remittances and financial liberalisation which
are the running concepts of this study. The definition of these concepts made explicit the
distinction between financial liberalisation and financial development. Whilst financial
liberalisation is de jure, and hence concerned with the actual policy reforms process of the
financial sector, financial development is essentially de facto, and thus concerned with the
outcome of the implementation of financial sector policy reforms. This chapter also discussed
existing alternative approaches and indicators of measuring international remittances and
financial liberalisation. Based on the underlying theories of each of these concepts, the
objectives informing this study, practicality and applicability, a specific indicator (or index) for
each concept was chosen with justifiable explanations. As is conventional in economics and,
indeed social sciences, there is no single measure of any of the core concepts of this study that
can be fully exonerated from theoretical criticisms and empirical limitations. As much as
24
A final normalisation score of 0.75, 1.50, 1.75, 2.50, 2.75 is possible only in the case of credit controls, aggregate
credit ceilings and reserve requirements because this dimension was originally treated as two separate dimensions
(directed credit and reserve requirements; and aggregate credit ceilings) in Abiad et al. (2008), therefore, the
application of the respective 3 4 and 1 4 sum weights and a deviation sum can result in a final code that may not
necessarily be exactly 0, 1, 2, or 3 on the 3-0 scale.
31
Stellenbosch University http://scholar.sun.ac.za
possible, however, reasonable explanations were assigned for the measurement of each
adopted indicator or index, which goes a long way to show that, taking into account the usual
problem of data constraint in developing countries, the specific measure of each concept is the
best applicable measure as per the scope and objectives of this study.
Therefore, throughout the remaining part of this dissertation, unless otherwise stated,
international (or migrant) remittances refer to the sum of “compensation of employees” and
“workers‟ remittances”. Financial liberalisation is measured as the normalised index embracing
seven dimensional variables viz. credit controls, reserve requirements and aggregate credit
ceilings, interest rate control, banking sector entry requirement, international capital flows
control, privatisation of banks, banking sector supervision and regulation, and stock market
development, which according to Abiad et al. (2010) reflect the essential components of
financial reforms. With the running concepts of this dissertation now defined and measured
within the context of this study, the stage is now set for an overview of the macroeconomy of
post-independent SSA. For the most recent years, viz. 2006-2009 for which data is not
available from Abiad et al. (2010), the author used the same sources of information to compute
the financial liberalisation indices for the sampled countries.
32
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 2
Box A2.1: Summary of New Measurement and Definition of Remittances*
Measurement of Concepts:
P Re m T Re m V VI
Total Remittances (TRem) P Re m IV
Personal Remittances (PRem) I II III
Total remittances and transfers to NPISHs
I
Personal transfers
II
Net compensation
of employees
III
Capital transfers
between
households
IV
Social
benefits
V
Current
transfers
NPISHs
to
VI
Capital
transfers
NPISHs
to
Definition of Core Concepts:
A: Personal Remittances: This is defined as current and capital transfers (in cash and in kind) between
resident individuals and non-resident households, plus net compensation of employees earned by persons
working in economies where they are not resident. In other words, personal remittances include all householdto-household transfers and net compensation of non-resident workers.
B: Total Remittances: This includes all household incomes earned from working abroad. Thus, personal
remittances plus social benefits. Intuitively, total remittances include income from individuals working abroad on
temporary basis, earnings of individuals residing abroad, and social benefits earned for working abroad.
C: Total Remittances and Transfers to Non-Profit Institutions Serving Households (NPISHs): This is to
include the total remittances plus the sum of current and capital transfers to Non-Profit Institutions Serving
Households (NPISHs).
Definition of Sub-Concepts:
I: Personal transfers (which now replace workers‟ remittances in BoPS) is defined to include all current transfers
in cash or in kind between resident households and non-resident households. Unlike workers‟ remittances, the
new concept is based neither on employment nor migration status and thus resolves inconsistencies associated
with measuring the previous concept which was linked strictly to residential status.
II: Net compensation of employees is to include gross compensation of employees less taxes, social security
contributions, and travel and passengers transportation related to short-term employment and paid to resident
entities in economies where they are not resident. It, thus, signifies “take-home compensation”.
III: Capital transfers between households are the “non-current” transfers in cash or in kind between resident and
non-resident households.
IV: Social benefits are the benefits payable under social security and pension funds.
V: Current transfers to NPISHs constitute all current transfers from governments and enterprises (in cash or in
kind) to NPISHs from any sector of the sending economy which directly or indirectly benefit households in
another economy (i.e. the receiving economy).
VI: Capital Transfers to NPISHs include all current transfers from governments and enterprises (in cash or in
kind) to NPISHs from any sector of the sending economy which directly or indirectly benefit households in
another economy (i.e. the receiving economy). It may include private and official donations, aid, sponsorships for
education and cultural festivities (including scholarships).
NOTE: Migrant transfers have been removed from the BoP Framework as the concept of migrant has been
abolished since the concepts of personal transfers and remittances are based on the concept of residence
rather than migration status. This is consistent with the use of residence criteria elsewhere in the BoP and
national accounts frameworks.
th
Source: IMF (2009) Balance of Payments and Investment Position Manual, 6 Edition (BPM6)
*
Effective implementation date is unknown but the new reporting system is likely to be formally used in reporting
2010 data as the new Remittances Compilation Guide and the programme aimed at improving Central Bank
reporting were launched during the International Technical Meeting on Measuring Remittances in June 2009.
33
Stellenbosch University http://scholar.sun.ac.za
Box A2.2: Coding Rules for the Financial Liberalisation Index (FLB)
To construct an index of financial liberalisation, codes were assigned along the seven dimensions below. Each
dimension has various sub-dimensions. Based on the score for each sub-dimension, each dimension receives a
„raw score.‟ The explanations for each sub-dimension below indicate how to assign the raw score.
After a „raw score‟ is assigned, it is normalized to a 0-3 scale. The normalization is done on the basis of the
classifications listed below for each dimension. That is, fully liberalised = 3; partially liberalised = 2; partially
repressed = 1; fully repressed = 0.
The final scores are used to compute an aggregate index for each country/year by assigning equal weight to each
dimension. For example, if the „raw score‟ on credit controls and reserve requirements totals 4 (by assigning a code
of 2 for liberal reserve requirements, 1 for lack of directed credit and 1 for lack of subsidised directed credit), this is
equivalent to the definition of Fully Liberalised. So, the normalisation would assign a score of 3 on the 0-3 scale.
I. Credit Controls and Reserve Requirements:
1) Are reserve requirements restrictive?
Coded as 0 if reserve requirement is more than 20 per cent.
Coded as 1 if reserve requirements are reduced to 10–20 per cent or complicated regulations to set reserve
requirements are simplified as a step toward reducing reserve requirements
Coded as 2 if reserve requirements are less than 10 per cent.
2) Are there minimum amounts of credit that must be channelled to certain sectors?
Coded as 0 if credit allocations are determined by the central bank or if mandatory credit allocations to
certain sectors exist.
Coded as 1 if mandatory credit allocations to certain sectors are eliminated or do not exist.
3) Are there any credits supplied to certain sectors at subsidised rates?
Coded as 0 when banks have to supply credits at subsidised rates to certain sectors.
Coded as 1 when the mandatory requirement of credit allocation at subsidised rates is eliminated or banks
do not have to supply credits at subsidised rates.
These three questions‟ scores are summed and coded as follows:
Fully Liberalised = [4], Largely Liberalised = [3], Partially Repressed = [1,2], Fully Repressed= [0]
4) Are there any aggregate credit ceilings?
Coded as 0 if ceilings on expansion of bank credit are in place. This includes bank-specific credit ceilings
imposed by the central bank.
Coded as 1 if no restrictions exist on the expansion of bank credit.
The final sub-index is a weighted average of the sum of the first three categories (with a weight of
last category (with a weight of
1
4
3
4
), and of the
).
II. Interest Rate Liberalisation
Deposit rates and lending rates are separately considered in coding this measure in order to look at the type of
regulations for each set of rates. They are coded as being government set or subject to a binding ceiling (code=0),
fluctuating within a band (code=1) or freely floating (code=2). The coding is based on the following description:
FL=4 [2, 2]
Fully Liberalised if both deposit interest rates and lending interest rates are determined at market rates.
LL = 3 [2, 1]
Largely Liberalised when either deposit rates or lending rates are freed but the other rates are subject to band or
only a part of interest rates are determined at market rates.
PR= 2/1 [2, 0] [1, 1] [1, 0]
Partially Repressed when either deposit rates or lending rates are freed but the other interest rates are set by
government or subject to ceiling/floor; or both deposit rates and lending rates are subject to band or partially
liberalised; or either deposit rates or lending rates are subject to band or partially liberalized.
34
Stellenbosch University http://scholar.sun.ac.za
FR= 0 [0, 0] Fully Repressed when both deposit rates and lending rates are set by the government or subject to
ceiling/floor.
Coding Matrix for Interest Rate Liberalisation
Deposit Rates
Lending Rates
0
1
2
0
FR
PR
PR
1
PR
PR
LL
2
PR
LL
FL
III. Banking Sector Entry
The following sub-measures were considered:
1) To what extent does the government allow foreign banks to enter a domestic market?
This question is coded to examine whether a country allows the entry of foreign banks into a domestic market;
whether branching restrictions of foreign banks are eased; and to what degree the equity ownership of domestic
banks by non-residents is allowed.
Coded as 0 when no entry of foreign banks is allowed; or tight restrictions on the opening of new foreign
banks are in place.
Coded as 1 when foreign bank entry is allowed, but non-residents must hold less than 50 per cent of the
equity share.
Coded as 2 when the majority of shares or equity ownership of domestic banks by non-residents is allowed;
or equal treatment is ensured for both foreign banks and domestic banks; or an unlimited number of
branching is allowed foreign banks.
Three questions look at policies to enhance competition in the domestic banking market.
2) Does the government allow the entry of new domestic banks?
Coded as 0 when the entry of new domestic banks is not allowed or strictly regulated.
Coded as 1 when the entry of new domestic banks or other financial institutions is allowed into the domestic
market.
3) Are there restrictions on branching? (0/1)
Coded as 0 when branching restrictions are in place.
Coded as 1 when there are no branching restrictions or if restrictions are eased.
4) Does the government allow banks to engage in a wide range of activities? (0/1)
Coded as 0 when the range of activities that banks can take consists of only banking activities.
Coded as 1 when banks are allowed to become universal banks.
The dimension of entry barriers is coded by adding the scores of these three questions.
Fully Liberalised= 4 or 5, Largely Liberalised= 3, Partially Repressed= 1 or 2, Fully Repressed = 0
IV. Capital Account Transactions
1) Is the exchange rate system unified? (0/1)
Coded as 0 when a special exchange rate regime for either capital or current account transactions exists.
Coded as 1 when the exchange rate system is unified.
2) Does a country set restrictions on capital inflow? (0/1)
Coded as 0 when significant restrictions exist on capital inflows.
Coded as 1 when banks are allowed to borrow from abroad freely without restrictions and there are no tight
restrictions on other capital inflows.
3) Does a country set restrictions on capital outflow? (0/1)
Coded as 0 when restrictions exist on capital outflows.
Coded as 1 when capital outflows are allowed to flow freely or with minimal approval restrictions.
35
Stellenbosch University http://scholar.sun.ac.za
By adding these three items,
Fully Liberalised = [3], Largely Liberalized = [2], Partially Repressed = [1], Fully Repressed= [0]
V. Privatisation
Privatisation of banks is coded as follows:
FL: Fully Liberalised if no state banks exist or state-owned banks do not make up a significant portion of
banks and/or if the percentage of public bank assets is less than 10 per cent.
LL: Largely Liberalised if most banks are privately owned and/or the percentage of public bank assets is
from 10 per cent to 25 per cent.
PR: Partially Repressed if many banks are privately owned but major banks are still state-owned and/or
the percentage of public bank assets is 25–50 per cent.
FR: Fully Repressed if major banks are all state owned banks and/or the percentage of public bank assets
is from 50 per cent to 100 per cent.
VI. Securities Markets
1) Has a country taken measures to develop securities markets?
Coded as 0 if a securities market does not exist.
Coded as 1 when a securities market is starting to form with the introduction of auctioning of T-bills or the
establishment of a security commission.
Coded as 2 when further measures have been taken to develop securities markets (tax exemptions,
introduction of medium and long-term government bonds in order to build the benchmark of a yield curve,
policies to develop corporate bond and equity markets, or the introduction of a primary dealer system to
develop government security markets).
Coded as 3 when further policy measures have been taken to develop derivative markets or to broaden the
institutional investor base by deregulating portfolio investments and pension funds, or completing the full
deregulation of stock exchanges.
2) Is a country‟s equity market open to foreign investors?
Coded as 0 if no foreign equity ownership is allowed.
Coded as 1 when foreign equity ownership is allowed but there is less than 50 per cent foreign ownership.
Coded as 2 when a majority equity share of foreign ownership is allowed.
By adding these two sub-dimensions,
Fully Liberalised = [4 or 5], Largely Liberalised = [3], Partially Repressed = [1, 2], and Fully Repressed = [0]
**NOTE** If information on the second sub-dimension was not available (as is the case with some low- income
countries), the measure was coded using information on securities market development. If information on securities
markets only was considered, a 0-3 scale was assigned based on the score on securities markets.
VII. Banking Sector Supervision
1) Has a country adopted a capital adequacy ratio based on the Basle standard? (0/1)
Coded as 0 if the Basle risk-weighted capital adequacy ratio is not implemented. Date of implementation is
important, in terms of passing legislation to enforce the Basle requirement of 8 per cent capital adequacy
ratio.
Coded as 1 when Basle capital adequacy ratio is in force. (Note: If the large majority of banks meet the
prudential requirement of an 8 per cent risk-weighted capital adequacy ratio, but this is not a mandatory
ratio as in Basle, the measure is still classified as 1). Prior to 1993, when the Basle regulations were not in
place internationally, this measure takes the value of 0.
2) Is the banking supervisory agency independent from executives‟ influence? (0/1/2)
A banking supervisory agency‟s independence is ensured when the banking supervisory agency can resolve banks‟
problems without delays. Delays are often caused by the lack of autonomy of the banking supervisory agency, which
is caused by political interference. For example, when the banking supervisory agency has to obtain approval from
different agencies such as the Minister of Finance (MoF) in revoking or suspending licenses of banks or liquidating
banks‟ assets, or when the ultimate jurisdiction of the banking supervisory agency is the MoF, this often causes
delays in resolving banking problems. In addition to the independence from political interference, the banking
36
Stellenbosch University http://scholar.sun.ac.za
25
supervisory agency also has to be given enough power to resolve banks‟ problems promptly .
Coded as 0 when the banking supervisory agency does not have an adequate legal framework to intervene
promptly in banks‟ activities; and/or when there is lack of a legal framework for the independence of the
supervisory agency such as the appointment and removal of the head of the banking supervisory agency;
or if the ultimate jurisdiction of the banking supervision is under the MoF; or when a frequent turnover of the
head of the supervisory agency is experienced.
Coded as 1 when the objective of the supervisory agency is clearly defined and an adequate legal
framework to resolve banking problems is provided (the revocation and the suspension of authorisation of
banks, liquidation of banks, and the removal of banks‟ executives etc.) but potential problems remain
concerning the independence of the banking supervisory agency (for example, when the MoF intervenes in
the banking supervision (in such a case that) or when the board of the banking supervisory agency is
chaired by the MoF, although the fixed term of the board is ensured by law); or when, although clear legal
objectives and legal independence are observed, the adequate legal framework for resolving problems is
not well articulated.
Coded as 2 when a legal framework for the objectives and the resolution of troubled banks is set up and if
the banking supervisory agency is legally independent from the executive branch and actually not interfered
with by the executive branch.
3) Does a banking supervisory agency conduct effective supervisions through on-site and off-site examinations?
(0/1/2)
Conducting on-site and off-site examinations of banks is an important way to monitor banks‟ balance sheets.
Coded as 0 when a country has no legal framework and practices of on-site and off-site examinations are
not provided or when no on-site and off-site examinations are conducted.
Coded as 1 when the legal framework of on-site and off-site examinations is set up and the banking
supervision agency has conducted examinations but in an ineffective or insufficient manner.
Coded as 2 when the banking supervisory agency conducts effective and sophisticated examinations.
4) Does a country‟s banking supervisory agency cover all financial institutions without exception? (0/1)
If some kinds of banks are not exclusively supervised by the banking supervisory agency or if offshore
intermediaries of banks are excluded from the supervision, the effectiveness of the banking supervision is seriously
undermined.
Coded as 1 when all banks are under supervision by supervisory agencies without exception.
Coded as 0 if some kinds of financial institutions are not supervised by the banking supervisory or are
excluded from banking supervisory agency oversights.
Enhancement of banking supervision over the banking sector is coded by summing up these four dimensions, which
are assigned a degree of reform as follows:
Highly Regulated = [6], Largely Regulated = [4-5], Less Regulated = [2-3], Not Regulated = [0-1]
Source: Abiad et al. (2008; 2010)
25
According to Omori (2004: 13), “Quintyn and Taylor (2002) categorise the independence of banking supervisory
agencies into four: regulatory independence, supervisory independence, institutional independence, and budgetary
independence. In this dataset, independence is measured by combining institutional independence and supervisory
independence. In the case of central bank independence, a legal framework of a central bank developed for
countries and/or the frequency of turnover of the governor of the central bank for developing countries are the often
used indicators. However, as discussed above, since the banking supervisory agency is not necessarily vested in
the central bank, legal documents for banking supervision are less available, and obtaining the information for
counting the frequency of the turnover of the head of the banking supervisory agency is much more difficult. In this
vein, we basically relied on experts or researchers‟ evaluation in coding the independence of the banking
supervisory agency. Lora (1997) also created the indicators based on subjective judgement of the quality of banking
supervision.”
37
Stellenbosch University http://scholar.sun.ac.za
CHAPTER THREE
MACROECONOMIC ENVIRONMENT AND EXTERNAL CAPITAL FLOWS TO SUBSAHARAN AFRICA (1960-2009)
3.0 INTRODUCTION
This chapter examines the general policy environment and macroeconomic performance of
post-independent SSA vis-à-vis the inflows of external financial resources to the sub-region.
The policy environment and performance of the financial sector as well as the foreign sector in
the sub-region are discussed to determine whether the various economic policies and reforms
implemented by the sub-region have had any significant impact on macroeconomic
performance and external capital flows. The focus of this chapter is on the period 1960-2009,
and this is based, essentially, on relevance and data availability. Rather than adopting a
rigorous econometric analytical framework, a set of simple descriptive statistical measures viz.
arithmetic mean, correlation coefficients and trend analyses are used to provide an insight into
the historical antecedents of the sub-region from a macroeconomic perspective. In doing this,
the chapter reviews the post-colonial political economy of SSA, and does a series of contextual
analyses of the macroeconomic performance and policy environment of SSA under the prereforms era (1960-1979), the reforms era (1980-1989), and the post-reforms era (1990-2009).
Following this, the chapter discusses macroeconomic performance and foreign capital flows to
SSA, and the trends and dynamics of remittance flows to SSA from a global perspective. The
chapter also outlines the stylised facts of migrant remittance flows to SSA, and the policy
imperatives of remittance inflows and macroeconomic policy stance for the sub-region.
3.1 BACKGROUND
Notwithstanding the Darfur crisis, which has been ongoing since February 2003, the Somalia
civil war, the Chad civil war and the hardly settled border and territorial dispute between Nigeria
and Cameroon over the right of ownership of the Bakassi Peninsula, largely, the political
economy of modern SSA appears to be relatively stable since post-independence. Today,
virtually every SSA country has what can be described as a multi-party democratically elected
president. Again, collectively, countries within the SSA sub-region now visibly frown on military
A paper from this Chapter titled, “Macroeconomic Environment and Remittances in Post-Independent SubSaharan Africa: Magnitudes, Trends and Stylised Facts,” was presented at IMF Staff Seminar, February 16, 2011,
Washington, DC, USA. This paper has been published in Journal of Studies in Economics and Econometrics, 36(2):
1-22.
38
Stellenbosch University http://scholar.sun.ac.za
take-overs, dictatorship, and the use of constitutionally unapproved means to assume political
leadership. More than ever before, the African Union (AU) and the international community,
especially western trade and donor partners26 now act very swiftly to sanction governments and
leaders that violate human rights, adopt extremist discrimination practices, and exploit
undemocratic means to assume political leadership in countries within the sub-region. For
instance, unprecedented in the history of the Economic Community of West African States
(ECOWAS), the sub-regional body acted very swiftly to review the political and security
situation in Côte d‟Ivoire after the declaration of certified second round Presidential election
results on November 28, 2010. Subsequently, ECOWAS issued a statement within ten days
after the elections to denounce the incumbent President, Mr. Laurent Gbagbo, and asked him
to concede defeat without delay. This is expected to continue to create the propitious
environment for some consistency in the formulation and implementation of pro-growth and
sustainable development policies under adopted national economic development programmes
towards the socioeconomic progress of the sub-region.
The strides being made in improving good governance and building stronger state institutions
provide a stable political environemnt necessary for creating the ideal investment atmosphere
required for the mobilisation of critical resources in SSA. Apart from the encouraging
developments on the political landscape of the sub-region, various macroeconomic policy
reform programmes have been adopted and implemented by SSA countries since political
independence in the 1960s. Each of these programmes was, among other things, centred on
mobilising domestic and external resources crucial to the socioeconomic development agenda
of the sub-region since SSA has been identified as the region most lacking critical resources
(Devarajan et al., 2002; Gupta, Powell and Yang, 2006). And as the sub-region is identified as
one of the leading net exporters of skilled and unskilled labour to the industrialised world
(Migration Policy Institute, 2006), the question that has remained unanswered is: Has the
implementation of various macroeconomic policies led to higher inflows of migrant remittances
during the post-independence era? In other words, are remittance inflows from SSA migrants
from abroad responsive to the changing macroeconomic policy environment? In order to find
the appropriate response to this question, this chapter seeks to explore the relationship
between the changing inflows of migrant remittances and the various macroeconomic policies
26
This refers essentially to those from Europe, the US and Canada. In recent years, however, China has emerged
as the leading trading partner and investor in SSA. Chinese companies invested US$ 1 billion in Africa in 2007
(Politzer, 2008).
39
Stellenbosch University http://scholar.sun.ac.za
implemented by SSA as a sub-region since independence. In doing so, rather than adopting a
rigorous macroeconometric analytical approach, this chapter uses a set of simple descriptive
statistics to provide insight into the performance of the sub-region in attracting remittances
during the post-independence era. A more rigorous econometric analysis of the determinants of
international migrant remittances under changing macroeconomic policy environment is
reserved for Chapter Four.
3.2 A CONTEXTUAL ANALYSIS OF POLICY ENVIRONMENT AND MACROECONOMIC
PERFORMANCE OF SSA (1960-2009)
Generally, the macroeconomy of SSA has not witnessed any significant and consistent
improvement since independence27, even though there was a major policy paradigm shift
across the length and breadth of the sub-region in the 1980s. The major policy reform the subregion underwent after independence has been the IMF/WB-led Economic Recovery
Programme (ERP) that embodied the Structural Adjustment Programme (SAP). Although there
has been some marginal progress in financial sector development and export growth, SSA still
lags behind in development, about fifty years after independence, with no significant gains in
terms of poverty reduction, food security, technological advancement and industrialisation,
production capacity and high productivity. From a macroeconomic perspective, these elements
of (under)development crop up from the structures of production, consumption, external trade,
technology, employment as well as the economic system and socio-political configuration of an
economy. It is, therefore, impossible to offer any functional understanding, remedies or policy
recommendations aimed at addressing these entrenched problems of SSA without a thorough
structural analysis of the political economy of the sub-region. Therefore, this chapter presents
structural analyses of SSA which broadly take into account a critical examination of the
enabling and disenabling internal and external macroeconomic factors which prevailed during
the period 1960-2009 under three main regimes – the pre-reforms, the reforms, and the postreforms regimes.
27
In recent years (i.e. between 2000-2009), however, available data show improvements in key macroeconomic
indicators such as real GDP per capita, and rate of inflation comparable to what was attained in the 1960s and the
1970s (see, for instance, Table 3.1).
40
Stellenbosch University http://scholar.sun.ac.za
3.2.1 The Pre-Reforms Era (1960-1979)
Many SSA countries, including inter alia Botswana, Democratic Republic of Congo, Ghana,
Kenya, Nigeria, Senegal, South Africa, and Uganda, were bequeathed reasonable international
reserves and social infrastructure in the form of schools, hospitals, and road networks
concentrated in the urban centres at the time of their political independence, albeit with poor
transport systems, unreliable energy supplies and
high levels of illiteracy, poverty, and
malnutrition, especially in the rural communities which constituted at least two-thirds of their
economies. The economy of SSA was largely subsistence and agrarian with low
industrialisation at the time of independence. The key structural features of the economy of
SSA at the time of independence can be described as including:
i.
over-dependency on the primary sector with raw agricultural products and exhaustible
natural resources28 dominating exports;
ii. a predominantly subsistence economy in production and domestic trade, and hence a
large informal sector;
iii. a fragmented economy that neglected the large rural economy, the informal sector and,
indeed, the private sector which was virtually crowded out by the public sector, hence
entrepreneurial spirit was not fostered.
iv. an absence of strong regional trade among SSA countries;
v. low production capacity that relied on high labour-intensive and non-scientific or outdated production techniques; and
vi. low technological and infrastructural base, and absence of a strong institutional
capacity.
Nevertheless, given the enthusiastic interest and commitment of the indigenous people in
taking over the political leadership and in managing their own affairs, there was high optimism
that independent SSA countries would quickly transform the structure of their economies and
attain higher-income status, once the available resources were put to optimal use.
Unfortunately, within the era under consideration (1960-1979), most of the economic policies
adopted by governments within the sub-region were restrictive with over-emphasis on exports
of raw primary products rather than commitment to value-addition and industrialisation. Most of
the industries inherited from the colonial masters were used mainly to produce import28
These include minerals like timber, gold, bauxite, manganese, diamond, iron ore, and uranium.
41
Stellenbosch University http://scholar.sun.ac.za
substituting consumer goods.
With economic policies being largely socialist in content, the state played a leading role in
investment and industrialisation which by the mid 1970s, had rapidly eroded all the reserves
inherited, especially as most of the heavy investments were not made prudently29. Thus, it can
be argued that, to a very large extent, bad governance and lack of vision and leadership
qualities resulting in ad hoc implementation of economic policies, mismanagement and corrupt
practices, together with political instability, drought, conflicts, and lack of adequate resources to
address the numerous socioeconomic problems facing the sub-region, led to stagnated growth
and underdevelopment of the sub-region. Many scholars and institutions, including Sachs and
Warner (1997) and AFRODAD (2007), blame the economic woes of the sub-region on external
factors such as declining prices of primary products, decreasing net inflows of official and
private financial resources, global recessions and oil price shocks of the 1970s and early1980s, and policy prescriptions by international financial institutions (mainly the IMF and the
World Bank). However, this assertion can be challenged insofar as good vision and other
quality leadership skills can play a decisive role in negotiations, the choice of a development
strategy, and the timing and sequencing of policy implementation. For example, according to
Sahn et al. (1994), consensus on the causes of the abysmal performance of postindependence Africa does not only end with external factors such as the collapse of
commercial lending to developing countries and worldwide economic recession, but also with
the fact that the implementation of economic policies was misguided. By the late 1970s, the
IMF in partnership with the World Bank concluded that the source of economic problems
confronting developing countries and, indeed, SSA is basically lack of structural transformation
and, hence, they advocated extensive economic reforms in these economies.
3.2.2 The Reforms Era (1980-1989)
By the end of the 1970s, it was evident that SSA could no longer rely on inward-looking
economic policies and industrialisation strategies, which were vulnerable to external shocks, to
resolve the deep-seated impediments to economic growth and prosperity and, subsequently,
put the sub-region on a sustainable path to reverse its worsening economic fortunes. Thus, two
decades after the political independence of SSA, its economic performance was worse and
29
In Ghana, for example, Nkrumah (the First President of the Republic) established about 200 state-owned
industries without taking into account the regular supply of raw materials required to feed these „state-protected
industries‟. As a result, many of these state monopolies never functioned whilst a majority of the remaining stateowned industries operated below full capacity but with large full-time workers on the state payroll.
42
Stellenbosch University http://scholar.sun.ac.za
characterised with declining per capita growth, galloping inflation, a worsening balance of
payments, a debilitating debt burden, and rising unsustainable fiscal deficits, unemployment,
poverty and deprivation (see Figure 3.1 and Tables 3.1 and 3.2). It became apparent that SSA
needed to adjust its economy to eliminate structural imbalances and react appropriately to
external shocks. In response, during the 1980s, many SSA countries accepted and pursued the
IMF/WB-recommended
economic
policy
reforms
towards
stabilisation
and
structural
transformation under a policy package known as the Economic Recovery Programme and
Structural Adjustment Programme (ERP/SAP)30. Essentially, economic reforms connote
economic openness and market fundamentalism for macroeconomic stability through policies
of privatisation, deregulation and liberalisation as prescribed by Bretton Woods Institutions,
notably the IMF and the World Bank, in the late 1970s. The broad sets of policy
recommendations by these institutions can be summarised as:
i.
austerity and fiscal policy discipline;
ii. redirection of policy spending from subsidies (especially indiscriminate subsidies)
toward broad-based provision of key pro-growth, pro-poor services like primary
education, primary healthcare and infrastructural investment;
iii. tax reforms via broadening the tax base and adopting moderate marginal tax rates;
iv. interest rates that are positive but moderate in real terms and freely determined by the
financial market forces;
v. competitive exchange rates;
vi. privatisation or divestiture of state-owned enterprises;
vii. trade liberalisation with special emphasis on elimination of quantitative restrictions
(licensing, etc.); any trade protection to be provided by low and relatively uniform tariffs;
viii. financial sector reforms to include the establishment or revitalisation of domestic capital
markets and the liberalisation of inward foreign direct investment;
ix. deregulation to include abolition of regulations that impede market entry or restrict
competition, except for those justified on safety, environment and consumer protection
grounds, and the prudent oversight of financial institutions; and
x. legal security for property rights.
30
According to OAU (1985), more than 30 African countries adopted ERP/SAP as of 1988 with support from the IMF
and the World Bank.
43
Stellenbosch University http://scholar.sun.ac.za
It is essential to point out that although both the IMF and the World Bank often lend to
economically depressed and developing countries, these loans are packaged to address
different problems confronting the beneficiary countries31. For instance, while the IMF mainly
lends to countries suffering from BoP problems and hence cannot repay their accumulated
external debts, the World Bank extends loans to enable countries, (mainly LDCs) to finance
specific development projects. Accordingly, traditional IMF Structural Adjustment Loans (SALs)
are due to be repaid within 30 and 48 months and are mainly directed at the temporary fixing of
problems that confront a country in crisis of an apparently perpetual macroeconomic
imbalance. The World Bank‟s SALs, however, are a longer-term credit package of loans and
grants to countries to finance critical development projects. In this regard, the World Bank is
functionally structured into two: the International Bank for Reconstruction and Development
(IBRD) and the International Development Association (IDA). Whereas IBRD focuses on
middle-income and credit-worthy poor countries, IDA focuses on the lowest and least creditworthy countries. Therefore, in the 1980s, most SSA countries contracted IMF/WB SALs to
tackle various economic problems concurrently.
The pursuit of ERP/SAP generally involved the implementation of stringent economic policies
and measures of demand management towards rationalising the overall expenditure pattern in
order to restore financial stability, fiscal balance and external equilibrium. These policies were
implemented through reliance on market forces and private sector-led growth. In effect, the
pursuit of ERP/SAP signified a radical departure from all previous socialist-oriented policies
adopted by the sub-region. By and large, the implementation of ERP/SAP resulted in the
abandonment of state controls and restrictions on prices, imports, and exchange rates. Broadly,
the various policy measures under the reforms programme were pursued in varying degrees of
implementation. Whilst countries like Benin, Ethiopia, Ghana and Malawi tried to follow the
prescriptions of the ERP/SAP strictly, others like Kenya, Nigeria ad Zimbabwe initiated a
number of complementary policies and programmes such as national employment
programmes, urban mass transit programmes, relief packages for public sector workers, and
food, road and rural infrastructural programmes alongside the ERP/SAP.
31
Aside the IMF and the World Bank as the main sources of institutional loans to developing countries, many of
these countries also rely on bilateral loans which can take the form of overdrafts, term loans and revolving credit
facility. Another form of credit available to developing countries is syndicated loans which are provided by a group of
lenders to developing countries.
44
Stellenbosch University http://scholar.sun.ac.za
The outcome of the ERP/SAP has, however, not shown which of these forms of implementation
is more successful. For instance, the major gains of ERP/SAP across the sub-region can be
noted as increased access of the private sector to foreign exchange and markets, export
diversification leading to the emergence of non-traditional commodities, increased international
trade openness, and improved access to financial services. Other achievements include
restructuring of public enterprises and freeing of prices. Despite these gains, the
macroeconomy failed to recover fully from the declining trend as reflected in real income per
capita growth, domestic savings and investment, basically because output per capita growth
failed to keep pace with population growth (see Table 3.1). Also, until quite recently, there were
no reversals in the rising trend of external debt accumulation and the rate of inflation. These
suggest that the formulation of various specific macroeconomic policies and their
implementation under the broad ERP/SAP programme were erratic, as most SSA governments
lacked full commitment to the underlying principles and short-term objectives of ERP/SAP. For
instance, in many SSA countries including Gambia, Ghana, Kenya, Senegal and Tanzania,
governments kept on subsidising some non-essentials and at the same time continued with
increased spending on non-productive activities in breach of the ideals of ERP/SAP in order to
gain political popularity. This has led to structural imbalances, macroeconomic instability and
economic retardation that still persist and abound among virtually all countries which adopted
the ERP/SAP in SSA.
Thus, by the mid-1980s, SSA countries were blaming the orthodox ERP/SAP for the poor
economic performance of the sub-region. According to these countries, the classical policy
instruments of credit control, tight money supply, flexible exchange rate and interest rate
adjustments, fiscal discipline and tax reforms, and trade liberalisation, by their very design, had
no human face to bring about the desired impact on their economies. Besides, the
„conditionalities‟ of these IMF SALs in particular had made it difficult for the countries within the
sub-region to attain the desired self-sufficiency and economic independence. Therefore, by
1986, with support from the UN, governments from the sub-region, through the OAU, had
designed three complementary/alternative SAP versions32 for selective implementation across
SSA.
32
This includes UN-led United Nations Programme of African Economic Recovery and Development (UN-PAAERD)
for 1986-1990 and the UN-led African Alternative Framework to Structural Adjustment Programmes for
Socioeconomic Recovery and Transformation (AAF-SAP) for 1989/1990. See Table A3.2 in the Appendix for details
on the two key alternative programmes, namely UN-PAAERD and AAF-SAP.
45
Stellenbosch University http://scholar.sun.ac.za
3.2.3 The Post-Reforms Era (1990-2009)
As shown in Figure 3.1 and Tables 3.1 and 3.2 below, the various policy reforms initiated by
SSA countries did not achieve much in the 1980s. The failure of the orthodox IMF/WB-led
ERP/SAP33 and other modified supportive SSA versions34 to address the key problems
confronting the sub-region meant that other purposive and positive actions were required to
break the structural bottlenecks and imbalances to position the economy of SSA towards
sustainable growth and to protect it against economic shocks.
30
90
25
80
70
20
60
15
50
10
40
5
30
0
20
-5
10
-10
0
External debt stocks, total (% of GDP)
Gross domestic savings (% of GDP)
External balance on goods and services (% of GDP)
Gross fixed capital formation (% of GDP)
external debt stocks (% of GDP)
all other indicators (in line graph)
Figure 3.1: Trends in Selected Macroeconomic Indicators in SSA, 1960-2009
GDP per capita growth (annual %)
Inflation, GDP deflator (annual %)
Source: Author based on WDI and GDF (April 2011). Note: The starting point of gross fixed capital formation and
external debt stock are 1970 and 1975 respectively because regional data for earlier years were non-existent.
By and large, these modified ERP/SAP programmes and supportive frameworks designed to
guide economic policy implementation and to break the shackles of underdevelopment have
not achieved the desired results. The extent of the lack of success of these policy initiatives can
be easily measured by the number of SSA countries that were forced into adopting the HIPC
Initiative at the beginning of this new millennium. By adopting the HIPC Initiative, the countries
have accepted the fact that they are poor and cannot manage their debts sustainably, and
hence need international support to address critical socioeconomic problems under a Poverty
Reduction Strategy (PRS) within the context of IMF‟s Poverty Reduction and Growth Facility
33
See Dollar and Svensson (2000) for evidence and reasons on why SAP was less successful in low-income and
African countries. Van de Walle and Johnston (1996) and Knight and Santaellah (1997) also provide other possible
reasons for the failure of ERP/SAP in most developing countries.
34
These are UN-PAAERD and AAF-SAP.
46
Stellenbosch University http://scholar.sun.ac.za
(PRGF)35. Thus, at present, SSA still remains the poorest region in the world and requires the
most resources to provide essential social amenities and to achieve the Millennium
Development Goals (MDGs).36
In Figure 3.1, it is shown that SSA has not made much economic progress since independence
in the 1960s. For instance, as a proportion of GDP, gross fixed capital formation and external
balance on goods and services are lower today than at the time of independence (see also
Table 3.1). Similarly, GDP per capita growth is lower in the 2000s than in the 1960s just like
gross domestic savings as a percentage of GDP, although showing a consistent upward trend
since the mid-1990s, and still lags behind the level attained in the mid-1970s.
Inflation, which averaged about three per cent in the 1990s, increased to an average of seven
per cent in the 2000s. In a similar fashion, external debt stock as a percentage of GDP
increased from about 15 per cent during the pre-reforms era to a high of over 70 per cent in the
1990s before debt cancellation and other reliefs under HIPC initiatives reduced it to about 22
per cent between 2006 and 2009.
3.2.4 Macroeconomic Performance and Policy Environment in SSA
Table 3.1 summarises the key macroeconomic performance indicators of SSA since
independence.
A key feature of the sub-region is that it is apparently trapped in a low-income equilibrium level
as its annual population growth rate consistently exceeds its annual GDP per capita growth
rate. In fact, even though the average annual population growth rate today is similar to what it
was at the time of independence, average yearly GDP per capita growth rate today is lower
than the rate attained in the 1960s. During the reforms era and in the first decade of the postreforms era, average growth in per capita income was negative whereas population growth was
2.81 per cent for the reforms era reaching an all-time high of 2.91 per cent in the 1980s. The
undesirable consequences for the higher dependency, unemployment, underemployment, net
savings and investment are obvious and unambiguous.
35
See Table A3.1 in the Appendix for the list of SSA countries that adopted the HIPC Initiative as of June 30, 2010
and Table A3.2 in the Appendix for details on HIPC Initiative as an economic development strategy.
36
Refer to Table A1.1 in Chapter 1, Brossard and Gacougnolle (2001), AfDB (2002), Mingat et al. (2002), World
Bank (2002) and World Bank (2003) for details on estimated resources required by SSA to meet MDGs and see Box
A3.1 for the list of MDGs.
47
Stellenbosch University http://scholar.sun.ac.za
Table 3.1: Macroeconomic Performance and Policy Environment in SSA, 1960-2009
Key Macroeconomic Indicators
GDP (constant 2000 US$ billions)
GDP growth (annual %)
GDP per capita (constant 2000 US$)
Pre-Reforms Era
Reforms
Post-Reforms Era
Overall Period
1960-69 1970-79 1960-79
1980-89
1990-99 2000-09 1990-09
1960-2009
120.64
190.97
155.81
243.36
293.83
422.66
358.25
254.29
4.64
4.07
4.34
2.17
2.02
4.54
3.28
3.46
469.57
573.66
521.62
552.17
504.61
558.08
531.35
531.62
GDP per capita growth (annual %)
2.06
1.26
1.64
-0.72
-0.67
1.97
0.65
0.75
External balance (% of GDP)
0.35
-1.51
-0.58
-0.37
-1.43
-1.41
-1.42
-0.87
External debt stocks, total (% of GDP)
-
Gross fixed capital formation (% of GDP)
15.64
15.64
39.72
65.22
39.56
52.39
40.03
-
18.83
18.83
18.33
17.16
16.87
17.02
17.80
19.40
28.87
24.13
22.23
14.48
23.50
18.99
21.70
Inflation (annual %)
3.06
9.18
6.28
10.68
10.00
7.03
8.51
8.09
Population growth (annual %)
2.53
2.78
2.66
2.91
2.71
2.52
2.62
2.69
Gross domestic savings (% of GDP)
Source: Author‟s computations from WDI (April 2011). Note: GGFCE and GFCF denote general government final
consumption expenditure and gross fixed capital formation respectively.
Again, on the average, annual GDP growth rate, external balance and domestic investment are
lower today than at the time of political independence. Annual average inflation is also higher
today than it was in the 1960s. These bear testimony to the fact that the economic fortunes of
the sub-region are worse today than 50 years ago, with lower aspirations for greater prospects
in the future. This surely is enough incentive for the active labour force to seek greener
pastures outside the sub-region, and this has actually been the case in recent years.
As far as the foreign sector is concerned, it can be argued that the sub-region chalked some
remarkable successes, even though the ultimate indicator, the international trade balance, is
still low and negative compared to what was recorded in the 1960s as shown in Table 3.2. For
example, as evident in Table 3.2, on the average, the current account balance turned positive
for the first time in the 2000s.
Table 3.2: International Trade Performance and Policy Environment in SSA, 1960-2009
Pre-Reforms Era
External Sector Policy Indicators
Current account balance (% of GDP)
Exports of goods and services (% of GDP)
Reforms
Post-Reforms Era
Overall Period
1960-69 1970-79 1960-79 1980-89 1990-99 2000-09 1990-09
-0.09
-
1960-2009
-1.56
-0.82
-2.31
-2.13
0.85
-0.64
-1.05
26.86
26.86
23.79
26.67
34.16
30.42
28.08
Exports of goods and services (annual % growth)
-
20.18
20.18
1.68
4.23
13.42
8.82
7.30
Exports of goods (% of GDP)
-
23.90
23.90
20.94
22.52
29.42
25.97
24.26
Imports of goods (% of GDP)
-
21.36
21.36
18.76
20.82
24.58
22.70
21.38
Import cover (goods exports/goods imports)
International trade balance (% of GDP)
-
1.12
1.12
1.13
1.09
1.20
1.14
1.14
0.35
-1.29
-1.22
-0.97
-1.80
-0.23
-1.02
-1.06
Openness to international trade (%)*
55.01
55.01
48.54
55.14
68.56
61.85
57.20
*
Source: Author‟s computation based on WDI (April 2011). Exports plus imports of goods and services as % of GDP
48
Stellenbosch University http://scholar.sun.ac.za
In addition, despite the fact that growth in exports and imports of goods as a percentage of
GDP can be described as virtually stagnant since political independence, foreign trade policy
indicators, notably import cover (measured as value of exports as a ratio of imports of goods)
and international trade openness have assumed affirmative and encouraging trends in recent
years. More specifically, as evident in Table 3.2, import cover of 1.20 for the 2000s averaging
1.14 for the post-reforms era is a slight improvement over the pre-reforms average of 1.12. In a
similar fashion, the degree of international trade openness which stood at 55.01 per cent during
the pre-reforms era, improved significantly to 68.56 per cent in the 2000s and averaged 61.85
per cent for the post-reforms period.
An overview of the performance of the financial sector has shown that, arguably, monetary
policy instruments have not succeeded in reducing financial risk as interest rate spread
increased from six per cent during the pre-reforms era to 11.76 per cent in the 2000s averaging
10.98 per cent for the post-reforms era (Table 3.3).
Table 3.3: Financial Sector Performance and Monetary Policy Environment in SSA, 1960-2009
Pre-Reforms Era
Monetary and Financial Sector Indicators
Bank liquid reserves to bank assets ratio (%)
Reforms
Post-Reforms Era
1960-69 1970-79 1960-79 1980-89 1990-99
4.72
5.00
Overall Period
2000-09 1990-09
4.90
9.50
13.79
14.03
1960-2009
13.91
9.82
Domestic credit by banks (% of GDP)
39.32
41.64
40.87
56.49
75.48
84.09
79.79
61.64
Domestic credit to private sector (% of GDP)
28.75
28.82
28.79
36.69
53.88
59.50
56.69
42.95
Interest rate spread (%)
-
6.00
6.00
6.59
10.35
11.76
10.98
8.65
Listed domestic companies, (end of period)
-
-
-
-
1,139.00
991.00
991.00
991.00
Market capitalization (% of GDP)
-
-
-
-
100.51
103.94
102.23
102.23
42.29
29.26
32.99
32.71
32.80
35.45
34.12
33.44
-
-
-
-
17.19
38.77
28.81
28.81
-
17.53
28.56
24.55
24.55
Money and quasi money (M₂) as % of GDP
Stocks traded, total value (% of GDP)
Stocks traded, turnover ratio (%)
Source: Author‟s computations from WDI and GDF (April 2011)
Financial depth, measured as broad money (M2) as a percentage of GDP, also declined
considerably from an average of 42.29 per cent in the 1960s with a pre-reforms average of
32.99 per cent to 35.45 per cent in the 2000s with a post-reforms average of 34.12 per cent.
When using this indicator, it is difficult to conclude that a significant amount of money still
circulates outside the banking system which points to low public confidence in the banking
system and the extent to which the economy is under-banked. The reason is that broad money
includes currency in circulation and demand deposits; and it is expected that as a financial
49
Stellenbosch University http://scholar.sun.ac.za
system develops, quasi money should increase as financial institutions mobilise more excess
funds, thereby reducing the currency in circulation. Unfortunately, data on narrow money (M1) is
not available on a regional basis.
However, a lot was achieved as liquid reserves to bank assets increased sharply from 4.90 per
cent in the pre-reforms era to 9.50 per cent during the reforms, and to 13.91 per cent during the
post-reforms era. Similarly, as a ratio of GDP, domestic credit by the banking sector increased
from 40.87 per cent to 79.79 per cent on the average during the same period. Also worth
noting, more importantly, is the upshot in domestic credit to the private sector as a ratio of
GDP. This ratio increased from 28.79 per cent during the pre-reforms era to as much as 36.69
per cent during the reforms era, further rising to 56.69 per cent for the post-reforms era, as
shown in Table 3.3.
3.3 EXTERNAL CAPITAL FLOWS TO SSA (1960-2009)
3.3.1 Composition and Trends in External Capital Flows to SSA: A Global Outlook
Conventionally, besides contractual loans, capital flows to SSA and other developing
economies are a composition of FDI, ODA, and portfolio equity. In recent years, however,
remittances have emerged as a complementary source of external capital for developing
countries. Figure A3.1 in the Appendix depicts the trends in external capital flows to the various
developing economies of the world. Figure 3.2 shows the trends in external capital flows to
SSA since 1970.
Figure 3.2: Trends in External Capital Flows to SSA, 1970-2009
8
Capital Flows (% of GDP)
7
6
5
4
3
2
1
0
-1
Years
FDI, net inflows
ODA, net inflows
Portfolio equity, net inflows
Source: Author‟s estimation based on WDI and GDF (April 2011)
50
Migrant remittances
Stellenbosch University http://scholar.sun.ac.za
Clearly, external capital flows to SSA were generally low during the pre-reforms era. From the
reforms era in the 1980s and onwards, there have been somewhat consistent low upward
trends in external capital flows to the sub-region even though these trends, except for migrant
remittances, have been fluctuating widely. On the average, the lowest external capital inflow is
portfolio equity, whereas, consistently, ODA has been the highest inflow to the sub-region since
independence. As depicted in Figure 3.2, the sub-region has not been successful in attracting
external capital inflows in a consistent manner, except probably for migrant remittances.
Arguably, the apparent consistency in the inflows of migrant remittances could be attributed to
the continuously growing poverty in the sub-region and the development gap between SSA and
the industrialised world37. The persistently high rates of unemployment and underemployment,
features of developing countries, have led to these poor SSA countries being trapped in lowincome equilibrium, with no reversibility in sight. The ever-increasing income gap has become a
recipe for emigration and subsequent inflows of remittances in SSA. In this case, driven by
altruism, migrants from SSA are compelled to continue to remit home, mainly for consumption
purposes, for so long as the economic conditions at home do not improve38. Miotti et al. (2010:
17) observe that given the severity of poverty within the sub-region, SSA migrants, unlike
migrants from other countries in the developing world, are compelled to “send money for
current expenditures rather than for investment purposes”.
Under this circumstance, international migrants from the sub-region are compelled to remit to
their families back home for altruistic motives and not in response to successful implementation
of economic policies by governments. As evident from various survey studies from different
parts of the world, migrant remittances are mainly used for consumption (Tongamoa, 1987;
Loomis, 1990; Hayes, 1993; Clark, 2004; Miotti et al. 2010), and most migrants from poor
countries are under social obligation to remit home (Morauta, 1985; Tongamoa, 1987; Boyd,
1990; Brown and Poirine, 2005). It should, however, be emphasised that as to whether a
migrant will patronise the formal financial market in remitting home or use the unofficial money
transfer channels is largely dependent upon the degree of financial efficiency, an aspect of
financial development which incorporates the cost of financial service delivery including the
cost of international money transfers. Unofficial money transfer channels are private and often
37
It is important to emphasise that although emigration of skilled labour from poorer regions to the industrialised
world often leads to higher remittance inflows in migrant-home countries, it is only migrant-home countries with
efficient institutions and financial markets that are more likely to record relatively higher remittances through official
channels. Therefore, the fact that SSA is poorer than LAC and SAS, but the former receives less officially reported
remittances should not be seen as counter-intuitive.
38
See van Dalen et al. (2005) for empirical evidence from Egypt and Turkey.
51
Stellenbosch University http://scholar.sun.ac.za
unorganised money transfer channels to target recipients in developing countries. These
channels may include funds sent by the migrant when travelling home through fellow
international migrants such as friends, relatives or drivers travelling home, and a network of
private individuals in a system commonly called the hawala or hundi. Also, easy access for
migrants to offshore banking services and online banking and the availability of innovative
international financial products, which are all aspects of financial system development noted in
Chapter Two, can be instrumental in attracting remittances through the formal financial sector.
Figure 3.3 reconfirms the information in Figure 3.2 that, generally, there has been a positive
growth trend in remittances received in SSA and other developing economies, when measured
as a ratio of international migrant stock. This trend is also consistent with what has been
revealed in Figure A3.1 as well as Figure A3.2 in the Appendix. With the exception of ECA
which recorded a sharp drop in remittance flows between 1970 and 1990, virtually all the
developing regions, including SSA, witnessed a consistent positive growth trend in migrant
remittances per migrant (Figure 3.3).
Figure 3.3: Remittances Received per Migrant (US$) in Developing Economies, 1970-2009*
Average Migrant Remittances per Migrant (US$), 1970-2009
3000
Trends in Migrant Remittances per Migrant (US$), 1970-2009
12000
2,845.70
2500
10000
2,020.65
2000
1500
8000
1,173.25
1,108.30
1000
756.74
500
178.66
6000
0
EAP
ECA
Year
EAP
ECA
1970
-
-
LAC
LAC
MNA
MNA
9.42
77.39
37.02
SAS
SAS
7.33
SSA
SSA
1975
10.14
2,840.48
309.10
28.08
36.68
1980
567.96
1,941.07
343.39 1,511.63
338.24
108.31
1985
761.07
1,443.75
455.76
359.29
90.45
877.43
1990
1,115.64
107.44
887.98 1,250.01
350.31
127.20
1995
2,928.37
218.92
2,442.82 1,477.42
754.72
191.68
2000
3,923.40
367.20
3,588.30 1,375.69
1,337.33
294.02
2005
10,613.30
839.20
8,400.53 2,507.37
2,878.66
578.51
TOTAL 19,919.87
7,758.08
16,155.80 9,308.64
6,046.63
1,426.86
MEAN
1,108.30
2,020.65 1,173.25
756.74
178.66
2,845.70
4000
2.45
2000
0
1970
EAP
1975
1980
ECA
1985
LAC
1990
MNA
1995
2000
SAS
2005
SSA
Source: Author based on WDI and GDF (April 2011) *Computation based on 5-year data point intervals for which
data is available on total international migration stock reported in WDI by the World Bank.
52
Stellenbosch University http://scholar.sun.ac.za
Although SSA receives the least remittances per migrant, both in total amount and in terms of
growth since 1970, the sub-region, like many other developing economies, has witnessed a
positive growth trend since the post-reforms era. What is clear from Figure 3.3 above is that, on
the average, international migrants have been increasing the amount they send home over
time. This may be due to the ever-increasing income gap between the industrialised countries
where migrants are resident and the developing countries, where migrants are natives. Another
possible reason is that developing countries which are the main „exporters of migrants‟, have
not been able to improve upon the livelihood of their citizens, hence the compelling need for
migrants to keep increasing remittances in support of their families left behind to enable them
access basic human needs such as food, clothing and healthcare.
Between the years 2000 and 2005, there has been about a 100 per cent rise in remittances per
migrant to SSA and, indeed, to other developing economies. A possible explanation is that the
US and many advanced countries strengthened regulations and clampdowns on unofficial
international fund transfers following the September-11 Al Qaeda attacks (Gupta, 2005). Thus,
unlike in the past, migrants are now obliged to transfer funds home using official channels. It is
also likely that more migrants from SSA might now be more interested in returning home.
Several studies, including those of Merkle and Zimmermann (1992), Brown (1997), Gubert
(2002), and Cai (2003) found that migrant intention to return home (or future migration plans)
has a strong positive impact on the probability of remitting and the amount of funds transferred
by migrants. The magnitude and trends in external capital flows to SSA appear quite different
from those in other developing economies of the world, as shown in Figure A3.1 in the
Appendix. For instance, as shown in Figure A3.1, migrant remittances are either the highest (as
is the case for MNA and SAS) or the second-highest (as is the case for EAP, ECA and LAC)
external capital inflows, but in the case of SSA alone, migrant remittance inflows are only
slightly higher than portfolio equity inflows. Again, in SSA, ODA has remained consistently the
highest type of external capital inflows, but for all other developing economies, ODA has either
been the lowest (as is the case for EAP and LAC) or the second lowest (as is the case for ECA,
MNA and SAS). Since the inflows of FDI are generally driven by profit motives whereas ODA
are mainly linked to humanitarianism of the donor country or institution, the fact that SSA
constantly receives ODA as the highest form of external capital inflows is a signal that the subregion has not been able to implement the appropriate economic policies to pull quality external
resources to advance its sustainable growth and development.
53
Stellenbosch University http://scholar.sun.ac.za
It is also of interest to note that migrant remittances have been the most consistently growing
external capital flows to developing economies in the world whilst ODA and portfolio equity
have been the most volatile, as shown in Figure A3.1. In addition to its high volatility, generally,
ODA to developing economies has been declining. This is an indication that it will be prudent
for developing countries, particularly those in SSA, to put policy measures in place to facilitate
an improved mobilisation of non-aid investment-related external capital to finance their
development projects, as well as to enable them to address their numerous underdevelopment
problems on a permanent basis. Evidently, migrant remittances are the least sensitive to
shocks, given the high and relatively smooth pattern of inflows to SSA and other developing
economies. Another important observation is that, whereas there seems to be a somewhat
general positive correlation between FDI and migrant remittance inflows to developing
economies, in contrast, there seems to be a negative relationship between migrant remittance
inflows and ODA across the developing world. Thus, developing economies that attract higher
migrant remittances also attract higher FDI but relatively lower ODA, and vice versa. For
instance, while being the sub-region that receives the least remittances, SSA also receives the
least FDI but the most ODA, whereas the opposite scenario commonly holds for the other
developing economies (see Figure A3.1).
According to the altruistic theory, given the net income of a migrant, remittances should
negatively correlate with the income level of target recipients (Lucas and Stark, 1985; Rapoport
and Docquire, 2006). Indeed, some macro-level studies, notably those by Bougha-Hagbe
(2004), Cartagena (2004), Gupta (2005), Mishra (2005), and World Bank (2006b) as well as
Giuliano and Ruiz-Arranz (2009) for countries with less developed financial systems, conclude
that remittances are countercyclical in recipient countries. Contrary to this highly held view of
remittance counter-cyclicality, it has been revealed that migrant remittance inflows generally
correlate positively with real GDP per capita, growth in real per capita GDP, real GDP growth
and even real GDP per person employed in developing economies39 as shown in Adenutsi et
al. (2012). Other empirical studies which found that remittances are largely pro-cyclical in the
recipient countries include studies by the IMF (2005a), Lueth and Ruiz-Arranz (2007a), and
Sayan (2006).
39
This does not necessarily imply the popularly held view has been invalidated since correlation does not
necessarily mean causation.
54
Stellenbosch University http://scholar.sun.ac.za
Table 3.4:
Correlation between Remittances and Selected Macroeconomic Indicators, 1990-2009
EAP
ECA
LAC
MNA
SAS
SSA
MRPC MRem MRPC MRem MRPC MRem MRPC MRem MRPC MRem MRPC MRem
Current account balance (% of GDP)
0.727 0.721
0.008 0.007 0.721 0.717
0.152
0.219
0.079 0.071 0.451
0.457
FDI, net inflows (% of GDP)
-0.185 -0.189 0.932 0.931 0.280 0.269
0.874
0.899
0.946 0.944 0.653
0.653
GDP (constant 2000 US$)
0.986 0.984
0.881 0.881 0.961 0.962
0.840
0.917
0.971 0.968 0.954
0.956
GDP growth (annual %)
0.197 0.196
0.389 0.387 0.193 0.184
0.128
0.124
0.428 0.418 0.489
0.475
GDP per capita (constant 2000 US$)
0.985 0.983
0.868 0.868 0.940 0.944
0.855
0.927
0.886 0.879 0.953
0.955
GDP per capita growth (annual %)
0.296 0.294
0.384 0.381 0.287 0.278
0.231
0.249
0.408 0.399 0.512
0.498
GDP per person employed (constant 1990 PPP $) 0.141 0.145
0.859 0.860 0.842 0.847
0.864
0.905
0.975 0.973 0.960
0.964
Exports of goods (% of GDP)
0.889 0.884
0.438 0.437 0.874 0.864
-
-
0.951 0.949 0.850
0.846
Imports of goods (% of GDP)
0.760 0.755
0.479 0.478 0.805 0.794
0.206
0.071
0.954 0.956 0.740
0.729
Gross fixed capital formation (% of GDP)
0.801 0.800
0.285 0.288 0.068 0.081
0.442
0.458
0.922 0.922 0.540
0.542
Gross domestic savings (% of GDP)
0.940 0.940
-0.018 -0.017 0.738 0.736
0.813
0.848
0.785 0.782 0.867
0.867
HFCE (% GDP)
-0.985 -0.984 -0.037 -0.038 -0.856 -0.851 -0.805 -0.822 -0.859 -0.853 -0.927 -0.929
Inflation, GDP deflator (annual %)
-0.223 -0.220 -0.386 -0.385 -0.522 -0.512 0.096
-0.016 -0.295 -0.284 -0.247 -0.245
Portfolio equity, net inflows (% of GDP)
0.429 0.424
0.029
0.243 0.240 -0.171 -0.162 0.004
0.199 0.196 0.095
0.078
Source: Author‟s computation based on BoPS and WDI (April 2011). Notes: HFCE denotes household final
consumption expenditure. MRPC and MRem represent migrant remittances per capita and total migrant remittances
received respectively.
It can also be noted that migrant remittances are more strongly and positively correlated with
gross domestic savings and investment rather than what is popularly believed, namely that
remittances are purely for consumption purposes and driven by altruism. If, indeed, remittances
are spent on consumption in developing countries, they are more likely spent on imported
consumer goods as found by Tongamoa (1987) rather than on domestically produced goods.
This is because, as revealed in Table 3.4, there is a strong negative correlation between
migrant remittances and household final consumption expenditure (HFCE) as they correlated
positively and robustly with the import of goods in developing countries. This is one of the most
consistent results across all the developing economies in the world. One should, however, be
cautious in concluding that remittances are likely to be used for imports, since the correlation
between the former and exports is also strong and positive for the developing economies. What
seems clear from the foregoing is that migrant remittance inflows are likely to be highly
associated with economic openness, given the high positive correlation coefficients of imports
and exports as a proportion of GDP. However, although SSA has higher trade openness than
ECA, the latter receives more remittances. In addition, there is no basis to conclude that
remittances are a substitute for exports in developing countries.
It is also observed that remittance inflows are positively associated with improved current
balance in migrant exporting developing countries. The correlation between migrant
55
Stellenbosch University http://scholar.sun.ac.za
remittances received and portfolio equity is low and generally insignificant (with none attaining
0.45). Conversely, the correlation between FDI and remittances is positive and robust for
developing economies, except for EAP where the relationship is negative but low. Further
evidence on the negative correlation between remittances and domestically produced
consumables in developing countries can be ascribed to the strong positive relationship
between remittances and gross fixed capital formation (a proxy for domestic investment) as
well as remittances domestic savings. For SSA, the correlation between remittances and
savings is 0.87 whilst that between remittances and investment is 0.54. One intriguing result
obtained in Table 3.4 is that, in SSA, just as is the case in all other developing regions,
remittance inflows are negatively correlated with the rate of inflation. This implies either that
migrant remittances contribute to reducing inflation in recipient countries or that price stability is
a sine qua non for migrant remittance inflows to developing countries. Finally, taking the
correlation coefficients into account, it can be concluded that there is no significant difference
between the volume of migrant remittance inflows and migrant remittances received per capita,
in the context of macroeconomic implications.
With regard to the relationship between the volume of migrant remittances received and the
degree of financial development, it can generally be argued that there is positive correlation
between the two in SSA, when bank-based indicators of financial sector development are used
(see Figure 3.4). Although these indicators might not be the best measures of financial
development (see Kar and Pentecost, 2000; World Bank, 2005), these are the only indicators
for which SSA regional data is consistently available over the past two decades.
The amount of migrant remittances received in SSA positively correlates with the proportion of
bank credit to the private sector in GDP and broad money as a ratio of GDP, as is the case for
all the other developing regions except LAC. The positive correlation between remittances and
financial depth could imply that, in SSA, official remittance inflows are likely to be deposited at
banks. This reaffirms the strong positive correlation between remittances received and
domestic savings in developing countries, as shown in Adenutsi et al. (2012). There is,
however, no clear-cut pattern of the magnitude of the correlation between remittances and
financial depth, which could be due to the variations in the relative magnitude of money to
quasi money across the various regions under consideration. For instance, with a coefficient of
0.92, SSA and ECA are the regions with the most financial depth, followed by EAP (0.84) and
LAC (0.80). However, with regard to attracting migrant remittances, LAC and EAP are the
56
Stellenbosch University http://scholar.sun.ac.za
highest recipients whilst ECA and SSA received the least (see Figure A3.2 Panel A and Figure
A3.3 in the Appendix).
Figure 3.4:
Correlation between Migrant Remittances and Financial Development Indicators, 1990-2009*
1.2
1
Correlation coefficient
0.8
0.6
0.96
0.92
0.83 0.84
0.92
0.87
0.80
0.62
0.46
0.4
0.34
0.2
0.38
0.26
0.26
0.15
0.06
0
-0.2
EAP
ECA
LAC
MNA
SAS
SSA
-0.17
-0.4
-0.49
-0.6
-0.8
-0.79
-1
Domestic credit to private sector (% of GDP)
Source: Author based on WDI (April 2011).
Listed domestic companies, total
M₂ as % of GDP
*Period chosen based on consistent data availability across regions
A contrasting revelation is that, unlike in the case of other developing regions, the correlation
between migrant remittances received and the number of listed firms is negative for SSA (the
region that received the least remittances) and LAC (the region that received the most).
Probably, in SSA and LAC, some of the firms which were listed at the establishment of the
stock markets in the 1990s under-performed and were eventually delisted. Thus, the number of
listed firms has been declining over time amidst increasing remittance inflows throughout the
post-reforms era. Although, there is a higher correlation between remittances received and
bank-based financial sector development indicators in SSA than in LAC, MNA and SAS, each
of these regions receives higher remittances than SSA in absolute, per capita, and per migrant
terms (see Figure A3.2 Panels A and B, and Figure A3.3 in the Appendix). In fact, even as a
proportion to GDP, SSA only managed to occupy the third spot, behind MNA and SAS.
Likewise, ECA recorded the highest correlation between remittances and bank-based financial
sector indicators, and was the least recipient of remittances as a ratio of GDP and in terms of
actual volume of remittances received; this sub-region struggled to outperform only SSA.
3.3.2 The Dynamics of Remittances and the Macroeconomic Environment in SSA
From a macroeconomic perspective, there seems to be sufficient evidence for three stylised
facts regarding the flow of migrant remittances to developing countries: (i) geographically
57
Stellenbosch University http://scholar.sun.ac.za
smaller countries are more likely to attract relatively higher migrant remittances probably
because, as a result of their small size, there is stronger social cohesion and ties; (ii) small
island countries such as Cape Verde, Comoros, Mauritius and Seychelles tend to attract higher
official remittances, probably, because, unlike non-island and landlocked countries, it is difficult
to remit through unofficial remittance service providers such as bus drivers and traders; and,
(iii) although poorer countries are more likely to attract higher remittances because harsh
economic conditions at home (e.g. high unemployment, poor working conditions, low wages)
are a recipe for higher emigration40, remittance flows through official channels to these lowincome countries are not automatic, but dependent on some macroeconomic fundamentals.
For instance, as shown in Table 3.5, in terms of real GDP per capita, Seychelles, Mauritius,
Botswana, Swaziland and Cape Verde can be described as relatively rich within the sub-region.
Yet, in relative terms, these countries were the leading recipients of migrant remittances in SSA
after Lesotho between 1980 and 2009. Similarly, the remaining countries in the top 10
remittance-recipient category such as Senegal, Comoros, Lesotho, Gambia, Lesotho and
Sudan are by far less poor than Niger, Sierra Leone, Ghana, Rwanda, Ethiopia and Malawi,
classified among the 10 least remittance-recipients, as depicted in Table 3.5. Thus, considering
real GDP per capita, it is observed that the very poorest countries in the sub-region (Ethiopia,
Malawi, Niger and Sierra Leone) are among the countries that receive the least remittances, as
shown in Table 3.5. Guinea-Bissau, Liberia, Uganda and Togo, however, managed to defy the
odds and are among the leading 10 remittance-receiving countries within the sub-region in
recent years, when measured relative to GDP or population size (see Table 3.5).
40
There is a consensus in the migration and remittance literature that it is actual or perceived income gap (or
differences in quality of life) that underlie the South-North migration leading to southwards remittance flows (Beijer,
1970; Lipton 1980; Clarke and Wallsten, 2003; 2004; Kapur, 2004; Yang, 2005; de Haas, 2007; UN, 2010).
58
Stellenbosch University http://scholar.sun.ac.za
Table 3.5:
Comparative Analysis of Top-10 and Bottom-10 Migrant Remittance per capita Recipients in SSA
Top-10 Migrant Remittance-Dependent Countries in SSA, 1980-2009
Migrant remittances per capita (US$)
Bank liquid reserves to bank assets ratio (%)
Lesotho
C-Verde
202.042
183.365
Mauritius Swaziland Seychelles Botswana Senegal
97.235
81.747
58.405
44.198
0.08
0.31
0.04
0.07
0.16
0.57
Comoros
Gambia
19.116
18.972
18.254
75.62
1.000
-
0.22
0.80
-
0.22
-0.461
32.834
Sudan
Mean
Cor_
Broad money (% of GDP)
37.69
57.07
67.64
24.25
60.23
27.49
24.80
16.97
21.48
29.86
36.75
0.514
Domestic credit to private sector (% of GDP)
14.93
32.08
47.30
18.58
18.98
13.73
24.21
7.31
12.24
15.07
20.44
0.384
7.70
4.06
3.79
0.455
321.79 1,605.73
-0.031
FDI, net inflows (% of GDP)
4.41
1.14
3.90
9.32
3.44
0.98
2.28
0.64
GDP per capita (constant 2000 US$)
347.59 1,072.30
3,110.32
1,213.09
6,116.13
2,654.25
498.06
336.69
387.07
GDP PPP per capita (constant 2005 US$)
999.77 1,977.75
7,374.04
3,545.54
14,721.73
7,895.60 1,539.00
1,300.26
37.63
62.06
20.35
Goods exports (% of GDP)
26.90
5.92
Goods imports (% of GDP)
112.14
43.39
45.59
70.81
56.58
37.54
45.20
30.86
23.75
19.71
26.41
26.96
409.75 1,073.98
1,961.61
953.28
3,918.43
958.12
4.78
8.24
10.81
5.17
Gross fixed capital formation (% of GDP)
HFCE per capita (constant 2000 US$)
Inflation, consumer prices (annual %)
11.20
Gross domestic savings (% of GDP)
46.82
17.89
1,131.72 1,147.73 4,163.31
-0.091
8.35
10.08
29.16
26.52
-0.006
26.95
11.24
21.95
45.06
47.13
0.723
21.10
13.92
16.56
19.63
24.41
0.856
383.40
290.04
349.67
10.11
4.48
41.76
3.31
9.85
251.90 1,055.02
0.074
10.97
-0.226
-0.607
-44.40
-3.79
21.29
5.03
19.71
36.86
6.20
10.71
-6.81
7.56
5.24
Real deposit interest rate (%)
1.94
-2.68
1.84
-2.33
2.24
-0.86
0.40
-28.07
-0.14
3.23
-2.44
0.244
Real lending interest rate (%)
4.66
7.64
10.21
4.86
7.22
3.81
9.97
-
6.19
14.19
6.87
-0.332
45.69
22.31
18.61
25.45
28.04
22.26
15.25
6.29
-
18.20
20.21
0.720
Tax revenue (% of GDP)
Bottom-10 Migrant Remittance-Dependent Countries in SSA, 1980-2009
Mauritania
Migrant remittances per capita (US$)
Niger
S-Leone
Congo
Rwanda
Ghana
Ethiopia Madagascar Tanzania
Malawi
Mean
Cor_
2.500
2.364
2.245
2.014
1.688
1.653
0.904
0.694
0.179
0.087
1.43
9.40
18.49
40.62
19.96
13.19
26.49
25.09
16.61
8.92
28.37
20.71
0.078
Broad money (% of GDP)
14.13
14.20
16.52
16.02
14.85
18.70
28.79
18.43
20.45
17.60
17.97
-0.547
Domestic credit to private sector (% of GDP)
16.93
10.39
4.32
11.39
8.11
7.65
15.06
13.33
9.00
6.53
10.27
0.117
4.20
1.25
0.57
6.59
0.68
1.57
2.01
1.86
2.25
1.17
2.21
0.242
336.15
0.302
Bank liquid reserves to bank assets ratio (%)
FDI, net inflows (% of GDP)
GDP per capita (constant 2000 US$)
1.000
424.96
188.78
236.44
1,136.80
248.12
247.66
135.27
267.16
327.95
148.34
1,610.38
684.55
653.61
3,454.06
765.47
1,017.28
572.23
956.57
834.22
637.46 1,118.58
Goods exports (% of GDP)
37.75
16.82
12.79
59.29
5.53
21.20
4.96
13.67
10.46
Goods imports (% of GDP)
37.38
19.89
21.04
25.56
16.04
29.48
14.66
17.67
22.31
22.08
22.61
0.407
Gross fixed capital formation (% of GDP)
22.14
12.20
10.28
26.98
15.69
16.93
18.36
15.99
21.78
16.49
17.68
-0.107
HFCE per capita (constant 2000 US$)
341.37
137.35
151.74
299.03
187.83
218.87
110.45
226.60
240.46
112.71
202.64
0.327
Inflation, consumer prices (annual %)
6.51
3.67
11.91
4.64
7.06
31.45
7.85
15.47
19.99
20.83
12.94
-0.513
0.048
GDP PPP per capita (constant 2005 US$)
22.51
20.50
0.338
0.375
Gross domestic savings (% of GDP)
2.14
5.22
3.43
36.34
0.65
6.05
8.58
5.44
8.30
7.18
8.33
Real deposit interest rate (%)
0.37
1.21
-23.81
4.84
3.87
-14.18
-2.60
0.61
-13.17
-3.47
-4.63
0.066
Real lending interest rate (%)
8.56
11.68
-1.113
10.49
8.22
-16.46
2.11
14.54
3.78
7.93
4.97
-0.050
-
10.71
9.83
9.26
9.03
15.22
8.70
10.54
-
-
7.33
0.066
Tax revenue (% of GDP)
Source: Author‟s estimation mainly based on MRF-2011, BoPS, WDI, and GDF (April 2011). Note: Averages were
computed for each country for only the years from 1980-2009 for which data was available. .
Broadly, the top-10 remittance-receiving SSA countries are those with relatively higher tax
revenue, exports and imports of goods, FDI, investment ratio, financial deepening (M2/GDP),
bank credit to private sector ratio and real interest rate, but lower rate of inflation and domestic
savings. These leading remittance-recipients in per capita terms also have relatively higher
household consumption expenditure, real deposit interest rate, real GDP per capita, and real
GDP per capita at purchasing power parity (PPP). However, taking a cursory look at countryspecific features, it is observed that, with the exception of Sudan, the top-10 remittancerecipients in per capita terms are those with a stronger fiscal policy stance in terms of tax
59
Stellenbosch University http://scholar.sun.ac.za
revenue mobilisation. Furthermore, countries which are successful in attracting higher
remittances in per capita terms are those with more developed financial systems when proxied
by M2/GDP and credit to the private sector (Table 3.5). This is also true for countries that lead
in attracting remittances in actual volumes and, even in terms of GDP (see Figure A3.4 in the
Appendix). It should also be noted that Ethiopia, Malawi, Niger, and Sierra Leone can be
described as the very poorest in the sub-region in terms of real per capita GDP (Table 3.5).
These countries also attracted very low migrant remittances as well as FDI during the postreforms era, 1980-2009 (Table 3.5). It can be argued that the success in implementing sound
macroeconomic policies, to a reasonable extent, may be necessary to attract international
migrant remittances through official channels. Alternatively, remittances could be important in
determining the success of macroeconomic policy implementation in remittance-recipients in
SSA.
For both top-10 and bottom-10 SSA remittance-recipients in Table 3.5, migrant remittances per
capita positively correlates with private sector credit, FDI, goods imports, tax revenue, real
deposit interest rate and HFCE, whilst for the rate of inflation and real lending interest rate, the
correlation is negative. With the exception of HFCE and inflation, in each of these cases, the
correlation is stronger for the top-10 than the for bottom-10 recipient countries. The
conspicuous differences, however, are that, whereas migrant remittances per capita positively
correlates with real GDP per capita, regarding real GDP PPP per capita, export of goods, gross
domestic savings, and bank liquid reserve to bank assets ratio among the lowest 10
remittance-recipients, the correlation is negative for the top 10 remittance-recipients (Table
3.5). Also, whereas there is a fairly strong positive correlation between remittances per capita,
and gross fixed capital formation and broad money to GDP ratio in the top 10 remittancerecipient countries, in the case of the bottom 10 remittance-recipients, the respective
correlation coefficients are negative.
With a coefficient in excess of 99 per cent (see Figure A3.3), there is a near perfect positive
correlation between migrant remittances received per capita and remittances per migrant
received in SSA and, indeed, for all other developing regions excluding ECA. For these other
developing regions, the correlation coefficients range between 97 per cent for MNA and 100
per cent for EAP and LAC. In the case of ECA, a correlation coefficient of -37 per cent (see
Figure A3.3) signifies a relatively low negative relationship between migrant remittances per
capita and remittances per migrant. Consequently, for the entire developing world, remittances
60
Stellenbosch University http://scholar.sun.ac.za
per capita and remittances per migrant are strongly positively correlated. Therefore, in the
absence of regular annual data on migrant stock, generally, migrant remittances per capita
should be seen as a perfect proxy for remittances per migrant because it is statistically evident
that the evolution of remittances per capita could be used to proxy the evolution of remittances
per migrant. In other words, Figure 3.5 is seen as providing good insight into the dynamics of
remittances sent by SSA migrants to their native countries over the past three decades, 19802009.
SSA as a sub-region has remained the least recipient of migrant remittances in the world, when
measured in actual volume of inflows or as a ratio of population or migrant (Figure A3.2).
Comparing SSA to other developing regions, it is quite apparent that the rate of growth in total
migrant remittance inflows as well as migrant remittances received by the sub-region relative to
population size or migrant stock is relatively slow (Figure A3.2 Panels A2 and B). On the basis
of individual SSA countries, however, (Table A3.3), it is evident that the trend in migrant
remittances received by the sub-region increased steadily and fairly robustly during the postreforms era as shown in Figure 3.5.
For the individual countries within the sub-region, the number of SSA countries that received an
annual minimum of US$1 in migrant remittances per capita increased from 25 in the 1980s to
31 in the 1990s and to 32 in the 2000s (Figure 3.5). In the 2000s, none of the 32 SSA countries
referred to above received less than US$2 in migrant remittances per capita (Figure 3.5). For
the period 1980-2009, 32 of the sampled 36 countries received at least US$1 in migrant
remittances per capita on annual basis. In per capita terms, Lesotho (US$202.04), Cape Verde
(US$183.36), Mauritius (US$97.24), Swaziland (US$81.75), Seychelles (US$58.40), Botswana
(US$44.20), and Senegal (US$32.83) are SSA countries that consistently received the highest
inflows of migrant remittances between 1980 and 2009 (Figure 3.5). Nevertheless, during the
most recent decade, (2000-2009), Cape Verde (US$249.28) and Mauritius (US$170.62)
displaced Lesotho (US$164.94) as the traditional leading recipient of migrant remittances in per
capita terms (Figure 3.5). Burkina Faso and Benin which ranked among the top 10 recipients of
remittances per capita during the 1980s and 1990s were dislodged by Gambia and Togo in the
2000s (Figure 3.5).
61
Stellenbosch University http://scholar.sun.ac.za
Figure 3.5: Migrant Remittances Received per capita by SSA Countries (in US$), 1980-2009
Tanzania
Sierra Leone
Malawi
Nigeria
Ghana
Gabon
Ethiopia
Madagascar
Rwanda
Guinea
Congo
Guinea Bissau
Niger
South Africa
Cameroon
Mauritania
Uganda
Côte d'Ivoire
Kenya
Togo
Mozambique
ST&P
Mali
Namibia
Comoros
Sudan
Gambia
Senegal
Benin
Mauritius
Burkina Faso
Seychelles
Botswana
Swaziland
Cape Verde
Lesotho
0.017
1980-1989
0.020
0.125
0.137
0.200
0.206
0.273
0.346
0.664
0.849
0.947
1.271
1.639
2.064
2.087
2.127
2.212
3.167
3.289
4.045
4.654
7.425
7.474
9.283
10.250
10.631
11.185
13.204
14.117
15.306
18.905
46.985
48.354
83.357
98.951
219.185
-
50.000
164.942
170.621
249.275
100.000
200.000
201.868
222.000
300.000
50.000
100.000 150.000 200.000 250.000
1980-2009
Malawi
Tanzania
Madagascar
Ethiopia
Ghana
Rwanda
Congo
Sierra Leone
Niger
Mauritania
Cameroon
Guinea
Gabon
Mozambique
Côte d'Ivoire
South Africa
ST&P
Guinea Bissau
Uganda
Namibia
Burkina Faso
Mali
Nigeria
Kenya
Togo
Benin
Gambia
Comoros
Sudan
Senegal
Botswana
Seychelles
Swaziland
Mauritius
Cape Verde
Lesotho
0.068
0.381
0.679
0.781
2.113
3.112
3.436
3.506
3.759
4.028
4.282
4.960
5.213
5.936
6.595
7.100
8.083
9.620
12.314
14.768
17.994
18.104
19.097
20.100
26.205
30.560
30.811
30.995
38.152
41.119
67.166
68.929
89.094
-
0.067
1990-1999
0.138
0.326
0.898
0.954
1.000
1.288
1.425
1.445
1.589
1.897
2.432
3.437
3.682
4.064
4.110
4.694
4.796
6.682
6.708
6.982
8.426
8.566
8.595
9.710
10.963
12.583
17.795
18.131
26.567
39.135
43.120
92.953
105.779
-
100.000 150.000 200.000 250.000
2000-2009
Malawi
Tanzania
Mauritania
Madagascar
Ethiopia
Rwanda
Mozambique
Congo
Ghana
Niger
Sierra_Leone
Burkina Faso
Cameroon
Gabon
Namibia
Guinea
Côte d'Ivoire
ST&P
South_Africa
Uganda
Mali
Guinea Bissau
Benin
Comoros
Kenya
Nigeria
Togo
Gambia
Sudan
Botswana
Senegal
Swaziland
Seychelles
Lesotho
Mauritius
Cape Verde
Malawi
Tanzania
Ethiopia
Guinea
Madagascar
Ghana
Rwanda
Niger
Cameroon
Congo
Guinea Bissau
Sierra Leone
Gabon
Mozambique
South Africa
ST&P
Mauritania
Togo
Uganda
Côte d'Ivoire
Nigeria
Kenya
Sudan
Namibia
Burkina Faso
Mali
Gambia
Benin
Senegal
Comoros
Seychelles
Botswana
Swaziland
Mauritius
Cape Verde
Lesotho
0.087
0.179
0.694
0.904
1.653
1.688
2.014
2.245
2.364
2.500
2.915
2.949
3.193
3.924
5.986
6.147
7.051
7.091
7.887
8.157
11.192
12.144
12.560
12.640
13.217
17.003
18.254
18.972
19.116
32.834
44.198
58.405
81.747
97.235
183.365
202.042
-
50.000
100.000 150.000 200.000 250.000
Source: Author based on MRF-2011, BoPS, WDI, and GDF (April 2011) and estimates from country-specific desks
of IMF and WB. Note: Only the 36 sampled countries are included due to data constraint.
62
Stellenbosch University http://scholar.sun.ac.za
A similar trend is discernible when the number of SSA countries that received at least one per
cent of remittances as a ratio of GDP is considered. From 15 countries in the 1980s, the
number of countries rose to 18 in the 1990s before reaching 22 in the 2000s, based on the 36
sampled countries for which consistent data are available over the past three decades41 (Figure
A3.4). For the overall period, more than 50 per cent of the sampled countries, specifically 19 of
the 36 sampled countries, received migrant remittances worth, at least, one per cent of their
GDP (Figure A3.4). Furthermore, as shown in Figure A3.5, although only 17 SSA countries
received an annual average of at least US$12 million, representing a minimum monthly
average of US$1 million in the 1980s, as many as 25 and 29 out of the 36 sampled SSA
countries received this minimum amount of remittances in the 1990s and 2000s respectively.
Overall, more than two-thirds, (specifically 25) of the sampled countries, received migrant
remittances representing not less than one per cent of GDP between 1980 and 2009 (Figure
A3.4). Additionally, on the average, the amount of international migrant remittances received by
each of the sampled SSA countries has been increasing over time, whether in absolute or
relative terms (Table A3.3).
Whereas Botswana, just like Burkina Faso, Mozambique, and Namibia, recorded a consistent
decline in remittances per capita over the past three decades (Table A3.3), Nigeria defied the
odds as the only country ranked among the bottom 10 recipients in the 1980s to occupy a
position among the top 10 in the 2000s (Figure 3.5). By this feat, Nigeria has not only managed
effectively to escape from the bottom 10 in per capita terms, but to progress from the 19th in
the 1980s to the first position since the 1990s as the largest recipient of actual volume of
remittances received (Figure A3.5). The situation at the opposite end of the migrant remittance
per capita ladder can be described as less competitive as six countries (Malawi, Tanzania,
Madagascar, Ethiopia, Rwanda and Ghana) never moved out of the bottom 10 category
throughout the past three decades (see Figure 3.5).
Regarding migrant remittances as a percentage of GDP, just as in terms of remittances per
capita, Lesotho (54.52 per cent), Cape Verde (15.98 per cent) and Swaziland (6.93 per cent)
maintained the top-three positions in the SSA for the period, 1980-2009, although Gambia
41
Despite this encouraging trend, however, the overall average of remittance/GDP percentage for the 36 sample
countries declined consistently between 1980 and 2009 due to the consistent fall in remittance/GDP ratios in some
leading recipients notably Botswana, Burkina Faso, Lesotho, and Swaziland. This might be due to a higher rate of
growth in GDP relative to migrant remittances as with the exception Burkina Faso, none of these countries recorded
consistent decline in actual volume of remittances received during the period under review (see Table A3.3).
63
Stellenbosch University http://scholar.sun.ac.za
dislodged Swaziland from the third spot in the 2000s. Here again, seven countries (Malawi,
Tanzania, Gabon, South Africa, Congo Republic, Madagascar, and Ghana) firmly remained
within the bottom 10 remittance-recipients relative to GDP over the past three decades, 19802009. However, Nigeria, which was ranked among the least 10 recipients, moved from the 33rd
position with a remittances/GDP per cent of 0.03 in the 1980s to the 15th with a remarkable
3.42 per cent in the 2000s (Figure A3.4). Other countries that made significant and consistent
progress on the migrant remittance-GDP ladder are Guinea-Bissau which moved from 19th
position with 0.77 per cent to 6th with 5.28 per cent, Kenya (17th, 0.91 per cent) to (9th, 4.58 per
cent), Senegal (10th, 2.29 per cent) to (5th, 8.43 per cent), and Togo (14th, 1.30 per cent) to (4th,
8.52 per cent) from the 1980s to the 2000s. In contrast, Botswana (6th, 4.05 per cent) to (26th,
0.75 per cent); Burkina Faso (4th, 7.39 per cent) to (19th, 1.42 per cent); and Swaziland (3rd,
10.82 per cent) to (12th, 3.61 per cent) experienced the most significant and consistent
retrogression on the remittance-GDP ladder (Figure A3.5).
As far as actual volume of migrant remittances received is concerned, based on 1980-2009
average in millions of US dollars, Nigeria (1,758.48), Sudan (675.97), Kenya (418.95), Lesotho
(341.67), Senegal (339.28), South Africa (269.19), Uganda (201.06), Mali (126.52), Mauritius
(115.33), and Benin (103.05) are the largest recipients in SSA. At the opposite end of this same
ladder, Malawi (0.84), São Tomé and Príncipe (0.92), Gabon (4.03), Seychelles (4.50),
Mauritania (5.22), Congo Republic (6.09), Tanzania (6.54), Comoros (9.57), Madagascar
(9.72), and Guinea-Bissau (10.18) each receiving an average migrant remittances of less than
US$100 million per annum, are the least recipients between 1980 and 2009. Thus, in actual
volume, Nigeria is the largest migrant remittance recipient in the sub-region with South Africa
being the 6th largest recipient, yet Nigeria is ranked 14th(17th) and South Africa as 21st(33rd) in
relative terms of population and GDP respectively. Equivalently, although Gambia, Comoros
and Seychelles are ranked 23rd, 29th and 33rd largest recipients of remittances in absolute
terms, these countries are ranked 10th, 9th and 5th respectively in per capita terms with Gambia
and Comoros occupying the 4th and 8th positions on the remittance-GDP ladder. This implies
that large and populous countries such as Nigeria and South Africa are likely to receive more
migrant remittances in absolute terms whilst smaller and less populous countries like Comoros
and Seychelles are more likely to be counted among the high remittance-recipients in relative
terms. Notwithstanding this observation, there are reasons to believe that there may be certain
conditions, policies and strategies that might be essential to attracting migrant remittances, as
some geographically small countries with relatively low population size are among the largest
64
Stellenbosch University http://scholar.sun.ac.za
remittance recipients in absolute terms. These countries include Lesotho (4th), Senegal (5th),
Mauritius (9th), Benin (10th), Cape Verde (13th), Swaziland (14th) and Togo (15th). In contrast,
some geographically large countries such as Mauritania (32nd), Tanzania (30th) and
Madagascar (28th), with relatively large population size, attracted relatively low migrant
remittances over the past 30 years (Figure A3.5).
For the 36 sampled SSA countries, migrant remittances received in absolute terms increased
steadily over the past three decades. From a low of US$40.24 million in the 1980s, migrant
remittances increased by over 100 per cent to US$87.36 million and by more than 350 per cent
to reach US$307.23 million in the 1990s and 2000s respectively (Table A3.3). Consistent with
this increasing trend, the 36 sampled SSA countries, migrant remittances per capita also
witnessed a steady rise since the 1980s. Between 1980 and 1989, the average migrant
remittances received by these sampled countries were US$17.92 per annum. This figure
increased to US$24.58 in the 1990s and to a further US$32.89 in the 2000s. This increasing
trend might be due to the fact that the growth in migrant remittances received was faster than
the population growth rate of the sub-region during the period under review. Another possible
reason attributable to this consistent positive growth trend in officially reported remittances
received in SSA is the pursuit of increasing financial liberalisation policies by the sampled SSA
countries over the past three decades.
Another important point worth noting is the fact that leading migrant remittance-recipient
countries (in actual volumes) such as Nigeria, Sudan, Kenya, South Africa, and Uganda, are
also countries with relatively very high GDP, an explanation for the reason why none of these
countries is listed among the top 10 migrant remittance-recipient countries when measured as
a percentage of GDP (see Figure A3.4). For example, out of the US$156,536.76 million migrant
remittances received by the 36 sampled SSA countries between 1980 and 2009, Nigeria, the
highest recipient with a total of US$52,754.25 million controlled more than one-third, specifically
33.70 per cent whilst the second-highest recipient, Sudan, received US$20,279.10 million in
total (accounting for 12.95 per cent). Thus, Sudan, together with Nigeria, received nearly half,
specifically 46.66 per cent of the total amount received by the sampled 36 SSA countries.
Kenya, the third-highest recipient, received a total of US$12,568.59 million, representing 8.03
per cent of the entire 36 sampled countries; hence, together with Nigeria and Sudan, these
three countries alone received more than 50 per cent, specifically, 54.68 per cent of the total
amount received by the group of 36 sampled countries. Therefore, as these three countries are
65
Stellenbosch University http://scholar.sun.ac.za
not ranked among the top 10 migrant remittance-GDP recipients in the sub-region, given the
relatively large size of the respective economies, the sample average of the remittance-GDP
was depressingly affected as manifested in Figure A3.4.
3.4 THE STYLISED FACTS OF MIGRANT REMITTANCE FLOWS TO SSA
Based on the evidence from the above expositions, the under-listed are presented as
constituting the stylised facts of international migrant remittance flows to SSA:
i.
There has been a fairly strong positive trend in international migrant remittance inflows
in SSA since the implementation of financial liberalisation in the 1980s.
ii.
SSA is the least recipient of migrant remittances when measured in terms of absolute
values and relative to population size and international migration stock.
iii.
In terms of remittance inflows as a percentage of GDP, SSA is the third highest
recipient after SAS and MNA in recent years. Although in absolute volumes and in
terms of population size and international migrant stock, SSA has consistently been the
least recipient of migrant remittances, yet the sub-region emerged as the third highest
recipient of remittances as a percentage of GDP, this goes to show that in relative
terms, SSA has been recording a lower rate of GDP growth than the rate of growth in
migrant remittance inflows, in contrast to what pertains in other developing economies.
iv.
The correlation between migrant remittance inflows per capita and (migrant) remittances
per migrant in SSA is positive and more than 99 per cent. Accordingly, remittances per
capita can be an excellent proxy for remittances per migrant in SSA when the
underlying evolution of each of these measures is taken into account.
v.
SSA is the only sub-region in the world today that receives more ODA than migrant
remittances.
vi.
As revealed by the changing trend in remittance inflows as a percentage of GDP in
Figure A3.1, SSA is the sub-region with the most sluggish but resilient growth rate in
remittance inflows.
66
Stellenbosch University http://scholar.sun.ac.za
vii.
Of all external capital flows to SSA, migrant remittances are the least volatile as is the
case in all other developing economies (see Figure A3.1).
viii.
Officially reported migrant remittances to SSA, in both absolute and relative terms,
stagnated throughout the pre-reforms era but with the inception of the reforms and,
especially, during the post-reforms era, migrant remittance inflows to SSA have been
growing at a faster rate.
ix.
On the average, richer SSA countries (when measured in terms of real per capita GDP)
are the recipients of higher official migrant remittances per capita. However, the
correlation between remittances and GDP per capita varies between the top 10
remittance-recipients (negative but approximately zero) and the bottom 10 remittancerecipients (positive).
x.
In SSA, countries that lead in attracting higher migrant remittances (when measured in
relative terms) also lead in attracting higher FDI as a ratio of GDP; and these are
countries with a higher real GDP per capita, investment/GDP ratio and a lower rate of
inflation. This again points to the fact that macroeconomic performance and migrant
remittance inflows are positively related within the sub-region.
xi.
Fiscal policy effectiveness seems to be crucial to attracting migrant remittance inflows
as tax revenue/GDP ratio positively correlates with migrant remittances received with
stronger correlation for the top 10 remittance recipients, which also have a higher tax
revenue/GDP ratio.
xii.
On the average, higher official migrant remittances are received in SSA countries with
relatively more developed bank-based financial market indicators compared to other
countries within the sub-region with relatively underdeveloped financial markets (Table
3.5). At this point, however, it may not be absolutely correct to conclude that financial
development directly impacts on official remittance inflows and vice versa since, in the
compilation of the remittance data, non-bank remittance service providers particularly
MTOs, which are the main agents in most SSA countries and other developing
countries, are recognised as formal money transfer service providers.
67
Stellenbosch University http://scholar.sun.ac.za
3.5 REMITTANCES AND MACROECONOMIC POLICY IMPERATIVES IN SSA
There is the need for economic policy re-orientation directed at attracting higher remittances
from abroad rather than over-relying on other volatile and unpredictable external capital
especially ODA, the inflows of which are largely exogenous to domestic economic policy design
in recipient economies. Without doubt, the under-listed policy initiatives could be relevant to
attracting international migrant remittances to SSA:
i.
Remittances received in excess of present consumption could be used for investment
purposes rather than spent on imported consumer goods. This will result in accelerated
economic growth since investment is an injection whereas imports are leakages.
ii.
Governments of SSA countries could establish special international relations with
foreign industrialised countries recognised as the main destinations of their migrants, so
that through an agreed framework (similar to what pertains under double taxation
agreements among nations), migrants could remit home without paying transfer fees
and any other charges more than once. For instance, it should be possible for a migrant
working abroad to remit home regularly towards the payment of his/her social security
and pension funds without paying fees for this purpose.
iii.
Implementation of macroeconomic policies aimed at stabilising the domestic prices and
currency in SSA to motivate migrants to remit home more regularly. With stabilised
domestic price and currency in SSA countries, it becomes easier for migrants to plan,
predict and regularise the amount of money to remit home.
iv.
Financial institutions could develop innovative financial products and incentive
packages aimed at enticing migrants abroad to remit home using approved routes more
regularly and conveniently at reduced cost. This can only be done sustainably when the
domestic financial market is open to competition and integrated at the domestic and the
international levels.
v.
Domestic banks could either directly go off-shore and open more branches in major
migrant „host‟ countries or collaborate with foreign banks in these migrant „host‟
countries so as to strategically increase banking convenience and access, thereby
motivating SSA migrants to remit home more regularly and at reduced costs using
68
Stellenbosch University http://scholar.sun.ac.za
official channels.
vi.
A stable macroeconomic environment with some consistency in positive real growth is a
signal of good economic fortunes in the future. Migrants who were compelled to seek
greener pastures abroad because they lost hope in the domestic economy are likely to
reconsider returning home in the future if there are better economic prospects. Such
migrants are less likely to spend significant proportions of their earnings in their host
countries, but rather remit more funds to their native countries towards investment
projects such as financing entrepreneurial ventures. Under this circumstance, improved
macroeconomic management could serve as a catalyst for receiving higher remittances
through official routes.
vii.
It is crucial for SSA to develop the appropriate policy framework for attracting
remittances through the formal transfer channels since continuous and colossal inflows
of foreign currencies through unapproved routes, which represent additional currencies
outside the banking system, could endanger currency stability and pose challenges for
effective macroeconomic management in the long run. This could have adverse effects
on economic growth and development prospects of the sub-region in the long run.
3.6 CHAPTER SUMMARY AND CONCLUSIONS
From a macroeconomic viewpoint, it is found that, generally, not much has been achieved by
the sub-region in terms of real per capita income growth, investment and resource mobilisation,
although some marginal gains have been made in recent years. Indeed, except in terms of
financial market development and international trade, there is no strong evidence that the
macroeconomic conditions of SSA have improved since the implementation of economic reform
policies in the 1980s. Thus, by and large, the unfavourable structural features of SSA which
existed at the time of independence are still prevalent today and there is no basis for any
strong argument that the standard of living today in SSA is an improvement on what prevailed
at the time of political independence in the 1960s.
Overall, FDI has remained the highest external capital inflows in developing economies, but
faces a strong challenge from migrant remittances which have been growing more consistently
in all developing economies. In other developing economies where FDI is not the leading
69
Stellenbosch University http://scholar.sun.ac.za
source of external capital inflow, migrant remittances have overtaken FDI in recent years.
However, in SSA alone, the leading capital inflow has been ODA with remittances being the
smallest in relative per capita and per migrant terms as well as in absolute volume. The
relatively high performance of SSA by emerging as the third-highest recipient of migrant
remittances in the world, when measured as a percentage to GDP, can be described as ironical
and misleading. This might be due to the fact that SSA, as a sub-region, has witnessed a
relatively slower rate of GDP growth than other developing economies like LAC and SAS during
the period under consideration. Portfolio equity flows have remained the least form of capital
inflows to developing economies as a whole. Across developing economies, portfolio equity,
FDI and ODA inflows are highly volatile but, whereas FDI exhibits an upward trend, ODA and
portfolio equity have been exhibiting a negative or a stagnated trend in recent years. This
makes migrant remittances the least volatile form of external capital in SSA in particular and in
other developing economies as a whole.
It has been observed that migrant remittances received by SSA as a sub-region have been
rising in both relative and absolute terms, but SSA still remains the least recipient of migrant
remittances with the lowest rate of growth. Besides, the dependency of the sub-region on
migrant remittances received is still very low with only five countries (Lesotho, Cape Verde,
Swaziland and Gambia) receiving more than five per cent of GDP. About 50 per cent of SSA
countries (17 out of the 36 sampled) received less than one per cent of GDP in migrant
remittances between 1980 and 2009 with a general improvement from 21 countries in the1980s
to 14 countries in the 2000s. It is also quite encouraging to observe that the number of SSA
countries receiving at least five per cent migrant remittances relative to GDP increased from
four in the 1980s and five in the 1990s to seven in the 2000s. Even though the growth trend in
migrant remittances per capita in SSA has been positive throughout the past 30 years, no SSA
country earns up to US$1 a day. This is evident from the fact that even the highest migrant
remittance per capita recipients, Lesotho (US$219.19 in the 1980s and US$222.00 in the
1990s), and Cape Verde (US$249.88 in the 2000s), received less than US$365.25 per annum,
an equivalent of US$1 per day.
Perhaps, the most fascinating conclusion that can be drawn from this chapter is that, five
countries, viz. Nigeria, Kenya, South Africa, Uganda, and Mauritius, which are classified by the
IMF as having emerging or frontier financial markets, have „coincidentally‟ dominated the list of
the top 10 migrant remittance recipients (in actual amounts through official channels) in recent
70
Stellenbosch University http://scholar.sun.ac.za
years. Even relative to population and GDP, the 13 SSA countries with emerging and frontier
financial markets still dominate as the leading recipients of official remittance. For instance, six
and five of these 13 SSA countries are listed among the top 10 recipients of remittances as a
ratio of population and GDP respectively. Virtually, all the countries ranked as least remittances
recipients are those with highly underdeveloped financial markets. The role of the financial
sector in attracting migrant remittances into the sub-region could be explored from various
facets such as analysing the sub-region as a bloc or a comparative analyses between various
cohorts in relation to financial market environment, together with other unique homogenous
macroeconomic policy environment, which have been varying over time since the adoption of
financial liberalisation programme in the 1980s.
71
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 3
Table A3.1: HIPC Status and Date of Political Independence of SSA Countries*
Country
Angola
+
Benin
Botswana
+
Burkina Faso
+
Burundi
+
Cameroon
Cape Verde
+
Central African Republic
+
Chad
+
Comoros
+
Congo, DR
+
Congo, Republic
+
Côte d'Ivoire
Equatorial Guinea
+
Eritrea
+
Ethiopia
Gabon
+
Gambia, The
+
Ghana
+
Guinea
+
Guinea-Bissau
Kenya
Lesotho
+
Liberia
Date of Independence
Country
November 11, 1975
August 1, 1960
September 30, 1966
August 5, 1960
July 1, 1962
January 1, 1960
July 5, 1975
August 13, 1960
August 11, 1960
July 6, 1975
June 30, 1960
August 15, 1960
August 7, 1960
October 12, 1968
May 24, 1993
4th Century, BC
August 17, 1960
February 18, 1965
March 6, 1957
October 2, 1958
September 24, 1973
December 12, 1963
October 4, 1966
July 26, 1847
Madagascar
+
Malawi
+
Mali
+
Mauritania
Mauritius
Mayotte
+
Mozambique
Namibia
+
Niger
Nigeria
+
Rwanda
+
São Tomé & Príncipe
+
Senegal
Seychelles
+
Sierra Leone
+
Somalia
South Africa
+
Sudan
Swaziland
+
Tanzania
+
Togo
+
Uganda
+
Zambia
Zimbabwe
Date of Independence
+
June 26, 1960
July 6, 1964
June 20, 1960
November 28, 1960
March 12, 1968
Territorial collectivity of France
June 25, 1975
March 12, 1990
August 3, 1960
October 1, 1960
July 1, 1962
July 12, 1975
April 4, 1960
June 29, 1976
April 27, 1961
July 1, 1960
May 31, 1961
January 1, 1956
September 6, 1968
December 9, 1961
April 27, 1960
October 9, 1962
October 24, 1964
April 18, 1980
Source: http:en.wikipidia.org/wiki/Decolonization_of_Africa. (Date posted/accessed: unknown/June 28, 2010).
+
HIPC countries (33 out of 48 countries as at June 30, 2010). See http://go.worldbank.org/4IMVXTQ090.
*Republic of South Sudan which was not part of the population from which the sampled was selected attained its
political autonomy and independence on July 09, 2011.
72
Stellenbosch University http://scholar.sun.ac.za
Table A3.2: Summary of Major Economic Policies Pursued in SSA since Post-Independence, 1960-2009
I: Inward-Looking Socialist Economic Policy (l960-1979)
Compelling Issues
Key Policy Objectives
Key Policy Instruments
Number/ List of Countries
Policy inherited from Provide
critical
social Government
spending All independent states
colonial masters
infrastructure
(infrastructural-based
of the sub-region as at
expansionary fiscal)
1979
Underdeveloped private Provide basic essential
sector
needs of life and simple Exchange rate (pegging /
farm
inputs
through
fixed regime)
Lack of adequate private
subsidies
capital and absence of
Money supply
entrepreneurial class
Promote
import Credit control
substitution
Openness
and
industrialisation
excessive dependence
on imports and external Create jobs especially
factor inputs
within the public sector
Excessive dependency Provide finance for key
on exports of primary
sectors of the economy
products with limited
capacity
for
export
expansion
II: Economic Recovery Programme / Structural Adjustment Programme (ERP/SAP) (1980-1989)
Compelling Issues
Key Policy Objectives
Key Policy Instruments
Number/ List of Countries
Huge and unsustainable Reduce the size of public Money supply and public Benin, Burkina Faso,
deficits
in
current
sector and improve upon
sector credit controls
Burundi, Central African
accounts of BoP
its management
Rep, Comoros, Congo
Fiscal discipline to reduce
Republic, Congo DR,
Imbalances
between Eliminate price distortions
government spending and
Côte d‟Ivoire, Equatorial
government
revenue Promote
deficit finance
economic
Guinea,
Ethiopia,
and
expenditure
liberalisation
with Privatisation of SOEs
Gambia,
Ghana,
resulting in huge deficits
emphasis on trade and the Exchange rate reforms
Guinea, Guinea-Bissau,
being financed through
financial sector
and liberalisation
Kenya,
Liberia,
printing of money
Promote deregulation and Interest rate reforms and
Madagascar,
Malawi,
Financial repression
price
mechanism
to
liberalisation
Mali,
Mauritania,
minimise the role of the Deregulation
of
credit
Mauritius,
Niger,
state in resource allocation
control
Nigeria,
Senegal,
Sierra
Promote domestic savings
Leone, Somalia, Sudan,
and investment in the
Tanzania,
Togo,
public and private sectors
Zambia, Zimbabwe
Increase tax revenue by
broadening the tax base
Increase the efficiency of
73
Main Policy Outcome
Excessive protection of State-Owned Enterprises
(SOEs) which resulted in losses arising from
production and managerial inefficiencies.
Excessive government spending resulting in high
inflation and depletion of international reserves
Fixed exchange rate regime led to currency
overvaluation
and
lack
of
international
competitiveness in exports of primary products
High and unsustainable external debts, and hence
absence of internal and external financial balances
High financial repression with low access of private
sector to credit
Main Policy Outcome
Some temporary improvements in macroeconomic
stability were achieved but these were inadequate
and below the desired levels
External sector dependency still prominent and many
countries had limited capacity to expand exports,
decreases in investment rate, and wider budget and
BoP deficits
Social issues ignored as governments reduced
spending on provision of social services especially
public healthcare, education, size of public sector and
parastatals with adverse consequences for
improvements in poverty, starvation, unemployment,
and malnutrition had not witnessed marked
improvements, and even worsened in some countries
like Liberia, Nigeria, Rwanda and Tanzania.
Stellenbosch University http://scholar.sun.ac.za
the financial system
42
III: African Alternative Framework to Structural Adjustment Programmes for Socioeconomic Recovery and Transformation (AAF-SAP) , 1989-99
Compelling Issues
Key Policy Objectives
Key Policy Instruments
Number/
List
of Main Policy Outcome
Countries
Absence of holistic Improve human resource Multiple
exchange
rates All SSA countries Under democratic governance, some gains were made
macroeconomic
capacity through higher
system in a rationalised
except
those
like
in reducing expenditure on military and defence in
framework and policy
incomes, motivation and
manner and/or creating and
South Africa which
favour of providing essential services to the citizens,
measures
and
empowerment as well as
streamlining such a system for
were under sanctions /
but these were still inadequate.
directions that takes
equitable distribution of
the purposes of resource
interdictions
from Governments continued to privatise key state
into
account
the
income
transfers,
resource
global
community
enterprises and seek foreign partnership in managing
dynamic
mobilisation and reversal of
including the UN and
Adjust the pattern of public
strategic SOEs due to lack of finance
interrelationships
capital flight and ensuring
OAU as at that time
expenditure to satisfy the
With high external debts accumulated and increasing
existing among the key
availability of essential imports
essential needs of citizens
social demand, governments did not succeed in
elements
of
the Strengthen scientific and Land reforms for better access
integrating
the
marginalised
into
adjustment
adjustment
with
and entitlement to land for
technological
base
to
programmes
transformation process.
productive use
enhance production and
Private sector contributed positively to exports of nonThus, to address the
Greater mass participation in
diversification
traditional commodities, but these enterprises lacked
missing
dichotomy Provide institutional support
governance (decision-making
adequate resources to meet high foreign demand.
between
structural
and
implementation
of
for
adjustment
with
adjustment
and
government programmes)
transformation towards less
sustainable
import dependency and Trade reforms with differential
development.
improved debt servicing
export
subsidies
and
Other
adjustment
and management
encouragement of barter trade
programmes ignored or Establish
to boost sub-regional trade
a
pragmatic
marginalised
the
balance between public Allocation of increasing share
people.
and private sectors of the
of foreign exchange for
No
justification
for
economy
imports of vital inputs for
orthodox
SAP
as
agriculture and manufacturing
privatisation failed in
Bilateral and multilateral trade
most countries mainly
agreements
on
primary
due to inefficiency, and
commodities
absence
of
robust
Supervised food production
private sector.
credit systems in rural areas
42
This is not a universal economic model, but a special framework applicable with selective emphasis according to the peculiar characteristics of the country in question and the
circumstances under which the country finds itself, as AAF-SAP is meant to be used for designing specific country programmes, selecting appropriate policy instruments and
measures as well as adopting the relevant implementation strategy. Also, as a human-centred framework, AAF-SAP is based on the assumption of full democratisation of all aspects
of economic and social activities and in all stages from decision making to implementation. This framework again requires intensified international co-operation in the formulation,
implementation and monitoring of national programmes for adjustments with transformation.
74
Stellenbosch University http://scholar.sun.ac.za
where collaterals are scarce
A danger of foreign
capital dominance over
Support for cottage industries
domestic
ownership
with emphasis on indigenous
under privatisation
technology
43
VI: Heavily Indebted Poor Countries (HIPC) Initiative / Millennium Development Programme (1999/2000 - 2015)
Compelling Issues
Key Policy Objectives
Debt burden of LDCs Bail countries out of
became unsustainable
unsustainable debt and
focus on building policy and
External debt servicing
institutional foundation for
prevents poor countries
sustainable
development
from addressing critical
and poverty reduction
social issues such as
poverty.
Greater emphasis on more
effective social policies like
Economic constraints
investing in human capital
forcing poor / SAP
(education and health) for
countries to reduce
long term growth
conservation allocation,
and use up natural Increased emphasis on
resources leading to
ownership,
transparency
environmental
and
broad-based
degradation, especially
participation
in
the
form
of Since 2001, achieve MDGs
deforestation and soil
by the year 2015
overuse
Debt-trapped
SSA
forced to cut back on
imports and increase
exports,
but
LDCs
suffer low export prices
whilst
industrialised
economies suffer fewer
market distortions.
Source: Author‟s compilation based on various sources
43
Key Policy Instruments
Fiscal
policy
(prudent
development-driven spending)
Public sector reforms towards
higher
transparency
and
accountability by public office
holders
Democratic governance and
collective
participation
in
decision-making as well as
policy
implementation
of
issues that directly affect the
ordinary man
Strategic
trade
and
international relations
Adopting non-debt and antiinflationary
approach
to
financing
development
programmes
Number/
List
of
Participating Countries
33 countries as of
June 30, 2010 (see
Table A2.1 above) but
all SSA countries have
consented to work
towards achieving the
MDGs
Main Policy Outcome
Increasing investment in essential social infrastructure
like schools and healthcare centres
External debts reduced temporarily as some countries
have started accumulating debts after reaching the
HIPC decision point.
Improvements in access to essential social services.
For example, many HIPC countries have now
introduced a free immunisation programme for
children, abolished user fees for primary education,
and cash-and-carry healthcare delivery system.
Improved consultation process in designing Poverty
Reduction Strategies has helped to increase the
potential of the poor to influence national resource
allocation
Many of the early beneficiaries of debt relief and
enhanced aid have consistently sustained annual
growth rates over 5 per cent.
There are fears that majority of SSA countries will miss
MDGs by 2015 (see Carceles et al., 2001; Bruns et al.,
2003; UNDP, 2003; White and Black, 2004; Fay et al.,
2005).
A country is defined as HIPC if its net present value of debt is above 150 per cent of exports or above 250 per cent of total government revenue.
75
Stellenbosch University http://scholar.sun.ac.za
Table A3.3: Remittances Received by Sampled SSA Countries, 1980-2009 (period averages)
Migrant Remittances Received (US$'m)
1980-89 1990-99 2000-09
Benin
Botswana
57.57
97.83
153.74
Migrant Remittances per capita (US$)
Migrant Remittances as % of GDP
1980-2009 1980-89 1990-99 2000-09 1980-2009 1980-89 1990-99 2000-09 1980-2009
103.05
14.12
17.80
19.10
17.00
4.32
5.06
3.45
4.27
54.11
64.19
76.43
64.91
48.35
43.12
41.12
44.20
4.05
1.47
0.75
2.09
145.07
94.82
68.17
102.69
18.90
9.71
4.96
11.19
7.39
3.72
1.42
4.18
Cameroon
21.36
20.21
94.89
45.48
2.09
1.45
5.21
2.92
0.25
0.19
0.54
0.33
Cape_Verde
Burkina Faso
31.39
79.49
118.82
76.57
98.95
201.87
249.28
183.36
17.17
18.02
12.67
15.96
Comoros
4.02
12.75
11.93
9.57
10.25
26.57
20.10
18.97
2.58
5.54
3.62
3.92
Congo, Rep
1.96
4.43
11.89
6.09
0.95
1.59
3.51
2.01
0.09
0.19
0.23
0.17
Côte d'Ivoire
32.48
101.80
155.52
96.60
3.17
6.71
8.08
5.99
0.38
0.91
0.99
0.76
Ethiopia
10.95
18.66
163.65
64.42
0.27
0.33
2.11
0.90
0.12
0.22
1.00
0.45
Gabon
0.16
3.77
8.16
4.03
0.21
3.44
5.94
3.19
0.00
0.08
0.10
0.06
Gambia
8.60
13.10
48.29
23.33
11.18
12.58
31.00
18.25
3.75
3.58
9.24
5.52
Ghana
2.65
17.62
83.17
34.48
0.20
1.00
3.76
1.65
0.06
0.26
0.68
0.33
Guinea
4.94
6.04
66.67
25.89
0.85
0.90
7.10
2.95
0.21
0.20
1.92
0.77
Guinea Bissau
1.17
2.25
27.11
10.18
1.27
1.90
18.10
7.09
0.77
0.96
5.27
2.33
63.89
235.16
957.81
418.95
3.29
8.43
26.20
12.64
0.91
2.27
4.58
2.59
Kenya
317.12
378.83
329.06
341.67
219.18
222.00
164.94
202.04
81.55
53.29
28.73
54.52
Madagascar
Lesotho
3.53
12.29
13.35
9.72
0.35
0.95
0.78
0.69
0.13
0.38
0.26
0.25
Malawi
0.92
0.69
0.92
0.84
0.13
0.07
0.07
0.09
0.07
0.04
0.03
0.05
Mali
58.82
103.06
217.69
126.52
7.47
10.96
17.99
12.14
3.61
4.20
3.84
3.88
Mauritania
3.72
9.95
1.99
5.22
2.13
4.69
0.68
2.50
0.44
0.76
0.12
0.44
Mauritius
15.67
119.50
210.82
115.33
15.31
105.78
170.62
97.24
1.03
3.12
3.47
2.54
Mozambique
60.53
56.64
72.11
63.09
4.65
3.68
3.44
3.92
1.99
2.15
1.12
1.76
Namibia
10.91
13.63
13.24
12.59
9.28
8.59
6.59
8.16
0.59
0.45
0.22
0.42
Niger
11.19
13.08
54.72
26.33
1.64
1.43
4.03
2.36
0.56
0.65
1.52
0.91
Nigeria
11.10
799.10 4,465.23
1,758.48
0.14
6.98
30.56
12.56
0.03
2.59
3.42
2.01
Rwanda
4.00
13.70
0.66
1.29
3.11
1.69
0.24
0.57
0.88
0.56
São Tomé & Prínicipe
Senegal
Seychelles
7.62
29.48
0.74
0.53
1.48
0.92
7.42
4.11
9.62
7.05
0.82
0.42
1.18
0.81
86.08
154.46
777.32
339.28
13.20
18.13
67.17
32.83
2.29
2.99
8.43
4.57
3.17
2.79
7.54
4.50
46.98
39.13
89.09
58.40
1.63
0.69
0.90
1.07
Sierra Leone
0.07
9.81
22.04
10.64
0.02
2.43
4.28
2.24
0.01
1.25
1.68
0.98
South Africa
63.05
160.79
583.74
269.19
2.06
4.06
12.31
6.15
0.07
0.12
0.27
0.15
275.24 1,504.21
675.97
10.63
8.57
38.15
19.12
2.26
2.52
5.22
3.33
Sudan
248.46
Swaziland
60.42
88.23
78.01
75.55
83.36
92.95
68.93
81.75
10.82
6.36
3.61
6.93
Tanzania
0.38
4.33
14.90
6.54
0.02
0.14
0.38
0.18
0.01
0.06
0.11
0.06
Togo
13.76
21.12
188.66
74.51
4.04
4.80
30.81
13.22
1.30
1.44
8.52
3.76
Uganda
34.57
140.99
427.62
201.06
2.21
6.68
14.77
7.89
0.88
2.94
4.62
2.81
Sample Average
40.24 87.36
307.23
144.94 17.92
24.58
32.89
25.13
4.23
3.60
3.46
3.77
Source: Author‟s computation base on MRF-2011, BoPS, WDI and GDF (April 2011) and estimates from countryspecific desks of IMF and WB. Note: Due to lack of consistent data, only the 36 sampled countries are listed.
76
Stellenbosch University http://scholar.sun.ac.za
Box A3.1: The Millennium Development Goals (MDGs)
Preamble: MDGs are eight international development goals commonly accepted as a framework for measuring the
pace of socioeconomic development progress by the World Bank, UN, IMF and other credible international
organisations since its unanimous adoption in September 2000. These goals, with 21 targets and a series of
measurable indicators for each target, are to be achieved by 2015.
MDG 1:Eradicate Extreme Poverty and Hunger
Target 1A: Halve the proportion of people living on less than US$1 a day
Target 1B: Achieve decent employment for women, men, and young people
Target 1C: Halve the proportion of people suffering from hunger
MDG 2: Achieve Universal Primary Education
Target 2A: By 2015, all children can complete a full course of primary schooling, girls and boys
MDG 3: Promote Gender Equality and Empower Women
Target 3A: Eliminate gender disparity in primary and secondary education preferrably by 2005, and at all levels by
2015
MDG 4: Reduce Child Mortality Rate
Target 4A: Reduce by two-thirds, between 1990 and 2015, the under-five mortality rate
MDG 5: Improve Maternal Health
Target 5A: Reduce by three-quarters, between 1990 and 2015, the maternal mortality ratio
Target 5B: Achieve, by 2015, universal access to reproductive health
MDG 6: Combat HIV/AIDS, Malaria, and Other Diseases
Target 6A: Have halted by 2015 and begun to reverse the spread of HIV/AIDS
Target 6B: Achieve, by 2010, universal access to treatment for HIV/AIDS for all those who need it
Target 6C: Have halted by 2015 and begun to reverse the incidence of malaria and other major diseases
MDG 7: Ensure Environmental Sustainability
Target 7A: Integrate the principles of sustainable development into country policies and programmes; reverse loss of
environmental resources
Target 7B: Reduce biodiversity loss, achieving, by 2010, a significant reduction in the rate of loss
Target 7C: Halve, by 2015, the proportion of people without sustainable access to safe drinking water and basic
sanitation
Target 7D: By 2020, to have achieved a significant improvement in the lives of at least 100 million slum-dwellers
MDG 8: Develop a Global Partnership for Development
Target 8A: Develop further an open, rule-based, predictable, non-discriminatory trading and financial system
Target 8B: Address the special needs of the Least Developed Countries (LDCs)
Target 8C: Address the special needs of landlocked developing countries and small island developing states
Target 8D: Deal comprehensively with the debt problems of developing countries through national and international
measures in order to make debt sustainable in the long term
Target 8E: In co-operation with pharmaceutical companies, provide access to affordable, essential drugs in
developing countries
Target 8F: In co-operation with the private sector, make available the benefits of new technologies, especially
information and communications
Source: UN MDGs website, retrieved 30 June, 2010.
77
Stellenbosch University http://scholar.sun.ac.za
Figure A3.1: Trends in External Capital Flows to Developing Economies, 1970-2009
External Capital Flows to ECA, 1970-2009
6
4
5
Capital Flows (% of GDP)
Capital Flows (% of GDP)
External Capital Flows to EAP, 1970-2009
5
3
2
1
0
4
3
2
1
0
-1
-1
Years
Years
External Capital Flows to LAC, 1970-2009
6
5
Capital Flows (% of GDP)
Capital Flows (% of GDP)
External Capital Flows to MNA
7
6
4
3
2
1
0
5
4
3
2
1
0
-1
-1
Years
Years
External Capital Flows to SAS, 1970-2009
External Capital Flows to SSA, 1970-2009
8
Capital Flows (% of GDP)
Capital Flows (% of GD)
5
4
3
2
1
0
-1
7
6
5
4
3
2
1
0
-1
-2
Years
FDI
ODA
Portfolio equity
Years
Migrant remittances
FDI
ODA
Portfolio equity
Source: Author based on BoPS as reported by the World Bank in WDI and GDF (April 2011)
78
Migrant remittances
Stellenbosch University http://scholar.sun.ac.za
Figure A3.2:
Migrant Remittance Flows to Developing Economies, 1970-2009 (actual, per capita & % of GDP)
Panel A2: Migrant Remittances Received, 1970-2009 (US$'m)
Panel A1: Total Migrant Remittances Received,
1970-2009 (US$'m)
90000
600000
594,632
590,501
500000
562,522
454,871
400000
324,962
300000
200000
159,150
Migrant Remittances (US$'m)
700000
80000
70000
60000
50000
40000
30000
20000
10000
0
100000
0
Year
ECA
LAC
MENA
SAS
SSA
LAC
MNA
SAS
SSA
1970
-
-
0.18
1.66
0.17
0.08
1975
0.02
3.87
0.63
6.63
0.55
1.08
1980
1.19
5.80
5.38
35.94
5.84
3.61
1985
1.40
4.57
6.60
31.25
5.73
2.63
1990
1.93
8.27
13.07
50.10
4.94
3.65
1995
5.21
16.28
28.06
52.20
8.04
5.42
2000
8.75
25.91
39.39
46.57
12.63
6.90
2001
11.51
25.77
47.06
53.62
13.83
6.75
2002
14.68
26.81
53.49
54.72
17.12
7.15
2003
17.39
29.01
68.95
68.96
21.19
8.23
2004
21.38
40.00
80.27
76.80
19.71
10.80
2005
26.66
58.12
91.72
81.41
22.95
12.37
2006
30.21
70.86
107.02
84.34
28.34
16.24
2007
37.10
97.94
113.04
100.65
35.49
23.24
2008
44.29
113.75
114.18
110.55
46.34
26.07
2009
43.71
87.63
99.40
101.32
47.74
24.80
Year
EAP
ECA
1970
EAP
ECA
EAP
120.00
Migrant Remittances per capita (US$)
EAP
ECA
LAC
MNA
SAS
SSA
Panel B: Migrant Remittances Received per Capita (US$)
100.00
80.00
60.00
40.00
20.00
-
EAP
LAC
MNA
SAS
SSA
5.00
-
-
0.03
0.75
0.14
0.03
4.50
0.01
-
0.05
1.05
0.32
0.26
0.43
-
0.26
3.01
2.25
0.51
1985
0.39
-
0.37
2.16
1.99
0.52
1990
0.46
0.34
0.52
4.29
1.39
0.62
1995
0.68
0.85
0.76
4.21
2.12
0.97
2000
0.92
1.48
0.99
3.03
2.85
1.36
2001
1.15
1.47
1.24
3.53
3.08
1.39
2002
1.33
1.34
1.61
3.75
3.68
1.39
2003
1.41
1.15
1.95
4.32
3.96
1.34
2004
1.51
1.20
1.99
4.25
3.15
1.45
0.50
2005
1.64
1.38
1.89
3.98
3.24
1.46
-
2006
1.55
1.37
1.90
3.58
3.59
1.70
2007
1.52
1.47
1.72
3.57
3.60
2.17
2008
1.45
1.39
1.51
3.21
4.69
2.18
2009
1.34
1.37
1.43
3.16
4.61
2.25
Migrant remittances (% of GDP)
1975
1980
ECA
LAC
MNA
SAS
SSA
Panel C: Migrant Remittances Received (% of GDP)
4.00
3.50
3.00
2.50
2.00
1.50
1.00
EAP
ECA
Source: Author‟s estimations based on data WDI and GDF (April 2011)
79
LAC
MNA
SAS
SSA
Stellenbosch University http://scholar.sun.ac.za
Figure A3.3: Remittances per capita vs per Migrant in Developing Economies, 1970-2009
5,000.00
4,500.00
Average Migrant per vs per Capita Migrant Remittances Received in Developing
Economies, 1970-2009
MRPC, MRPM (US$)
4,000.00
3,500.00
3,000.00
2,500.00
2,000.00
1,500.00
1,000.00
500.00
-
EAP
ECA
LAC
Migrant Rem ittances per Capita (US$)
MNA
SAS
SSA
Migrant Remittances per Migrant (US$)
LAC
MNA
SAS
SSA
Migrant Rem ittances per Capita (US$)
5,794.13
EAP
771.57
4,113.08
1,633.43
1,431.86
310.72
Migrant Rem ittances per Migrant (US$)
2,845.70
1,108.30
2,020.65
1,173.25
756.74
178.66
Correlation_MRPC,MRPM
0.99672
ECA
-0.37250
0.99938
0.96681
0.98672 0.99144
Source: Author based on WDI and GDF (April 2011). Note: 5-year data ranging 1970, 1975,….2005 was used as
data on total international migration stock was not reported on annual basis by the World Bank in its WDI. MRPC
represents migrant remittances per capita whilst MRPM denotes (migrant) remittances per migrant.
80
Stellenbosch University http://scholar.sun.ac.za
Figure A3.4:
Migrant Remittance-Recipient Countries in SSA (average, based on % of GDP), 1980-2009
1980-1989
Gabon
Tanzania
Sierra Leone
Nigeria
Ghana
South Africa
Malawi
Congo, Rep
Ethiopia
Madagascar
Guinea
Rwanda
Cameroon
Côte d'Ivoire
Mauritania
Niger
Namibia
Guinea Bissau
ST&P
Uganda
Kenya
Mauritius
Togo
Seychelles
Mozambique
Sudan
Senegal
Comoros
Mali
Gambia
Botswana
Benin
Burkina Faso
Swaziland
Cape Verde
Lesotho
0.005
0.006
0.008
0.030
0.056
0.071
0.073
0.094
0.125
0.129
0.206
0.241
0.251
0.376
0.442
0.560
0.593
0.768
0.816
0.883
0.909
1.034
1.302
1.625
1.994
2.258
2.286
2.584
3.606
3.755
4.050
4.318
7.390
10.819
17.173
81.551
-
20.000
40.000
60.000
Malawi
Tanzania
Gabon
South Africa
Congo, Rep
Madagascar
Cameroon
Ghana
Namibia
Mauritania
Ethiopia
Rwanda
Côte d'Ivoire
Guinea
ST&P
Niger
Sierra Leone
Seychelles
Mozambique
Nigeria
Botswana
Guinea …
Mauritius
Kenya
Uganda
Sudan
Togo
Mali
Comoros
Burkina Faso
Benin
Senegal
Gambia
Swaziland
Cape Verde
Lesotho
28.727
-
10.000
20.000
30.000
1990-1999
0.038
0.062
0.077
0.121
0.193
0.193
0.197
0.221
0.260
0.377
0.419
0.448
0.573
0.647
0.692
0.759
0.906
0.962
1.254
1.445
1.466
2.153
2.270
2.522
2.587
2.936
2.994
3.125
3.583
3.722
4.198
5.057
5.544
6.362
18.025
53.286
-
80.000 100.000
2000-2009
0.033
0.099
0.111
0.120
0.215
0.227
0.255
0.272
0.539
0.680
0.748
0.880
0.904
0.989
1.000
1.118
1.183
1.419
1.522
1.677
1.921
3.424
3.446
3.466
3.613
3.617
3.841
4.576
4.623
5.216
5.275
8.428
8.522
9.236
12.675
Malawi
Gabon
Tanzania
Mauritania
Namibia
Congo, Rep
Madagascar
South Africa
Cameroon
Ghana
Botswana
Rwanda
Seychelles
Côte d'Ivoire
Ethiopia
Mozambique
ST&P
Burkina Faso
Niger
Sierra Leone
Guinea
Nigeria
Benin
Mauritius
Swaziland
Comoros
Mali
Kenya
Uganda
Sudan
Guinea Bissau
Senegal
Togo
Gambia
Cape Verde
Lesotho
Malawi
Tanzania
Gabon
South Africa
Congo, Rep
Cameroon
Guinea
Ethiopia
Ghana
Madagascar
ST&P
Namibia
Rwanda
Niger
Seychelles
Mauritania
Côte d'Ivoire
Guinea Bissau
Sierra Leone
Togo
Botswana
Mozambique
Kenya
Sudan
Nigeria
Uganda
Senegal
Mauritius
Gambia
Burkina Faso
Mali
Benin
Comoros
Swaziland
Cape Verde
Lesotho
40.000
-
10.000 20.000 30.000 40.000 50.000 60.000
1980-2009
0.048
0.060
0.060
0.154
0.171
0.254
0.328
0.332
0.419
0.440
0.449
0.565
0.757
0.775
0.806
0.910
0.979
1.074
1.755
2.014
2.088
2.335
2.542
2.585
2.814
3.332
3.757
3.882
3.915
4.177
4.274
4.570
5.525
6.932
15.957
54.521
20.000
40.000
60.000
Source: Author‟s estimation from MRF-2011, BoPS, WDI, and GDF (April 2011) and estimates from country-specific
desks of IMF and World Bank. Note: Only the 36 sampled countries are included due to data constraint.
81
Stellenbosch University http://scholar.sun.ac.za
Figure A3.5:
Migrant Remittances Received in SSA Countries (period average in US$‟m), 1980-2009
Sierra Leone
Gabon
Tanzania
ST&P
Malawi
Guinea…
Congo, Rep
Ghana
Seychelles
Madagascar
Mauritania
Rwanda
Comoros
Guinea
Gambia
Namibia
Ethiopia
Nigeria
Niger
Togo
Mauritius
Cameroon
Cape Verde
Côte d'Ivoire
Uganda
Botswana
Benin
Mali
Swaziland
Mozambique
South Africa
Kenya
Senegal
Burkina Faso
Sudan
Lesotho
Malawi
ST&P
Mauritania
Seychelles
Gabon
Congo, Rep
Comoros
Namibia
Madagascar
Tanzania
Sierra Leone
Guinea Bissau
Rwanda
Gambia
Niger
Guinea
Burkina Faso
Mozambique
Botswana
Swaziland
Ghana
Cameroon
Cape Verde
Benin
Côte d'Ivoire
Ethiopia
Togo
Mauritius
Mali
Lesotho
Uganda
South Africa
Senegal
Kenya
Sudan
Nigeria
1990-1999
1980-1989
0.07
0.16
0.38
0.74
0.92
1.17
1.96
2.65
3.17
3.53
3.72
4.00
4.02
4.94
8.60
10.91
10.95
11.10
11.19
13.76
15.67
21.36
31.39
32.48
34.57
54.11
57.57
58.82
60.42
60.53
63.05
63.89
86.08
145.07
248.46
317.12
100.00
200.00
300.00
400.00
2000-2009
0.92
1.48
1.99
7.54
8.16
11.89
11.93
13.24
13.35
14.90
22.04
27.11
29.48
48.29
54.72
66.67
68.17
72.11
76.43
78.01
83.17
94.89
118.82
153.74
155.52
163.65
188.66
210.82
217.69
329.06
427.62
583.74
777.32
957.81
1,504.21
4,465.23
-
ST&P
Malawi
Guinea Bissau
Seychelles
Gabon
Tanzania
Congo, Rep
Guinea
Rwanda
Sierra Leone
Mauritania
Madagascar
Comoros
Niger
Gambia
Namibia
Ghana
Ethiopia
Cameroon
Togo
Mozambique
Botswana
Cape Verde
Swaziland
Burkina Faso
Benin
Côte d'Ivoire
Mali
Mauritius
Uganda
Senegal
South Africa
Kenya
Sudan
Lesotho
Nigeria
2,000.00
4,000.00
0.53
0.69
2.25
2.79
3.77
4.33
4.43
6.04
7.62
9.81
9.95
12.29
12.75
13.08
13.10
13.63
17.62
18.66
20.21
21.12
56.64
64.19
79.49
88.23
94.82
97.83
101.80
103.06
119.50
140.99
154.46
160.79
235.16
275.24
378.83
799.10
-
Malawi
ST&P
Gabon
Seychelles
Mauritania
Congo, Rep
Tanzania
Comoros
Madagascar
Guinea Bissau
Sierra Leone
Namibia
Rwanda
Gambia
Guinea
Niger
Ghana
Cameroon
Mozambique
Ethiopia
Botswana
Togo
Swaziland
Cape Verde
Côte d'Ivoire
Burkina Faso
Benin
Mauritius
Mali
Uganda
South Africa
Senegal
Lesotho
Kenya
Sudan
Nigeria
6,000.00
500.00
1,000.00
0.84
1980-2009
0.92
4.03
4.50
5.22
6.09
6.54
9.57
9.72
10.18
10.64
12.59
13.70
23.33
25.89
26.33
34.48
45.48
63.09
64.42
64.91
74.51
75.55
76.57
96.60
102.69
103.05
115.33
126.52
201.06
269.19
339.28
341.67
418.95
675.97
1,758.48
-
500.00
1,000.00
1,500.00
2,000.00
Source: Author‟s estimation from MRF-2011, BoPS, WDI, and GDF (April 2010) and estimates from country-specific
desks of IMF and World Bank. Note: Only the 36 sampled countries are included due to data constraint.
82
Stellenbosch University http://scholar.sun.ac.za
CHAPTER FOUR
MACROECONOMIC DETERMINANTS OF INTERNATIONAL REMITTANCES IN SUBSAHARAN AFRICA,
4.0 INTRODUCTION
This chapter aims at identifying the core macroeconomic factors responsible for explaining the
changing levels of international migrant remittances received by sub-Saharan Africa (SSA)
countries since the implementation of financial liberalisation programme in the 1980s. A set of
annual panel data on 36 SSA countries, covering 1980-2009, was used in a system
Generalised Method of Moments (GMM), following Blundell and Bond (1998) dynamic panel
data estimation technique. In order to provide a more detailed insight into the possible
dynamics of the varying impact of macroeconomic variables that explain the inflow of
remittances in SSA, decade-based (1980-89, 1990-99 and 2000-09), as well as an overall
study period 1980-2009, estimations were carried out. Furthermore, efforts were made to
explore the determinants of migrant remittances at the disaggregated level - workers‟
remittances and compensation of employees. This chapter proceeds with a background
discussion to motivate the study and to outline its objectives. This is followed by selected
stylised facts in Section 4.2. In Section 4.3, a review of the theoretical and empirical literature is
presented, whilst Section 4.4 presents the theoretical framework. Section 4.5 formulates the
empirical model and the methodology adopted in analysing the data.
A presentation and
discussion of the empirical results can be found in Section 4.6, whilst Section 4.7 concludes the
chapter with policy implications.
4.1 BACKGROUND
Over the past two to three decades in particular, international migration from low-income
countries to high-income countries has been rising steadily. From a low 75 million international
Papers based on this chapter were presented at African Economic Research Consortium (AERC) bi-annual
conferences (May/June 2010; November/December 2010; May/June 2011) at Mombasa and Nairobi, Kenya. Also,
at IMF Staff Seminar (March 2, 2011), Washington, DC, USA; and Economic Society of South Africa (ESSA) biannual conference, September 5-7, 2011, Cape Town, South Africa.
A paper based on this chapter, entitled “Macroeconomic Determinants of Remittances in Sub-Saharan Africa,” has
been accepted for publication as a chapter in The Macroeconomics of Africa‟s Recent Growth edited by Shanta
Devarajan and Ibi Ajayi. Also published from this chapter are: “The Changing Impact of Macroeconomic Environment
on Remittance Inflows in Sub-Saharan Africa,” Journal of Academic Research in Economics (2011), 3(2): 136-167.
“Macroeconomic Determinants of Workers‟ Remittances and Compensation of Employees in Sub-Saharan Africa,”
Journal of Developing Areas, 48(1): 337-360.
83
Stellenbosch University http://scholar.sun.ac.za
migrant stock in 1965, the figure gradually rose to 120 million in 1990, and to more than 176
million in the year 2000, before attaining a high of 191 million and 214 million 44 in 2005 and
2010 respectively (IOM, 2010; UN, 2010). There is no compelling reason to expect a reversal
trend in international migration in this era of increasing globalisation and widening income-gap
between low income migrant-home countries and high income migrant-host countries45. Indeed,
in various Human Development Reports since 2005, the United Nations (UN) attributes this
trend of migration, involving both skilled and unskilled labour from developing countries to the
industrialised world, to low living standards and poor working conditions in developing
countries. In the case of sub-Saharan Africa (SSA), the Migration Policy Institute (2006) reports
that more than 20 per cent of tertiary graduates from the sub-region compared to less than 10
per cent of their counterparts from the Middle East and North Africa (MNA) were working in the
industrialised countries as at the end of 2006. During this same period, Angola, Guinea-Bissau
and Mozambique had at least 50 per cent of their tertiary graduates working in advanced
countries. According to UN (2009), Europe, with 32.6 per cent of international migrant stock,
leads as the main host of emigrants, followed by Asia (28.6 per cent), North America (23.4 per
cent), Africa (9 per cent), Oceania (2.8 per cent) and Latin America (2.4 per cent).
Although migrant-home countries may suffer from brain drain, these low-income countries have
been benefiting directly and quite instantaneously from their citizens who migrate abroad
through the receipt of remittances. This could be the most obvious reason why developing
countries are the main destination of migrant remittances with the industrialised world
maintaining their status as the main source of remittances46. Parallel to the recent upsurge of
cross-border migration, international remittances received by developing countries have been
rising rapidly and incessantly since the 1980s. The significant and consistent growth trend in
remittance flows in recent years obviously has important implications for economic growth and
development in the recipient countries. For instance, some macro-level studies have shown
that official remittance inflows promote long-run growth (Faini, 2003; Ahortor and Adenutsi,
2009; Adenutsi, 2011) and socioeconomic development (Özden and Schiff, 2005; Adenutsi,
44
According to UN (2010), the stock of international migrants represents more than three per cent of the world‟s
population in 2010. 128 million persons, being 60 per cent of international migrant stock, were residing in
industrialised countries of which 74 million representing 57.8 per cent were nationals from developing countries.
45
High search frictions in the labour market due to low value addition in production (de-industrialisation) in the
developing world are the most obvious explanations for the low wages and poor living standards in this part of the
world.
46
Developing countries receive at least 75 per cent of reported migrant remittances. In 2009, developing countries
alone received as much as US$ 316 billion out of the world‟s total of US$ 414 billion, representing 76.3 per cent
even though the amount they received in 2009 fell by about six per cent of the amount they received in 2008
(Author‟s computation based on World Bank‟s WDI, April 2011).
84
Stellenbosch University http://scholar.sun.ac.za
2010a, 2010b). Also, it has been proven that, in the long run, remittances foster price and
currency stability or appreciation (Katseli and Glytsos, 1986; Adenutsi and Ahortor, 2008), and
reduce poverty but do not necessarily improve income inequality (Adams and Page, 2005;
Azam and Gubert, 2005; Adams, 2006; Ratha and Mohapatra, 2007; Nguyen, 2008).
Remittances also create employment in recipient countries through increased investment and
productivity (El-Sakka and McNabb, 1999; Ratha, 2003; IMF, 2005). In effect, in recent years,
remittances have emerged as an important source of external development finance and a
mitigating
factor
for
economic
imbalances
and
financial
instability,
particularly
in
underdeveloped countries, including those in SSA (Ratha, 2003; IMF, 2005; World Bank,
2006a).
Furthermore, in some developing countries such as India, Mexico, Philippines, and Lesotho,
remittances have far exceeded other international capital flows in the form of Official
Development Assistance (ODA) and Foreign Direct Investment (FDI) in recent years47.
Accordingly, remittances have become a crucial source of foreign exchange in most developing
countries. Consistent with the trend in international migration, official migrant remittances
received by developing countries reached US$116 billion in 2003 representing more than 1.5
per cent of their gross domestic product (GDP). In 2004, migrant remittances of US$126 billion
became the second most important source of foreign exchange earnings to developing
countries (World Bank, 2006a,b). This was the year in which FDI to developing countries stood
at US$165 billion with gross ODA amounting to US$79 billion (World Bank, 2006a). Recorded
migrant remittances received by developing countries rose to US$194.2 billion in 2005,
reaching an all-time high of US$336 billion in 2008 before plummeting slightly to US$316 billion
in 2009, in response to the global financial crisis of 2007-2009 (World Bank, 2010). Yet, the
relative importance of migrant remittances over other capital inflows in developing countries,
with respect to the size, growth rate and stability, remains unchanged over the past four
decades as the decline in 2009 is only the second after the first was recorded in 1985.
Even though remittances received by developing countries have more than doubled during the
last decade in terms of absolute volume, Africa experienced only a marginal rise. For instance,
official migrant remittances to Africa amounted to US$9 billion (out of which SSA received
$1.86 billion) in 1990; and by 2003, migrant remittance flows to Africa had reached US$14
47
Migrant remittances are the second largest form of non-debt capital inflows in developing countries. In the Middle
East and North Africa (MNA) just as in South Asia (SAS), migrant remittances are now the leading source of external
capital (see Figure 3A.1 in Chapter Three).
85
Stellenbosch University http://scholar.sun.ac.za
billion (out of which SSA received $5.96 billion) 48. During this period, Egypt and Morocco were
the largest recipients of remittances in Africa with Northern Africa as a sub-region receiving
over 60 per cent of the total remittance flows to the continent. The rising trend in official migrant
remittance flows to SSA continued, reaching US$10 billion in 2005; attaining an all-time high of
US$21.6 billion in 2008 before dropping slightly to US$ 20.7 billion in 2009 following the global
financial crisis which led to 2007-2009 economic recession that hit the industrialised world.
Despite this positive growth trend SSA remains the least recipient of migrant remittances,
receiving only five per cent of global remittances compared to East Asia and the Pacific (20.7
per cent), South Asia (18 per cent), LAC (13.7 per cent), Europe and Central Asia (11.0 per
cent) and MNA (7.7 per cent). In fact, as at the end of 2009, SSA as a sub-region received far
less official remittances (US$20.74 billion) than any of the world‟s top-three migrant remittancerecipient countries - India (US$49.26 billion), China (US$47.55 billion) and Mexico (US$22.16
billion)49. It is acknowledged that the officially reported value of migrant remittances received by
developing countries is far lower than the actual amount received which is estimated to be at
least 50 per cent higher than the officially reported amount (World Bank, 2006a,b). Freund and
Spatafora (2005) posit that SSA receives the highest informal remittances, representing 45-65
per cent of what is officially reported, unlike 5-20 per cent in the case of Latin America. The
adverse repercussions of the increasing flow of migrant remittances to SSA, and the
developing world as a whole, through informal channels cannot be underestimated. These
include money laundering, sponsorship of anti-government groups for self-centred interest,
financing terrorist activities, creation or expansion of existing informal financial markets such as
the „underground‟ foreign exchange market, de facto dollarisation, and arbitrary growth in
money supply in remittance-receiving countries. Ultimately, the continuous inflows of
remittances through the informal channels can undermine the economic and political stability of
the remittance-receiving countries and, at the same time, threaten the peace and security of
the world.
Certainly, several factors, ranging from micro to macro, might have accounted for the relatively
low receipt of official migrant remittances (or high receipt of informal remittances) by SSA. This
chapter explores the factors that inhibit the optimal inflows of migrant remittances through
official channels to SSA as a sub-region from a macroeconomic perspective. The fundamental
question is: What role can macroeconomic factors play under liberalised financial market
48
49
Author‟s calculation based on World Bank‟s WDI (April, 2011)
Author‟s compilation from World Bank (2011b) Remittance Database
86
Stellenbosch University http://scholar.sun.ac.za
regime in attracting official migrant remittances to SSA? The interrelated pertinent research
questions with regard to SSA are:
i.
What macroeconomic factors explain variations in official migrant remittance inflows?
ii.
Are there any time variations in the impact of these macroeconomic factors on official
migrant remittance inflows over the past three decades?
iii.
Do macroeconomic factors impact differently on migrant remittances and workers‟
remittance inflows?
iv.
To what extent do macroeconomic factors explain variations in the inflow of
compensation of employees?
In response to the above research questions, this study attempts to explain broadly the
macroeconomic factors behind migrant remittance flows to SSA. It seeks to find the long-run
macroeconomic determinants of remittance flows to SSA. More specifically, with respect to
SSA, the study seeks to:
i.
determine the impact of macroeconomic factors on official migrant remittance inflows;
ii.
examine if the impact of the macroeconomic factors identified in (i) vary on migrant
remittances inflows over time;
iii.
verify if macroeconomic factors have any unique impact on workers‟ remittances; and,
iv.
explore the influence of macroeconomic factors on compensation of employees inflows.
Based on the empirical findings, appropriate policy recommendations are made to guide
macroeconomic policy formulation towards attracting a higher inflow of official migrant
remittances in SSA. As far as the sub-region is concerned, this study is novel in the
measurement of migrant remittances and in providing an insight into the time-dependent
changing role of macroeconomic factors affecting migrant remittances over the past three
decades. Also of unparalleled contribution is the fact that this study identifies the
macroeconomic factors that explain migrant remittances at the disaggregated levels.
4.2 SELECTED STYLISED FACTS ON REMITTANCE FLOWS TO SSA
This section presents some stylised facts on the cyclical behaviour and the composition of
migrant remittance inflows, as well as the destination of SSA migrants outside the sub-region.
87
Stellenbosch University http://scholar.sun.ac.za
4.2.1: The Cyclical Behaviour of Remittance Flows to SSA, 1980-2009
In line with the altruistic theory, migrant remittance inflows are expected to be countercyclical;
pro-cyclical in conformity with the self-interest motive, and acyclical in manifestation of the
mixed motive (or tempered self-interest) in the recipient countries.
Figure 4.1:
Trends in Migrant Remittances, Household Consumption and Income in SSA, 1980-2009
6
50
30
2
20
0
10
0
-2
-10
HFCE and GDP per capita (annual %)
Migrant remittances per capita (annual %)
40
4
-4
-6
GDP per capita (annual %)
HFCE per capita (annual %)
Migrant remittances per capita (annual %)
Source: Author based on WDI (April 2011).
-20
-30
Note: HFCE denotes household final consumption expenditure
Figure 4.1 shows the trends in migrant remittances received, household consumption and
income in SSA between 1980 and 2009. With reference to Figure 4.1, there is fairly strong
evidence of pro-cyclicality in the growth of migrant remittances per capita and GDP per capita
in SSA in the 1980s and in the 2000s. In the 1990s, there appears to be countercyclicality in
the inflow of migrant remittances per capita as against GDP per capita growth in SSA. This
trend is notwithstanding the fact that over the past three decades, migrant remittances (both in
actual volumes and per capita terms) to the sampled 36 SSA countries have been increasing
consistently taking into account the group mean for each decade as shown in Table A3.3 in
Chapter Three.
The trends in the annual growth in household final consumption expenditure per capita and
migrant remittances per capita confirm the pro-cyclicality in the flow of remittances to SSA in
the 1980s and in the 2000s. The pro-cyclicality in the flow of migrant remittances in relation to
household final consumption expenditure can be attributed to altruistic motive driving
remittances. Therefore, with reference to the trends in per capita income growth and migrant
remittances per capita growth, it can be argued that migrant remittances received by SSA are
generally pro-cyclical during „good times‟ (i.e. the 1980s and the 2000s). Furthermore, as
88
Stellenbosch University http://scholar.sun.ac.za
revealed in Figure 4.1, in terms of growth in migrant remittances per capita, the flow of migrant
remittances to SSA cannot be described as unwavering but rather as erratic, especially in the
1990s. This suggests that in understanding the cyclical behaviour of migrant remittance inflows
the use of the growth rate in the flow of remittances per capita rather than the popularly used
absolute volume (as in Figure 4.2 Panel A) or relative to nominal GDP should be seen as more
appropriate (cf. Chami et al., 2005; Gupta, 2005; Lueth and Ruiz-Arranz, 2007a).
Figure 4.2:
Trends in Components of Migrant Remittances and GDP per capita in SSA, 1980-2009
MREPC, WREMPC (%)
200
150
1200
100
800
15
10
600
5
Real GDP per capita (US$)
1000
400
0
-5
200
-10
-15
Migrant remittances (MREMPC)
Workers' remittances per capita (WREMPC)
Compensation of employees per capita (COMPPC)
Real GDP per capita
0
50
0
-50
-100
50
1998
519.2179
6.81751
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
508.7881
485.5772
514.0512
615.8291
743.5645
846.5879
966.4
1085.824
1214.148
1126.819
20
6.901955
6.754287
15
7.159268
10
8.224343
10.79062
5
12.37861
16.25646
0
23.31172
-5
26.37124
24.71215
-10
25
Panel B: Trends in REMPC, WEMPC and Real GDP per capita
1999
504.9608 6.784137
Growth Rates
Real GDP per capita (%)
250
-15
Migrant remittances per capita (MREMPC)
Workers' remittances per capita (WREMPC)
Real GDP per capita
Panel C: Trends in MREMPC, COMPPC and Real GDP per
capita Growth Rates
40
100
0
30
-100
20
-200
10
-300
0
-400
-10
-500
-20
-30
Real GDP per capita
Migrant remittances per capita (MREMPC)
Compensation of employees per capita (COMPPC)
-600
Source: Author based on WDI (April 2011)
Figure 4.2 presents the cyclical behaviour of migrant remittance inflows per capita (MREMPC),
as well as the components of MREMPC – the inflows of workers‟ remittances per capita
(WREMPC) and compensation of employees per capita (COMPPC) in SSA between 1980 and
2009. Figure 4.2 Panel A, reveals that the actual values of per capita migrant remittances and
workers‟ remittances received in SSA are highly pro-cyclical with respect to real GDP per capita
over the past three decades. During this same period, actual values of COMPPC were acyclical
to real GDP per capita prior to the year 2005. Beyond 2005, however, COMPPC became
countercyclical relative to real GDP per capita (Panel A).
89
COMPPC (%)
1400
GDP per capita, MREMPC (%)
MREPC, WREMPC and CMPPC (US$)
7.097719
4.116677
2.981042
40 Panel
A: Trends
in Remittance Inflows per capita and Real
6.81751 4.17488 2.642631 GDP per capita
6.784137 4.524295 2.259842
35
6.901955 4.715698 2.186257
6.754287 4.653501 2.100786
7.159268
5.058056 2.101213
30
8.224343 5.629988 2.594355
10.79062 8.273497 2.517126
25
12.37861 24.0194 -11.6408
16.25646 27.34948 -11.093
20
23.31172
29.89979 -6.58808
26.37124 33.41858 -7.04735
24.71215 30.64306 -5.93092
Stellenbosch University http://scholar.sun.ac.za
Figure 4.2 Panel B shows that between 1980 and 2009 the trend in the annual growth rate of
per capita workers‟ remittances received in SSA was pro-cyclical in the 1980s and 2000s but
countercyclical in the 1990s. Coincidentally, it was in the 1990s that SSA as a sub-region
recorded its worst macroeconomic performance as reflected in reduced real GDP per capita,
domestic savings, higher external imbalance and debt stock (see Table 3.1). It can be seen
that the inflows of workers‟ remittances and migrant remittances follow a similar growth trend,
understandably because the former constitutes a significant component of the latter. In other
words, the cyclical behaviour of migrant remittance inflows is dependent upon the growth trend
of workers‟ remittances which is the dominant component of migrant remittances. From Figure
4.2 Panel C, it is apparent that the annual growth in compensation of employees received per
capita (COMPPC) has been acyclical to the growth in real GDP per capita in SSA since 1980.
Whilst this trend analysis cannot be interpreted as akin to or underscored by causal effects, two
main conclusions are possible from the observations based on Figure 4.2. Either, (i)
remittances from „permanent‟ migrants are positively responsive to macroeconomic conditions
at home, whilst remittances from „temporary‟ migrants are irresponsive to home-country
macroeconomic conditions; or (ii) workers‟ remittances can contribute more positively to homecountry macroeconomic performance whilst the impact of compensation of employees on
macroeconomic performance in SSA is relatively less important, given that workers‟
remittances form an integral part of migrant remittance inflows in SSA (see Figure 4.3).
4.2.2 The Composition of Migrant Remittances Received in SSA, 1980-2009
Figure 4.3 shows the composition and the degree of dependency on migrant remittances in the
36 SSA countries sampled for the empirical analysis. Figure 4.3 Panel A reveals that, with the
exception of Cape Verde, countries in the southern part of the sub-region viz. Lesotho,
Mauritius, Swaziland, Seychelles and Botswana, dominate the top-six migrant remittancerecipient countries. The remaining top-12 migrant remittance-recipients (Cape Verde, Senegal,
Sudan, Gambia, Benin and Togo) are predominantly West African countries. Comoros is the
only country from the eastern part of the sub-region listed among the top-12 remittancerecipients. Also, although the majority of the top-12 leading remittance-recipient countries are
small in geographical size, Botswana and Sudan are relatively large.
90
Stellenbosch University http://scholar.sun.ac.za
Figure 4.3: Composition of Migrant Remittances Received by SSA Countries, 1980-2009
Migrant Remittances per capita (US$)
250
Panel A: Top-12 Migrant Remittances-Recipient SSA Countries
200
150
100
50
0
Migrant Remittances per capita (US$)
3.5
Panel B: Bottom-12 Migrant Remittances-Recipient SSA Countries
3
2.5
2
1.5
1
0.5
0
REMPC
WREMPC
COMPPC
Source: Author based mainly on WDI (April 2011). Note: Only the 36 sampled countries included.
The geographical background of the bottom-third of migrant remittance-recipient countries is
quite heterogeneous. All the same, West African countries dominate with five countries
(Guinea, Mauritania, Niger, Sierra Leone and Ghana) in this bottom 12 category. Central and
Eastern Africa are represented by four countries (Cameroon, Congo, Rwanda and Ethiopia)
with Southern Africa having three countries (Madagascar, Tanzania and Malawi) among
countries which received the least migrant remittances. Again, Panel B is dominated by
countries with relatively large geographical size such as Cameroon, Mauritania, Niger, Congo,
Madagascar and Tanzania. Despite this, Guinea, Sierra Leone, Rwanda and Malawi, with
relatively small geographical size are also included in this category of countries which receive
the least migrant remittances.
91
Stellenbosch University http://scholar.sun.ac.za
With the exception of Congo Republic, only countries in the southern part of the sub-region
(Lesotho, Botswana and Tanzania) depend more on compensation of employees than how
they depend on workers‟ remittances. This implies that relatively high frequency of temporary or
circular migration is likely to be more common among countries in the southern part than
elsewhere in the sub-region. It is likely that unlike the nationals of other countries in the subregion the nationals of these Southern African countries might find it relatively easier, cheaper
and more convenient to migrate temporarily to neighbouring South Africa, the country that can
be described as industrialised, at least, by the standard of the sub-region.
In terms of income status, there is no distinctive pattern of dominance in either category as
relatively high-income countries such as Cameroon, Congo and Ghana are listed among the
least migrant remittance recipients just as other high-income countries like Seychelles, Cape
Verde, Mauritius and Botswana are listed among the high migrant remittance recipients. Thus,
migrant remittances flow to both high-income and low-income SSA countries; and the inflow of
remittances does not depend necessarily on the geographical size or location of the country.
This implies that some macroeconomic fundamentals and policies could be responsible for the
changing and unequal flow of migrant remittances received by the various SSA countries.
4.2.3 Migratory Patterns in SSA: Main Destinations and Sources of Remittances
Theoretical as well as empirical literature suggests the inclusion of both home-country and the
host-country factors in identifying the macroeconomic factors that explain migrant remittances
received by developing countries (see Section 4.4 and Table A4.1 in the Appendix). As
reported in Table A4.1, most empirical works on macroeconomic determinants of remittances
tend to use the USA as the migrant-host country. Some authors including Elbadawi and Rocha
(1992), Lianos (1997), Bouhga-Hagbe (2004), Akkoyunlu and Kholodilin (2006) and Akkoyunlu
(2010) made attempts at using countries other than the USA as the migrant-host nations in
macro-level country-specific studies with focus on bilateral remittances. In the case of SSA
countries, however, the majority of their migrants, at least 70 per cent, migrate to reside in other
SSA countries as shown in Table A4.3 in the Appendix.50 This makes the pattern of migration
among citizens of SSA unique compared to the rest of the world. Notwithstanding the fact that
SSA still serves as the main host of its „own migrants‟, the most important source of
international remittances to the various SSA countries is the SSA migrants residing in countries
50
This confirms earlier estimate by Ratha and Shaw (2007). For Africa as a whole, Barajas et al. (2010) observe that
more than 50 per cent of African migrants reside in Africa.
92
Stellenbosch University http://scholar.sun.ac.za
outside the sub-region (Ratha and Shaw, 2007; Bollard et al., 2010). Table 4.1 presents a list of
the 36 sampled SSA countries and the main host of their citizens residing outside Africa.
Table 4.1: Host Countries of SSA Migrants Resident outside SSA
Country Code Country Name
Non-SSA Host-Country
Country Code Country Name
Non-SSA Host-Country
1
Benin (BEN)
France (FRA)
19
Mali (MLI)
France (FRA)
2
Botswana (BSW)
Great Britain (GBR)
20
Mauritania (MRT)
France (FRA)
3
Burkina Faso (BFA)
Pakistan (PAK)
21
Mauritius (MRS)
France (FRA)
4
Cameroon (CAM)
France (FRA)
22
Mozambique (MZQ)
Portugal (POR)
5
Cape Verde (CPV)
Portugal (POR)
23
Namibia (NAM)
Great Britain (GBR)
6
Comoros (COM)
France (FRA)
24
Niger (NGR)
Germany (GER)
7
Congo Republic (CON) France (FRA)
25
Nigeria (NIG)
United States of America (USA)
8
Côte d'Ivoire (CIV)
France (FRA)
26
Rwanda (RWA)
Belgium (BEL)
9
Ethiopia (ETH)
United States of America (USA)
27
São Tomé & Príncipe (ST&P) Portugal (POR)
10
Gabon (GAB)
France (FRA)
28
Senegal (SEN)
France (FRA)
11
Gambia (GAM)
Spain (ESP)
29
Seychelles (SEY)
Great Britain (GBR)
12
Ghana (GHA)
United States of America (USA)
30
Sierra Leone (SLE)
United States of America (USA)
13
Guinea (GUI)
Great Britain (GBR)
31
South Africa (RSA)
Great Britain (GBR)
14
Guinea-Bissau (GBS) Portugal (POR)
32
Sudan (SUD)
Saudi Arabia (SAU)
15
Kenya (KEN)
Great Britain (GBR)
33
Swaziland (SWZ)
Great Britain (GBR)
16
Lesotho (LSO)
Germany (GER)
34
Tanzania (TNZ)
Great Britain (GBR)
17
Madagascar (MAD)
France (FRA)
35
Togo (TOG)
France (FRA)
18
Malawi (MLI)
Great Britain (GBR)
36
Uganda (UGA)
Great Britain (GBR)
Source: Author based on Parson et al. (2007).
A key feature in the pattern of SSA international migration as shown in Table 4.1 is that most of
its citizens outside the sub-region reside in Europe rather than in the Americas. It is also logical
to think that factors such as distance or travelling cost, geopolitical history or former colonial
relationship, lingual Franca and religious affinities underlie the choice of destination of SSA
international migrants. For instance, international migrants from Francophone SSA countries
such as Benin, Cameroon, Comoros, Congo, Côte d‟Ivoire, Mauritania and Senegal are hosted
by France with which they have a common language. These SSA countries were also
colonised by France in the past. The same trend is easily visible in the case of migrants from
Portuguese speaking SSA countries (Cape Verde, Mozambique, São Tomé and Príncipe, and
Guinea-Bissau) and migrants from English speaking SSA countries such as Kenya, Botswana,
Malawi, Namibia, South Africa and Uganda. On religious affinities, SSA migrants from Muslimdominated countries such as Burkina Faso, Benin, Niger and Sudan are mostly resident in
countries like Jordan, Pakistan and Saudi Arabia with which these SSA countries have a
common dominant religion. Evidence of proximity can be traced to Australia as an important
host country where many international migrants from southern SSA countries, notably
Botswana, Mauritius and Seychelles are resident (see Table A4.3).
93
Stellenbosch University http://scholar.sun.ac.za
From Table 4.1, France (12), Great Britain (10), United States of America (4), Portugal (4) and
Germany (2) lead as the first-choice host of SSA migrants outside the sub-region. This implies
that Europe leads as the main host of SSA international migrants51. Clearly, using the USA as
the main host of SSA migrants in an empirical study cannot be considered as appropriate.
Nevertheless, when the first three leading hosts of SSA international migrants are taken into
account as presented in Table A4.3, the USA emerges strongly as one of the leading hosts. In
fact, the USA hosts migrants from 24 SSA countries and ranks second only after Germany (27)
and is followed by France (17), Great Britain (16) and Portugal (3). From this perspective using
USA as the main host of international migrants from SSA could be considered a fairly good
proxy. This is because in this context, with 24, USA leads Germany in number of SSA migrants
that chooses the former as the first (or the most preferred) destination outside the continent.
Some important conclusions can be drawn from the stylised facts. First, the changing
macroeconomic policy environment in SSA is likely to impact on remittances received by the
sub-region in view of the fact that the cyclicality in the flow of migrant remittances were found to
vary over time – pro-cyclical in the 1980s and the 2000s but countercyclical in the 1990s. It is
for this reason that this chapter seeks to investigate, among other issues, the impact of
changing macroeconomic policy environment on remittance inflows in SSA by undertaking a
decade-by-decade analysis. Second, whereas just like migrant remittances, workers‟
remittances were largely pro-cyclical in the 1980s and in the 2000s but countercyclical in the
1990s, compensation of employees received in SSA were acyclical, hence less responsive to
the changing macroeconomic policy environment of SSA. This study took this observation into
account by analysing the determinants of migrant remittances at the aggregated and
disaggregated levels. Third, contrary to popular perception, some SSA countries, mainly SADC
countries other than South Africa, receive more compensation of employees than workers‟
remittances. This study did not probe this unique characteristic of SADC countries because as
explained earlier, there is the likelihood that proximity to a „big brother‟ industrialised country (in
this particular case, South Africa) could explain this phenomenon, requiring the inclusion of
physical distance (a non-macroeconomic variable) into the model as in (Lueth and Ruiz-Arranz,
2007b). Finally, because the facts clearly show that it is Europe and not the USA that leads as
the host of SSA migrants, it is the leading non-SSA migrant-host country of each sampled SSA
country rather than the USA that was used as the migrant-host country in this study.
51
Sander and Maimbo (2003), and Barajas et al. (2010) also identify Europe rather than North America as the main
host of African migrants.
94
Stellenbosch University http://scholar.sun.ac.za
4.3 LITERATURE REVIEW
The literature reviewed in this chapter covers the microeconomic foundation underlying the flow
of migrant remittances which discusses the reason why migrants remit, the uses of migrant
remittances, and the characteristics of potential remitters and potential recipients of
remittances. Also covered are the theories regarding remittance flows from a macroeconomic
perspective and the empirical studies on macroeconomic determinants of remittance inflows.
4.3.1 The Microeconomic Foundation and Theoretical Underpinnings of Remittances
4.3.1.1 Motivations to Remit and Uses of Migrant Remittances
Cross-border remittances are usually small value non-debt-creating monetary transfers from
income-earning migrants or benevolent organisations resident abroad and sent to family
members, other close associates, or social welfare institutions resident in native countries
directed at meeting a specific need. These flows are called migrant remittances if they strictly
involve interpersonal transfers from an international migrant to his/her close relation(s) resident
in his/her native country. In other words, international migrant remittances exclude the transfer
of funds from institutions to persons or social institutions in less privileged and vulnerable
economic environment. This study is centred on migrant remittance flows as they directly relate
to migration of labour, an important resource for which the SSA sub-region is well endowed.
From a microeconomic viewpoint, it can be observed that the motivation for a migrant to remit
part of his/her earnings to his/her native country is, either directly or indirectly, influenced by the
end use of remittances. For instance, in order to thoroughly understand the motives behind
migrant remittance inflows as well as the magnitude, regularity and volatility of these flows to
SSA it is essential to be acquainted with how remittances are used within the sub-region.
Although, admittedly, the uses of remittances can only be studied appropriately and
comprehensively at the micro level (which is outside the scope of this study), a review of the
available survey studies on the uses of remittances could offer some important insights into the
understanding of the dynamics and trends in migrant remittance flows at the macro level.
Rapoport and Docquier (2006) identify altruism, exchange, strategic behaviour, co-insurance,
inheritance, investment and mixed factors as the motives behind migrant remittance flows at
the microeconomic level. This was after the debate on motivations to remit was initiated by
Lucas and Stark (1985) who identified pure altruism, pure self-interest and tempered altruism
95
Stellenbosch University http://scholar.sun.ac.za
(or enlightened self-interest) as the microeconomic determinants of remittances using evidence
from Botswana. Becker (1974) argues that altruism is the most fundamental reason for
remitting. The altruistic motive of remittances is driven by natural love and concern for
improving the living standards of the other family members and close associates left behind in
the migrant‟s home country. In this case, a migrant derives positive utility from sending funds
home to improve upon the distressing economic condition of the target recipients (often close
relatives and friends) in his/her home country knowing very well that these recipients are in a
less advantageous economic environment. It is expected that a rise in migrant income, a
negative economic shock in the home country, a decrease in the real disposable income of the
target recipients and the migrant‟s intention to return to his/her home country after some time
should positively impact on remittance flows driven by altruism. From a migrant‟s perspective,
however, the number of international migrants in a target household should inversely relate the
regularity and the size of the remittances per migrant received by a household over time.
The pure self-interest theory of remittances as proposed by Lucas and Stark (1985) generates
three critical motives from the perspective of the remitting migrant. These are inheritance,
assets accumulation and intention to return home at a future date. Thus, under the theory of
pure self-interest, a migrant‟s motivation to remit is driven essentially by the migrant‟s intention
to return home after some time and, hence, the need to save at home in advance as well as to
earn respect among his/her family and close associates; and the aspiration to inherit a family
property like land, chieftaincy reign, and even sometimes to galvanise support for a political
position upon return. De la Brière et al. (2002) find evidence for this proposition in Dominican
Sierra. With regard to the intention to return home in the future, the migrant can then use a
member of his/her family or a close associate as a trustworthy supervisor and well-informed
agent who will monitor his/her children and spouse left behind as well as capital-intensive
investment projects such as construction of an apartment, commercial farming and other
entrepreneurial initiatives (Bernheim et al. 1985; Cox, 1987). Cox and Stark (1994) note that, in
a three-generational setting, a migrant may be motivated to remit to his/her parents as a
demonstration to his/her children how he/she (the migrant) should also be taken care of in old
age by them (the children). For this demonstrative effect to be effective under this
circumstance, the migrant makes sure the transfers of funds (i.e. remittances) are visible to
his/her children and even, in some cases, to his/her grandchildren. The net earnings of the
migrant and the intention to return home after some time rather than the negative economic
shocks at home and the number of emigrants in a household are the contributing factors that
96
Stellenbosch University http://scholar.sun.ac.za
are expected to have a significant positive impact on remittances. Also, the stability and growth
prospects of a migrant‟s home country as reflected in good democratic and accountable
governance, low inflation, higher per capita income and improved access to bank credit by the
private sector can positively influence a migrant‟s return and, hence, higher inflow of selfinterest driven remittances.
Tempered altruism (or enlightened self-interest) is the mixed motive of migrant remittances
representing the less extreme cases of pure altruism or pure self-interest (Lucas and Stark,
1985). This motive of remittances is informed by an implicit contractual framework of mutual
benefit from international migration which involves the migrant and his/her family resident in
his/her country of origin. The implicit contractual agreements would normally include coinsurance, loan repayment, exchange for services and strategic behaviour (Bernheim, et al.
1985; Poirine, 1997). For instance, a household may agree to mobilise funds to finance the
initial cost of migration of a family member to a country where the probability of job acquisition
and earning higher real wages is relatively high. The migrant is expected to remit part of his/her
income to the household left behind to offset the debt acquired in sponsoring his/her trip, and
thereafter, remittances are expected to continue to flow especially during periods of negative
economic shocks. A migrant could also enter into an agreement with his/her family to be
sponsored abroad so that in return, he/she will pay the airfare of an agreed number of
economically active family members to travel abroad for greener pastures. Besides, both
parties (the migrant and his/her family) might agree to invest the remittances received by the
household into an agreed investment project that could be mutually beneficial to both parties.
The investment project could serve as a hedge against uncertain future misfortunes such as illhealth and deportation of the migrant and negative shocks at home or in the country of
residence of the migrant. Furthermore, in economies where the extended family systems and
social ties are strong migrants may be compelled to remit home regularly as a compensation
for the loss of his/her personal services to his/her family and community.
It should be obvious from the foregoing that the uses of international migrant remittances at the
microeconomic level can be many and varied over time. Besides, the use to which remittances
are put can be influenced by the gender of the recipient (Russell, et al. 1990). In addition, the
educational status, marital status, family size, age, level and regularity of income, and the type
of employment of the recipient are some other obvious personal characteristics that can
influence the uses of remittances. Aside these personal features, the season, the value of the
97
Stellenbosch University http://scholar.sun.ac.za
amount received and the frequency of the flow of remittances could influence the use to which
these funds are put. It is very likely that for whichever underlying motive remittances are sent,
in most average homes, consumption and loan repayment towards improved living standards
will lead the uses of remittances at the initial stages. Over time, the uses of remittances in a
typical household are expected to switch in favour of investment in education and
entrepreneurial ventures. As Gupta (2005: 4) puts it, evidence from different parts of the world
shows that “remittances are mostly used for consumption and for investment in land and
property”. Connell (1980) and Ahlburg (1991) posit that remittances are used primarily for
consumption rather than for financing investment projects because of the barriers and
inconveniences attached to private investment in most developing countries. An overriding
conclusion from survey studies conducted by Morauta (1985) and Boyd (1990) in New Papua
Guinea, Tongamoa (1987) for Tonga, Cox and Jimenez (1992) for Peru, and Ilahi and Jafarey
(1999) for Pakistan supports the view of Poirine (1997), and Brown and Poirine (2005) that
most migrants are under social obligation to remit home and that these remittances are firstly
used to settle family debts incurred in financing their trips and/or education.
Of all the consumption purposes of using migrant remittances clearly driven by altruism,
available evidence shows that food and other general living expenses constitute the largest
proportion of the uses to which remittances are put. For example, cross-sectional studies
conducted by Tongamoa (1987) in Tonga, Georges (1990) and Pessar and Grasmuck (1991)
in Dominican Republic, Loomis (1990) in Cook Islands, Hayes (1993) in New Papua Guinea,
Rensel (1993) in Fiji, Durand et al. (1996a,b) in Mexico, Dennis (2003), and Clark (2004) in
Tuvalu and Kiribati, and Miotti et al. (2010) in Africa, find that between 67 and 88 per cent of
remittances received are spent instantaneously on basic needs especially food items.
Regarding the rural southern districts of Zimbabwe, Maphosa (2005) finds that 98.8 per cent of
remittance-receiving families spend these funds primarily on food. Also, Pendleton et al. (2006)
find that household consumption represents 93 per cent of migrant remittance usage by
recipient households in the Southern African Development Community (SADC). Generally, the
food items comprise imported tinned and processed foods, beverages and tobacco
(Tongamoa, 1987; Dennis, 2003). Usually, the non-food consumables on which migrant
remittances are spent in developing countries include healthcare services, clothes, telephones,
household electronic appliances such as television sets, sound and video systems, simple tools
and equipment, and housing and construction materials (Shankman, 1976; Loomis, 1990;
James, 1991; Scott, 2003).
98
Stellenbosch University http://scholar.sun.ac.za
Investment in human capital, housing and business ventures also benefit from migrant
remittance usage (Özden and Schiff, 2005). In Tuvalu and Kiribati, for instance, Borovnik
(2003) and Clark (2004) find that education appears to be the second most important reason
for remitting after food. The usage of remittances in human capital accumulation can take many
forms namely finance of schooling, vocational training, and emigration. Özden and Schiff
(2005) observe a similar trend in the use of foreign remittances in SSA. Brown (1995) observes
that 75 per cent of Samoan migrants and 33 per cent of Tongan migrants have had their
airfares paid by a family member who migrated earlier. Shankman (1976) finds that in Western
Samoa, remittances were seldom used for capital investment projects because emigration was
regarded as a more lucrative investment than capital investment in businesses and other
entrepreneurial initiatives. In contrast, Walker and Brown (1995), and Muliaina (2001) find out
that in Tonga and Samoa, a significant proportion of remittances are used for investment in
farm inputs and simple tools and business establishments (see also Georges, 1990; Pessar
and Grasmuck, 1991; Brown and Connell, 1993; Faeamani, 1995; Taylor, 1996). In a study on
Mali, Findley and Sow (1998) also substantiated this finding by noting that remittances are not
only used for consumption purposes but also for investment into mechanisation in agriculture.
Similarly, Gubert (2000) observes that migrant remittance-receiving households in the Kayes
region of Mali do not only have higher income per capita but also these households use more
mechanised farming techniques and sophisticated farm implements than their counterparts who
have no family member resident abroad. Ahmed (2000) reveals that Somaliland remittancereceiving households use these funds mainly for financing productive activities even in periods
of harsh economic and political conditions. Similarly, for Zambia, Chilivumbu (1985) observes
that remittances are largely used to finance agricultural inputs. In the case of southern rural
districts in Zimbabwe, Maphosa (2005) discovers that migrant remittances in excess of
immediate consumption are also used to purchase agricultural farm inputs.
If remittances are meant for small and medium-scale capital investment projects they are likely
to be initially saved until the target working capital is obtained. James (1991) and Borovnik
(2003) observe that young families in Tonga save a portion of remittances received towards
future use. In LAC, a small proportion of remittances in excess of basic subsistence needs is
used for investment and business initiatives, as found by Gorges (1990) and Pessar and
Grasmuck (1991) for the Dominican Republic, and Durand et al. (1996a,b) for Mexico.
McCormick and Wahba (2001) observe that literate returnees to Egypt have a higher probability
99
Stellenbosch University http://scholar.sun.ac.za
of using their savings from abroad to become entrepreneurs, while a longer stay abroad has no
influence on a returnee migrant‟s chances of becoming an entrepreneur. Miotti et al. (2010)
point out that the general motivation to remit for investment purposes other than owning a
house in a native country is a major concern for uneducated migrants and those who have
stayed in France the longest. In contrast migrants from SSA “send money for current
expenditures rather than for investment” (Miotti et al., 2010: 17). This finding confirms the
conclusion by Mophosa (2005) on Zimbabwe and Pendleton et al. (2006) from a survey report
on SADC that, apart from consumption, remittances are used for financing transportation, fuel,
utilities, education and medical services by recipient households.
International migrants also remit to native communities and religious bodies and leaders
according to Shankman (1976), Brown (1995) and Scott (2003), whilst Walker and Brown
(1995) claim that remittance-recipient households sometimes spend these funds (often only in
excess of consumption) on payment of marriage expenses, funerals and other social, cultural
and religious ceremonies. For SSA, Diatta and Mbow (1999) indicate that besides household
consumption, remittances were used to finance development projects in migrant‟s home
communities in Senegal. Gubert (2002) explains that household members52 who fall ill during
the year are the most significant reason why Malian migrants remit and migrant remittances to
Mali increase at once when a family member dies. On the average, one death and one sick
family member induce an increase in migrant remittances by 124 per cent (ibid).
Given the above scenarios, there are no clear-cut uses of migrant remittances across recipient
families in the developing world. The implication is that there is no universal answer to the
question as to whether remittances are spent on „productive‟ or „unproductive‟ activities. For
instance, even though consumption remains the most important use of remittances in many
remittance-recipient homes as the evidence shows, the microeconomic impact of remittances
on welfare is obviously positive as far as household access to basic essential needs of life are
concerned. At the macroeconomic level, however, the composition of the consumption basket
(whether or not, remittances are spent on imported or locally produced consumables) is critical
to economic growth and stability.
52
The number of relatives and the severity of the sickness should influence the regularity and magnitude of
remittance flows.
100
Stellenbosch University http://scholar.sun.ac.za
4.3.1.2 Potential Remitters and Sources of Remittances to SSA
At the household level, there are a litany of factors directly relating to a migrant, including net
average income, educational status, economic status of the family at home, number of closely
related emigrants in a family, marital status, and job security. Other factors are gender, age, the
desire to return home in the future, duration of stay, and the strength of social ties between the
migrant and the potential recipients. These can affect not only the willingness and the ability to
remit, but also the amount and the regularity of the flow of remittances. Additional factors that
can influence the willingness and the ability as well as the magnitude and the regularity of
remittance flows are migrant co-habitation with spouse, children and parents, number of
younger siblings at home, motive for migration, future supportive expectations from family and
migrant‟s legal status in host country. Durand et al. (1996a,b) identify labour market experience
in the host country, homeownership status, access to capital, duration of trip and cost of
migration as other factors that can play an important role in this respect.
Migrants on regular income and those with an accumulated stock of wealth are easily the most
likely to remit home since remittances are a portion of a migrant‟s earnings transferred to meet
a specific purpose in his/her native country. In this regard, a migrant with no accumulated
wealth and who is a full-time student and, therefore, does not engage in any income-generating
activity cannot be seen as a reliable potential remitter at the contemporary time. This is
consistent with the definition of migrant remittances discussed in Chapter Two. However, there
is no theoretical consensus on the question as to whether the regularity in the flow of migrant
remittances is dependent upon the income level of the migrant even if the taxonomy provided
by Wahba (1991)53 is taken into account. The reason is that where altruism strongly dominates
as the motive for remitting a migrant may keep remitting home irrespective of the level of
his/her disposable income, even though under this circumstance income level is likely to be
positively related to the amount of funds remitted. Blom and Henriksen (2008) find out that in
Norway, Somali immigrants are by far the most regular remitters although they are in a weaker
financial condition than other immigrants.
53
In this taxanomy, Wahba (1991) argues that: (i) the maximum transferrable income of a migrant at any given time
is the excess income available to a migrant after meeting his/her basic needs in the host country; (ii) fixed
remittances are the minimum funds a migrant is required to transfer to enable his/her family back home to meet the
basic needs and other contractual obligations; (iii) discretionary remittances are transfers in excess of fixed
remittances, which together with fixed remittances represent the level of actual remittances. Discretionary
remittances vary according to financial risks and rewards on savings and investment in home and host countries;
and (iv) saved remittances are the difference between potential remittances and the amount remitted during a given
period.
101
Stellenbosch University http://scholar.sun.ac.za
Probably, the most important factor that influences the decision of migrants to remit is related to
marital status and number of children left behind in migrants‟ home country. For instance, it
should be very easy to find that married migrants whose spouses and children are living with
them abroad are less likely to remit home than their counterparts who leave their spouses and
children in their home countries. The reason is simple and straightforward: migrants who are
motivated to remit home in support of their spouses, children and parents will remit larger
amounts and more regularly when these direct dependants are staying behind. And as can be
expected these migrants will be less inclined to remit more and more regularly when their
spouses, children and parents who were previously staying behind now unite with them in the
host country. Also, all other things being equal, migrants who stay with their spouses, children
and parents abroad are less likely to remit and to remit larger amounts on regular basis,
because such migrants often spend a higher proportion of their disposable incomes on
essential basic necessities of life, with very little left over for savings and remittance-related
investment.
There is evidence that remittances tend to decrease in value with duration of migrant stay
(Lucas and Stark, 1985). In the case of Germany, migrants who intend to stay abroad longer
remit less Merkle and Zimmermann (1992). Similarly, among the migrants from Pacific Islands,
those who intend to return home remit more (Brown, 1997) in consistency with the observation
by Glytsos (1997) that Greek temporary migrants remit more than Greek permanent migrants.
If the conclusion drawn from the various studies that remittances per migrant decrease over
time is anything to go by, then the less educated migrants are more likely remit than the more
educated migrants, notwithstanding the finding by Rodriguez and Horton (1995) that the level
of education of migrants has no effect on the amount of migrant remittances. This is because
various research reports by Borjas (1989), Knerr (1994), and Reagan and Olsen (2000)
suggest that, generally, the higher the level of education of a migrant, the higher the duration of
stay abroad. Earlier reports by Johnson and Whitelaw (1974), and Rempel and Lobdell (1978),
however, suggest that remittances tend to increase with the level of formal education and skills
of migrants (cf. Faini, 2006b).
In Somalia, Lindley (2007: 12) found that women are better remitters, especially in terms of
reliability and consistency than men, to the extent that, “…it is better to have one daughter
abroad than ten sons”, notwithstanding that the empirical evidence shows that Somali men
resident abroad remit higher amounts than their women do. Thus, on the average, in Somali
102
Stellenbosch University http://scholar.sun.ac.za
communities, men who remit home, remit higher amounts, but there is a higher probability that
a Somali female migrant will remit home more than what her male counterpart will. This finding
supports the observation by Ponsel (2001) on South Africa that women migrants remit larger
proportions of their net earnings although men generally remit larger amounts than women.
Exceptions to this finding can be found in studies by Crãciun (2006) and Osaki (2003) in the
case of Moldova and Thailand respectively.
Bollard, McKenzie and Morton (2010), in a survey study involving a sample of 12,875 SSA and
North African migrants in nine Organisation for Economic Co-operation and Development
(OECD) countries conclude that:
i.
African migrants remit twice more (an average of US$1,263) than migrants from other
developing countries (an average of US$668), with migrants from poorer African
countries more likely to remit than those from richer African countries.54
ii.
Among migrants with spouses left behind at home, male migrants are more likely to
remit and actually remit more (an average of 42 per cent) than female migrants (with an
average of 26 per cent probability to remit). Men with spouses left behind remit an
average of US$3,879 per annum more than their female counterparts.
iii.
Migrants with higher education remit more than migrants with less education. This
pattern is stronger among African migrants compared with non-African migrants.
Bollard, McKenzie, Morton and Rapoport (2009) found a similar result. This finding
could, however, be explained by the fact that migrants with a higher level of education
are more likely to find jobs and earn higher incomes. Notwithstanding this, it must be
borne in mind that most migrants from developing countries seldom find jobs that are
directly related to their level of education in the industrialised countries where these
migrants are hosted.
iv.
In relative terms, high income earners remit more than low income earners, but this
relationship is quite flat over middle ranges of income and steeper at the tails of the
distribution frequency. On the average, an increase in a migrant‟s income by 10 per
cent is associated with an additional US$110 remitted per annum.
v.
There is no strong evidence that remittances decrease with time spent abroad in
relation to the policy debate as to whether or not the episode of temporary or permanent
54
Given that this finding clearly contradicts the facts presented in Chapter Three that SSA is the least recipient of
migrant remittances in per capita and in absolute terms, it is possible either that most Africa migrants remit via
unofficial channels, or that there are relatively fewer African migrants in OECD countries (as shown in Table A4.3),
or that African migrants in non-OECD countries do not remit regularly.
103
Stellenbosch University http://scholar.sun.ac.za
migration impacts on remittance inflows to migrant-home countries. It was observed that
the likelihood to remit and the amount remitted increased over the first five years to ten
years of migration. The likelihood to remit begins to fall only after 20 years, but at this
stage the total amount remitted does not fall since migrants with longer years of stay
abroad, remit more on the few occasions that they do remit.
vi.
Out of the over 12,000 respondents, only a third remit, and with those who remit
sending an average of US$2,638 annually (with a medium of US$1,088), “an amount far
exceeding SSA‟s per capita GNI of US$1,082” (op. cit. 2010: 5).
vii.
Only a few migrants remit amounts worth US$5,000 or more per annum. This is five per
cent to SSA and nine per cent to North Africa. The majority of African migrants remit
US$500 or less in a year.
viii.
Migrants with excellent legal status are approximately 12 per cent more likely to remit
than those with illegal status. “This may reflect differences in access to formal financial
institutions such as banks, between legal and illegal migrants” (ibid. 2010: 17). This
finding is consistent with the finding of Konica and Filer (2005) on Albanian emigrants
who remit less if they do not have the appropriate documents.
Ratha and Shaw (2007) observe that international migration in SSA occurs predominantly
within the sub-region, but the major source of remittance flows to the sub-region are
industrialised countries outside the sub-region. Various World Bank reports on remittances
corroborate this fact by noting that about 75 per cent of all remittances received in SSA are
sent from the USA and Western Europe. Major migrant-host countries in Western Europe are
former colonial powers such as Great Britain, France, Netherlands, Portugal and Spain as well
as Germany (see Table 4.1). The obvious reasons are that these countries have the largest
economies in Europe, common language, or close historical and political ties with most SSA
countries. Bollard, McKenzie and Morton (2010) also identify OECD countries as the main
source of remittances received by developing countries.
4.3.1.3 Potential Receivers of Migrant Remittances
Frontline potential recipients of migrant remittances are low-income households55 with an
economically active adult nuclear family member resident and working in a foreign high-income
country. Other potential recipients of migrant remittances are low-income communities and
55
Itzigsohn (1995) in comparative analysis of four countries finds that household income had a positive effect on
remittances received in Guatemala, negative effect in the Dominican Republic but no effect in Haiti and Jamaica.
104
Stellenbosch University http://scholar.sun.ac.za
countries with a relatively large segment of their active adult citizens working in a relatively
developed country. Among these households and communities, those that practise communal
inheritance and give prestigious awards or respectable social titles to their members/citizens
are more likely to receive higher remittances than those that do not. Moreover, in a typical
household with a migrant abroad, spouses, children of school going age, economically inactive
parents, and younger siblings are, in descending order of importance, the most likely to receive
the highest and most regular remittances. Lucas and Stark (1985) report evidence for this
hypothesis in Botswana where male migrants who are prospective heirs, competitively remit
home in order to receive favour, respect and inheritance from their family and community back
home. This suggests the value of assets such as land, plantation and animals owned by a
household or a community is a positive determinant of migrant remittance inflows. Similarly, in
societies where assumption of a chieftaincy reign is directly linked to donations and exhibits of
wealth and generosity, migrants are more likely to remit, remit larger amounts and remit
regularly. Osili (2007), however, argues that the relationship between household assets and
remittances received may not necessarily be direct or simple as it is possible that the current
assets owned by remittance-receiving households might be financed by previous remittances
received.
Notwithstanding the above it must be pointed out that although „migrant-exporting‟
households/communities are the main recipients of migrant remittances, whether driven by
altruism or otherwise, the amount and the regularity in the flow of remittances received is highly
contingent on the social ties between the remittance-sending migrant and the target recipient
as well as the economic conditions at home. Conceivably, social ties between a migrant and
his/her target recipients of remittances can be positively influenced by the premium the migrant
places on the type of family and/or social support the migrant received prior to his/her
migration. Lucas and Stark (1985) and Pleitez-Chavez (2004) upon studying the pattern of
remittance inflows in Botswana and El Salvador respectively, conclude that negative income
shocks in native countries significantly increased remittances received from relatives abroad.
However, Lozano-Ascencio (1993) concludes from a study on Mexican migrants in the USA
that remittances often decrease after the first or second generation of migrants, so that the
stability in the flow of remittances is mainly sustained by a new generation of migrants.
Although, social ties between a migrant and his/her family at home is the nucleus of altruism, to
a very large extent, migrants who remit for self-interest motives are also indirectly influenced by
105
Stellenbosch University http://scholar.sun.ac.za
social ties. For example, the stronger the social tie between a migrant and his/her family at
home, all other factors being equal, the more likely the migrant will be willing to return home in
the future. Naturally, migrants who plan to return home in the future are those who will be more
concerned about economic opportunities at home and consequently, more motivated to initiate
income-generating projects in their home countries, hence more likely to remit regularly.
4.3.2 Theoretical Review of Macroeconomic Determinants of Remittances
There are no distinctive macroeconomic theories underlying the flow of remittances.
Macroeconomic models on the determination of remittances are, thus, based directly on the
microeconomic theories of altruism and self-interest. Russell (1986a,b), Elbadawi and Rocha
(1992), Lianos (1997), Bouhga-Hagbe (2004), Vargas-Silva and Huang (2006), and Coulibaly
(2009) argue from the theoretical viewpoint that macroeconomic factors can play an influential
role in the determination of international remittance flows. From the perspective of altruism at
the macro level, remittances are higher when negative shocks associated with higher rates of
underemployment and unemployment occur in the migrant‟s native country as the desperate
macroeconomic conditions compel active labour to travel abroad in search of greener
pastures56. In this context of pure altruism, lower growth in real income (or economic
recession), higher rate of inflation, bad governance and weak institutions, exchange rate
instability and limited access to private sector credit in migrant-home countries stimulate a
higher inflow of migrant remittances (Wahba, 1991; Rapoport and Docquier, 2006; Vargas-Silva
and Huang, 2006; Coulibaly 2009).
Generally, the level of economic activities in the migrant‟s resident country is important
because improved economic conditions in the host country boost the ability of migrants to
increase their employment and earnings prospects, which puts them in a better position to be
able to remit more. For example, a migrant who is willing to remit for whichever purpose, but
was unable to remit in the past because he/she was unemployed should have a higher
propensity to remit once he/she is gainfully employed. Specifically, however, pure altruistic
theory and pure self-interest investment theory predict different impacts of relative
improvements in host-country income level on remittances received in migrant-home countries.
According to the theory of altruism, when the real income of the migrant-host country improves
56
This is in line with the Lewis (1954) theory of excess supply of labour, resulting in high unemployment and low
wages in underdeveloped economies, which forces the nationals of these underdeveloped economies to migrate to
the industrialised world where there is a higher prospect of being engaged in relatively higher income jobs.
106
Stellenbosch University http://scholar.sun.ac.za
relative to that of his/her home country, the migrant is motivated to remit more because he/she
sees his/her family back home as being worse off (Lucas and Stark, 1985; Rapoport and
Docquier, 2006; Vargas-Silva, 2006). Thus, host-country income positively determines
remittance inflows in the migrant-home country. In contrast, self-interest investment theory
contends that as the real income level of the migrant-host country improves relative to that of
the migrant-home country worsening the North-South income gap, migrants remit less because
with improved income conditions abroad, it is more lucrative to invest in the host country than
at home (Lucas and Stark, 1985; Vargas-Silva, 2006; Coulibaly, 2009). Also, conceivably
because the incentive to return home in the future reduces as the economic conditions in
migrant-host country improve relative to the migrant-home country conditions, there is a higher
tendency that remittances from permanent migrants may fall. However, because this widening
income gap encourages further South-North migration the newly arrived migrants who often
have stronger social ties with their families back home during the initial stages of migration are
likely to keep the tempo of remittances active.
In accordance with the altruistic theory, remittances received in migrant-home countries fall
when, in real terms, the income level of developing countries improves to narrow the income
gap between the migrant-home country and the migrant-host country (Lucas and Stark, 1985;
Rapoport and Docquier, 2006; Vargas-Silva and Huang, 2006). The decline in remittances in
response to a relatively improved economic condition of the target recipients is underscored by
a reduction in the pressure on migrants to remit home to lessen income constraints faced by
their direct dependants. The altruistic theorists, thus, argue that during periods of economic
recession, as the real income level of the home country declines, migrants are compelled to
increase remittances in a bid to mitigate the adverse effects of the negative economic shocks at
home (Swamy, 1981; Brown, 1997, Vargas-Silva and Huang, 2006; Coulibaly 2009). With
regard to home-country economic conditions and self-interest theory of remittances, the
microeconomic theory can be transformed and directly related into a portfolio choice theory at
the macro level. The portfolio choice theory implies that as economic conditions in migranthome countries improve relative to the rest of the world, more remittances are received in the
home country through higher savings and investment by migrants (Russell, 1986a,b; Wahba,
1991; Coulibaly, 2009). Higher real average income growth in the migrant-home country signals
improved economic conditions and bigger potential markets which are required for increased
private investment and the emergence of a vibrant entrepreneurial society. Consequently, self-
107
Stellenbosch University http://scholar.sun.ac.za
interest investment-oriented migrants tend to remit more for business or investment purposes
when the potential market size in their home countries expands.
Return on financial assets having accounted for inflation is usually proxied by real deposit
interest rates (Elbadawi and Rocha, 1992). According to the self-interest theory of remittances,
increases in the real return on investment and savings in a migrant-home country relative to the
migrant-host country induce higher inflows of investment-driven remittances in the migranthome country (Schiopu and Siegfried, 2006; Vargas-Silva and Huang, 2006; Coulibaly, 2009).
This is because if, for example, real deposit interest rates are relatively more attractive in
migrant-home countries than migrant-host countries, non-altruistic migrants (especially those
with an intention of returning home in the future) may increase their marginal propensities to
save and invest at home in a bid to augment their expected lifetime utility at home. Therefore, a
migrant-home country that creates a stable macroeconomic environment and an unrestricted
opportunity for earning a higher return on domestic financial assets relative to the rest of the
world can attract higher inflows of remittances, most likely through the formal financial system
when, for example, the expected returns on portfolio assets in the home-country equity market
become relatively more attractive. The pure altruistic theory of remittances does not predict any
relationship between remittance inflows and real deposit interest rate57 in the migrant-home
country suggesting that remittances may flow to migrant-home countries irrespective of the
level of deposit interest rate.
Real exchange rate can have an ambiguous effect on remittance inflows depending upon
whether or not it is a fixed amount in the home-country denominated currency or host-country
denominated currency that is remitted by a migrant, irrespective of the underlying motive for
remitting. On the one hand, real exchange rate can positively influence remittance inflows as
depreciation of the home-country currency against the host-country currency raises the
purchasing power of the foreign currencies remitted, thereby making remittances more
economically valuable in the home country as far as locally produced goods and services are
concerned. An altruistic-oriented migrant who remits an equivalent fixed amount of local
57
Altruistic theory seems more concerned with real lending rates which are considered as the cost of borrowing
funds in the migrant-home country. The lower the cost of borrowing from financial institutions, the lower the expected
amount of remittances received and vice versa, as altruistic migrants become less worried about the cost of living
and constraints to financing entrepreneurial ventures at home. Because data on lending rates in most of the sampled
SSA countries are at best incomplete; and consistent with most previous studies (see Table A4.1) and with the
Vargas-Silva and Huang (2006) theoretical framework adopted by this study, only real deposit interest rates were
used in the empirical analysis.
108
Stellenbosch University http://scholar.sun.ac.za
currency to his/her dependants58, can under this circumstance, decide to remit less in terms of
host-country currency but which in the eyes of the ordinary recipients represents the same
amount in nominal terms. Indeed, where these remittances are spent on domestically produced
goods, the purchasing power of the amount received remains unchanged when inflation and all
other related factors remain constant. However, in an import-dependent migrant-sending
country or where the purpose of remittances is to purchase some essential necessities such as
imported medication, under the same conditions, an altruistic migrant may be compelled to
remit more foreign currency to enable recipients of remittances to fulfil the “life-smoothing”
purpose behind the sending of the remittances.
Likewise, migrants who remit for non-altruistic motives may either respond positively or
negatively to changes in real exchange rate depending upon their level of financial literacy and
their specific objective of remitting. For example, a rational migrant with a reasonable level of
financial literacy who remits an equivalent local currency fixed amount for the payment of
wages and rent at home will hedge by remitting more59 to cover a reasonable given period of
time when real exchange rate of the home-country currency depreciates against the currency of
his/her host country. Ceteris paribus, when the home-country currency depreciates, migrants
who remit for the purposes of investment in locally denominated financial assets will have a
higher incentive for sending more remittances for investment purposes (say the purchase of
land or listed shares), but as domestic currency appreciates, the incentive for investment at
home, say the purchase of stocks, falls. Here, non-altruistic migrants see remittances as more
profitable at home due to home-country currency depreciation. Under the same circumstances,
a migrant who does not have the capacity to hedge will remit less when his/her home-country
currency depreciates against his/her host-country currency as less foreign currency may now
be required to meet the same fixed expenditure budget denominated in the home-country
currency. The migrant, in this case, will have to remit more in the event of real exchange rate
appreciation in order to meet the same expenditure budget in his/her home country.
Investment-oriented migrants who consider stronger domestic currency as improvement in
macroeconomic conditions at home and, hence, have higher confidence in the economy of the
home country may be inclined to remit more when the home-country currency appreciates, and
58
Equivalent local currency denominated fixed value remittances can be easily applicable in situations where
migrants remit to their non-working spouses and children who attend school at home.
59
Here, although the total amount remitted might appear higher even in the foreign currency, in actual fact, the
migrant has remitted less in terms of the host-country currency, if the normal average value of remittances per
transaction were to be taken into consideration.
109
Stellenbosch University http://scholar.sun.ac.za
as the domestic currency depreciates, the amount of foreign currency denominated remittances
are reduced. Finally, Gupta (2005) argues that, in general, there is a high tendency for officially
channelled remittances to increase when home-country currency depreciates in developing
countries where dual exchange rate exists, as depreciation of the home-country currency raises
the official rate to comparable levels with the black market rate, or even raises migrants‟
expectations of a future appreciation.
The impact of home-country rate of inflation on remittances received is not straightforward and
it is dependent upon which of the main motives behind remittance inflows dominates the other.
Generally, higher inflation is analogous to harsh and unstable macroeconomic conditions in a
country, therefore, given the rate of unemployment, the incentive for international migration
increases. Increased migrant stock, all other factors remaining equal, increases total migrant
remittances received from abroad as revealed by Elbadawi and Rocha (1992), Liano (1997),
and Freund and Spatafora (2005) inter alia. Besides, the altruistic motive of remittances
predicts that remittances meant to lessen constraints to consumption and general living
expenses of recipients will increase when the cost of living soars in the migrant‟s home country
(Lucas and Stark, 1985; Rapoport and Docquier, 2006). As the rate of inflation in the migrant‟s
home country falls, the altruistic theory predicts a decline in the amount of remittances received
because migrants now regard the economic condition of target recipients as less deplorable.
According to the self-interest investment theory, higher rates of inflation and higher price
uncertainty are deterrent to remittance inflows as migrants anticipate a lower rate of return on
investment alongside higher investment risk under inflationary conditions at home (Schiopu and
Siegfried, 2006). Nevertheless, this will depend upon the prevailing price level and the rate of
inflation as a moderate rate of inflation at manageable levels in SSA could signal the
opportunity for making higher profits at home, which subsequently induces investment-oriented
SSA migrants to remit more. Therefore, where altruism dominates self-interest investment
motive, a positive effect of inflation on remittances received is expected because higher
inflation erodes the purchasing power of the target recipients. Be that as it may, where
investment motive dominates, inflation is expected to impact negatively on remittance inflows
because of higher investment risk at home. In the very long run, however, it is expected that
lower rates of inflation should impact positively on remittance inflows as migrants who remit for
purely altruistic motives become more investment-oriented at home due to improved
macroeconomic conditions and, hence, reallocate remittances in favour of investment at home
even as altruistic remittances fall.
110
Stellenbosch University http://scholar.sun.ac.za
According to the theory of altruism, when there is restricted private sector access to quality
credit in migrant-home countries, migrant remittances increase to ameliorate liquidity
constraints in the migrant-home country (Schrooten, 2005; Giuliano and Ruiz-Arranz, 2009).
Altruistic theory, thus, predicts that when private sector access to competitive credit improves in
migrant-home countries, remittances received decline not only because the pressure on
migrants to remit to offset limited access to credit reduces, but also because the motivation for
international migration in search of working capital to finance business falls. In contrast, the
self-interest investment theory envisages higher inflows of remittances when there is less
restricted private sector access to credit and working capital in migrant-home countries. This is
because reduction in liquidity constraint is regarded as an improved development of the
financial sector which is required for private sector participation in economic activities at home.
Russell (1986a,b) and Funkhouser (1995) argue that political risk factors in migrant-home
countries can determine the inflow of migrant remittances. The impact of political risk in a home
country on remittances received is dependent upon the motive behind the remittances. Whilst it
may be positive or zero when remittances are driven by altruism, the impact of political risk is
expected to be negative when the underlying motive is self-interest. Thus, institutional quality
which embodies democratic governance and geopolitical conditions of the migrant-home
country is expected to have an ambiguous effect on international remittance inflows. Going by
the theory of altruism, remittances, in the short run, are expected to respond negatively to
higher quality home-country institutions for two main reasons: (i) poor institutions and bad
governance at home encourage higher international emigration, hence higher remittance
inflows, as is the case in Somalia; and, (ii) when institutions become weaker, more remittances
are expected to be received as international migrants become more sympathetic towards their
relatives at home. In the long run, quality institutions at home can impact negatively on
remittances received when more migrants (most likely temporary migrants) return home in
response to improved political conditions in the home country. Furthermore, the self-interest
investment theory of remittances, predicts a positive effect of institutional quality on remittance
inflows because financial assets such as bank deposits, stocks and real estate are adversely
affected by the risks associated with investment return and geopolitics. For example, a rise in
political risk and uncertainty in a migrant-home country which adversely affects its credit rating
and economic stability also deters the inflow of investment-seeking remittances as migrants
111
Stellenbosch University http://scholar.sun.ac.za
become pessimistic about investment security at home. In this respect, quality institutions in the
country enhance higher remittance inflows.
The McKinnon (1973) and Shaw (1973) theory of financial liberalisation postulates that
deregulation and abandonment of repressive financial policies breed competition and efficiency
in the financial market, thereby enabling financial institutions to pay attractive returns on
deposits. Through the pursuit of financial liberalisation policies, therefore, financial institutions
are encouraged to develop cost-saving strategies and innovative products for resource
mobilisation from domestic and external sources. Consistent with the McKinnon-Shaw theory,
Alberola and Salvado (2006), Ghosh (2006), Suki (2007) and Singer (2010) assert that the
implementation of restrictive economic policies such as exchange rate restrictions in migranthome countries do not attract
a higher inflow of international remittances. Conversely, a
liberalised financial sector and improved financial development in migrant-home countries are
imperative if a country is to attract higher remittances from its migrants through official
channels.
It is, however, important to stress that, at the macro level, migrant remittances are often driven
by mixed motives in so far as the altruistic and self-interest theories are not mutually exclusive.
Furthermore, altruism underlies all kinds of remittances. This might be the main reason why in
macro-level studies on the determinants of remittance inflows (see Table A4.1) analysts do not
often attribute their findings strictly to the validity of any particular remittance theory.
4.3.3 Empirical Review of Macroeconomic Determinants of Remittances
The motivation for providing empirical evidence has increased since Lucas and Stark (1985)
formally initiated60 the debate on the determinants of remittances. Though the motives behind
remitting might differ across time, households and countries, it is generally believed that
improvement in migrant income and negative shocks in the migrant-home country have a direct
relationship with remittances. For instance, with respect to home-country‟s economic
performance many studies, including those of El-Sakka and McNabb (1999), de la Brière et al.
(2002), Bouhga-Hagbe (2006), Yang and Choi (2007), and Singh et al. (2010) provide evidence
on the countercyclical property of remittances. In sharp contrast, Aydaş et al. (2004) and
Higgins et al. (2004) conclude that remittances exhibit pro-cyclical behaviour as they tend to
60
Prior to the contribution of Lucas and Stark in 1985, Johnson and Whitelaw (1974) argued that the incentive
behind migrant remittances is income disparity between the resident countries of migrants and their home countries.
112
Stellenbosch University http://scholar.sun.ac.za
rise with improvements in per capita income and the growth rate of remittance-receiving
economies. Sayan (2006) also observes that in most cases remittances tend to be acyclical or
pro-cyclical.
Russell (1986a), and Chipeta and Kachaka (2005) reveal that the decision to remit depends on
different factors over the business cycle rather than on the altruistic motive of smoothing
consumption of recipients. In particular, Chipeta and Kachaka (2005) reaffirm an earlier
observation by Russell (1986a,b) and Russell et al. (1990) that, from a macroeconomic
perspective, the inflow of remittances depends on deposit interest rate differentials of the home
country and the main host country of migrants, the rate of inflation, political atmosphere, the
level of economic activity in the host country and the exchange rate in the home country.
El-Sakka and McNabb (1999), in an attempt to explain remittances received by Egypt, included
real income levels of the sending and receiving countries, interest rate differentials, rate of
inflation in Egypt, and the black market premium for foreign exchange as regressors in a single
equation following the Ordinary Least Squares (OLS) procedure. The empirical results show
that whereas remittances increase with the Egyptian rate of inflation and income abroad, they
decline with the black market premium. Bouhga-Hagbe (2004) analyses workers‟ remittance
flows to Morocco using cointegrating and error-correction models. Bouhga-Hagbe finds that,
consistent with the altruistic theory, remittance inflows are, in the long run, positively correlated
with wage levels in the source country proxied by wage levels in France whilst they negatively
correlate with real GDP growth in Morocco.
Empirical literature suggests that the number of migrant workers outside the home country61,
differences in wage rates at home and abroad, economic condition in the migrant-native
country, exchange rate fluctuations, interest rates, political risk, facilities or mechanisms of
international money transfer and the economic conditions in the country of residence influence
remittance flows (see Table A4.1). With respect to official flow of remittances, the level of
financial development as reflected in the cost of funds transfer, existence or absence of dual
exchange rate, and the availability of financial infrastructure and innovative products in migranthome countries are also important (Orozco, 2002; 2003; Ratha, 2003; Gupta, 2005; Terry and
61
Freund and Spatafora (2005) find that a 100 per cent rise in migrant stock causes a 75 per cent rise in remittance
inflows, but Elbadawi and Rocha (1992), and Aydaş et al. (2004) observe that the importance of migrant stock in
determining remittance inflows declines over time as a result of ageing labour force.
113
Stellenbosch University http://scholar.sun.ac.za
Wilson, 2005). For instance, Hadjimichael et al. (1998) and Pinto et al. (2000) conclude from
various empirical studies that financial development and economic growth in migrant-home
countries are crucial positive determinants of remittance inflows. With particular reference to
financial infrastructure, Wahba (1991) concludes that financial institutions play a crucial role in
attracting higher remittances through the formal money transfer channels. And as found by
Orozco (2002), and Freund and Spatafora (2005), migrant-home countries that have relatively
developed financial systems as reflected in reduced transaction costs on remittances, attract
more remittances because the propensity to remit increases at lower money transfer cost.
Jadhav (2003) analyses the determinants of workers‟ remittance inflows in India using a loglinear regression specification involving oil prices as an indicator for level of economic activities
in the Middle East and the Gulf region considered as the hosts of Indian migrants, US GDP as
proxy for economic activities for non-oil India migrant hosts, interest rate differentials measured
as the difference between nominal domestic interest rate and LIBOR, and exchange rate
depreciation as explanatory variables. The estimated results show that whereas interest rate
differentials do not affect remittance flows to India, the level of economic activities in both
categories of migrant-host countries and exchange rate depreciation positively impact on
remittance flows to India. In a similar fashion, Gupta (2005), in an attempt to analyse a more
complete model to unearth the determinants of remittances in India, included trend, number of
migrants, changes in country rating, and return on domestic stock market. The findings lend
strong support for altruism as the main determinant of remittances. It was found out that
migrant stock, migrant earnings, economic environment in migrant resident country, and Indian
drought dummy variable each has positive impact on the cyclical component of remittances in
India. Gupta (2005) did not find a statistically significant impact of political uncertainty, interest
rates, and exchange rate depreciation on the flow of remittances to India.
Elbadawi and Rocha (1992) and Aydaş et al. (2004) find that migrant stock loses its importance
as a determinant of remittances over time due to the ageing of the labour force. Therefore, in
relative terms, it is not countries with the largest Diaspora population that attract the most
remittances, but rather the countries where migrants are more sensitive to the economic
conditions at home (as is the case for small island countries such as Cape Verde, Comoros
and Cook Islands inter alia) and developing countries that are closer to industrialised countries
(as is the case for Lesotho and Mexico) that attract more remittances (see also Buch et al.,
2002 for details). Possible reasons that can be attributed to these findings are closer family ties
114
Stellenbosch University http://scholar.sun.ac.za
between migrants and the family back home due to proximity resulting in frequent home visits
by migrants, and the generally small geographical size of island countries.
From a more general perspective, World Bank (2006a) underscores the fact that government
policies clearly affect remittance inflows. The World Bank (2006a: 93) concludes that “in the
remittance-receiving countries, these policies include tax exemptions for remittance income;
improved access to banking services by recipients; incentives to attract investments by the
Diaspora; access to foreign exchange or lower duties on imports; support for the projects of
migrant associations; and help for migrants in accessing financial systems. In the remittancesource countries, they include policies affecting access to banks, access to foreign exchange,
support to migrant groups, types of immigration regimes, and co-operation with receiving
countries”.
In summary, empirical results from various macroeconomic studies62 on remittance inflows
reveal that at the initial stage of migration, remittances are likely to be: (i) countercyclical in so
far as they increase during economic downturns in recipient countries; (ii) driven more by an
altruistic motive than by an investment motive; (iii) stimulated by life-sustaining motives, for
which reason they are more for transactions motive (consumption) than for investment motive;
and (iv) relatively insensitive to interest rate differentials between home and abroad. At the later
stages of migration when the self-interest investment motive is more likely to emerge stronger
than altruism, remittances flow pro-cyclically; or acyclically, because altruism and self-interest
are of equal importance to the remitting migrant. Other macro variables that have been of
empirical relevance to remittance flows to developing countries include the rate of inflation as a
measure of financial instability in the home country, private sector access to bank credit, and
exchange rates.
4.4 THEORETICAL FRAMEWORK
Using a two-period scenario, Vargas-Silva and Huang (2006) analyse the flow of remittances to
developing countries under the assumptions that period one represents an initial stage of
international migration of an individual, typically from a less developed country (the migranthome country) to a more developed country (the migrant-host country). During this period, the
individual (the migrant) does not migrate with his/her direct dependants (family). Thus, the
economically active migrant resides in a relatively industrialised country where he/she is
62
See Table A4.1 in the Appendix for the review of other empirical studies on the determinants of remittances.
115
Stellenbosch University http://scholar.sun.ac.za
engaged in an income-generating economic activity whilst his/her dependants continue to live
in his/her low-income native (home) country. In this context, the utility of the migrant depends
on his/her personal consumption in the host country (C f ) and the consumption of his/her
family (Ch ) living in his/her native country. Accordingly, Vargas-Silva and Huang (2006) specify
that the utility function of the migrant in the initial period of migration is given as U (C f , Ch ) with
U1 0 , U11 0, U 2 0 , U 22 0; and under the assumption that utility is additively separable63.
The consumption expenditure of the migrant‟s household living in his/her home country is
dependent upon the income and remittances received ( r ) with which is the cost of
transferring funds from the host country to the home country being of the form, 1 . This
implies that although a migrant remits
r amount of money only a fraction of this amount r is
received by his/her family.
The income received by the migrant‟s household living at home is made up of two components
and given as Yh Yh with capturing the relationship between the economic conditions in the
migrant‟s native country and the average income earned by his/her family living at home. Yh is
the fraction of the household income that is not susceptible to changes in the macroeconomic
conditions of the home country, whilst
Yh is that part of the household income that is
predisposed to changes in the macroeconomic environment of the home country. It is assumed
that 0 which implies that an improvement in the economic conditions of the home country is
generally associated with an improvement in the household‟s (the family left behind‟s) income,
even though the magnitude of may differ across households. The consumption of the
migrant‟s household living at home is given by Ch ((Yh Yh ), r ) . This consumption function is
additively separable with Ch1 0, Ch2 0, Ch11 0 and Ch22 0. Likewise, the income of the
migrant is in the form y f Y f such that reflects the relationship between the economic
conditions in the host country and the income the migrant earns in the host country. Here
again, y f is that portion of the migrant‟s disposable income in the initial stages of migration
that is not susceptible to varying macroeconomic conditions of the host country. Similarly, Y f
is the portion of household income that is susceptible to changes in the economic condition of
63
In this case U1 is the derivative of utility with respect to home-country consumption.
116
Stellenbosch University http://scholar.sun.ac.za
the host country (Y f ). The income constraint of the migrant during this first or initial period of
migration can, thus be given as y f Y f C f r s , in which case 0 with r being the
proportion of the migrant‟s disposable income which he/she remits home, whilst s represents
the fraction of migrant‟s income saved in the home country.
During the second period (i.e. at the later stages) of migration, the migrant‟s household
migrates to a foreign industrialised country, often joining the migrant in his/her host country 64. If
this assumption holds, then the migrant‟s optimisation problem is given as follows:
MaxU (C f , Ch ) V (Cz )
(4.1)
{C , r , s}
subject to
y f Yf C f r s
(4.2)
Cz yz Yz (1 i)s
(4.3)
and
where V (Cz ) denotes the utility from second-period consumption so that V1 0 and V11 0 ,
is a discount factor, i is the interest rate (intuitively the deposit interest rate) of the home
country, with y z and Yz having similar interpretations as y f and Y f but for the second period.
By finding the first-order conditions of this problem, Vargas-Silva and Huang (2006) obtained
Equations (4.4) and (4.5) below:
U1 V1 (1 i)
(4.4)
U 2Ch V1 (1 i)
(4.5)
r
From Equations (4.4) and (4.5), it is possible to derive
r with respect to host-country income
(Y f ) as shown in Equation (4.6) below:
U11V11 (1 i )2
r
0
Y f
D
(4.6)
64
According to Vargas-Silva and Huang (2006: 86), “similar results can be obtained assuming that, in the second
period, the emigrant returns to the home country and joins the household”.
117
Stellenbosch University http://scholar.sun.ac.za
where
D,
the
determinant
of
the
matrix
of
the
second
derivatives,
D U11 V11 (1 i)2 2 V11 (1 i)2 U 22Chr U 2Chrr U11 2 ((Chr )2U 22 Chrr U 2 ) 0
for
is
a
maximum (Vargas-Silva and Huang, 2006). The central implication of Equation (4.6) is that an
improvement in the economic conditions of the host country positively affects remittance flows
from the migrant-host country to the migrant-home country. This is so because an increase in Y
implies improvements in economic conditions in the migrant-host country which enables a
migrant to send more money home as 0; given that households spend their incomes on
normal goods. In Equation (4.7), it is also shown that an improvement in the economic
conditions of the migrant-home country is associated with a decrease in remittance inflows in
the remittance-receiving (migrant‟s home) country.
U 22Ch Ch U11 V11 (1 i )2
r
r
y
h
( )
0
Yh
D
(4.7)
Impliedly, Equation (4.7) is non-positive when a migrant is remitting for altruistic purposes.
Under this assumption, the migrant remits less funds to his/her family in the home country
because the target household is better off, 0. To conclude, Vargas-Silva and Huang (2006)
prove that changes in remittances as a result of changes in the rate of interest in the host
country could have two contrasting effects for which reason the sign of Equation (4.8) is
indeterminate unless further assumptions are made.
r U11 V1 V11 (1 i ) s
/0
i
D
(4.8)
Thus, from Equation (4.8), it is evident that, on the one hand, if there is a higher real rate of
interest on deposits in the host country a rational migrant who is driven by a self-interest
investment motive, will reduce the amount of funds remitted to his/her home country and
increase his/her savings in the host country. On the other hand, if the real deposit interest rate
in the host country increases, a migrant can now consume more in the future and since
remittances form part of the consumption basket of the consumer, funds remitted home during
the second period may increase.
118
Stellenbosch University http://scholar.sun.ac.za
However, a typical non-altruistic migrant from a developing country is more likely to react more
to changes in real interest rate in his/her home country than in the host country. The logical
reasoning being that the average price level in a migrant-host country often tends to be lower
and more stable than what pertains in the migrant‟s home country (a typical SSA country),
which makes real deposit interest rates lower and more erratic at home than abroad. Apart from
this, migrants under implicit social contract would still be compelled to remit under duress
irrespective of changes in the returns on financial assets in their respective host countries. With
a higher real deposit interest rate at home, migrants with the intention of returning home after
some time may be motivated to remit through the formal financial system in which banks and
stock markets are prominent. Real interest rates on deposits at home are an indication of
improved financial sector development through competition and risk diversification. In other
words, commercial banks in the migrant-home country are likely to mobilise more private funds
in the form of remittances from migrants living abroad if real deposit interest rates are positive
and attractive. Accordingly, in an attempt to respond to what macroeconomic policy SSA
countries must implement to attract higher inflows of migrant remittances through the official
channels, the role of the domestic financial sector and, for that matter, the level of real deposit
interest rate in the home countries should be more imperative. This is why with the same level
of interest rates in France (the leading host nation of SSA migrants) some SSA countries
(Benin, Comoros, Mauritius, Senegal and Togo) receive more official remittances than other
SSA countries (Congo, Madagascar, Mauritania, and Niger) in per capita terms (see Figure
3.5).
Again, it is imprudent to assume that a rational pure self-interest investment-driven migrant, in
taking investment decisions in this globalised world, will restrict such decisions to his/her home
country conditions in comparison with the prevailing and/or anticipated conditions in his/her
host country only, and completely ignore the relevant investment conditions of the rest of the
world. Therefore, the key modification made in this study to the Vargas-Silva and Huang (2006)
theoretical framework is to, rather than emphasising the real interest rate differential between
the migrant-host country and the migrant-home country, the migrant-host country real interest
rate is held constant whilst attention is given to the real deposit interest of the migrant-home
country. This is imperative because in devising an effective policy strategy for attracting higher
migrant remittances through the formal transfer channels to the sub-region, policy makers in
potential remittance-receiving SSA countries can influence only the domestic factors such as
119
Stellenbosch University http://scholar.sun.ac.za
interest rates, bearing in mind the exogenous prevailing and expected global economic
conditions.
4.5 EMPIRICAL MODEL, METHODOLOGICAL APPROACH AND DATA ISSUES
This section presents the empirical model, methodology and data issues under in sub-sections
4.5.1, 4.5.2 and 4.5.3 respectively.
4.5.1 The Empirical Model
In line with the modified Vargas-Silva and Huang (2006) theoretical framework, Equation (4.9)
constitutes the baseline specification of the empirical dynamic panel-data model, involving the
36 sampled SSA countries over a 30-year period, 1980-2009:
R it Ri ,t l xit it
(4.9)
where is a scalar, and xit is the i th observation on the k explanatory variables. And
because Equation (4.9) is a model with a one-way error component it has unobservable
country-specific effect ( i ) and the remaining residuals ( it ) , such that
it i it . More
explicitly, the estimated model is of the form:
R it Ri ,t l xit i it
(4.10)
where the residuals ( it ) are white-noise such that it
IID(0, 2 ) , i
IID(0,2 ) and is a
scalar such that, generally, 1 ; i 1, 2,3,...., N is an index for individual sampled SSA
countries, implying N 36 ; t 1, 2,3,....,T is an index for time-variant periods, in this case,
years, so that T 10 for decade-based estimations such as 1980-89, 1990-99, and 2000-09;
whilst T 30 for the estimations involving the overall study period, 1980-2009. The countryspecific effect and the disturbance term are independent of each other and among themselves.
xit as row vector of explanatory variables, excluding the lagged dependent variable, has the
dimension k where k n 1 with n being the number of exogenous variables, but it is
acknowledged that these variables may not be strictly exogenous. is the unknown parameter
of the lagged endogenous variable; is the unknown parameter vector of the k exogenous
variables; l is the number of significant lags carried by the dependent variable to capture “the
entire history of the right-hand side variables, so that any measured influence is conditioned on
120
Stellenbosch University http://scholar.sun.ac.za
this history”65 (Greene, 2003: 307); and i country-specific fixed effects. This model is also
based on the assumptions that: (i) the error term is orthogonal to the exogenous variables so
that E ( xit it ) 0; (ii) the independently and identically distributed error term is uncorrelated
with the lagged endogenous variable implying that E ( Ri ,t 1 it ) 0; (iii) the exogenous variables
might be correlated with the individual effect for which reason, E ( xit i ) 0; and (iv) there is the
need to capture the dynamic effects of remittances because either altruistic migrant remittances
could decay often by the second generation as family and social ties become weak (Lucas and
Stark, 1985; Lozano-Ascencio, 1993; Glytsos, 1997), or the value of migrant remittances
increase over time for self-interest investment motive as the legal residential and income status
of migrants improve.
The endogenous variable R is a measure of remittances either as migrant remittances (MRem),
workers‟ remittances (WREM) or compensation of employees (COMP). When deflated by
population, the endogenous variable is redefined as REMPC, WREMPC and COMPPC
respectively in the specific estimated models. In estimating the model on the total migrant
remittance inflows in SSA, the natural logarithmic form of the dependent variable (lnMRem)
was used. Macroeconomic factors influencing migrant remittance inflows as a percentage of
GDP (REMGDP) were also explored to make room for comparison with the preferred REMPC
results. The explanatory variables are real deposit interest rate of a typical SSA country (RIR),
f
h
real bilateral exchange rate (RXR), host-country income (Y ) , home-country income (Y ) ,
home-country CPI-based inflation rate (INF), bank credit to the private sector as a percentage
of GDP in the home country (PSC), and the quality of institutions in the home country (INS). For
the entire sample period analysis, a dummy (D9_11) was introduced for post-September 11,
2001 such that D9_11=0 for 1980-2001, and D9_11=1 for 2002-2009 to reflect the era of
improved enforcement of regulations and tougher laws on international money transfers in a bid
to clampdown on money laundering, official corruption and other illegal activities that threaten
global security. The introduction of this time dummy (D9_11) is also important as it helps to
prevent any possible cross-individual correlation or contemporaneous correlation.
65
Because of the implicit effect of altruism behind all kinds of migrant remittances and because of the continuous
influence of newly-arrived migrants on remittance flows, the number of significant lags can be very high, as many as
eight, as noted in this study (the results are available but are not reported due to space), but the study restricted the
number of significant lags to two as the key motivation is not to determine the historical trend in remittance flows in
completeness. The lag of two is also consistent with the migrant remittances inter-generational effects theory
(Elbadawi and Rocha, 1992; Lozano-Ascencio, 1993) and the optimal number of lags for stationarity (Table A4.5).
121
Stellenbosch University http://scholar.sun.ac.za
From Equation (4.10), there are T 2 orthogonality restrictions in levels which are exploited;
hence that observation t in levels was used for the estimation, where differences are used as
valid instruments, when it is assumed that x is at least predetermined66. For instance, for the
i iT , the instruments used are
last observation, T , specified as RiT Ri ,T 1 xiT
dyi ,1 , dyi ,2 , dyi ,3 ,...., dyi ,T 1 , dxi1 , dxi2 , dxi3 ,...., dxi,T . The corresponding matrices used for the
estimation as given by Behr (2003) are shown in Box A4.1 in the Appendix.
The choice of the dynamic panel-data model is informed by the fact that data on remittance
inflows in most SSA countries have been more consistently available across countries only in
recent years and, therefore, the panel has small fixed T and large N . The small T large N
dimension of the panel data is also underscored by the fact that a decade-by-decade analysis
was carried out prior to estimating for the overall study period so as to find out, whether or not,
the macroeconomic factors that attract remittances to SSA have been consistent over the past
three decades. Another justification for the dynamic panel-data estimation approach is that the
relationship under consideration is linear; the left-hand side variable is singular and dynamic;
the explanatory variables are not strictly exogenous; there are fixed individual effects; and there
are heteroskedasticity and autocorrelation within the cross-sectional units but not across them
(Behr, 2003; Blundell and Bond, 1998; Roodman, 2009a). A unique advantage of dynamic
panel-data models is that, by allowing for empirical modelling of dynamics alongside the
individual-specific dynamics, they provide the necessary platform to account for past behavioureffect directly on current behaviour, whilst recognising the fact that individual cross-sectional
units have a predilection to behave in any particular way.
4.5.2 The Methodological Approach
The methodological approach is presented under five sub-themes namely, the system
Generalised Method of Moment (GMM) estimator, the Sargan test for over-identifying
restrictions, the Arellano-Bond test for second-order serial correlation, the decade-based
parameter evolution and instability test, and the panel-data unit root test.
66
Further discussions and insights can be found in Arellano and Bond (1991), Blundell and Bond (1998), Wooldridge
(2002), Behr (2003), and Greene (2003).
122
Stellenbosch University http://scholar.sun.ac.za
4.5.2.1 The System Generalised Method of Moment (sys-GMM) Estimator
The Generalised Method of Moment (GMM) estimation procedure is the preferred choice over
other alternative panel-data estimation techniques such as instrumental variable panel-data
fixed (or random) effects (IV FE/RE) and 2SLS panel FE/RE, because: (i) some of the
explanatory variables viz. PSC, RXR, Y and Y are known not to be strictly exogenous,67
f
h
given the possibility of a two-way causality (see Section 4.3.2); (ii) time-invariant individual
features such as geography, demography and economic policy may correlate with the
explanatory variables so that the error term is influenced by unobserved country-specific effects
and the observation-specific errors; (iii) the dynamic nature of the model naturally gives rise to
serial correlation; and (iv) the panel data setting is structured by small time dimension (T ) and
large cross-sectional dimension ( N ) . In the presence of the aforementioned econometric
problems, neither IV FE/RE nor 2SLS FE/RE panel-data estimators can be efficient, whilst
GMM estimation technique was specially developed to handle these problems more
efficiently.68 For instance, instead of using only exogenous instruments as is in panel FE/RE
and 2SLS FE/RE, the lagged endogenous regressors are also instrumented with their past
levels, thereby making endogenous variables pre-determined and, hence, uncorrelated with the
error term. Also, to circumvent the problems associated with time-invariant country-specific
characteristics, GMM uses a unique difference transformation process to eliminate these
effects (Arellano and Bover, 1995; Blundell and Bond, 1998; Greene, 2003). Furthermore,
Roodman (2009a) explains with experimental evidence that within a panel data setting of
N T as is conditional to the application of GMM technique, where T is large but not in
relation to N (i.e. both T and N are large, but N is still larger than T ), a shock to a countryfixed effect which is captured in the idiosyncratic error term will decline over time.
The Blundell and Bond (1998) system GMM estimation technique rather than the „difference‟
GMM proposed by Arellano and Bond (1991) and the „deviation‟ GMM suggested by Arellano
and Bover (1995) is employed in this study. The system GMM (henceforth sys-GMM) is the
preferred choice because within this framework it is possible to include non-country-specific
time-invariant regressors such as D9_11, which tend to disappear in „difference‟ GMM (Baltagi,
2008; Roodman, 2009a). Additionally, the Arellano and Bond (1991) and the Arellano and
67
Apart from simultaneity, endogeneity in Equation (4.9) can arise from period effects which occur due to systematic
shocks after a change in any of the explanatory variables; unobserved heterogeneity due to omitted variables; and
variable measurement errors (Woodridge, 2002; Greene, 2003; Balgati, 2008).
68
For proof or further insight, see Hansen (1982), Holtz-Eakin et al. (1988), and Arellano and Bond (1991).
123
Stellenbosch University http://scholar.sun.ac.za
Bover (1995) GMM techniques are known to be inefficient when the instruments are weak as
they make use of the information contained in differences only, but in the sys-GMM framework,
additional level information besides the differences are used (Blundell and Bond, 1998;
Roodman, 2009a). More specifically, the sys-GMM estimator uses the levels equation to obtain
a system of two equations – a differenced equation and a levels equation – such that by adding
the second equation, additional instruments are generated. Thus, by allowing for more
instruments the estimated coefficients of the Blundell and Bond (1998) sys-GMM are not only
more efficient, but also more consistent than other alternative techniques of dynamic paneldata analysis. The Blundell and Bond (1998) GMM is popularly referred to as the sys-GMM
because it is composed of moment restrictions for differences and levels resulting in a common
efficient estimator. This study adopted the two-step sys-GMM estimator which combines the
T 1 first-differenced equations and the average level equation, and has been proven69 to
produce a standard covariance matrix that is robust to panel-specific autocorrelation and
heteroskedasticity.
The main econometric concern of GMM methodological approach is the problem of instrument
proliferation (Roodman, 2009b). The instrument proliferation problem is more severe in large
samples70 whereby a large collection of instruments, even if individually valid, can be invalid as
a group in infinite samples because the instruments over-fit the endogenous variables
(Roodman, 2009b). According to Roodman (2009b), because models involving sys-GMM are
almost always over-identified, and because the Hansen J statistic71 theoretically detects any
violation of the instrument validity assumption, econometricians using sys-GMM are relieved of
the need to probe this further. The problem, however, is that there are contexts in which
instrument proliferation weakens the Hansen J test statistic (Baum et al. 2003; Roodman,
2009b). Proliferating purely by increasing T to prevent covariates, Roodman (2009b) concludes
that the symptoms of proliferation became noticeable only when T 15 , implying a longer time
dimension reduces instrument invalidity in a simulation exercise. At T 20, the full-instrument
variant never detects it, with an average -value on the Hansen J test of 1.00.
69
See Roodman (2009a,b) for this proof.
Although „large samples‟ in this context are not explicitly defined, various Monte Carlo experiments in this regard
seem to suggest the severe manifestation of this problem as N is about 50 in rare cases but often when N 100
(see, for example, Roodman, 2009b; Chan et al. 2012).
71
The Hansen J statistic is equivalent to the Sargan test statistic for over-identifying restrictions computed from
robust estimates (Baum et al., 2003; Roodman, 2009a,b; Chan et al. 2012). This is why some scholars refer to the
Sargan statistic from robust estimates as the Hansen-Sargan statistic or the Sargan-Hansen statistic or as in the
case of this study, simply the Sargan statistic.
70
124
Stellenbosch University http://scholar.sun.ac.za
The two conventional techniques to addressing instrument over-fitting in dynamic GMM panels
is limiting. One of the techniques that is often used to overcome the problem of instrument
proliferation is to limit the lag depth by selecting some of the lags to be included in the
instrument set, making the instrument count linear in T . Another approach is to collapse the
instrument set to the extent that the orthogonality condition no longer needs to be valid for any
one time period but still for each lag, again making the instrument linear in T . A combination of
both techniques makes the instrument count invariant to T (Mehrhoff, 2009). Currently, there is
no widely accepted rule of thumb for optimal instrumentation whilst the decision to choose any
of the aforementioned approaches to instrument proliferation treatment has been done
arbitrarily stirring further controversy. Studies are currently underway to establish how the data
can decide how the transformation matrix should look (see, for example, Mehrhoff, 2009). So
far, the available preliminary results from work-in-progress on using Monte Carlo simulation to
investigate model performance with instrument over-fitting has been for samples of minimum of
N 100 and T 30 at best (see, for example, Mehrhoff, 2009; Roodman, 2009b). What
remains unclear and unresolved in the literature of GMM estimation is how large or small T and
N have to be, relative to each other, for the GMM estimator to be „perfect‟ (Roodman, 2009b).
As has been widely documented in the literature, it must be emphasised that the instrument
proliferation problem is not peculiar to GMM estimators (Tauchen, 1986; Anderson and
Sørensen, 1996; Altonji and Segal, 1996; Ziliak, 1997; Bowsher, 2002). For instance, the poor
performance of IV estimators when instruments are too many has been long identified by
Hayashi, 2000; Ruud, 2000; Wooldridge, 2002; Arellano, 2003). Moreover, over-fitting is still a
problem even at low instrument counts (Roodman, 2009b).
Furthermore, if heteroskedasticity is present the GMM estimator is more efficient than the
simple IV estimator, whereas “if heteroskedasticity is absent, the GMM estimator is no worse
asymptotically than the IV estimator” (Baum et al. 2003: 11). There is no proof suggesting,
however, that in the presence of homoscedasticity other panel-data estimators including the IV
estimator in a dynamic context are more efficient than the sys-GMM estimator. According to
Balgati (2008: 87) “homoskedastic disturbances when heteroskedasticity is present will still
result in consistent estimates of the regression coefficients, but these estimates will not be
efficient” because of biased standard errors. This requires the econometrician to compute
robust standard errors to correct for possible heteroskedasticity (Baum et al. 2003; Balgati,
125
Stellenbosch University http://scholar.sun.ac.za
2008; Roodman, 2009b; Chan et al. 2012). Robust estimation by two-step GMM automatically
generates homoscedastic standard errors (ibid). In consequence of this, to exonerate the
empirical results from criticisms of heteroskedasticity, a two-step robust sys-GMM estimation
procedure was followed in this study.
To demonstrate empirically that in spite of the fact that there is no universally accepted perfect
estimator in econometric modelling, the dynamic panel-data sys-GMM estimator is more
appropriate, efficient and reliable than both conventional and robust static panel-data
estimators in the general context of this study, the researcher proceeded to estimate the
empirical static panel-data models involving the complete sample size, in other words, where
( N T ) (30 36) .
4.5.2.2 The Sargan-Hansen Test for Over-Identifying Restrictions
If, and only if, a dynamic panel-data model is over-identified, it should be possible to verify if the
excluded instruments are correctly independent of the residual process. Therefore, to test for
the joint validity of the instruments used, the study adhered to the suggestions by Arellano and
Bond (1991), and Roodman (2009a) inter alia by conducting the Sargan-Hansen test for overidentifying restrictions after the sys-GMM estimation. Like the conventional Sargan test, the
Sargan-Hansen test can only be performed when a model is estimated using instrumental
variable techniques and the estimated model is actually over-identified (Arellano and Bond,
1991). The “robustified Sargan statistic is numerically identical to the Hansen J statistic
computed from feasible efficient two-step GMM for that model” which is commonly referred to
as the Hansen-Sargan or the Sargan-Hansen statistic (Baum et al., 2003: 18). In fact, in robust
estimation, stata reports the Hansen J statistic in the place of the usual Sargan statistic with the
same null hypothesis (Baum et al., 2003; Roodman, 2009a,b). For simplicity sake, however, the
test statistics for the joint validity of the instruments used in the sys-GMM estimations are
reported as the Sargan test in this dissertation.
The routine Sargan-Hansen test is formulated on the null hypothesis that the instruments used
as a group are exogenous, therefore, the higher the p-value of the test statistic, the better. The
Sargan-Hansen test, which has an asymptotic distribution with degrees of freedom equal to
2
the number of over-identifying restrictions (i.e. number of instruments used minus number of
endogenous variables), is often weak in small N where the number of instruments is large and
exceeds the number of groups. Roodman (2009a,b) cautions that the Sargan-Hansen test
126
Stellenbosch University http://scholar.sun.ac.za
should not be relied upon too faithfully, as it is prone to weakness” because “large instrument
collection can over-fit endogenous variables, but notes also that this does not compromise the
consistency of the coefficient estimates.
According to Roodman (2009b), although the Sargan-Hansen test requires homoskedastic
errors in order to be consistent, it is not vulnerable to instrument proliferation because it does
not depend on an estimate of the optimal weighting matrix. The Sargan-Hansen test is based
on the observation that the residuals should be uncorrelated with the set of exogenous
variables if the instruments are truly exogenous (Sargan, 1958; Hansen, 1982). The Sargan
test statistic can be calculated as nR 2 (the number of observations n multiplied by the
coefficient of determination R 2 ) from the OLS regression of the residuals (from IV estimation)
onto the set of exogenous variables (Sargan, 1958; Hansen, 1982; Wooldridge, 2002; Baum et
al. 2003). This test statistic will be asymptotically chi-squared with m k (where m is the
number of instruments and k is the number of endogenous variables) degrees of freedom
under the H 0 that the error term is uncorrelated with the instruments.
4.5.2.3 The Arellano-Bond Test for Second-Order Serial Correlation
Following the recommendations by Arellano and Bond (1991) which were later substantiated by
a host of other panel data experts, including Blundell and Bond (1998), Behr (2003), Greene
(2003), Roodman (2009a,b), and Baltagi (2008), the Arellano-Bond test (henceforth the A-B
test) was performed. The A-B test is specially designed to detect second-order serial
correlation (AR(2)) in the idiosyncratic disturbance term within a GMM framework, a situation
which rendered some lags invalid as instruments. Arellano and Bond (1991) show that the A-B
test for AR(2) in first differences is more relevant than that of the AR(1) because the former
specifically examines the presence of autocorrelation in levels. Arellano and Bond (1991) prove
that the A-B test is critically important because the consistency of the GMM estimator is
dependent upon the realism of the condition that E ( it i ,t 2 ) 0 .
The A-B test for autocorrelation is based on a null hypothesis of no autocorrelation and it is
applied to the differenced residuals. It is for this reason that the A-B test for AR(1) process in
first differences usually rejects the null hypothesis essentially because it it i ,t 1 and
i ,t 1 i ,t 1 i ,t 2 both have i ,t 1 (Arellano and Bond, 1991; Blundell and Bond 2000). This
127
Stellenbosch University http://scholar.sun.ac.za
hypothesis holds true if
it is serially uncorrelated or does not follow a random walk to
undermine the appropriate specification of an estimated dynamic panel-data model by GMM.
Technically, therefore, the test for AR(2) in first differences is more important because, as
noted earlier, it will detect autocorrelation in levels. This study transgresses accordingly by
testing and reporting the A-B statistics for second-order autocorrelation.
Based on the outcome of experimental evidence, Roodman (2009a), however, observes that in
small N dynamic panels the A-B test may not be reliable, but so far a more efficient alternative
test for this purpose has not been developed.
4.5.2.4 The Decade-Based Parameter Evolution and Instability Test
Generally, all available sample observations are used for estimation to enhance the possibility
of an econometric specification that best fits a given dataset. This does not allow the
econometrician to test for parameter constancy, stability and robustness of the estimated
relationship (Brown et al. 1975). In panel-data econometric modelling, the econometrician will
have to go by either the observations in time (T ) or the observations in cross section ( N ) and
use a subset of either panel-data dimension (i.e. T or N ) for the parameter evolution and
instability testing. Consistent with the time-dependent decade-based analysis in Chapter Three
and the revealed cyclical pattern of migrant remittances received by SSA, there is a reasonable
basis to expect a priori a structural break by the end of each decade (in this case, 1989 and
1999).
In testing for the decade-based structural break, this study fits the specified panel-data
econometric model separately for each decade in order to verify whether there are statistically
significant differences in the estimated decade-based parameters. A statistically significant
difference indicates a parameter evolution, hence a validated evidence of a changing impact of
the explanatory variables on the dependent variable over time according to the specific
decades. Accordingly, to investigate the instability of the estimated coefficients obtained from
the three decades (1980-89, 1990-99 and 2000-09), the full sample data was further partitioned
into two overlapping decades - 1985-1994 and 1995-2004. Within the confines of this study,
only two additional time-variant sub-samples (1985-1994 and 1995-2004) are possible because
as shown by Chow (1960), Gujarati (1970a,b), Fomby et al. (1984) and Chan et al. (2012),
each sub-sample must necessarily have the same dimension.
128
Stellenbosch University http://scholar.sun.ac.za
Empirical econometric parameters are unlikely to be stable over time in response to policy
regime changes because rational economic agents are likely to change their decisions
according to the prevailing or the anticipated policy environment. In view of this hypothesis, the
parameter instability test procedure proposed by Andrews (2003) and Chan et al. (2012) was
adopted to examine the degree of instability of the estimated decade-based parameters across
the three decades, 1980-89, 1990-99 and 2000-09. The justification for the Andrews (2003)
and Chan et al. (2012) test for structural breaks and parameter instability is that the well-known
econometric tests for structural breaks in the literature including the celebrated Chow (1960;
1984) test are often applied in cases of a single structural break. The modified related
alternative statistical tests proposed by Andrews (1993), Andrews and Ploberger (1994), and
Bai and Perron (1998) for an unknown structural breakpoint or the multiple breakpoints are
appropriate only when the break is relatively long lasting and occurs in the midpoint of the
sample distribution (Chan et al. 2012). As shown in Mancini-Griffoli and Pauwels (2006) the
application of the Andrews (2003) coefficient stability test under the assumption of fixed effects
panel data is straightforward.
The standardised „diferential‟ Z statistic for evolution and instability in panel-data model
essentially amounts to comparing two average statistics taken from a pre-break sub-sample
and a post-break sample (Andrews, 2003; Chan et al. 2012). The determination of the average
statistic for a pre-break and the post-break sub-samples requires the computation of the test
statistic s times for each individual sub-sample panel in addition to any overlapping subsample of equal dimension if the interest is to include verifying coefficient stability by rolling
over time. In general terms, Andrews (2003) and Chan et al. (2012) specify the standardised Z
statistic to test for evolution and instability in panel-data models which involves taking the
difference of the post- and pre-break average statistics as:
Z
ˆ 1 ˆ 0
Var ( S ˆ1 S ˆ 0 )
S
ˆ 1 ˆ 0
2
ˆ1
S 2Sˆ1 S ˆ 0
2
ˆ 0
(6.11)
where ˆ 1 and ˆ 0 are the estimated coefficient for the post- and pre-break samples
respectively, S ˆ 0 and S ˆ1 are the corresponding standard errors, and Z follows an asymptotic
distribution
A
N Z N (0,1) . Intuitively, if the H 0 is true then Z will be centred around zero.
However, under the alternative, the Z will centre further away from zero, indicating more
129
Stellenbosch University http://scholar.sun.ac.za
evidence against the H 0 . The sample size that was used to determine ˆ 0 is the same as in
the case of ˆ 1 , implying that, with reference to this study, each of the comparing decade-based
estimated coefficient was obtained from a panel dimension of ( N T ) (36 10). According to
Chan et al. (2012), it is recommended to use the first i observations to estimate ˆ 0 in order
to effectively minimise the potential impact of serial correlation in the errors by maximising the
distance between any two sub-samples.
Using j , j , C j , D j and E j to represent a decade-based estimated parameter for a given
explanatory variable for the periods, 1980-89, 1990-99, 2000-09, 1985-94, and 1995-2004
respectively, the „differential‟ Z ij test for each of the three decades of interest i and a specific
explanatory variable j was computed by finding the following differences:
i.
j j to determine how each of the estimated coefficients of the results for 1980-89
is statistically different from the corresponding estimated coefficients of 1990-99.
ii.
j C j to determine how each of the estimated coefficients of the results for 1990-99
is statistically different from the corresponding estimated coefficients of 2000-09.
iii. j C j to determine how each of the estimated coefficients of the results for 1980-89
is statistically different from the corresponding estimated coefficients of 2000-09.
Similarly, to establish the degree of time-variant evolution stability of the respective estimated
parameters for each of the three decades of interest (i.e. 1980-89, 1990-99, and 2000-09) the
„differential‟ Z ij test was computed for:
i.
j D j to determine how each of the estimated coefficients of the results for 1980-89
is statistically different from the corresponding estimated coefficients of 1985-1994.
ii.
B j D j to determine how each of the estimated coefficients of the results for 1990-99
is statistically different from the corresponding estimated coefficients of 1985-1994.
iii. j E j to determine how each of the estimated coefficients of the results for 1990-99
is statistically different from the corresponding estimated coefficients of 1995-2004.
iv. C j E j to determine how each of the estimated coefficients of the results for 2000-09
is statistically different from the corresponding estimated coefficients of 1995-2004.
130
Stellenbosch University http://scholar.sun.ac.za
Therefore, to establish, whether or not, the estimated parameter ( inf ) of a given explanatory
variable (say, inflation abbreviated as inf) for a decade (say, 1980-89) is statistically different
from the estimated parameter (inf ) of the same explanatory variable (inf) for another decade
(say, 1990-99) the „differential‟ Z ij statistic was computed by evaluating the expression:
Z c Z inf
S
inf
2
S2 2S S
(6.12)
inf
where Z c denotes the computed „differential‟ Z statistic so that in case Zc Zcritical value at 95
per cent confidence interval, then the conclusion would be that the difference between the two
estimated coefficients ( inf and inf ) is not statistically different from zero, hence the
H 0 : inf inf 0 which implies inf inf would not be rejected. Thus, from an econometric
viewpoint, the impact of inflation on the dependent variable (say, remittances) in the 1980s and
in the 1990s is actually the same. Otherwise, if Zc Zcritical value , then the H 0 : inf inf 0 or
H 0 : inf inf is rejected at the conventional statistical levels.
4.5.2.5 The Panel-Data Unit Root Test
Concerning the possible problems associated with data non-stationarity, Phillips and Moon
(2000) dismiss earlier arguments by Kao (1999), and Phillips and Moon (1999) that the use of
panel data naturally evades spurious regression and produces efficient, reliable and consistent
parameter estimates even when both N and T are large and approach infinity. According to
Phillips and Moon (2000), unlike in the case involving the use of survey data, it is often
impossible to have a panel setting where N is sufficiently large in macro panel-data models.
Consequently, many macro cases actually involve (in relation to each other) large N and large
T even where N T . And in the case of panels where both N and T are large, there is the
need to pay serious attention to verifying the asymptotic properties of the data. Therefore, to
exonerate this study from criticisms of spurious regressions, unit root tests for panel data were
performed on each of the variables included in the empirical analyses despite the fact that the
panel data structure is of N T . This is because for the estimations based on the overall study
period, whilst N still remained at 36, T increased to 30, compared to T 10 in the case of
decade-based regressions. This study assumes that since T 30 is generally considered large
131
Stellenbosch University http://scholar.sun.ac.za
enough for a country-specific time series analysis, issues related to spurious regression cannot
be downplayed, though N T .
Consequently, out of the various panel-data unit root tests available (see Asteriou, 2006;
Baltagi, 2008), this study essentially relied on the Fisher Phillips-Perron (Fisher P-P) chi-square
test of individual root test and the Hadri Heteroskedasticity Consistent z-statistic (HHC) test,
whilst the Levin-Lin-Chu (LLC) adjusted t test was evoked when the Fisher P-P and HHC tests
were in conflict. The Fisher P-P test is essentially the panel version of the P-P test used in time
series econometrics, and in reality and similarity, the average of the P-P t-test statistics of the
individual cross-sectional units where the disturbance term is serially correlated with different
autocorrelation properties across the cross-sectional units (Im et al. 2003). The null hypothesis
of Fisher P-P is that each series in the panel has a unit root whilst the alternative hypothesis
permits some of the individual series to be non-stationary. The main advantage of the Fisher PP over the other alternative tests, (notably, HHC, LLC and Breitung t-test), is its ability to handle
both balanced and unbalanced panel data, including balanced panels with missing data points.
The HHC residual-based Lagrange Multiplier (LM) panel unit root test was developed by Hadri
(2000) from the Kwiatkowski, Phillips, Schmidt and Shin (KPSS) stationarity test for time series
data. The null hypothesis of HHC test is that there is no unit root in any of the series in the
panel against the alternative of a unit root in the panel. The key advantage of HHC is its ability
to handle panels with heteroskedasticity, but it is not applicable where a series suffers from
omission(s). The LLC panel unit root test is particularly designed for panels with size structure
where 10 N 250 and 25 T 250 , under the assumptions of no cross-sectional
correlations, and all cross-sectional units having or not having a unit root (Levin et al. 2002).
The LLC test, which is essentially the panel version of the popular Dickey-Fuller or Augmented
Dickey-Fuller test in time series econometrics, is formulated on the hypothesis that each
individual time series has a unit root against the alternative that each time series is stationary in
levels (Levin et al. 2002). The main limitation of LLC is its inability to handle heterogeneous
panels and panel series with missing data points72.
4.5.3 Data Measurement, Sources and Expected Impact on Remittances
Low frequency balanced panel data from secondary sources was used in this study. The
relevant annual series were collated on 36 SSA countries for the period, 1980-2009. The
72
Further information on panel unit root tests, including the formula for the Fisher P-P, HHC, and LLC test statistics
can be found in Asteriou (2006) and Baltagi (2008).
132
Stellenbosch University http://scholar.sun.ac.za
sample size was determined strictly by the availability of consistent data on the relevant
variables particularly migrant remittances. The choice of the start date was contingent upon the
desire to cover as many countries as possible for higher representation of the population rather
than covering longer periods with limited coverage of the target population. Besides, the
relevance of migrant remittance inflows in SSA became evident only in the 1980s as shown by
the stylised facts presented in Chapter Three. Among other reasons, the choice of the upper
limit of the study period is based on the availability of consistent data as data on some of key
variables notably financial liberalisation, poverty and income inequality proxied by Gini
coefficient are not currently available beyond 2009. In fact, the World Bank publishes data on
poverty and income inequality after every five years. Coincidentally, the upper limit of 2009 has
also enabled the researcher to be consistent with the objective of undertaking a decade-based
analysis.
For this study, migrant remittances as defined in Chapter Two constitute the sum of workers‟
remittances recorded in the current account of IMF‟s Balance of Payments Statistics (BoPS)
under the heading “current transfers”; and compensation of employees recorded under the
“Income” sub-category of the current account. Mathematically, compensation of employees is
the net of migrant remittances less workers‟ remittances. Migrant remittances and workers‟
remittances were obtained mainly from the World Development Indicators (WDI) published by
the World Bank based on the Balance of Payments Statistics Yearbook (BoPS) of the IMF, and
the Migration and Remittances Factbook 2011 published by the World Bank. Other sources
such as estimates based on IMF country-specific desk official information were used to fill in
some of the missing data points where possible73. Compensation of employees received
(COMP) was obtained by subtracting workers‟ remittances received (WREM) from migrant
remittances (MRem) received but due to problems related to the classification of reported
remittances from migrants, there are instances where the implied COMP received was less
than zero because the reported WREM was greater than migrant remittances. Consistent with
the definition of migrant remittances and underlying BoP double-entry accounting principle
where a credit transaction cannot be negative74, for the estimations based on COMP, only data
points for which COMP is non-negative were used in this study.
73
74
In a recent study, Singh et al. (2010) used a similar approach to obtain remittances data on 36 SSA countries.
See Chapter Two of this dissertation, and IMF‟s BoPS Yearbook and Balance of Payment Textbook for details.
133
Stellenbosch University http://scholar.sun.ac.za
The measurement of home-country income and host-country income adopted in this study is
quite rare in empirical studies. Real GDP per capita in US dollars at purchasing power parity
h
(PPP) of a typical SSA country ( Y ) was used as a proxy for the migrant-home country income
which also represents the average annual earnings of the family left behind who are considered
as the main prospective recipients of migrant remittances. Similarly, real GDP per capita PPP
f
in US dollars of a typical migrant-host country ( Y ) was used as a proxy for the average
annual earnings of an SSA international migrant who is considered as a potential remitter. The
use of real GDP per capita PPP instead of the commonly used real GDP per capita as a proxy
for host-country (migrant) income and home-country (family) income in this study is second
only to Schiopu and Siegfried (2006), Moore and Greenidge (2008), and Adams (2009). The
justification for using the PPP-based real income includes, that: (i) in making decisions on how
much and how often to remit, migrants take into consideration the quantity of goods and
services the assets transferred (remittances) can actually buy in the recipient country at the
prevailing market conditions; (ii) inflating the income gap between countries are resolved as
non-tradables
are
accounted
for;
and
(iii)
any
perceived
existing
country-specific
misrepresentation of facts arising especially from monetary and exchange rate policies such as
redenomination of a national currency that might be misinterpreted as leading to currency
overvaluation is adequately accounted for.
Human capital accumulation (HCA) is a more comprehensive concept than just acquisition of
formal education by a portion of the population of a particular country. However, most macrolevel empirical studies are restricted to the measures of education (see, for example, Table
A4.1) so that education and human capital are often conflated and used interchangeably in the
literature. For the sake of easy access to adequate data which is its clear advantage over other
alternative measures in macro-level cross-country studies, this study transgresses similarly by
using secondary school enrolment to proxy for human capital accumulation (Mankiw et al.
1992). Barror and Lee (1996) used average years of schooling. Kalaitzidakis et al. (2001) used
the proportion of government expenditure devoted to education. Bils and Klenow (1998)
suggest the use of life expectancy to proxy for human capital accumulation. Barro (1998) and
Barro and Sala-í-Martin (1995) assert that the gender disparity component of school attainment
can be an alternative measure of human capital accumulation in a typical growth equation.
Measurement of institutional quality (INS) is a daunting task because it is both complex and
subjective. It is complex not only because institutions have two main dimensions viz. the formal
134
Stellenbosch University http://scholar.sun.ac.za
and the informal, but also because each of these aspects has various components. For
example, formal institutions cover bureaucracy, corruption, law, property rights protection,
political environment, governance, access to information, freedom, and economic and political
participation. Informal institutions cover and wide range of indicators including perceptions of
life, environment, work, family, politics and society, religion and morality, and national identity
(Knack, 2001; Knack and Keefer, 1997; Knowles and Weatherston, 2006). It is subjective
because an objective measure of institutions does not exist (Duc and Lavallee, 2004)
necessitating measurement according to the feelings and perceptions of economic actors.
There are alternative proxies to the institutional quality index constructed by Marshall and
Jaggers (2011) used in this study. A typical alternative is the worldwide governance indicators
compiled by the World Bank based on the governance research index developed by Kaufmann
et al. (2003; 2008) for six dimensions namely, voice and accountability, political stability and
presence of violence, government effectiveness, regulatory quality, rule of law and control of
corruption. However, a major limitation of the World Bank‟s worldwide governance indicators is
that data are only available for the years 1996, 1998, 2000 and from 2002-2007. Like the WB‟s
worldwide governance indicators, the governance research index by Kaufmann et al. (2003) is
not available on annual basis neither is it available for periods prior to 1996 nor after 2002.
Similarly, although the International Country Risk Guide (ICRG) reports a composite political
risk index that comprises 12 institutional measures, it is limited in terms of time coverage. The
12 dimensions of the ICRG index include bureaucratic quality, corruption, internal conflict,
external conflict, ethnic tensions, democratic accountability, government stability, investment
profile, law and order, military in politics, socioeconomic conditions, and religious tensions. The
ICRG data is not available prior to 1984 and while a good number of the sampled SSA
countries do not have the ICRG data prior to the 1990s, some of the countries are not either
consistently covered or they are completely excluded. There is also the worldwide index of
human freedom constructed by the Fraser Institute (2008). The Fraser freedom index cover five
categories, namely the size of government, the legal structures and security of property right,
the access to sound money, the regulation of credit, labour and business, and the freedom to
trade internationally. However, only 25 of the 36 sampled SSA countries are covered by the
Fraser index. Moreover, the 10-category based economic freedom index developed by
Heritage Foundation covers periods starting from 1995 at best and it is available for only 12 out
of the sampled 36 SSA countries. It is for these reasons that the author relied on the Marshall
and Jaggers polity2 index.
135
Stellenbosch University http://scholar.sun.ac.za
Meanwhile, Rodrik (2004) cautions that because the composite governance research index and
the ICRG proxies for institutional quality are generally based on surveys of domestic and
foreign investors they capture perceptions rather than the true formal aspects of the institutional
setting. This creates problems of reverse causality and endogeneity. Moers (1999), however,
observes that the use of subjective institutional measures instead of objective indicators in
growth empirics is quite consistently verified.
Liquid liabilities of the financial system are the sum of currency plus demand and interestbearing liabilities of banks and non-bank financial intermediaries divided by nominal GDP.
Broad money ratio which is often considered as the broadest measure of financial
intermediation (King and Levine, 1993) encompasses the overall lending capacity of the
financial system which covers the Central Bank, deposit money banks and other financial
institutions. The merits and demerits of M2/GDP and PSC are discussed quite extensively in
Chapter Seven.
Whilst the degree of trade openness (OPN) is defined and is measured as the percentage of
the volume of cross-border trade flows (exports plus imports of goods) undertaken by a typical
migrant-home country to its nominal GDP, the rate of inflation (INF) is measured as the annual
percentage variations in consumer price index (CPI) of a typical SSA home country. Trade
openness index shows the degree of economic liberalisation and participation in a globalised
world whilst inflation rate measures the speed of adjustment in general price level, which
reflects macroeconomic uncertainty and investment risk in the „labour-exporting‟ SSA country.
Real bilateral exchange rate (RXR) is the annual average value of the national currency of a
typical SSA migrant-sending country in real terms of the national currency of its leading nonSSA migrant-host country. Mathematically, it is computed as the multiplication of the nominal
bilateral exchange rate by the ratio of the migrant-host country CPI to the migrant-home country
CPI. Real deposit interest rate (RIR) is the annual average bank deposit interest rate less the
annual average CPI-based rate of inflation of a typical migrant-sending country.
Apart from these traditional macroeconomic variables which were obtained essentially from the
International Financial Statistics Yearbook / CD-ROM (IFS) and World Economic Outlook
(WEO), as well as the World Bank‟s WDI and AfDB‟s African Development Indicators (ADI), a
dummy variable for capturing the global regulatory environment in connection with cross-border
136
Stellenbosch University http://scholar.sun.ac.za
money transfer following September 11, 2001 was also introduced into the empirical model. In
Table A4.2 in the Appendix, a more concise and exact information on the description,
measurement and sources of the specific variables used in this study is provided. The a priori
sign of each of the explanatory variables is also indicated in Table A4.2.
4.6 EMPIRICAL RESULTS AND DISCUSSIONS
This section is organised into four sub-sections. The results on robustness test models and the
diagnostic tests are discussed in Section 4.6.1. Then, the main results upon which policy
recommendations are made follow in this order: the empirical results on the determinants of
migrant remittance inflows are presented and discussed in Section 4.6.2; those related to
workers‟ remittance inflows are presented and discussed in Section 4.6.3; while those on
compensation of employees are presented and discussed in Section 4.6.4.
4.6.1 Results of Robustness Models and Diagnostic Tests
To be able to compare the results from this study with those obtained from previous related
studies, estimations were carried out following the common practice where international
migrant remittance inflows as a percentage of GDP (REMGDP), and where the USA was
assumed to be the main host country of SSA migrants. The results of these estimations are
presented in Table A4.4 in the Appendix. These results show that the use of the SSA countryspecific leading migrant-host countries compares very closely with using the USA as the
migrant-host nation in many respects except for the differences in the magnitude of the
economic and the statistical significance75. In relative terms, in the estimation involving the use
of the USA as the migrant-host of SSA migrants, the test statistics (both the z-statistics and the
Wald statistic) are generally lower, with only two exceptions, home-country income and real
2
exchange rate (see Table A4.4). Ignoring the dynamic effects of migrant remittances, the
robustness results on USA versus country-specific migrant-host nations show that apart from
institutional quality and the migrant-host country income the estimated coefficients of the results
based on the USA as the common host nation of SSA migrants are economically more
75
Many reasons can be assigned to the close comparison of the two results. The reasons include: (1) the USA
might, for security reasons, have tougher rules and regulations on international money transfers compared with other
migrant-host nations, following the 9-11 terrorist attack on the former; (2) Money transfer costs might be cheaper in
the USA than in the other nations hosting SSA migrants; (3) The USA might have a relatively more advanced
financial infrastructure with wider migrant access to alternative official cross-border money transfers than the other
SSA migrant-host nations such as Pakistan, Saudi Arabia, Portugal, Spain, Belgium, France, Germany and Great
Britain; (4) SSA migrants in USA might be more skilful and economically viable, and hence with higher incomes than
their counterparts in Europe and the rest of the world; and (5) The cost of living in USA might be relatively less than
the average cost of living in Europe, Pakistan and Saudi Arabia.
137
Stellenbosch University http://scholar.sun.ac.za
significant than those obtained from country-specific migrant-host countries. On a lesser note,
the Wald statistics76 of 1400000 and the Arellano-Bond autocorrelation test on the firstdifference errors at order-2 probability value of 0.9152 for the specific migrant-host nation,
compared with 454424.61 and 0.9043 respectively obtained in the estimation involving USA as
the SSA migrant-host nation shows that the former estimation is relatively more efficient.
Therefore, for policy relevant studies, the use of „true‟ migrant-host country rather than the
generally assumed use of the USA as the migrant-host nation should be more appropriate.
Regarding the appropriateness of migrant remittances in per capita (REMPC) as against
migrant remittances as a percentage of GDP (REMGDP), the estimated results presented in
Table A4.4 justify REMPC in a number of ways. Generally, with the exception of home-country
income and without taking into account the dynamic effects, each of the estimated coefficients
from the REMPC model is both economically and statistically more robust in comparison with
the estimated REMGDP model. The fact that the results from the REMGDP model suggest
that, in the long run, as the income of SSA migrants increases in real terms, migrant
remittances received by the sub-region decline, which is in sharp contrast to the fundamental
theories of remittances and a priori expectation, the use of REMGDP instead of REMPC cannot
be described as the more appropriate. The low value of the reported Wald statistic and the
parameter estimates together with the high probability values of the majority of the explanatory
variables is an attestation that the estimated lnMRem is the least efficient in comparison with
REMGDP and REMPC. This is notwithstanding the fact that the signs of the estimated
parameters of the lnMRem model, in contrast to the modelled REMGDP, are more consistent
with those obtained from REMPC and in conformity to the underlying theories of remittances.
To demonstrate empirically that dynamic panel-data modelling by GMM is more appropriate
than static panel-data estimations either by Fixed (within) Effects (FE) or by Generalised Least
Squares Random Effects (RE), both the conventional and the robust FE and RE estimations
were carried out for the overall study period. The empirical static panel-data modelling results
for REMPC, WREMPC and COMPPC are presented in Table A4.6, Table A4.7 and Table A4.8
respectively. As extensively discussed under sub-section 4.5.2.1, these empirical static panel-
76
Under H 0 , in large samples, the Stata-automatically generated Wald statistic has a chi-squared distribution with
degrees of freedom equal to the number of coefficient restrictions imposed on a model. According to Baltagi (2008),
the Wald test for restrictions has its power dependent upon the number of instruments used and the degree of serial
correlation and heterogeneity in the residuals. It, thus, is the test statistic interpreted in determining the joint
statistical significance of the explanatory variables in GMM models.
138
Stellenbosch University http://scholar.sun.ac.za
data models are not expected and, in fact, did not produce unbiased estimators due to the
presence of obvious endogeneity among the explanatory variables and the omission of the
dynamic components. For each of the estimated static panel-data models (see Table A4.62
A4.8), the reported overall coefficient of determination ( R ) did not exceed 10 per cent whilst
the Breusch-Pagan statistic suggest the presence of serious heteroskedasticity. Besides, there
is a serious concern for multicollinearity because in the midst of very low coefficient of
determination R 2 , the computed z-statistics are relatively high for the conventional static paneldata estimators. In the case of the heteroskedastic-corrected standard errors obtained from the
robust static panel-data estimations, the reported z-statistics and
R 2 are statistically
insignificant in each case.
The main results upon which policies are prescribed in this chapter are presented in Tables 4.2,
4.3 and 4.4 in Sections 4.6.2, 4.6.3 and 4.6.4 respectively. Given the distinctiveness of each
component of migrant remittances, three main sets of estimations were carried out. Each set of
estimations is either on migrant remittances, workers‟ remittances or compensation of
employees received by SSA countries during the period 1980-2009. An overall study period,
1980-2009, as well as decade-by-decade (1980-89; 1990-99 and 2000-09) estimations was
carried out. In each of these estimations international migrant remittances were measured in
per capita terms which represent the closest proxy for remittances per migrant as revealed in
Chapter Three. Also, for each of these estimations, the USA was not chosen as the common
host country for SSA migrants as done in many previous studies. In other words, this study
used non-SSA countries with the highest percentage of migrants from the various SSA
countries as the migrant-host countries.
Each of the estimated results presented in Tables 4.2, 4.3 and 4.4 comprised of the 36
sampled SSA countries with 51, and 441 or 442 valid instruments for decade-based analysis
and the overall study period analysis respectively. The number of observations for the various
decade-based analyses ranges from 217 to 288, whilst 1006 was reported for the overall study
period analyses except in the case of the compensation of employees estimated model for
which 823 was reported due to missing observations. The reported probability value of the
Wald statistic for each estimated model was 0.000, suggesting that for the balanced panel-data
empirical models, each regressand was jointly explained by the set of regressors at one per
cent level of statistical significance. The various statistics reported by the Arellano-Bond test
139
Stellenbosch University http://scholar.sun.ac.za
point to the fact that at five per cent level of statistical significance the idiosyncratic residuals
generated from the two-step sys-GMM estimation do not suffer from second-order
autocorrelation in any of the estimated results. To provide further proof that each estimated
model is statistically efficient with unbiased and reliable estimated parameters, the Sargan test
for over-identifying restrictions was conducted. The reported Sargan test statistic for each
estimated model confirms the selected instruments for each empirical model as valid, while
none of the estimated models suffered from endogeneity bias. Meanwhile, a pre-estimation
examination of the asymptotic properties of the relevant variables included in the estimated
model indicate that each variable is integrated of order zero (see Table A4.5 in the Appendix),
hence proving the existence of a cointegrating relationship77. It also shows that none of the
estimated results from this chapter is spurious whilst the estimated coefficients are
cointegrating parameters. Essentially, because the motivation for this study is not just to
determine the particular decade in which macroeconomic factors contributed more to migrant
remittances received in SSA but more importantly to verify if as the liberalisation of the SSA
financial market improves, the macroeconomic environment has had a changing impact on
remittance inflows in SSA, in discussing the results presented in Sections 4.6.2, 4.6.3, and
4.6.4, more emphasis is laid on the overall study period rather than on the various decades.
4.6.2 Macroeconomic Determinants of Migrant Remittances
The estimated results on the macroeconomic determinants of migrant remittance inflows in
SSA are presented in Table 4.2. The results show that the current amount of remittances sent
by SSA migrants resident outside the sub-region is determined by past remittances, migranthost country income and migrant-home country factors viz. institutional quality, real „family‟
income, rate of inflation, credit to the private sector, real bilateral exchange rate and real
deposit interest rate. The tightening rules and regulations aimed at clamping down on the use
of informal money transfer channels by migrants have since 2002 contributed substantially to
the increasing inflow of migrant remittances received in SSA through official channels.
The size of the amount remitted by a migrant over the immediate past two years affect the
current inflow of official remittances. Whilst the immediate past year amount remitted positively
impacts on the current level of migrant remittance inflows, the impact of the past two years is
largely negative on current level of official remittances received by the sub-region. This seems
to support the view that new migrants often remit more in the initial stages of their migration
77
For the proof of this assertion, refer to Engle and Granger (1987).
140
Stellenbosch University http://scholar.sun.ac.za
when the social ties are strong or when they are under implicit social contractual obligation to
remit. Therefore, from an individual migrant‟s perspective, remittances are likely to decrease
over time and, indeed, after a second generation of migrants, so that the stability in the flow of
remittances is mainly sustained by a new generation of migrants, as observed by Elbadawi and
Rocha (1992) on North African migrants in Europe, and Lozano-Ascencio (1993) in the case of
Mexican migrants in USA.
Table 4.2: Estimated Results of Migrant Remittances (REMPC) Flows to SSA, 1980-2009
Group variable: code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel Data Procedure
1980-89
1990-99
2000-09
1980-2009
REMPC lag 1 (REMPC_1)
1.0998
(2875.64)***
0.6586
(361.51)***
0.8896
(325.32)***
0.9477
(359.71)***
REMPC lag 2 (REMPC_2)
-0.3488
(-303.75)***
0.1011
(185.00)***
-0.2315
(-99.26)***
-0.1267
(-107.93)***
Institutional quality (INS)
0.9483
(34.62)***
-0.2308
(-9.08)***
-0.6938
(-6.62)***
-0.3107
(-8.17)***
3.1063
(6.95)***
-13.0805
(-58.22)***
-0.6076
(-0.87)
-1.0084
(-2.82)***
Host-country income (lnY )
31.6778
(87.42)***
1.71530
(1.98)**
142.5704
(42.29)***
18.3925
(13.93)***
Rate of inflation (INF)
-0.0507
(-2.07)**
0.1407
(10.91)***
-0.1395
(-5.35)***
0.0449
(4.75)***
Bank credit to private sector (lnPSC)
0.7645
(4.19)***
8.9322
(36.86)***
-4.6438
(-9.92)***
2.3553
(7.12)***
Real exchange rate (lnRXR)
-9.0491
(-112.66)***
-2.4873
(-14.36)***
-3.0266
(-9.92)***
-4.9314
(-12.98)***
Real deposit interest rate (RIR)
-0.0300
(-1.14)
0.0760
(5.58)***
0.0263
(0.89)
0.0347
(4.46)***
Regulatory environment (D9_11)
……….
……….
………
………
……….
……….
4.6614
(71.91)***
Constant term (constant)
-285.9995
(-101.71)***
71.8419
(7.79)***
-1414.6010
(-43.12)***
-157.9818
(-12.40)***
Number of observations
288
286
288
1006
Number of groups (N)
36
36
36
36
Number of instruments
Wald [2 ],
51
51
51
442
[9],5520000***
[9],1260000***
[9],1230000***
[10],1400000***
h
Home-country income (lnY )
f
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-1.0780{0.281}
-1.5868{0.113} -0.6783{0.498}
-0.1065{0.915}
Sargan test for over-identifying restrictions:
[2],
Source: Author‟s estimation
[41],28.1669
[41],27.3812
[41],28.6736
denotes 5(1) per cent respectively.
2-step robust z-statistics in ( ), z-probabilities in { }
**(***)
141
[431],27.1849
Stellenbosch University http://scholar.sun.ac.za
Over the 1980-2009 period, the most important factor that positively impact on migrant
remittance inflows in SSA through official channels is the improvement in real income of
countries hosting SSA migrants. In particular, a percentage rise in the real GDP per capita PPP
in migrant-host country resulted in, on the average, about a US$18.39 increase in remittances
per capita received in SSA when the entire study period is taken into consideration. Although
the positive effect of migrant income on remittances received in SSA was unswerving in each of
the past three decades considered in this study, one striking revelation from this study is that
during good economic times in the home country, thus, in the 1980s and the 2000s (see Table
3.1 in Chapter Three) SSA migrants remitted more. This finding is consistent with both altruism
and self-interest theories of migrant remittance inflows which predict that as the economic
status of migrants improve, migrants will remit more of their incomes home ceteris paribus. This
result validates the findings from a host of previous related studies, notably those of Elbadawi
and Rocha (1992) for six North African and European countries, Bouhga-Hagbe (2004) for
Morocco, Vargas-Silva and Huang (2006) in a global study, Coulibaly (2009) for 16 LAC
countries, and Singh et al. (2010) for 36 SSA countries. At the same time, however, this result
contradicts the result obtained by Freund and Spatafora (2005) for 104 countries in SSA, EAP
and ECA. The sources of this contradiction could include the use of the FE model by Freund
and Spatafora (2005) which is less efficient than the sys-GMM used in this study. Also, in
contrast to this study, Freund and Spatafora (2005) defined migrant remittances beyond the
two current account elements (WREM and COMP) by adding migrant transfers; and covered
104 countries from various continents over the study period of 1995-2003 compared to only 36
SSA countries over the period 1980-2009 in this study. Obviously, differences in economic
conditions in the sampled countries during the study periods in the two studies could affect the
results obtained.
Turning to the domestic macroeconomic environment, overall, the leading puller of official
migrant remittances is improved macroeconomic performance as reflected in a stronger
national currency. In fact, apart from having the highest economic impact, as far as domestic
factors are concerned, currency appreciation is also the most consistent determinant of migrant
remittance inflows through official channels in SSA. Overall, between 1980 and 2009, the real
appreciation of a typical SSA migrant-home country currency against the national currency of
its migrant-host country by one percentage point increased the official inflow of migrant
remittances by at least US$4.93 in the SSA migrant-home country. In fact, in the 1980s, a
similar rate of currency appreciation could attract as much as US$9.05 per capita migrant
142
Stellenbosch University http://scholar.sun.ac.za
remittances compared to US$2.49 and US$3.03 in the 1990s and the 2000s respectively.
Though this finding is in affirmation of the result obtained by Lueth and Ruiz-Arranz (2007a) for
Sri Lanka, it is in contrast with the results obtained by Jadhav (2003) for India, Adenutsi and
Ahortor (2008) for Ghana, and Singh et al. (2010) for 36 SSA countries who found depressing
effects of appreciation of local currencies on remittances received. Apart from differences in
estimation techniques and scope as regards the sample size and/or the study period, and
unlike the real bilateral exchange rate used in this study, nominal bilateral exchange rate was
used in Jadhav (2003) and Adenutsi and Ahortor (2008) whilst Singh et al. (2010) used real
effective exchange rate (REER). These could be the main sources of the inconsistencies in the
results across the various studies. In other studies, Lianos (1997) on Greece, Aydaş et al.
(2004) on Turkey, Gupta (2005) on India, and Moore and Greenidge (2008) on 15 Caribbean
islands, failed to find exchange rate as a determinant of remittance inflows. Here again,
differences in the measurement of remittances, study period, sampled countries, sample size
as well as the methodology could be the underlying sources of this contradiction.
In terms of economic significance, improved private sector access to bank credit in SSA is the
second most important domestic factor that impacts on migrant remittance flows to the subregion. A one per cent increment in private sector credit as a ratio of GDP in SSA leads to at
least a US$2.35 rise in migrant remittances received per capita in SSA between 1980 and
2009. With the one percentage increase in private sector access to bank credit in SSA, migrant
remittances per capita increased by about US$0.76 and US$8.93 in the 1980s and the 1990s
respectively. In the 2000s, however, improved access to private sector credit in SSA impacted
negatively, (as much as US$4.64), on migrant remittances per capita received by the subregion. This finding seems to support the altruistic hypothesis of migrant remittances formulated
by Schrooten (2005), and Giuliano and Ruiz-Arranz (2009). According to this hypothesis, in
periods of improved but limited access to credit by the private sector in SSA migrant-home
countries as witnessed in the 1980s and the 1990s (Table 3.3), migrants are compelled to remit
more to ease the liquidity constraints at home. In the case of SSA, as there was more restricted
access to private sector credit in the 1980s and the 1990s than in the 2000s (see Table 3.3 in
Chapter Three), SSA international migrants were probably compelled to remit more funds to
their families during the first two decades of the financial liberalisation and as access to private
sector credit improved in the 2000s, migrant remittances declined. To this extent, the findings
support earlier results obtained by Schrooten (2005) for 24 transition countries, Niimi and
Özden (2006) for 85 countries, and Singh et al. (2010) for 36 SSA countries.
143
Stellenbosch University http://scholar.sun.ac.za
Overall, improvements in home-country income had depressing effects on migrant remittances
received in SSA between 1980 and 2009. Over the entire study period, this depressing effect of
home-country income was more pronounced in the 1990s when SSA as a sub-region recorded
its worst economic performance as evident in the negative growth rate in real GDP per capita,
the least real GDP per capita, and the highest external debt stock (see Table 3.1 in Chapter
Three). In the 1980s, home-country income impacted positively on migrant remittance inflows,
whilst in the 2000s its impact was statistically insignificant. Coincidentally, with reference to the
study period, the 1980s and the 2000s were the periods in which the sub-region recorded its
best economic performance as measured in the level of real GDP per capita, domestic savings,
and external debt stock (Table 3.1). This suggests that, although, home-country income has an
overall negative impact on migrant remittance inflows in SSA, under sound macroeconomic
environment, migrant remittances became less altruistic, hence SSA migrants are likely to be
motivated to remit more as the economic prospects of their home country improve.
The general finding that migrant remittances received in SSA increase as home-country income
falls is consistent with the results obtained in all previous related studies reviewed in this study
except those obtained by Freund and Spatafora (2005) for 104 developing countries, and Lueth
and Ruiz-Arranz (2007b) for 11 developing countries in Asia, Europe and the Middle East. The
reasons for the contrasting results obtained in this study with those aforementioned can be
ascribed to differences in the measurement of remittances, choice of methodology, and scope
of study (see Table A4.1). For instance, whereas a dynamic panel-data sys-GMM estimation
procedure was followed in this study which effectively accounted for all possible endogeneity
problems associated with remittances, Freund and Spatafora (2005), and Lueth and RuizArranz (2007b) estimated static panel-data models without IV procedures which are less
efficient in this particular context. Also, while this study, just as that of Freund and Spatafora
(2005) relied on global remittance data from the World Bank and the IMF, Lueth and RuizArranz (2007b) used bilateral remittances data as reported by the various Central Banks.
Finally, because the results of this study suggest that the response of remittance inflows to
home-country income is dependent upon the soundness of the macroeconomic environment of
the recipient-countries; probably, the countries studied in the Freund and Spatafora (2005), and
Lueth and Ruiz-Arranz (2007b) had superior macroeconomic conditions which enabled them to
attract more non-altruistic remittances.
144
Stellenbosch University http://scholar.sun.ac.za
Generally, SSA migrants tend to remit home as a result of loss of confidence in their
governments, given that institutional quality negatively impacts on migrant remittances. In the
1980s, quality institutions impacted positively on remittance inflows in SSA, but in the 1990s
and the 2000s the effect of institutional quality on migrant remittance inflows was increasingly
negative. This could imply that the sub-region actually suffers from „forced‟ migration in more
recent years, due to weak institutions and poor governance, so that these „forced migrants‟ are
under pressure to remit more funds home as institutions become weaker. In this case, the
altruistic motive of remittances outweighs the self-interest investment motive. This finding
invalidates previous results obtained by Lueth and Ruiz-Arranz (2007b) for 11 developing
countries from Asia, Europe and the Middle East, and Singh et al. (2010) who found a positive
effect of institutional quality on 36 SSA countries as well as Schrooten (2005) who found no
impact of political uncertainty on remittance inflows for 42 countries in transition. The main
possible source of this contrast could be the differences in measurement of institutional quality.
Whereas previous studies used international country credit risk rating, this study used a more
comprehensive index constructed by Marshall and Jaggers (2011) to capture the qualities in
governance and institutions directly. In addition, the differences in the prevailing economic
conditions in the sampled countries and the differences in the study periods covered could be
the underlying reason for this discrepancy.
For the entire study period, 1980-2009, the results of this study show that SSA countries
receive more migrant remittances as the rate of inflation in migrant-home countries increases.
This study further reveals that higher rates of inflation in migrant-sending SSA countries led to a
lower inflow of remittances in the 1980s and in the 2000s, but in the 1990s, higher rates of
inflation in SSA were required for the receipt of higher migrant remittances in the sub-region.
This implies that during good economic times (e.g. periods with higher real per capita incomes
in home countries) such as the 1980s and the 2000s in the context of this study (see Table
3.1), SSA countries with lower rates of inflation are more likely to receive higher migrant
remittances, but during periods of economic recession more migrant remittances are received
in SSA even as inflation rate escalates. This result confirms the earlier findings reported by ElSakka and McNabb (1999) for Egypt, and Moore and Greenidge (2008) for 15 Caribbean
countries. This finding, however, contradicts the results obtained by Elbadawi and Rocha
(1992) for six North African and European countries, and Adenutsi and Ahortor (2008) for
Ghana. Whilst sys-GMM was used in estimating a dynamic panel-data model in this study
which is also on SSA countries, Elbadawi and Rocha (1992) used a panel FE model.
145
Stellenbosch University http://scholar.sun.ac.za
Additionally, differences in sampled countries and the study periods could account for this
nonconformity because as revealed by the results of this study, macroeconomic fundamentals
at any particular time could affect the magnitude and sign of the impact of inflation on migrant
remittances received.
By and large, the real deposit interest rate of the home country is a positive factor that explains
variations in per capita migrant remittances received by SSA between 1980 and 2009. This is
notwithstanding the fact that, in the 1980s and the 2000s, real deposit interest rate of the
migrant-sending SSA country was statistically insignificant in the determination of migrant
remittance inflows. The implication of this finding is that a typical SSA migrant-home country
can attract more migrant remittances through the formal transfer channels by offering higher
returns on savings and investment. This result validates the findings of all previous related
studies, viz. those of Lianos (1997) for Greece, Adenutsi and Ahortor (2008) for Ghana, and
Adams (2009) in a cross-sectional analysis involving 76 developing countries, which also used
home-country interest rates rather than interest rate differentials.
The results in Table 4.2 show that macroeconomic factors that influence migrant remittance
inflows in SSA migrant-home countries have a varying impact over time and, to a very large
extent, in accordance with the macroeconomic environment of the migrant-home countries.
This might be the reason behind the close comparison between the results obtained for the
1980s and those for the 2000s. For example, even though migrant-host country income had a
consistently positive impact on migrant remittances received in SSA in each of the past three
decades, the impact was statistically and economically more robust in the 1980s and the 2000s
than in the 1990s. A similar conclusion can be drawn regarding the impact of real bilateral
exchange rate on migrant remittances received in SSA. Evidently, the changing impact of
home-country macroeconomic factors on migrant remittance flows to SSA was for the most part
different in the 1990s and consistent with the stylised fact revealed by Figure 4.1, that migrant
remittance flows to SSA were most volatile and countercyclical in the 1990s.
The empirical results of the test verifying, whether or not, the estimated decade-based
parameters of the macroeconomic determinants of migrant remittances are statistically different
from decade to decade and stable over time are presented in Table 4.2.1. In columns A-B, B-C
and A-C of Table 4.2.1, the results validate the hypothesis that the estimated decade-based
coefficients of the various explanatory variables reported in Table 4.2 actually differ from
decade to decade. Given the significance of each of the explanatory variables reported in
146
Stellenbosch University http://scholar.sun.ac.za
column A-B in Table 4.2.1 at the conventional statistical levels, it is concluded that, with
reference to Table 4.2, the estimated parameters for the 1980-89 decade are statistically
different from the corresponding estimated coefficients of the 1990-99 decade. Similarly,
because the computed „differential‟ z-statistic of each of the explanatory variables reported in
columns B-C and A-C is statistically significant within 95 per cent confidence interval, the
hypothesis that the individual explanatory variables have a decade-based changing impact on
migrant remittance inflows in SSA during the 1980s, the 1990s and the 2000s is upheld. In
other words, the estimated decade-based coefficients of the macroeconomic determinants of
migrant remittance inflows reported in Table 4.2 are statistically different from each other, so
that the impact of any particular explanatory variable on international migrant remittance inflows
evolves from one decade to another in an apparent response to the cyclical behaviour of
remittance inflows depicted in Figure 4.1. The implications of these results are that the
macroeconomic determinants of migrant remittance inflows in SSA have a changing impact
according to macroeconomic fundamentals and policy environment as explained in Chapter
Three. This result has also lend credence to the proposition that macroeconomic policy
environment plays a crucial role in attracting official remittances in a migrant-home country. In
fact, the results also suggest the negative impact of harsh macroeconomic environment on
remittance inflows in SSA can outweigh the potential strength of financial liberalisation policy in
attracting remittances through the formal channels.
147
Stellenbosch University http://scholar.sun.ac.za
Table 4.2.1: Results of Decade-Based Parameter Evolution and Instability Tests for Migrant Remittances
Estimated Decade-Based Results
A
B
C
1980-89
REMPC lag 1 (REMPC_1)
REMPC lag 2 (REMPC_2)
Institutional quality (INS)
1990-99
2000-09
Decade-Based Rolling
Estimated Results
D
E
1985-1994
A-B
B-C
A-C
Overlapping Decade-Based Coefficient Stability
Test Results
A-D
B-D
0.8896
[0.0027]
0.8088
[0.0013]
1.0725
[0.0018]
0.4411
[0.0014]
-0.2310
[0.0009]
0.2102
[0.0023]
{2875.64}***
{361.51}***
{325.32}***
{607.78}***
{598.37}***
{306.33}***
{-253.80}***
{89.43}***
-0.3488
[0.0011]
0.1011
[0.0005]
-0.2315
[0.0023]
-0.1528
[0.0037]
-0.3019
[0.0016]
-0.4499
[0.0006]
0.3325
[0.0018]
-0.1174
[0.0012]
-0.1960
[0.0025]
0.2539
[0.0031]
0.4029
[0.0010]
0.0704
[0.0008]
{-303.75}***
{185.00}***
{-99.26}***
{-41.29}*** {-192.64}***
{-749.85}***
{186.81}***
{-99.47}***
{-76.88}***
{80.59}***
{395.04}***
{92.64}***
0.9483
[0.0274]
-0.2308
[0.0254]
-0.6938
[0.1048]
-0.8735
[0.0284]
0.4469
[0.0134]
1.1791
[0.0020]
0.4630
[0.0794]
1.6420
[0.0774]
1.8217
[0.0010]
0.6427
[0.0030]
-0.6777
[0.0120]
-1.1406
[0.0914]
{33.32}***
{21.21}*** {1751.65}*** {213.51}***
{-56.42}***
{-12.48}***
{-9.08}***
{-6.62}***
{-30.72}***
3.1063
[0.4469]
-13.0805
[0.2247]
-0.6076
[0.6984]
0.9956
[0.6914]
{6.95}***
{-58.22)***
{-0.87}
{1.44}
Host-country income (lnYf)
31.6778
[0.3624]
1.7153
[0.8663]
142.5704
[3.3713]
0.3230
[4.0380]
{87.42}***
{1.98}**
{42.29}***
-0.0507
[0.0245]
0.1407
[0.0129]
-0.1395
[0.0261]
{598.50}***
{5.83}***
?
?
16.1868
[0.2222]
-12.4729
[0.4737]
?
3.7139
[0.2515]
-0.1502
[0.0005]
C-E
0.6586
[0.0018]
{34.62}***
0.2909
[0.0010]
B-E
1.0998
[0.0004]
Home-country income (lnYh)
Rate of inflation (INF)
1995-2004
Non-Overlapping Decade-Based
Coefficient Stability Test Results
{306.24}*** {-306.49}***
-0.4139
[0.0003]
{-1379.53}*** {-194.57}***
2.1108
[0.2444]
-14.0761
[0.4667]
?
?
?
{72.84}***
{-26.33}***
{14.77}***
{8.64}***
{-30.16}***
-0.1107
[0.0441]
29.9625
[0.5039]
-140.8550
[2.5049]
-110.8930
[3.0089]
31.3547
[3.6756]
1.3923
[3.1717]
{0.08}
{-2.51}**
{59.46}***
{-56.23}***
{-36.85}***
{8.53}***
-0.0146
[0.0331]
-0.2744
[0.0166]
-0.1914
[0.0116]
0.2802
[0.0132]
0.0889
[0.0016]
-0.0361
[0.0086]
-0.1829
[0.00094]
?
?
?
1.8260
[0.8222]
142.6810
[3.3271]
{0.44}
{2.22}**
{42.88}***
0.1553
[0.0202]
0.4151
[0.0037]
0.1349
[0.0095]
{-2.07}**
{10.91}***
{-5.35}***
{-0.44}
{-16.52}***
{-16.53}***
{21.26}***
{55.53}***
{-4.18}***
{7.68}***
{111.89}***
{14.24}***
Bank credit to private sector (InPSC)
0.7645
[0.1825]
8.9322
[0.2423]
-4.6438
[0.4681]
2.5136
[0.4997]
6.3097
[0.2173]
-8.1677
[0.0599]
13.5761
[0.2258]
5.4084
[0.2857]
-1.7490
[0.3172]
6.4187
[0.2574]
2.6226
[0.0250]
-10.9535
[0.2508]
{4.19}***
{36.86}***
{-9.92}***
{5.03}***
{29.04}***
{104.69}***
{-43.67}***
Real exchange rate (InRXR)
-9.0491
[0.0803]
-2.4873
[0.1732]
-3.0266
[0.3051]
-14.9534
[0.6270]
{-112.66)***
{-14.36}***
{-9.92}***
{-23.85}***
-0.0300
[0.0263]
0.0760
[0.0136]
0.0263
[0.0296]
0.0163
[0.0297]
-0.2326
[0.0171]
{5.58}***
{0.89}
{0.55}
{-13.61}***
71.8419 -1414.6010
[9.2223]
[32.8062]
61.9620
[33.6750]
?
?
?
Real depositt interest rate (RIR)
{-1.14}
Constant term (constant)
-285.9995
[2.8119]
{-136.45}***
{60.12}***
{18.933}***
{-5.51}***
{24.94}***
?
?
-6.5618
[0.0929]
0.5393
[0.1319]
-6.0225
[0.2248]
5.9043
[0.5467]
12.4661
[0.4538]
?
{-70.64}***
{4.09}***
{-26.79}***
{10.80}***
{27.47}***
-0.1060
[0.0127]
0.0497
[0.0159]
-0.0563
[0.0033]
-0.0463
[0.0034]
0.0597
[0.0161]
0.3086
[0.0035]
0.2589
[0.0125]
{3.12}***
{-17.27}***
{-13.74}***
{3.72}***
{88.93}***
{20.76}***
{-8.36}***
-357.8410
[6.4104]
1486.4429 1128.6015 -347.9615
9.8799
[23.5838] [29.9942] [30.8631] [24.4527]
?
?
?
?
?
?
?
?
?
?
{-101.71}***
{7.79}***
{-43.12}***
{1.84}**
{-55.82}***
{63.03}***
{37.63}***
{-11.27}***
{0.40}
288
286
288
288
286
287
287
288
288
287
286
287
Number of groups
36
36
36
36
36
36
36
36
36
36
36
36
Number of instruments
51
51
51
51
51
51
51
51
51
51
51
5520000***
1260000***
1230000***
Number of observations
1245000***
3375000***
5130000***
?
3000000*** 2320000***
51
Wald statistic
A-B 2ⁿᵈ-order autocorrelation test
-1.078(0.28) -1.587(0.11) -0.678(0.50)
1.192(0.23)
n/a
-
-
-
-
-
-
-
Sargan over-identifying restrictions
28.167(0.94) 27.381(0.95) 28.674(0.93) 33.784(0.78)
n/a
-
-
-
-
-
-
-
Source: Author‟s estimation
4740000*** 2320000*** 3390000***
?
2324000***
*/**/***denotes significant at 10/5/1 per cent statistical levels respectively. Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
148
Stellenbosch University http://scholar.sun.ac.za
With reference to columns A-D, B-D, B-E and C-E in Table 4.2.1, it can be generally concluded
that the statistical evolution of the estimated decade-based parameters associated with the
explanatory variables in the empirical model involving migrant remittances are statistically
consistent but evolutionarily unstable over time. The only isolated exceptions to the affirmation
of the instability of the estimated decade-based coefficients are the constant term and the hostcountry income in column B-D where the parameter estimates of the 1990s was compared with
the corresponding estimated parameters of the 1985-1994 overlapping decade. In effect, a
statistical justification is hereby given to the extent that the computed „differential‟ z -statistics
reported in columns A-D, B-D, B-E and C-E are statistically significant suggesting that, at the
conventional levels of statistical significance, the estimated decade-based coefficients are
centred further away from zero. Therefore, statistical evidence is hereby established that the
estimated decade-based evolving coefficients reported in Table 4.2 are generally consistent
and statistically stable over time. The evidence also favours instability in the estimated
parameters of the decade-based empirical migrant remittance model.
149
Stellenbosch University http://scholar.sun.ac.za
4.6.3 Macroeconomic Determinants of Workers’ Remittances
The empirical results on the macroeconomic determinants of workers‟ remittance inflows in
SSA are presented in Table 4.3. The results suggest that, aside the asynchronous effects,
workers‟ remittance inflows are driven by both home-country and host-country macroeconomic
factors. For the overall 30-year study period, the two most important factors which positively
impact on the inflow of workers‟ remittances in SSA are the host-country factors viz. hostcountry income, and the enforcement of rules and regulations directed at clamping down on the
use of informal channels to transfer funds globally. Among the home-country factors, „family‟
income, real bilateral exchange rate, quality of institutions, and bank credit to the private sector
are the most important. Besides, real deposit interest rate and inflation did have some
significant impact on remittances received in SSA at various times over the past three decades,
albeit the overall individual impact of each of these variables on workers‟ remittances was
statistically insignificant. Only previous empirical studies on macroeconomic determinants of
remittances in which remittances were exclusively measured as workers‟ remittances like the
studies by Jadhav (2003), Amuedo-Dorantes et al. (2007), and Shahbaz and Aamir (2009) are
considered relevant when comparing the results reported in Table 4.3 with those obtained in
the past.
The empirical results suggest that improvements in host-country income, and regulations
aimed at clamping down on the activities of informal money transfer agents were crucial in
attracting higher workers‟ remittances through official channels to SSA between 1980 and
2009. At one per cent level of statistical significance, a one percentage rise in the real per
capita income of a typical host country of SSA migrants had the tendency of increasing
WREMPC to SSA by roughly US$19.11 during the 1980s, US$18.34 in the 1990s and
US$40.34 in the 2000s. This result points to the fact that during periods of relatively sound
macroeconomic environment in SSA (i.e. in the 1980s and 2000s as shown in Table 3.1), a
typical remittance-receiving SSA country could attract more remittances from its permanent
migrants than in times (such as in the 1990s) when macroeconomic conditions at home are
unfavourable. To this extent, this result is consistent with the self-interest investment motive
and confirms earlier findings by Jadhav (2003) for India, and Shahbaz and Aamir (2009) for
Pakistan. In the study by Amuedo-Dorantes et al. (2007) migrant-host income was not included
in the analysis.
150
Stellenbosch University http://scholar.sun.ac.za
Table 4.3: Estimated Results of Workers‟ Remittances (WREMPC) Flows to SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel Data Procedure
1980-89
1990-99
2000-09
1980-2009
WREMPC lag 1 (WREMPC_1)
0.8856
(952.27)***
0.5810
(440.56)***
0.9734
(365.67)***
0.9805
(343.77)***
WREMPC lag 2 (WREMPC_2)
-0.1240
(-145.23)***
0.0803
(76.17)***
-0.2868
(-172.32)***
-0.1151
(-40.47)***
Institutional quality (INS)
2.0974
(68.13)***
-1.3010
(-51.60)***
-0.1695
(-3.17)***
-0.8428
(-64.21)***
10.6169
(49.61)***
14.0312
(28.06)***
-0.6866
(-1.05)
2.7488
(6.67)***
Host-country income (lnY )
19.1087
(72.93)***
18.3370
(12.30)***
40.3369
(22.17)***
7.2828
(6.37)***
Rate of inflation (INF)
-0.0483
(-2.55)**
-0.2398
(-5.01)***
0.0910
(8.07)***
-0.0302
(-1.51)
Bank credit to private sector (lnPSC)
2.0462
(9.84)***
-1.2281
(-2.52)**
8.4104
(14.45)***
0.6757
(1.75)*
Real exchange rate (lnRXR)
-8.2534
(-163.05)***
14.3902
(41.23)***
-1.7684
(-11.07)***
-0.8803
(-1.98)**
Real deposit interest rate (RIR)
-0.0469
(-2.45)**
-0.2161
(-4.49)***
0.0898
(8.26)***
-0.0321
(-1.59)
Regulatory environment (D9_11)
……….
……….
………
………
……….
……….
5.4166
(72.67)***
Constant term (constant)
-218.4456
(-91.29)***
-346.9176
(-26.55)***
-413.6602
(-25.20)***
-99.1562
(-12.55)***
Number of observations
288
286
288
1006
Number of groups (N)
36
36
36
36
Number of instruments
Wald [2 ],
51
51
51
442
[9],2610000***
[9],4410000***
[9],1270000***
[10],1520000***
h
Home-country income (lnY )
f
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-0.4067{0.684} -1.7114{0.088}*
1.1479{0.251}
0.2257{0.821}
Sargan test for over-identifying restrictions:
[2],
Source: Author‟s estimation
[41],28.2255
[41],33.2958
[41],28.3559
denotes 5(1) per cent respectively.
2-step robust z-statistics in ( ), z-probabilities in { }
[431],28.3907
**(***)
Consistent with the result obtained by Shahbaz and Aamir (2009) for Pakistan, this study finds
that for the overall study period, 1980-2009, workers‟ remittance inflows were positively
influenced by the level of real per capita income in SSA migrant-home countries. A one
percentage rise in the real per capita GDP PPP in a typical „labour-exporting‟ SSA country
increased WREMPC by about US$$10.62, US$14.03 and US$2.75 for the periods, 1980-1989,
1990-1999, and 1980-2009 respectively. During the 2000s, the impact of migrant-home country
income on workers‟ remittance inflows was statistically insignificant probably because in the
opinion of permanent migrants, the growth of real per capita income recorded in the 2000s was
151
Stellenbosch University http://scholar.sun.ac.za
too low to attract more investment-oriented remittances from them. Overall, this finding seems
to support the self-interest economic theory of remittances. In this case, permanent SSA
migrants from whom the largest proportion of remittances is received are generally driven by
self-interest economic motives to remit. Again, permanent migrants are likely to remit more with
the slightest improvement in income level at home during difficult economic times as was the
case during the 1990s in SSA.
As far as the home-country factors are concerned, variation in real bilateral exchange rate of
the domestic currency relative to the currency of the migrant-host country is the second most
important reason why permanent SSA migrants remit. At five per cent level of statistical
significance, a one per cent rate of depreciation of the currency of a typical migrant-home
country led to a decrease in WREMPC by US$0.88 during the past three decades. Although it
is difficult to clearly attribute this result to the dominance of any particular remittance theory, the
evidence seems to support the self-interest economic theory when the results of the decadeby-decade analysis are taken into account. This is because during periods of favourable
economic conditions at home (i.e. the 1980s and 2000s in the context of this study), permanent
migrants remit less as home-country currency depreciates against host-country currency.
However, during periods of unfavourable macroeconomic conditions at home (e.g. the 1990s)
depreciation of the home-country currency increased remittances received from permanent
SSA migrants. Perhaps, permanent migrants regard weaker home-country currency as an
outcome of poor economic management at home, hence the need to remit less for investment
purposes. However, altruistic-driven permanent migrants who remit an equivalent homecountry currency denominated fixed amount without taking into account the purchasing power
of the domestic currency, remit less equivalent foreign currency when the home-country
currency depreciates. The overall positive impact of real exchange rate on workers‟ remittance
inflows in SSA validates previous results reported by Jadhav (2003) for India, AmuedoDorantes et al. (2007) for 111 developing countries, and Shahbaz and Aamir (2009) for
Pakistan.
At one per cent level of statistical significance, a unit improvement in the index of institutional
quality could increase WREMPC by US$2.10 in 1980s, but decrease WREMPC by US$1.30 in
the 1990s, US$0.17 in the 2000s and US$0.84 over the entire study period, 1980-2009. The
negative impact of institutional quality on workers‟ remittance inflows in the 1990s and the
2000s might be due to the increasing exodus of SSA citizens in recent years due to poor
152
Stellenbosch University http://scholar.sun.ac.za
governance, because with higher stock of migrants, the possibility of receiving more altruistic
remittances increases. In other words, consistent with the predictions of the altruistic theory,
the more SSA permanent migrants lose confidence in the rule of law and governance at home,
the more these migrants will remit home to support their families left behind. It is also possible
that as institutions become weaker and corruption becomes more prevalent, permanent
migrants take advantage to invest more at home since it is then relatively easier for the elite
class to set up businesses and to evade taxes on returns on investment. The changing trend in
the magnitude of institutional quality on WREMPC received in SSA points to the fact that
improvements in the quality of institutions could enhance the chances of a migrant-home
country to receive more non-altruistic remittances from its permanent migrants. Previous
related studies did not analyse the effects of institutional quality on workers‟ remittances.
Consistent with the predictions of the self-interest remittance investment theory, this study finds
that for the overall study period 1980-2009, official workers‟ remittance inflows are positively
influenced by the level of financial development as reflected in the access of the private sector
to bank credit. A one per cent improvement in private sector access to bank credit as a ratio of
nominal GDP in the migrant-home country stimulated about a US$2.05 rise in WREMPC in the
1980s, a US$1.23 decrease in WREMPC in the 1990s, and a US$8.41 increase in WREMPC
in the 2000s. Thus, under favourable macroeconomic conditions, an SSA migrant-home
country stands a higher chance of attracting more remittances from its permanent migrants by
promoting private sector access to bank credit. In other words, workers‟ remittances were
complementary rather than a substitute for private sector credit in SSA between 1980 and
2009. This finding appears somehow more consistent with the self-interest investment motive
than to the altruistic motive because improved private sector access to bank credit could
encourage investment-oriented permanent SSA migrants to remit more through official
channels for self-interest economic motive. None of the previous related studies reviewed in
this study analysed the effects of private sector credit on workers‟ remittance inflows.
Overall, there was a statistically zero-effect of home-country real deposit interest rate on
WREMPC in SSA between 1980 and 2009. However, the same cannot be said for the decadebased analysis. A one percentage rise in home-country real deposit interest rate decreased
WREMPC by US$0.05 and US$0.22 in the 1980s and the 1990s respectively; but in the 2000s,
a similar change in the real deposit interest rate increased WREMPC by US$0.09. This implies
that under similar macroeconomic conditions, with the passage of time, permanent SSA
migrants have been becoming more and more positively responsive to increases in real deposit
153
Stellenbosch University http://scholar.sun.ac.za
interest rate at home since the implementation of financial liberalisation in the 1980s. This
seems to justify the self-interest investment theory of remittances among permanent SSA
migrants in recent years. Previous related studies failed to consider the effect of home-country
interest rate on workers‟ remittance inflows.
In the 1980s and the 1990s, migrant-home country inflation had a depressive effect on
WREMPC received (-US$0.05 and -US$0.24 respectively), but a positive effect of US$0.09 in
the 2000s. For the entire study period, however, the effect of domestic inflation on WREMPC
was statistically insignificant in SSA. This result suggests that workers‟ remittances were
inversely related to home-country inflation in the early years of financial liberalisation in SSA
when the rate of inflation was relatively higher (see Table 3.1). Nonetheless, as the homecountry macroeconomic environment became fairly stable and the tendencies for rising price
levels in SSA decreased in the 2000s, a marginal rise in the rate of inflation stimulated higher
inflows of remittances from permanent migrants probably because this was considered by
investment-oriented migrants as an opportunity for making higher profits. As shown in this
study that the overall impact of inflation on workers‟ remittances received by SSA countries is
statistically insignificant, it is contradictory to the findings from the only relevant previous study
in which home-country inflation was found to have negatively impacted on workers‟ remittance
inflows. Contextually, the relevant previous study is that by Shahbaz and Aamir (2009) on
Pakistan over the period, 1971-2006. A possible reason for this conflicting result could be the
differences in the macroeconomic fundamentals such as differences in the rates of inflation
between Pakistan and the SSA migrant-home countries analysed in this study. Besides
methodological differences, the contradiction could also be attributed to differences between
the relatively high WREMPC received by Pakistan as one of the all-time leading recipients of
workers‟ remittances in the world compared to the sampled SSA countries, none of which
received REMPC up to US$1 per day during the period under investigation.
The estimated results on the determinants of workers‟ remittances in SSA suggest that optimal
remittances from permanent SSA migrants cannot be received unconditionally as permanent
SSA migrants are sensitive to home-country macroeconomic conditions when making
remittance decisions. It is the migrant-sending SSA countries with conducive investment
climate as reflected in higher real per capita income, stronger domestic currency, lower rate of
inflation, and higher private sector access to bank credit, inter alia that have the chance of
mobilising optimal remittances from their permanent migrants.
154
Stellenbosch University http://scholar.sun.ac.za
Table 4.3.1: Results of Decade-Based Parameter Evolution and Instability Tests for Workers‟ Remittances
Decade-Based Rolling
Estimated Results
Estimated Decade-Based Results
WREMPC lag 1 (WREMPC_1)
WREMPC lag 2 (WREMPC_2)
Institututional quality (INS)
Home-country income (lnYh)
Host-country income (InYf)
Rate of inflation (INF)
A
B
C
D
E
1980-89
1990-99
2000-09
1985-1994
1995-2004
Non-Overlapping Decade-Based
Coefficient Stability Test Results
A-B
B-C
0.3047
[0.0004]
A-C
Overlapping Decade-Based Coefficient
Stability Test Results
A-D
-0.3922
[0.0013]
-0.0875
[0.0017]
{781.15}*** {-292.68}***
{-50.60}***
B-D
0.9731
[0.0027]
0.7754
[0.0009]
0.9731
[0.0025]
{952.27}***
{440.56}***
{365.67}***
{819.38}***
{397.3}***
-0.1240
[0.0008]
0.0803
[0.0011]
-0.2868
[0.0017]
0.0079
[0.0031]
-0.0979
[0.0013]
{-145.23}***
{76.17}***
{-172.32}***
{2.51}**
2.0974
[0.0308]
-1.3010
[0.0252]
-0.1695
[0.0535]
-0.5978
[0.0120]
-0.6550
[0.0637]
3.3983
[0.0056]
-1.1315
[0.0282]
2.2668
[0.0227]
2.6952
[0.0188]
-0.7031
[0.0132]
-0.6459
[0.0385]
0.4856
[0.0103]
{68.13}***
{-51.60}***
{-3.17}***
{-49.98}***
{-10.28}***
{610.11}***
{-40.05}***
{99.95}***
{143.21}***
{-53.07}***
{-16.77}***
{47.33}***
10.6169
[0.2140]
14.0312
[0.5000]
-0.6866
[0.6539]
-13.1980
[0.3407]
-11.3476
[0.5582]
-3.4143
[0.2860]
14.7178
[0.1539]
11.3035
[0.4399]
23.8149
[0.1267]
27.2291
[0.1594]
25.3787
[0.5000]
10.6610
[0.6539]
{49.61}***
{28.06}***
{-1.05}
{-38.74}***
{-20.33}***
{-11.94}***
{95.67}***
{25.70}***
{188.01}***
{170.87}***
{50.75}***
{16.30}***
19.1087
[0.2620]
18.3370
[1.4908]
40.3369
[1.8194]
-9.4318
[0.7102]
51.6566
[0.7455]
0.7717
[1.2288]
-21.9999
[0.3286]
-21.2283
[1.5574]
28.5405
[0.4482]
27.7688
[0.7806]
-33.3196
[0.7453]
-11.3197
[1.0739]
{72.93}***
{12.30}***
{22.17}***
{-13.28}***
{69.29}***
{0.63}
{-66.95}***
{-13.63}***
{63.67}***
{35.57}***
{-44.71}***
{-10.54}***
-0.0483
[0.0189]
-0.2398
[0.0479]
0.0910
[0.0113]
0.0559
[0.0180]
-0.1129
[0.0279]
0.1915
[0.0289]
-0.3308
[0.0366]
-0.1393
[0.0077]
-0.1042
[0.0009]
-0.2957
[0.0298]
-0.1270
[0.0199]
0.2038
[0.0167]
{-18.14}*** {-112.03}***
{-9.91}***
{-6.37}***
{12.23}***
-1.5776
[0.0048]
-5.3311
[0.2147]
4.3073
[0.3094]
{5.97}*** {-327.29}***
{-24.83}***
{13.92}***
11.7268
[0.1357]
-4.4318
[0.0116]
{-75.77}*** {-1021.35}***
0.3671
[0.0006]
-0.1944
[0.0004]
-0.3921
[0.0011]
{5512.5}*** {-525.41)*** {-347.02}***
0.1628
[0.0008]
-0.1319
[0.0023]
{601.79}*** {201.01}***
{-57.84}***
0.0724
[0.0021]
{-2.55}**
{-5.01}***
{8.07}***
{3.10}***
{-4.04}***
{6.62}***
{-9.04}***
2.0462
[0.2079]
-1.2281
[0.4873]
8.4104
[0.5820]
0.3494
[0.4922]
4.1030
[0.2726]
3.2743
[0.2794]
-9.6385
[0.0947]
-6.3642
[0.3741]
{9.84}***
{-2.52}**
{14.45}***
{0.71}
{15.05)***
{11.72}*** {-101.80}***
{-17.01}***
Real exchange rate (InRXR)
-8.2534
[0.0506]
14.3902
[0.3490]
-1.7684
[0.1597]
-12.0783
[0.2149]
2.6634
[0.1481]
-22.6436
[0.2984]
16.1586
[0.1893]
-6.4850
[0.1091]
3.8249
[0.1643]
26.4686
[0.1341]
{-163.05}***
{41.23}***
{-11.07}***
{-56.20}***
{17.98}***
{-75.88}***
{85.37}***
{-59.43}***
{23.28}***
{197.38}***
-0.0469
[0.0192]
-0.2161
[0.0481]
0.0898
[0.0109]
0.0468
[0.0183]
-0.1165
[0.0288]
0.1691
[0.0290]
-0.3059
[0.0372]
-0.1367
[0.0083]
-0.0938
[0.0009]
-0.2629
[0.0298]
{-16.49}*** {-107.76}***
1.6968
[0.2842]
0.1782
[0.0002]
0.0001
[0.0002]
{0.29}
-0.1889
[0.0004]
{34.81}*** {742.33}*** {-510.62}***
Bank credit to private sector (InPSC)
Real deposit interest rate (RIR)
C-E
0.5810
[0.0013]
-0.2043
[0.0002]
0.1103
[0.0000]
B-E
0.8856
[0.0009]
{86.42}*** {-381.72}***
-0.0996
[0.0194]
0.2063
[0.0179]
{-8.81}***
{-5.15}***
{11.53}***
66.7430 195.2150 -469.9220 -598.3940
[3.3485] [14.0222]
[4.2670]
[6.4067]
116.2220
[4.7487]
49.4793
[8.0972]
{-2.45}**
{-4.49)***
{8.26}***
{2.56}***
{-4.05}***
{5.84}***
-218.4456
[2.3929]
-346.9176
[13.0666]
-413.6602
[16.4151]
251.4770
[6.6599]
-463.1400
[8.3179]
128.4720
[10.6737]
{-91.29}***
{-26.55}***
{-25.20}***
{37.76}***
{-55.68}***
{12.04}***
{19.93}***
{-93.40}***
{24.47}***
{6.11}***
288
286
288
288
286
287
287
288
288
287
286
287
Number of groups
36
36
36
36
36
36
36
36
36
36
36
36
Number of instruments
51
51
51
51
51
51
51
51
51
51
51
51
2610000***
410000***
127000***
407000*** 998838.91***
1510000***
268500*** 1368500*** 1508500***
224000***
704420***
562919***
Constant term (constant)
Number of observations
{-8.21}***
{13.92}*** {-110.13}***
Wald statistic
A-B 2ⁿᵈ-order autocorrelation test
-0.407(0.68) -1.711(0.09)*
1.148(0.25) -0.459(0.65)
-0.411(0.68)
-
-
-
-
-
-
-
Sargan over-identifying restrictions
28.223(0.94) 33.296(0.80) 28.356(0.93) 26.984(0.95)
26.903(0.96)
-
-
-
-
-
-
-
Source: Author‟s estimation
*/**/***denotes significant at 10/5/1 per cent statistical levels respectively. Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
155
Stellenbosch University http://scholar.sun.ac.za
In Table 4.3.1, the results of the statistical inquiry into the extent to which the estimated
parameters of the decade-based estimations actually differ from decade to decade are
reported. The extent to which the estimated coefficients of each of the explanatory variables for
the 1980-89 decade differ from the corresponding estimated coefficients of the 1990-99 decade
as well as those of the 1990-99 and the 1980-89 decade respectively differ from those of the
2000-09 decade are reported in columns A-B, B-C and A-C respectively.
With reference to column A-B, the results of the reported „differential‟ z -statistics suggest that
with the exception of host-country income, each of the estimated parameters for the 1980-89
decade is statistically different from the corresponding estimated parameters for the 1990-99
decade at five per cent level of statistical significance. Furthermore, the statistical significance
of the computed „differential‟ z -statistics reported in column B-C validate the hypothesis that
each explanatory variable actually had a decade-based varying impact on workers‟ remittances
in the 1990-99 decade compared with the 2000-09 decade. In much the same manner, with 95
per cent statistical confidence, each explanatory variable had a decade-based evolving impact
on workers‟ remittance inflows in SSA when the 1980-89 decade is compared with the 2000-09
decade. Therefore, in statistical terms, the macroeconomic factors determining workers‟
remittance inflows in SSA between 1980 and 2009 had a changing impact according to the
macroeconomic conditions and policy environment in each of the three identified decades
1980-89, 1990-99 and 2000-09.
Each of the computed „differential‟ z -statistics reported in columns A-D, B-D, B-E and C-E is
statistically significant at five per cent level. The only exception is the immediate past value of
workers‟ remittances (WREMPC_1) in column C-E where the estimated parameter of the 200009 decade statistically compares to the corresponding parameter estimate of the overlapping
decade, 1995-2004. Therefore, overall, a statistical basis has now been established that there
is statistical evolution and instability in each of the estimated decade-based parameters over
time. Needless to say, the reported results in Table 4.3.1 suggests that the extent to which the
macroeconomic factors influence workers‟ remittance inflows in SSA between 1980 and 2009
differ across the three decades (1980-89, 1990-99 and 2000-09), which can possibly be
attributed to the changing macroeconomic policy environment.
156
Stellenbosch University http://scholar.sun.ac.za
4.6.4 Macroeconomic Determinants of Compensation of Employees
The estimated results on the determinants of compensation of employees received in SSA are
presented in Table 4.4. As expected, these results show that remittances sent by temporary
migrants follow a historical process with an evidence of decay by the second generation of
temporary migrants. Econometrically, the estimated results are reliable having passed all the
relevant diagnostic tests discussed under Section 4.6.1. The empirical results show that in
descending order of economic significance of the estimated parameters, host-country income,
home-country income, real bilateral exchange rate, tougher rules and regulations prohibiting
unofficial remittance channels, bank credit to the private sector, institutional quality, real deposit
interest rate, and inflation are crucial to understanding the changing levels of compensation of
employees per capita (COMPPC) received in SSA between 1980 and 2009. Over the past
three decades, host-country income, regulations discouraging illegal remittances, the amount
of COMPPC received last year, institutional quality, real deposit interest rate, and inflation
impacted positively on current COMPPC received in the migrant-sending SSA countries.
Conversely, home-country income, real exchange rate, bank credit to the private sector, and
the amount of COMPPC received two years ago impacted negatively on current COMPPC
received in SSA. In the absence of any known previous related studies on macroeconomic
determinants of compensation of employees, it is impossible to compare the results reported in
Table 4.4 with others.
A one percentage increase in the real GDP per capita PPP of a migrant-host country had a
US$2.85 depressing effect on COMPPC in the 1980s. The impact of host-country income,
however, turned positive thereafter and became even more economically significant over time.
For example, with a one percentage rise in migrant-host country real per capita GDP PPP,
there was the propensity for per capita remittances received from temporary migrants to
increase by US46.57 in the 1990s, US$89.30 in the 2000s, and for the overall study period (i.e.
1980-2009), by US$15.80. This shows a consistent increasing impact of host-country income
on remittances received from temporary SSA migrants. One possible reason could be the everincreasing income gap between migrant-sending countries and the more industrialised migranthost countries, which has led to unusually increasing temporary migration in recent years.
Another possible explanation for the negative impact of migrant-host country income on
COMPPC received in SSA in the 1980s could be the saving and investment constraints faced
by investment-oriented temporary SSA migrants that might favour migrant investment in the
host country rather than in the home country in the early years of financial liberalisation.
157
Stellenbosch University http://scholar.sun.ac.za
Table 4.4: Results on Compensation of Employees (COMPPC) Flows to SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel Data Procedure
1980-89
1990-99
2000-09
1980-2009
0.6704
0.7143
0.6996
0.8388
(424.00)***
(816.56)***
(352.48)***
(324.16)***
0.0111
-0.0993
0.0016
-0.0363
(23.07)***
(-67.68)***
(1.30)
(-26.09)***
-0.0171
0.5569
0.3930
0.5401
(-1.00)
(189.29)***
(7.22)***
(18.58)***
-6.9860
-17.9083
-5.1970
-8.6096
(-27.27)***
(-75.68)***
(-15.92)***
(-36.15)***
-2.8463
46.5742
89.3016
15.8024
(-5.31)***
(31.91)***
(89.31)***
(20.16)***
0.5289
0.3423
-0.5403
0.1525
(56.50)***
(14.58)***
(-25.50)***
(10.24)***
2.1798
1.1275
-14.2033
-2.2669
(12.98)***
(11.07)***
(-50.18)***
(-17.40)***
-4.8357
-2.6922
-10.1267
-3.7227
(-182.02)***
(-38.34)***
(-74.26)***
(-26.70)***
0.5229
0.3512
-0.2282
0.1539
(70.89)***
(15.71)***
(-8.76)***
(10.19)***
……….
………
……….
2.3746
……….
………
……….
(29.40)***
89.1156
-328.6384
-778.5007
-72.0586
(24.70)***
(-24.67)***
(-88.61)***
(-8.80)***
Number of observations
217
237
251
823
Number of groups (N)
34
35
35
35
Number of instruments
Wald [2 ],
51
51
51
441
[9],3220000***
[9],7030000***
[9],1820000***
[10],8170000***
-1.1818{0.237}
-0.8738{0.382}
COMPPC lag 1 (COMPPC_1)
COMPPC lag 2 (COMPPC_2)
Institutional quality (INS)
h
Home-country income (lnY )
f
Host-country income (lnY )
Rate of inflation (INF)
Bank credit to private sector (lnPSC)
Real exchange rate (lnRXR)
Real deposit interest rate (RIR)
Regulatory environment (D9_11)
Constant term (constant)
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
0.9107{0.362}
-0.9071{0.364}
Sargan test for over-identifying restrictions:
[2],
Source: Author‟s estimation
[41],21.0029
[41],29.2380
[41],30.4453
denotes 5(1) per cent respectively.
2-step robust z-statistics in ( ), z-probabilities in { }
[431],31.3146
**(***)
The reason is that the majority of the SSA countries that embarked upon financial liberalisation
in the 1980s actually started implementing the programme in the latter years of the decade
(see Table 5.1 in Chapter Five). In this regard, the possibility of SSA temporary migrants
finding it relatively more costly and frustrating to remit through official channels and to invest at
158
Stellenbosch University http://scholar.sun.ac.za
home is high. Also, a fall in the real per capita income in migrant-host countries could dampen
the aspiration of temporary migrants to seek permanent residential status and rather encourage
higher return of temporary migrants, hence the need to remit more in the 1980s. Finally, selfinterest seeking temporary migrants who remit mainly because they consider their family at
home as insurance in the event of undesirable economic shocks in the host country, are likely
to remit more even as their real earnings fall and the general economic prospects in the host
country become bleak.
In line with the altruistic theory, migrant-sending SSA countries received more remittances from
temporary migrants when the real per capita GDP PPP declined at home. This finding is
consistent throughout the three decades. In fact, during periods of economic recession as
reflected in the reduced real per capita GDP in the 1990s (see Table 3.1), a one percentage
decrease in home-country income stimulated as much as a US$17.91 increase in COMPPC
received compared to US$6.96 in the 1980s and US$5.20 in the 2000s. Thus, unlike
permanent SSA migrants, temporary SSA migrants seem to remit more for altruism rather than
for self-interest investment purposes.
With a one percentage real depreciation of the national currency of a typical migrant-sending
SSA country against the currency of the migrant-host country, COMPPC received in SSA
declined by US$4.84 in the 1980s, US$2.69 in the 1990s, US$10.13 in the 2000s, and
US$3.72 for the overall period, 1980-2009. Although this result is applicable to both altruism
and self-interest economic theories, given the consistency with which it is parallel to the sign of
home-country income, altruism seems the more likely underlying reason behind remittances
received by SSA from its temporary migrants since 1980. The altruistic theory of migrant
remittances predicts that where the amount remitted is fixed in home-country denominated
currency, migrants tend to remit less when the home-country currency depreciates against the
host-country currency because a smaller amount of the foreign currency (say, the French franc)
would be equivalent to the usual nominal amount of remittances sent by migrants in the homecountry currency.
In conformity with the altruistic theory, between 1980 and 2009, the overall impact of private
sector access to bank credit is negative on COMPPC received in SSA. This is notwithstanding
the fact that, in the 1980s and the 1990s, increased access to private sector credit was a
positive determinant of COMPPC received in SSA. In the 2000s, a one percentage increase in
159
Stellenbosch University http://scholar.sun.ac.za
private sector access to bank credit in migrant-home country reduced COMPPC received by
the sub-region by about US$2.27. A possible explanation for this result within the context of
altruistic theory is that in the early years of financial liberalisation, private sector credit in SSA
was very low due to the underdevelopment of the financial market. Therefore, initially,
temporary migrants might have remitted more to alleviate credit constraints at home since the
marginal improvement in private sector access to credit in migrant-home countries might still be
low in the early years of financial liberalisation. However, as the financial markets developed
with improved private sector access to competitive credit over time, the incentive for remitting
by temporary migrants for this specific purpose might have waned.
Generally, an improvement in the quality of institutions impacted positively on remittances
received by SSA from its temporary migrants between 1980 and 2009, although the effect of
institutions was not statistically significant at the conventional statistical levels during the initial
years of financial liberalisation, 1980-1989. Holding all other factors constant, with a one
percentage improvement in governance and democratic dispensation in SSA, COMPPC
received by migrant-sending countries within the sub-region increased by US$0.56 in the
1990s, US$0.39 in the 2000s, and US$0.54 over the entire study period, 1980-2009. The
possible reason for this finding is that as political risks are reduced and state governance
improves at home, temporary migrants are more likely to return home instead of using the
geopolitical tensions at home to seek asylum or permanent residential status abroad. With the
higher return rate of temporary migrants, all other things remaining equal, COMPPC received
by migrant-sending SSA countries are increased for both altruistic and non-altruistic reasons.
For the entire study period, the impact of home-country real deposit interest rate on COMPPC
received was positive in SSA. In the 1980-89 period, a one percentage increase in the homecountry real deposit interest rate increased COMPPC received in SSA by US$0.52. In the
1990s, this positive impact decreased to US$0.35. In the 2000s, a similar rise in the homecountry real deposit interest rate reduced COMPPC received in SSA by US$0.23. These
results suggest that the apparent self-interest investment motive that might have stimulated
temporary migrants in the 1980s steadily faded away and by the 2000s, altruism seems to have
emerged more dominantly. It is also possible that the self-interest motive might still be a
dominant reason behind the compensation of employees even in the 2000s except that
temporary migrants now invest more in the host country than at home due to higher rate of real
returns on investment in the migrant-host country.
160
Stellenbosch University http://scholar.sun.ac.za
A one percentage increase in the home-country rate of inflation increased COMPPC received
by US$0.53 in the 1980s, US$0.34 in the 1990s, and US$0.15 for the overall period, 19802009, in migrant-sending SSA countries. In the 2000s, however, the impact of inflation on
COMPPC received in SSA was negative. This seems to suggest that in the 1980s and the
1990s, temporary SSA migrants might have remitted more for altruism, but in more recent
years, these migrants are becoming more self-interest investment driven. Here again, the
evidence seems to suggest that unlike in the initial years of implementing financial liberalisation
in SSA temporary migrants are becoming increasingly motivated by the self-interest investment
motive rather than by altruism.
The varying impact of home-country macroeconomic factors on the amount of COMPPC
received in SSA is evident in the decade-based analysis. It is apparent that besides homecountry income, host-country income, and real exchange rate, the estimated parameters of the
remaining factors failed to carry the same sign from the 1980s to the 2000s. There seems to be
sufficient evidence to conclude that temporary migrants are generally altruistic, but the extent of
this altruism seems to be fading in favour of the self-interest economic motive.
In columns A-B, B-C and A-C of Table 4.4.1, the empirical results on the extent to which the
decade-based parameter estimates of the inflows of compensation of employees in SSA from
1980 to 2009 differ over time are reported. Column A-B reports the results on comparing the
parameter estimates of the 1980-89 decade with the corresponding parameter estimates of the
1990-99 decade. In column B-C of Table 4.4.1, the results on comparing the parameter
estimates of the 1990-99 decade with the corresponding parameter estimates of 2000-09
decade are reported, whilst the results on the statistical difference between the parameter
estimates of the 1980-89 decade and the 2000-09 decade are reported in column A-C. The
reported „differential‟ z -statistics in columns A-B, B-C and A-C show that each of the estimated
parameters reported in Table 4.4 is statistically different from decade to decade. Stated
differently, at one per cent level of statistical significance, each of the macroeconomic factors
influencing the inflows of COMPPC received in SSA between 1980 and 2009 had a decadebased evolving impact in an apparent response to the macroeconomic policy environment in
the remittance-receiving SSA country.
161
Stellenbosch University http://scholar.sun.ac.za
Table 4.4.1: Results of Decade-Based Parameter Evolution and Instability Tests for Compensation of Employees
Estimated Decade-Based Results
COMPPC lag 1 (COMPPC_1)
COMPPC lag 2 (COMPPC_2)
Decade-Based Rolling
Estimated Results
A
B
C
D
E
1980-89
1990-99
2000-09
1985-1994
1995-2004
0.6704
[0.0016]
0.7143
[0.0009]
0.6996
[0.0020]
0.4992
[0.0018]
{424.00}***
{816.56}***
{352.48}***
0.0111
[0.0005]
-0.0993
[0.0015]
0.0016
[0.0013]
-0.0982
[0.0021]
-0.2497
[0.0063]
Non-Overlapping Decade-Based
Coefficient Stability Test Results
A-B
B-C
A-C
Overlapping Decade-Based Coefficient Stability
Test Results
A-D
B-D
B-E
C-E
1.1879
[0.0017]
-0.0440
[0.0007]
0.0147
[0.0011]
-0.0293
[0.0004]
0.1712
[0.0003]
0.2151
[0.0010]
-0.4736
[0.0008]
-0.4883
[0.0003]
{272.01}*** {708.15}***
{-61.90}***
{13.23}***
{-73.15}***
{658.42}***
{221.79}***
0.1103
[0.0010]
-0.1009
[0.0002]
0.0095
[0.0008]
0.1093
[0.0016]
-0.0011
[0.0006]
0.1504
[0.0049]
0.2513
[0.0051]
{-584.65}*** {-1627.53}***
{23.07}***
{-67.68}***
{1.30}
{-47.29}***
{-39.50}***
{111.45}***
{-458.55}***
{12.29}***
{68.31}***
{-1.72}*
{31.01}***
{49.56}***
Institutional quality (INS)
-0.0171
[0.0171]
0.5569
[0.0029]
0.3930
[0.0544]
-0.2142
[0.0046]
-0.3961
[0.0477]
-0.5740
[0.0142]
0.1639
[0.0515]
-0.4101
[0.0373]
0.1971
[0.0126]
0.7711
[0.0016]
0.9530
[0.0447]
0.7891
[0.0068]
{-1.00}
{189.29}***
{7.22}***
{-47.02}***
{-8.31}***
{-40.54}***
{3.18}***
{-10.99}***
{15.70}***
{478.94}***
{21.31}***
{116.55}***
Home-country income (lnYh)
-6.9860
[0.2562]
-17.9083
[0.2366]
-5.1970
[0.3264]
-9.2130
[0.1209]
1.6488
[0.5646]
10.9223
[0.0195]
-12.7112
[0.0898]
-1.7889
[0.0703]
2.2271
[0.1352]
-8.6952
[0.1157]
-19.5570
[0.2366]
-6.8458
[0.3264]
{-27.27}***
{-75.68}***
{-15.92}***
{-76.17}***
{2.92}***
{558.75}***
{-141.52}***
{-25.46}***
{16.47}***
{-75.17}***
{-82.65}***
{-20.97}***
-2.8463
[0.5360]
46.5742
[1.4595]
89.3016
[0.9999]
39.4113
[0.7525]
30.1835
[0.6356]
-49.4205
[0.9235]
-42.7274
[0.4596]
-92.1479
[0.4639]
-42.2576
[0.2165]
7.1630
[0.7070]
16.3907
[0.8240]
59.1181
[0.3643]
{-5.31}***
{31.91}***
{89.31}***
{52.37}***
{47.49}***
{-53.51}***
{-92.96}***
{-198.65}***
{-195.17}***
{10.13}***
{19.89}***
{162.26}***
0.5289
[0.0094]
0.3423
[0.0235]
-0.5403
[0.0212]
0.1089
[0.0152]
-0.0775
[0.0446]
0.1866
[0.0141]
0.8826
[0.0023]
1.0692
[0.0118]
0.4200
[0.0059]
0.2333
[0.0082]
0.4198
[0.0211]
-0.4628
[0.0234]
{56.50}***
{14.58}***
{-25.50}***
{7.15}***
{-1.74}*
{13.22}***
{385.40}***
{90.38}***
{71.42}***
{28.32}***
{19.92}***
{-19.80}***
2.1798
[0.1679]
1.1275
[0.1018]
-14.2033
[0.2831]
3.7781
[0.0892]
-9.7486
[0.3211]
1.0523
[0.0661]
15.3308
[0.1812]
16.3830
[0.1151]
-1.5983
[0.0787]
-2.6506
[0.0127]
10.8761
[0.2192]
-4.4547
[0.0381]
{12.98}***
{11.07}***
{-50.18}***
{42.36}***
{-30.36}***
{15.92}***
{84.61}***
{142.31}***
{-20.30}***
{-209.37}***
{49.61}***
{-117.07}***
-4.8357
[0.0266]
-2.6922
[0.0702]
-10.1267
[0.1364]
-4.8532
[0.1075]
-1.9484
[0.0661]
-2.1434
[0.0436]
7.4344
[0.0662]
5.2910
[0.1098]
0.0175
[0.0810]
2.1609
[0.0373]
-0.7438
[0.0041]
-8.1783
[0.0702]
{-182.02}***
{-38.34}***
{-74.26}***
{-45.13}***
{-29.46}***
{-49.10}***
{112.39}***
{48.19}***
{0.22}
{57.90}***
{-182.31}***
{-116.45}***
0.5229
[0.0074]
0.3512
[0.0224]
-0.2282
[0.0261]
0.0813
[0.0117]
-0.0540
[0.0446]
0.1716
[0.0150]
0.5794
[0.0037]
0.7510
[0.0187]
0.4415
[0.0044]
0.2699
[0.0106]
0.4052
[0.0222]
-0.1742
[0.0185]
{6.92}***
{-1.21}
{11.46}***
{157.03}***
{40.23}***
{101.04}***
{25.44}***
{18.23}***
{-9.40}***
-314.0904 -282.4071
[8.0930]
[9.2079]
417.7540
[9.7135]
449.8620
[4.5357]
867.6163
[5.1778]
403.2060
[4.4851]
-14.5480
[5.2284]
-46.2313
[4.1135]
-496.0940
[0.4222]
Host-country income (InYf)
Rate of inflation (INF)
Bank credit to private sector (InPSC)
Real exchange rate (lnRXR)
Real deposit interest rate (RIR)
{70.89}***
{15.71}***
{-8.76}***
89.1156
[3.6079]
-328.6384
[13.3214]
-778.5007
[8.7857]
{24.70}***
{-24.67}***
{-88.61}***
{-38.81}***
{-30.67}***
{43.01}***
{99.18}***
{167.57}***
{89.90)***
{-2.78}***
217
237
251
237
286
227
244
234
227
237
262
268
Number of groups
34
35
35
35
35
35
35
35
35
35
35
35
Number of instruments
51
51
51
51
51
51
51
51
51
51
51
51
3220000***
7030000***
1820000***
407000*** 2320000***
5125000***
4425000***
2520000***
1813500***
3718500***
4675000***
2070000***
0.911(0.36)
-0.907(0.36)
-1.182(0.24)
-0.459(0.65) 1.192(0.23)
-
-
-
-
-
-
-
21.002(0.99)
29.238(0.92)
30.45(0.89)
26.984(0.95)33.784(0.78)
-
-
-
-
-
-
-
Constant term (constant)
Number of observations
Wald statistic
A-B 2ⁿᵈ-order autocorrelation test
Sargan over-identifying restrictions
Source: Author‟s estimation
{-11.24}*** {-1174.94}***
*/**/***denotes significant at 10/5/1 per cent statistical levels respectively. Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
162
Stellenbosch University http://scholar.sun.ac.za
In the strict sense of stability of the estimated decade-based coefficients over time, the results
as reported in columns A-D, B-D, B-E and C-E of Table 4.4.1 suggest that, generally, the
statistical differences in each of the estimated decade-based parameters is statistically
consistent over time and this provides the statistical evidence for evolution and instability in the
estimated decade-based parameters. The only exception to this finding is the reported z statistic associated with real bilateral exchange rate reported in column A-D of Table 4.4.1
suggesting that there is evidence against statistical instability of the estimated coefficient when
the 1980-89 decade and the 1985-94 decade are compared.
4.7 CONCLUSIONS, POLICY IMPLICATIONS AND RECOMMENDATIONS
In harmony with the objectives specified, this chapter examined at both the aggregated and
disaggregated levels, the macroeconomic factors that influence the flow of migrant remittances
to SSA. To verify if the impact of the macroeconomic factors that influence migrant remittance
inflows in SSA vary over time, separate empirical analyses were carried out for each of the past
three decades along with the overall study period analysis. Given the results obtained and in
response to the underlying research questions, the study concludes that, generally:
i.
Both host-country and home-country macroeconomic factors play a crucial role in
determining the amount of officially reported remittances received in SSA between 1980
and 2009. Of these factors, however, host-country factors viz. migrant-host country
income, and the enforcement of laws and regulations prohibiting the use of informal
channels in remitting were found to be the most positive determinants of remittances
received in SSA. Concerning home-country macroeconomic factors, overall, real
bilateral exchange rate, real income per capita PPP and institutional quality impacted
negatively on migrant remittances received, whilst bank credit to private sector, inflation,
and real deposit interest rate had positive impact on remittance inflows in SSA. Apart
from these factors, the amount of remittances received over the past two years also
influence how much remittances are received at any particular point in time.
ii.
The impact of macroeconomic factors on migrant remittances received in SSA varied
over time, but the pattern of this varying effect is largely dependent upon the general
macroeconomic performance rather than on any specific programme such as financial
liberalisation. For example, in the 1980s and the 2000s when real income levels were
relatively high, migrant remittances were pro-cyclical, and seemed to be driven more by
163
Stellenbosch University http://scholar.sun.ac.za
the self-interest investment motive rather than by altruism. Unlike in the „bad times‟ of
the 1990s, during the „good times‟ of the 1980s and the 2000s, it was migrant-sending
SSA countries with lower rates of inflation, higher income growth, stronger currencies
and higher real deposit interest rates that mobilised more remittances through the
official channels. This implies that though the pursuit of financial liberalisation has a
substantial potential of enhancing the mobilisation of international migrant remittances
in SSA through official channels, this ambition can only be successful provided
favourable macroeconomic environment exists in the migrant-sending country. Stated
differently, financial liberalisation should be seen only as a necessary but not a
sufficient condition for mobilising remittances from SSA migrants outside the sub-region
since its impact on remittance inflows is contingent upon the macroeconomic
fundamentals of the migrant-sending countries.
iii.
Macroeconomic factors, to a reasonable extent, impacted differently on workers‟
remittances and migrant remittances received in SSA over the past three decades even
though the results from these two measures of remittances are more consistent than in
comparison with compensation of employees. Both workers‟ remittances and migrant
remittances seem to be driven by the self-interest economic motive. For the entire study
period, real deposit interest rate and inflation were not statistically significant in
determining workers‟ remittances received in SSA, but these two variables had
significant positive impact on migrant remittances received in the sub-region. The
impact of home-country income on migrant remittances was negative but positive in the
case of workers‟ remittances. The effects of the remaining macroeconomic variables,
viz. host-country income, bank credit to the private sector, regulatory environment,
institutional quality and real exchange rate on migrant remittances and workers‟
remittances received in SSA were the same in terms of statistical direction. However, in
terms of economic importance, generally, each of these variables exerted higher impact
on migrant remittances than on workers‟ remittances, except for the „political economy‟
variables - institutional quality and regulatory environment. This finding implies that it
may not be appropriate to use only workers‟ remittances in studies that aim at trying to
provide a complete insight into remittances from international migrants.
iv.
Macroeconomic factors were important determinants of compensation of employees
received in SSA between 1980 and 2009. Overall, whereas host-country income,
164
Stellenbosch University http://scholar.sun.ac.za
regulatory environment, quality institutions, real deposit interest rate and rate of inflation
had positive effects, home-country income, bank credit to private sector, and real
exchange rate had negative effects on compensation of employees. This finding
appears to lend support for the validity of the altruistic theory of remittance inflows.
Thus, while permanent migrants from SSA seem to be influenced by the self-interest
investment motive, temporary migrants from the sub-region are likely to be more
altruistic.
The conclusions of this study are imperative with a number of policy implications for strategies
aimed at attracting optimal migrant remittances to SSA through the formal financial sector. The
key policy implication is that since host-country factors are exogenous to remittance-recipient
countries and because these industrialised countries host migrants from other countries of the
developing world that have been receiving higher remittances than SSA as a sub-region, SSA
cannot attribute its low receipt of officially reported remittances to factors in the migrant-host
countries. By implication, the low receipt of migrant remittances by SSA countries should be
ascribed to the absence of relevant and effective macroeconomic policies for the mobilisation
of remittances from their citizens living abroad. Explicitly, countries that are receiving higher
official migrant remittances today are doing so largely because these countries have put some
policy measures in place for this specific purpose. Therefore, to mobilise increased remittances
from SSA migrants through official channels, it is recommended that SSA countries should
design attractive policies that will induce its nationals living and working outside the sub-region,
to remit home conveniently. More specifically, policy makers in SSA should:
i.
advance stable and credible macroeconomic policy environment through reduction in
the rate of inflation, improvement in economic performance which reflects in higher real
per capita income, and stronger national currency in the international financial market
so as to encourage private sector savings and investment. Self-interest seeking
migrants may be encouraged to remit more funds home for investment purposes if the
macroeconomic conditions at home are favourable or investment friendly. For instance,
with higher growth in home-country income, not only migrants but recipients of
remittances be will encouraged to invest as the domestic market expands;
ii.
ensure that they encourage stronger institutions through improved democratic
governance and freedom from strife as in more recent years, since quality institutions
165
Stellenbosch University http://scholar.sun.ac.za
impact positively on the inflows of workers‟ remittances and compensation of
employees. To achieve this, pragmatic measures must be put in place to reduce
corruption (or the perception thereof), improve national security and peace, and create
a conducive environment through the enactment of laws that protect the interest of
investors and entrepreneurs, whether resident at home or abroad;
iii.
continue to rigorously pursue prudent financial market liberalisation programmes which
are expected to deregulate exchange rates, promote competition among banks and
other like-service providers, including post offices working with Money Transfer
Operators (MTOs). In SSA, the most notable MTOs are Western Union and
MoneyGram for which many financial institutions, especially commercial banks and post
offices, act as agents or intermediaries. When financial liberalisation leads to
competition in the financial market, financial institutions are expected to become more
efficient, resulting in reduced money transfer fees, the introduction of innovative and
diversified financial products and services, and expansion and wider coverage with
more outlets at home and abroad. This is critical because for as long as SSA migrants
find the patronage of informal money transfer channels cheaper, safer, more convenient
and accessible, the sub-region cannot improve upon the mobilisation of remittances
from its nationals living abroad through the formal transfer channels.
iv.
roll out strategic policies under the pursuit of financial liberalisation programmes that will
motivate commercial banks to reach out to migrants in their host countries. For
instance, commercial banks can open outlets in major migrant-host nations, offer
preferential interest rates on remittances saved, convert asset holdings in local
currencies at a premium rate, and invest saved remittances in high-yielding financial
instruments. It should also be feasible for local banks to open a joint account for
migrants and their main target remittance-recipients. Banks can even pay „assured
remittances‟ on behalf of migrants under special terms and conditions;
v.
not only stabilise the local currency in the international markets under the pursuit of
financial liberalisation programme but also integrate foreign exchange markets so as to
abolish the existence of dual exchange rates, which hitherto, create incentive for
migrants to use unofficial channels for transferring funds. Normally, under a dual
exchange rate regime, there is a wide disparity between a relatively lower official rate
166
Stellenbosch University http://scholar.sun.ac.za
and a relatively higher parallel rate. Under this condition, more local currencies are
received in any given amount of foreign currency as higher premiums exist in „black‟
foreign markets where operators do not pay any commission on their earnings. Also,
high exchange rate volatility can provide an incentive for currency hedging or hoarding
which can ultimately reduce the patronage of official channels to remit; and
vi.
design special incentive packages, including zero tax on remittances received, special
remittance agreements with major migrant-host countries, the regulation of informal
intermediaries in the money transfer market, the issuing of special foreign currency
denominated bonds targeted at the Diaspora communities, establishing „remittance
banks‟ at home with branches, and creating opportunities for social security
contributions from abroad, in order to attract SSA citizens resident abroad to remit funds
home through the formal money transfer channels.
167
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 4
Table A4.1: Summary of Empirical Studies on Macroeconomic Determinants of Remittances
Author(s), Year
Case Study
Elbadawi
and
Rocha (1992)
Six
North
African
and
European
countries
(Algeria,
Morocco,
Portugal,
Tunisia, Turkey,
Yugoslavia)
Study
Period
1977-1989
Model & Estimation
Method
Correlation analysis for
six countries.
Fixed-Effects
model
involving a panel of five
countries
(excluding
Algeria)
Variables Included
Key Finding(s)
Dependent: Logarithm of remittances; remittances per
migrant worker; logarithm of remittances per capita.
(Remittances here mean remittances from both
categories of foreign residents either for 12 or more
months or less than 12 months).
Explanatory: Migrant stock, native-country real GDP,
black market premium, inflation, duration of stay in host
country, dummy for political stability, host-nation GDP
and deposit interest rate differential between home and
host country.
Migrant stock and real GDP in host
country have positive effects on
remittance
inflows.
Black
market
premium, domestic inflation rate, and
length of stay negatively impact on
remittance inflows. Thus, although
migrant stock may positively affect
remittances in migrant-home countries,
an
ageing
labour
force
abroad
decelerates remittance inflows. Also,
special incentive schemes cannot
substitute for a stable and credible
macroeconomic policy.
Remittance
inflows
are
positively
influenced by migrants‟ income, inflation,
real discount (or deposit) interest rate
and number of migrants. Exchange rate,
Unemployment rate and home country
income have no statistical significant
impact
Land dummy positively impact on
remittances; GDP per capita and female
activity have negative effects on inward
remittances
whilst
inflation
was
statistically insignificant.
Lianos (1997)
Greece
1961-1991
for Germany;
1981-1991
for Belgium;
1980-1991
for Sweden.
A set of single equation
OLS
models
on
bilateral
remittances
from Belgium, Germany
and Sweden to Greece
Dependent: Unpublished data obtained from the Bank
of Greece on funds sent home by Greek migrants in
Belgium, Germany, Sweden)
Explanatory: migrant‟s income, family income, rate of
interest, rate of inflation, exchange rate, rate of
unemployment, number of migrants
Buch et al. (2002)
145 countries
1970-99 for
Correlation.
1990-99
averages for
OLS
Correlations coefficient.
Cross-sectional
OLS
based
on
each
country‟s averages of
the 1990s
Jadhav (2003)
India
1988(2)2003(1)
(quarterly
data)
Single equation OLS
estimation
Dependent (for OLS): Remittance (WREM plus private
capital flows) ratio to GDP. For correlation analysis,
remittances/GDP and growth rates of remittances were
used.
Explanatory: GDP per capita or Human Development
Indicator (HDI) or index of economic freedom
representing the level of country development,
macroeconomic instability, inflation, dummy for Island
states, and female economic participation
Dependent: Log of WREM
Explanatory: Price of international crude oil used as an
indicator of economic activities in oil exporting Middle
East and Gulf regions as Indian migrant hosts, US GDP
168
Improved economic activities of migrant
host countries (both oil exporting and
non-oil exporting hosts) and exchange
rate depreciation are positive drivers of
Stellenbosch University http://scholar.sun.ac.za
Aydaş
(2004)
as economic activity indicator of non-oil Indian migrant
hosts, interest (deposit) rate differential (domestic
minus LIBOR), nominal exchange rate.
Dependent: Remittance data restricted to only cash
transfers
Explanatory: Log of: migrant stock, host-country per
capita income, home-country per capita income,
domestic growth, and black market premium. Also,
dummy for military regime to capture political instability,
interest rate differential, domestic inflation, and real
currency overvaluation.
Turkey
1964-1993
Single equation OLS
Bouhga-Hagbe
(2004)
Morocco
(Bilateral study
with France as
host nation for
Moroccan
migrants)
1993(1)2004(4)
(quarterly
data)
Co-integration graphing
and
error-correction
(ECM) modelling by
OLS
Dependent: Logarithm of remittances (defined as total
migrant worker transfers) an equivalence of
WREM+COMP+ migrant transfers (MT).
Explanatory: Wages in France, financial assets held by
Moroccan emigrants, interest rate differential, bilateral
exchange rate, and real GDP
Freund
and
Spatafora (2005)
104 countries in
SSA, EAP and
ECA
1995-2003
Fixed effects
data estimation.
panel
Gupta (2005)
India
1990-2003
A
set
of
equations OLS
single
Dependent: Remittances (WREM+COMP+MT with
many adjustments in absence of data) in logarithmic
levels, per capita, and per emigrant.
Explanatory: Domestic output, domestic per capita
income, stock of migrant workers, dummy for dual
exchange rate, transfer or service fees, exchange rate
spread, host country (the country which has the largest
share of a sampled country‟s migrant workers) per
capita income
Dependent: Logarithm of real remittances defined as
private transfers on current accounts of BoPS
Explanatory: Trend, migrant stock, lagged dependent
variable, migrant earnings, US non-agricultural
employment, credit rating of India, political uncertainty,
stock market return, exchange rate variation, drought,
annual changes in LIBOR, changes in oil prices,
issuance of bonds, dummy for Asian financial crisis,
et
al.
169
remittance flows to India. Deposit interest
rate differential does not affect remittance
flows to India.
Results reported for 1979-1993 show that
stock of migrants and real overvaluation
of domestic currency were insignificant.
Interest rate differential and host-country
income
positively
determines
remittances.
Home-country
income,
inflation, black market premium and
political instability negatively affect
remittance inflows.
Wages in France and interest rate
differential and are positive long-run
determinants. No evidence of portfolio
diversification
motives
behind
remittances in the long run. In the short
run, volume of real estate construction in
Morocco by Moroccan migrants is a
positive driver of remittances into
Morocco. Exchange rate depreciation
enhances remittance inflows in the short
run.
Transfer fee, host-country per capita
income and dual exchange rate
adversely affect remittance inflows.
Migrant stock and home-country per
capita income positively impact on
remittance inflows. Exchange rate spread
not significant.
Migrant stock, earnings of migrants, and
host-country economic environment are
positive determinants of remittances
received by India. Interest differential not
statistically significant.
Stellenbosch University http://scholar.sun.ac.za
and dummy for post-September-11, 2001
Dependent: REMGDP and REMPC where remittances
are defined as WREM+COMP+MT.
Explanatory: Lagged dependent variable, nativecountry GDP per capita, unemployment rate, domestic
credit to private sector/GDP, dummy for wars,
institutional development measured by an index that
takes the value between 1 and 4.5, openness, per
capita income growth rate.
Schrooten (2005)
24
transition
countries
1990-2003
Dynamic panel data
estimation
technique
following
firstdifferenced
Generalised Method of
Moment (GMM)
Akkoyunlu
and
Kholodilin (2006)
Turkey
(focus
on
bilateral analysis
involving Turkish
migrants
in
Germany)
1962-2004
Cross-correlations,
unrestricted
bivariate
VARs and Grangercausality test
Two forms of remittances were tried. Bilateral real total
remittances and bilateral real remittances per capita.
Bilateral remittances (measured directly from official
source – Deutsche Bundesbank) in this study imply
remittances from Germany to Turkey.
Real German GDP, real Turkish GDP, exchange rate,
CPI and number of migrants
Niimi and Özden
(2006)
85 countries
2000 (due to
limited time
series data
on
migrant
stock)
Cross-country analysis
involving
single
equation OLS (ignoring
endogeneity)
and
Instrumental Variable
(IV) estimation (taking
into
account
endogeneity problems)
Schiopu
and
Siegfried (2006)
21
Western
European
countries
(as
source
countries) 7 EU
neighbouring
Differ across
nations but
averagely
between
2001-2003
A set of 11 single
equations OLS (With
the exception of income
differential and return
on financial assets the
other
explanatory
Dependent: Three forms of remittances (measured as
WREM+COMP+MT)
were
tried.
These
are
remittances/GDP ratio, logarithm of real remittances,
and the logarithm of real remittances per capita.
Explanatory: Level of migration, educational level of
migrants, financial development (either bank deposit or
private sector bank credit to GDP), and economic
conditions in native-country (GDP growth and GDP per
capita)
Dependent: Logarithm of bilateral remittances (which
were not specifically defined) per migrant obtained from
Central Banks in recipient countries
Explanatory: Rate of return on financial assets, income
differentials (real per capita GDP PPP of host country
minus real per capita GDP PPP of home country),
170
Remittances per capita and remittances
per GDP are driven by similar factors.
Remittances are highly driven by
unemployment rate in native country,
GDP per capita, and higher international
integration of host country‟s real sector
decrease
remittances.
Institutional
development has no impact on
remittances per capita. War dummy has
no effect on remittances/GDP but
remittances per capita increase during
war times. Remittances act as substitute
for well-performing banking sector.
Remittances response more positively to
changes in economic activity in host
country.
No
causality
between
remittances and real GDP per capita or
with real growth in German GDP. No
correlation between growth rate of
German real GDP and growth rate of real
remittances per migrant. No correlation
between annual growth rates of Turkish
real GDP and annual growth rates of real
remittances per migrant.
Stock of migrants is the main determinant
of remittances. Education level of migrant
relative to population in home country,
size of economy and level of economic
development adversely affect remittance
inflows. Financial development positive
but largely insignificant and where private
sector credit is occasionally significant, it
is generally positive.
Increases in migrant skills, GDP
differential between home and host
countries promote remittances received.
Large informal sector in migrant resident
country and fund transfer fees depress
official
remittances.
Interest
rate
Stellenbosch University http://scholar.sun.ac.za
countries
Vargas-Silva and
Huang (2006)
Two cases: (1)
Net remittances
between US and
rest of the world
(ROW).
(2)
Remittances to
Mexico from US
1981(1)2003(4):
Quarterly
data
AmuedoDorantes
(2007)
111 developing
countries with a
comparative
analysis of 19
Small
Island
Developing
States
(SIDS)
and
92
developing
countries
Sri Lanka
1990-2003
11
developing
countries
in
Asia,
Europe
Differ across
countries, but
generally
et
al.
Lueth and RuizArranz (2007a)
Lueth and RuizArranz (2007b)
1996-2004
(Quarterly
observations)
variables
were
not
included at the same
time)
Vector Error Correction
Model
(VECM)
supported by Variance
Decomposition,
Impulse
Response,
Granger-causality
Panel
VAR
with
variance
decompositions,
and
impulse
response
functions.
Three
estimations: 19 SIDS
only, 92 developing
countries only, and 111
full sample developing
countries
Co-integrating
single
equation OLS for longrun
parameter
estimations and
Vector Error Correction
Model (VECM) for short
run
parameter
estimations.
Pooled OLS, Fixed
Effects and Random
Effects
panel
data
migration, skill level, income inequality, remittance
transfer cost, unofficial economic activity, and rate of
return on real estate
US study: Net aggregate remittances (i.e. private
remittances plus other transfers) with ROW). US
Federal Funds Rate (US FFR), US money supply (US
M2), US CPI, US unemployment, two indices of
economic conditions (exchange with the US$ and home
78
country inflation) from ROW .
Mexico study: Inward remittances (i.e. Mexico‟s credit
current transfers)). US FFR, US M2, US CPI, US
unemployment; and Mexican GDP, CPI and domestic
currency exchange rate with the US$ (to represent
home country economic conditions). All variables in real
terms or logarithm.
Dependent: WREM as ratio of GDP
Explanatory: Multilateral real effective exchange rate,
natural disasters and official foreign aid.
differential not significant determinant.
Dependent: Remittances (WREM+COMP+MT) in actual
volumes and not seasonally adjusted
Explanatory: Real GDP, CPI, exchange rate, and oil
price (simple average of UK Brent, Dubai and West
Texas crude oil prices).
Remittances and oil price positively
correlate. Real GDP and exchange rate
of home country adversely affect
remittance inflows.
Dependent: Logarithm of total bilateral remittances (as
reported by various Central Banks)
Explanatory: Nominal GDP of home country, physical
Motive to remit is mixed, but altruism
appears less of a factor than is widely
believed. Remittances are positively
78
Generally,
host-country
economic
conditions are relatively important in
driving remittances. For Mexico, none of
the home-country economic factors was
significant. US M2 explains the largest
percentage of remittance variance. A
positive shock to US M2 elicits a positive
response from both measures of
remittances. US FFR, US inflation, US
unemployment Granger-cause inward
remittances to Mexico.
Real effective exchange rate depreciation
enhances remittance inflows. 2.4% of
remittances‟ error forecast variance
explained by the remaining system
variables. Remittances explain 27.5% of
foreign aid‟s error forecast variance.
Remittances increase following a disaster
shock but that of aid is more robust.
ROW is proxied by Brazil, Colombia, the Dominican Republic, El Salvador and Mexico because these were the largest five recipients of remittances from the US at the time of this
study.
171
Stellenbosch University http://scholar.sun.ac.za
and the Middle
East
ranged 19802004.
Average
period is 9
years
estimation
Adenutsi
and
Ahortor (2008)
Ghana
1983(4)2005(4)
Static and dynamic
VAR,
Vector
Error
Correction
Model
(VECM),
Impulse
Response
Functions
and
Variance
Decompositions
Elkhider
(2008)
Morocco
1970-2006
VAR and Vector Error
Correction
Model
(VECM)
Dependent:
Logarithm
of
total
remittances
(WREM+COMP+MT) sent by Moroccan migrants
abroad
Explanatory: Agricultural GDP, exchange rate, time
trend
Moore
and
Greenidge (2008)
15
Caribbean
countries
1987-2008
Panel
Generalised
Least Squares (Panel
GLS)
Adams (2009)
76
developing
countries
Crosssectional
data
with
each country
having a data
point
between
1995-2001
2-stage
Instrumental
variable (IV) estimation.
Instruments for migrant
skills, migrant stock
and poverty.
Dependent: Logarithm of remittances (WREM+COMP)
as a ratio of GDP.
Explanatory: Real interest rate differential, real GDP per
capita differential (the PPP measure), home country
inflation, real effective exchange rate, age dependency
ratio
Dependent: Log of REMPC where remittances are new
undefined national data assembled by the author.
Explanatory: Skill composition of migrants, home
country poverty, per capita GDP, per capita GDP
square, real deposit interest rate, exchange rate
spread, cost of remitting money, GINI coefficient,
percentage of population under 14 years and dummy
for war
Coulibaly (2009)
16
1980-2006
Panel
et
al.
LAC
VAR,
impulse
distance between a sampled country and source
countries, and a vector of potential factors influencing
remittance flows which includes GDP per capita of
home and source countries, separate dummies for
share border, common language, and political risks
proxied by country international credit rating.
Dependent: Log of remittances (WREM+COMP)
Explanatory: Interest rate (treasury bill rate), monetary
aggregate (M1), exchange rate and domestic price
level.
Dependent:
172
Remittances
(WREM+COMP+MT)
per
driven by oil prices and native-country
GDP. Remittances reduce as exports
decline, investment and political climate
of home country worsen.
Static long-run model reveals that
monetary aggregates, exchange rate,
and interest rate impact positively on
remittance inflows whilst domestic price
level negatively affect the inflow of
remittances. This result holds for the
dynamic long-run model except that
exchange rate has a negative influence
on remittances under this circumstance.
Agricultural GDP has a positive effect on
remittances received whereas exchange
rate negatively affects remittance inflows
in the long run. In the short run, however,
exchange rate positively influences
remittances. Thus, over the long run,
exchange rate depreciation does not
positively impact on the Moroccan
resident abroad.
Interest
rate
differential,
income
differential, inflation and dependency
ratio have significant positive impact on
remittances. Real effective exchange rate
(REER) negative but insignificant impact.
Skills of migrants play a crucial in
remittance determination. Higher-skilled
labour exporters receive less REMPC
than countries which export a larger
proportion of low-skilled migrants. Real
deposit interest rate impact positively on
REMPC. Poverty incidence at home,
exchange rate spread, cost of remitting,
and periods of war do not affect REMPC.
Host-country
GDP
and
interest
Stellenbosch University http://scholar.sun.ac.za
countries
Shahbaz
and
Aamir (2009)
Pakistan
1971-2006
Akkoyunlu (2010)
Turkey
(Bilateral study
involving Turkish
migrants
in
Germany)
1962-2005
Singh
(2010)
36
countries
1990-2005
et
al.
SSA
response
functions,
variance
decompositions
Autoregressive
Distributive Lag (ARDL)
model
capita.
Explanatory: Growth rates in host country (US) GDP
and home-country GDP, interest rate differential
Dependent: WREMGDP and WREMPC.
Explanatory: Manufacturing output used as proxy for
economic activity at home country, world GDP, inflation,
REER, world interest rate, and secondary school
enrolment as proxy for skilled labour at home.
Single equation cointegration and errorcorrection models by
OLS.
General-tospecific approach for a
parsimonious
unrestricted
general
model.
Fixed Effects Panel
Data (Single Equation
and 2SLS) Estimation
Dependent: Remittances from Germany (obtained from
Balance sheets of Bundesbank) as a ratio of GDP.
Explanatory: Real Turkish GDP per capita, real German
GDP per capita, stock of Turkish migrant workers in
Germany, real exchange rate, government instability.
Dependent: Measurement of remittances was
inconsistent. It ranged from constructing new dataset,
WREM+COMP+MT to using only “other current
transfers” due to data unavailability. In the empirical
model, logarithm of remittances/GDP was used as the
dependent variable.
Explanatory: Real GDP per capita of home country, real
GDP per capita of host country, migrant
stock/population, dual exchange rate regime,
institutional quality (proxied by international country risk
index which is on financial risk as far as the debt
payment of a country is concerned), financial
development (M2/GDP or domestic credit/GDP), real
bilateral exchange rate against the US dollar, interest
rate differential
Source: Author‟s compilation from various sources.
173
differential positively drive remittances to
LAC whilst increases in native-country
GDP dampens inward remittances.
In the long run, increases in world GDP,
REER and inflation impact positively on
remittance inflows, whereas world
interest
rate,
home-country
manufacturing output and secondary
school enrolment adversely affect
remittances in the long run.
Turkish income negatively impact on
remittance inflows in the long run but
stock of migrants, real exchange rate,
and political instability have positive
effects.
In the short run, German GDP had a
positive but Turkish GDP had a negative
effect on remittance inflows.
Stock of migrants residing in wealthier
nations, quality institutions and hostcountry income positively impacted on
remittance
inflows.
Also,
financial
development proxied by M2/GDP and
domestic credit/GDP impacted positively
on remittance inflows with M2/GDP
having a more robust effect. Homecountry income and interest differential
have depressing effects on remittances
inflows. Real exchange rate and dual
exchange regime had no significant
impact on remittance inflows.
Stellenbosch University http://scholar.sun.ac.za
Box A4.1: Matrices Corresponding to the Instruments Used in the Estimation
Ri 3 Ri ,2
R R
i ,3
i4
Ri Ri ,T Ri ,T 1
Ri 3
RiT
Ri 2 Ri1
R R
i ,2
i3
X i Ri ,T 1 Ri ,T 2
Ri ,2
Ri ,T 1
X ( R1 , X ), ( , ),
W (W1,W2,...,WN )
0
[ Ri1 , xi1 , xi2 ]
0
[ Ri1 , Ri 2 , xi1 , xi2 , xi3 ]
D
0
0
Wi
0
0
xi3 xi2
xi4 xi3
xiT xi,T 1
xi2
xiT
0
[ Ri1 , Ri 2 ,...Ri ,T 2 , xi1 , xi2 ,..., xiT 1 ]
0
0
0
[dRi1 , dxi1 , dxi2 ]
0
[dRi2 , dRi3 , dxi2 , dxi3 , dxi4 ]
0
0
Wi L
0
0
0
[dRi2 ,....dRi ,T 2 , dxi2 ,..., dxiT 1 ]
0
0
Wi D 0
Wi
L
0 Wi
The first-step estimator uses a covariance matrix taking this autocorrelation into account
enlarged for the level equations.
N
V W GW WiGTWi
i 1
174
Stellenbosch University http://scholar.sun.ac.za
where G I N G D, L
1
0
GL
0
0
2 1
1 2
and G D
1
1 2
0
0
1
0
0
0
1
W D 0
G D,L i
L
0 Wi
The two-step GMM estimator used the residuals of the first-step to estimate the covariance
matrix as suggested by White (1980):
N
Vˆ WiFT ˆiˆi FT Wi
i 1
Hence, finally, the resulting estimator is ˆ sys GMM ( XWVˆ 1W X )1 X WVˆ 1W R.
Source: Behr (2003: 13-14).
175
Stellenbosch University http://scholar.sun.ac.za
Table A4.2: Data Description, Measurement and Sources
Variable
Notation
Dependent Variables
Remittances
lnMRem
Remittances
as
percentage of GDP
REMGDP
Remittances
capita
per
REMPC
Workers‟
remittances
capita
per
WREMPC
Compensation of
employees
per
capita
COMPPC
Description, Measurement and Main Sources
The sum of workers‟ remittances and compensation of employees. Source:
Mainly WDI based on BoPS, Migration and Remittances Factbook 2011
(MRF-2011), and author‟s compilations from country desks under the African
Department of the IMF and WB.
Source: Author‟s computation as total migrant remittances received as a
percentage of nominal GDP of a recipient SSA country based on the
sources of migrant remittances cited above and nominal GDP data reported
mainly in WDI and WEO.
Source: Author‟s computation as total migrant remittances received as a
ratio of total population of a recipient SSA country based on the sources of
migrant remittances cited above and population data reported mainly in WDI
and WEO.
Workers‟ remittances are the remittances sent by migrants who spent 12
months or more in the host country. Source: Author‟s computation as gross
workers‟ remittances received as a ratio of total population of a recipient
SSA country based on the sources of migrant remittances cited above and
population data reported mainly in WDI.
Compensation of employees are the remittances received from citizens who
spent less than 12 months in the host country. Source: Author‟s
computation as total compensation of employees received as a ratio of total
population of a recipient SSA country based on the sources of migrant
remittances cited above. The population data was collated mainly from WDI.
Explanatory Variables
+
(•)_l
Lagged dependent
The immediate past values of the dependent variable. Source: Author‟s
computation.
+/Rate of growth in annual average consumer price index. Source: WDI, IFS
INF
Inflation rate
and WEO.
The annual average value of the national currency of a sampled SSA
country in real terms of the national currency of the migrant-host country.
+/Real
exchange lnRXR
Computed as a multiplication of nominal exchange rate by the ratio of hostcountry CPI to home-country CPI. Source: Author‟s computation based on
rate
WDI, IFS and WEO.
Real per capita GDP at purchasing power parity (PPP) in US dollars
Host-country
f
ln Y
(constant 2000 prices) of a typical non-SSA migrant host country. Source:
income
WDI.
Real per capita GDP PPP in US dollars (constant 2000 prices) of a typical
Home-country
h /
ln Y
SSA country. Source: WDI.
income
Real
Deposit
Average annual deposit rate of a typical SSA country less minus average
+/annualised CPI-based inflation rate. Source: Author based on WDI, IFS,
RIR
Interest Rate
WEO and Central Banks of selected countries.
Domestic credit to
Total domestic credit to the private sector by the home-country financial
+/system as a ratio of nominal GDP. Source: WDI, ADI and the Central Bank
lnPSC
private sector
website of selected sampled countries.
A polity2 index used to capture the qualities of democratic governance and
institutions in a typical home SSA country. It ranges between -10 for low
+/INS
Institutional quality
democratic governance (including dictatorship and autocratic regimes) and
weak institutions, and +10 for high democratic governance and strong
institutions. Source: Marshall and Jaggers (2011)
A dummy (=1 for post-2001 and 0 elsewhere) to capture post-September 11,
+
D9_11
Regulatory
2001, when the US and other migrant-host countries improved regulations
environment
on international money transfers, which has discouraged migrants from
using informal channels to remit. Source: Author‟s construction.
Source: Author. Note: The a priori sign is indicated by +/- in the notation column of each variable. WDI, ADI, WEO,
IFS and BoPS refer to April 2011 CD-ROM and e-database editions. The prefix notation „ln‟ denotes natural
logarithm.
176
Stellenbosch University http://scholar.sun.ac.za
Table A4.3: Host Countries of SSA Migrants
Migrant Home-Country
1
Top-5 SSA Migrant Host Countries in the World
2
3
4
TOG (12.45) CIV (10.60)
5
Top-3 Non-SSA Host Countries
1: Host
2
3
Benin (BEN)
NIG (27.71)
BFA (12.68)
Botswana (BSW)
RSA (18.01)
NAM (16.96) GBR (12.47) ZIM (11.64)
USA (9.58) GBR (12.47) USA (9.58)
Burkina Faso (BFA)
CIV (72.47)
GUI (11.03)
DRC (1.38)
PAK (1.06) PAK (1.06)
GER (1.03) FRA (0.37)
Cameroon (CAM)
FRA (22.62)
GAB (17.74) NIG (9.91)
USA (7.53)
GER (5.43) FRA (22.62)
USA (7.53)
Cape Verde (CPV)
POR (22.52)
USA (13.83) FRA (6.65)
GER (6.61)
MZQ (8.44) POR (22.52) USA (13.83) FRA (6.65)
Comoros (COM)
FRA (43.14)
UGA (22.47) GER (7.01)
TZA (4.68)
LBY (2.59) FRA (43.14)
GER (7.01) USA (0.50)
Congo Republic (CON) SUD (47.91)
TZA (16.03)
FRA (8.34)
BEL (2.68)
GER (2.29) FRA (8.34)
BEL (2.68)
Côte d'Ivoire (CIV)
FRA (27.38)
BFA (19.72)
GER (5.95)
BEN (6.56)
GUI (5.40)
FRA (27.38)
GER (5.95) USA (4.57)
Ethiopia (ETH)
USA (25.65)
ISR (20.67)
SAU (7.72)
CAN (5.08)
SWE (3.40) USA (25.65)
ISR (20.67) SAU (7.72)
Gabon (GAB)
SUD (47.30)
FRA (19.69)
SLE (4.39)
GER (3.44)
KEN (3.66) FRA (19.69)
GER (3.44) USA (0.92)
Gambia (GAM)
ESP (12.18)
USA (11.95) NIG (8.45)
GBR (7.73)
SEN (3.67) ESP (12.18)
USA (11.95) GBR (7.73)
Ghana (GHA)
CIV (31.91)
NIG (13.06)
BFA (9.74)
GUI (8.69)
USA (7.31) USA (7.31)
GBR (5.97) GER (2.27)
Guinea (GUI)
CIV (23.74)
LIB (12.11)
SEN (18.69) BFA (11.17) GAM (6.57) GBR (1.45)
GER (1.34) USA (1.15)
Guinea-Bissau (GBS)
SEN (25.45)
GAM (13.36) POR (16.72) FRA (6.34)
BFA (5.81) POR (16.72) FRA (6.34)
GER (4.45)
Kenya (KEN)
GBR (28.54)
TZA (27.04)
UGA (7.35)
GER (5.38) GBR (28.54) USA (9.85)
GER (5.38)
Lesotho (LSO)
MZQ (46.02)
ZIM (19.45)
RSA (16.18) MWI (2.38)
TZA (1.72)
PAK (0.71)
Madagascar (MAD)
FRA (51.55)
REU (12.88) GER (8.64)
Malawi (MWI)
ZAM (19.03)
TZA (18.93)
GHA (5.13)
USA (9.85)
ZIM (4.22)
RSA (17.65) ZIM (12.91)
GAB (5.59) FRA (2.22)
GER (0.82)
GER (1.30) PAK (1.07)
USA (0.81)
AUS (4.27)
GER (5.43)
GER (2.29)
COM (3.82) FRA (51.55)
REU (12.88) GER (8.64)
GBR (8.30) GBR (8.30)
GER (1.93) USA (1.38)
Mali (MLI)
CIV (30.82)
BFA (27.74)
GUI (10.32)
NIG (5.77)
GHA (5.22) FRA (2.73)
GER (1.37) USA (0.45)
Mauritania (MRT)
SEN (34.21)
NIG (10.32)
FRA (8.78)
BFA (7.68)
GUI (7.56)
ESP (3.21)
Mauritius (MRS)
FRA (18.41)
RSA (18.39) GBR (15.63) AUS (9.70)
Mozambique (MZQ)
TZA (22.78)
ZIM (19.98)
Namibia (NAM)
MZQ (23.07)
Niger (NGR)
BFA (27.84)
Nigeria (NIG)
FRA (8.78)
GER (2.72)
GER (6.53) FRA (18.41)
GBR (15.63) AUS (9.70)
RSA (17.74) POR (8.97) POR (8.97)
GER (2.07) GBR (0.50)
RSA (17.28) ZIM (13.13)
TZA (1.50)
GBR (5.14) GBR (5.14)
USA (4.19)
GER (1.88)
CIV (26.25)
NIG (11.89)
GUI (10.79)
GHA (5.16) GER (1.10)
PAK (1.06)
FRA (0.73)
SUD (23.76)
USA (13.74) GBR (8.60)
CAM (8.39)
GHA (5.14) USA (13.74)
GBR (8.60) GER (2.91)
Rwanda (RWA)
UGA (42.17)
TZA (27.94)
KEN (4.89)
BEL (2.83)
GER (1.87) BEL (2.83)
GER (1.87) USA (1.16)
São Tomé & Príncipe
POR (54.97)
CPV (15.94) GER (9.30)
BFA (3.00)
GUI (2.93)
Senegal (SEN)
GAM (20.56)
FRA (18.32)
MRT (8.48)
Seychelles (SEY)
GBR (17.40)
RSA (18.69) AUS (14.55) ZIM (6.24)
TZA (6.18)
Sierra Leone (SLE)
USA (22.87)
LIB (18.31)
GER (4.50) USA (22.87)
South Africa (RSA)
GBR (18.15)
MZQ (16.04) AUS (10.12) USA (8.99)
ZIM (7.37)
Sudan (SUD)
SAU (32.05)
UGA (24.31) JOR (3.78)
USA (3.43)
EGY (2.64) SAU (32.05)
JOR (3.78)
USA (3.43)
Swaziland (SWZ)
MZQ (28.48)
RSA (17.05) ZIM (14.53)
GBR (7.43)
USA (5.46) GBR (7.43)
USA (5.46)
GER (2.30)
Tanzania (TNZ)
UGA (20.46)
RSA (18.31) GBR (11.48) ZIM (8.81)
CAN (6.95) GBR (11.48) CAN (6.95)
USA (4.47)
MWI (19.16)
ITA (9.58)
GBR (18.18) GHA (5.00)
POR (54.97) GER (9.30) FRA (1.02)
GER (5.30) FRA (18.32)
ITA (9.58)
GBR (18.18) GER (4.50)
GBR (18.15) AUS (10.12) USA (8.99)
Togo (TOG)
NIG (36.10)
BEN (12.05) BFA (8.75)
GUI (8.84)
GAM (6.61) FRA (6.38)
Uganda (UGA)
GBR (32.41)
TZA (23.82)
CAN (6.48)
GER (6.06) GBR (32.41) USA (7.38)
USA (7.38)
GER (5.30)
GBR (17.40) AUS (14.55) CAN (6.15)
GER (2.02) USA (1.63)
CAN (6.48)
Source: Author‟s compilation from Parson et al. (2007). Note: AUS, CAN, ISR, ITA, LIB, DRC, EGY, REU, SWE,
ZAM, ZIM, JOR and LBY stand for Australia, Canada, Israel, Italy, Liberia, Democratic Republic of Congo, Egypt,
Reunion, Sweden, Zambia, Zimbabwe, Jordan and Libya respectively.
177
Stellenbosch University http://scholar.sun.ac.za
Table A4.4:
Robustness Test Results of International Migrant Remittance Flows to SSA, 1980-2009
Group variable: Country Code
Number of obs = 1006
Time variable: Year
Number of groups (N)= 36
Number of instruments = 442
Two-Step Estimation by Blundell-Bond System Dynamic Panel Data Procedure
Migrant remittances lag 1 (•_1)
Migrant remittances lag 2 (•_2)
Institutional quality (INS)
h
Home-country income (lnY )
f
Host-country income (lnY )
Rate of inflation (INF)
Bank credit to private sector (PSC)
Real exchange rate (RXR)
Real deposit interest rate (RIR)
Regulatory environment (D9_11)
Constant term (constant)
2
Wald [10],
REMPC
lnMRem
REMGDP
REMPC_USA
0.9477
0.6438
1.0306
0.9359
(359.71)***
(13.19)***
(203.61)***
(331.25)***
-0.1267
0.0851
-0.1009
-0.1248
(-107.93)***
(2.98)***
(-18.52)***
(-46.02)***
-0.3107
-0.0054
-0.0469
-0.1087
(-8.17)***
(-1.33)
(-14.64)***
(-1.96)**
-1.0084
-0.0887
-1.4763
-4.2271
(-2.82)***
(-0.17)
(-5.58)***
(-8.68)***
18.3925
1.2693
-0.9862
15.2761
(13.93)***
(2.63)***
(-2.13)**
(7.03)***
0.0449
-0.0008
0.0088
0.0781
(4.75)***
(-0.25)
(2.41)**
(3.80)***
2.3553
-0.0848
0.1612
2.7793
(7.12)***
(-1.58)
(4.02)***
(5.29)***
-4.9314
-0.0131
-0.2125
-6.8819
(-12.98)***
(-0.08)
(-5.84)***
(-26.01)***
0.0347
-0.0024
0.0101
-0.0615
(4.46)***
(-0.81)
(2.71)***
(-2.75)***
4.6614
0.1905
0.6862
4.8291
(71.91)***
(8.59)***
(72.01)***
(37.86)***
-157.9818
-7.6688
21.3176
-100.3526
(-12.40)***
(-2.71)***
1.40e+06
***
18200.39
(5.11)***
***
4.29e+06
(-4.80)***
***
454424.61
***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-0.1065{0.915}
-1.0463{0.295}
-1.0233{0.306}
-0.1202{0.904}
Sargan test of over-identifying restrictions:
2
[431],
Source: Author‟s estimation
27.1849
31.4014
32.2858
21.5861
denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
REMPC_USA reports estimates where the USA is assumed as the host country
for SSA migrants
*/**/***
178
Stellenbosch University http://scholar.sun.ac.za
Table A4.5: Results of Panel Unit Root Tests
VARIABLES
PANEL UNIT ROOT TEST STATISTICS
Fisher P-P chi-square
Hadri HC z-stat
LLC Adjusted t-stat
Conclusion
At Level
At Level
At Level
REMPC
2.7415***
37.9585***
I(0)
{0.0031}
[0.0000]
WREMPC
2.5115***
38.1682***
I(0)
{0.0060}
[0.0000]
COMPPC
n/a
9.0253***
I(0)
n/a
{0.0000}
INS
4.0808***
38.2215***
I(0)
{0.0000}
[0.0000]
h
lnY
2.4244***
49.0222***
I(0)
{0.0000}
[0.0000]
f
lnY
-4.8303
23.2519***
-4.2924***
I(0)
{1.0000}
[0.0000]
(0.0000)
INF
n/a
n/a
25.8682***
I(0)
n/a
n/a
{0.0000}
lnPSC
0.6143
38.0580***
-1.5210*
I(0)
{0.2695}
[0.0000]
(0.0641)
lnRXR
-0.2351
30.8740***
-2.0439**
I(0)
{0.5929}
[0.0000]
(0.0205)
RIR
25.3546***
11.6515***
I(0)
{0.0000}
[0.0000]
Source: Author‟s computations
Note: Figures in brackets are respective probability values. ***/**/* significant
at 1/5/10 level statistical level respectively. Constant and trend included. n/a
means not applicable due to omitted data.
179
Stellenbosch University http://scholar.sun.ac.za
Table A4.6: Static Panel-Data Modelling of Migrant Remittance Inflows in SSA, 1980-2009
Fixed
Effects (FE)
Random GLS
Effects (RE)
Robust FE
Robust Random
++
GLS (RE)
-0.6913
(-3.57)***
-0.5852
(-3.08)***
-0.6913
(-1.68)*
-0.5852
(-1.58)
17.6715
(4.45)***
14.3954
(4.16)***
17.6715
(1.04)
14.3954
(1.08)
Host-country income (lnY )
36.9054
(5.32)***
28.7328
(4.79)***
36.9054
(1.76)*
28.7328
(1.70)*
Rate of inflation (INF)
0.1267
(1.12)
0.1162
(1.02)
0.1267
(1.08)
0.1162
(1.07)
Bank credit to private sector (lnPSC)
5.6051
(3.69)***
6.1008
(4.02)***
5.6051
(1.75)*
6.1008
(1.73)*
Real exchange rate (lnRXR)
-9.6136
(-5.18)***
-7.1883
(-4.87)***
-9.6136
(-1.69)*
-7.1883
(-1.88)*
Real deposit interest rate (RIR)
0.1577
(1.34)
0.1480
(1.24)
0.1577
(1.34)
0.1480
(1.38)
Regulatory environment (D9_11)
6.5054
(3.14) ***
7.9939
(3.98) ***
6.5054
(1.65)*
7.9939
(2.40)**
Constant term (constant)
-447.9099
(-6.81)***
-354.5817
(-6.09)***
-447.9099
(-1.59)
-354.5817
(-1.63)
Number of observations
1078
1078
1078
1078
Number of groups (N)
36
36
36
36
Institutional quality (INS)
h
Home-country income (lnY )
f
Overall R
2
F-statistics
Hausman_FE
Breusch-Pagan (B-P) statistics
Source: Author‟s estimation
0.0962
27.32{0.000}
-10.19
0.0998
***
ᴥ
208.93{0.000}
0.0962
***
n/a
2.74{0.019}
n/a
***
0.0998
**
20.76{0.008}
n/a
n/a
n/a
n/a
8618.63{0.000}
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
ᴥ
robust z-statistics in ( ), probabilities in { }, probabilities not available
++
n/a denotes not available or required, most efficient and reliable results
180
***
Stellenbosch University http://scholar.sun.ac.za
Table A4.7: Static Panel-Data Modelling of Workers‟ Remittance Inflows in SSA, 1980-2009
Fixed
Effects (FE)
Random GLS
Effects (RE)
Robust FE
-1.2893
(-5.50)***
-1.0818
(-4.84)***
-1.2893
(-1.39)
-1.0818
(-1.32)
16.0950
(3.35)***
12.4477
(3.24)***
16.0950
(0.86)
12.4477
(0.99)
Host-country income (lnY )
43.8241
(5.22)***
28.6914
(4.31)***
43.8241
(1.90)*
28.6914
(1.68)*
Rate of inflation (INF)
0.0691
(0.51)
0.0558
(0.41)
0.0691
(0.56)
0.0558
(0.49)
Bank credit to private sector (lnPSC)
5.3290
(2.90)***
6.0171
(3.34)***
5.3290
(1.49)
6.0171
(1.52)
Real exchange rate (lnRXR)
-9.3396
(-4.17)***
-5.2262
(-3.30)***
-9.3396
(-1.65)*
-5.2262
(-1.69)*
Real deposit interest rate (RIR)
0.0750
(0.53)
0.0626
(0.44)
0.0750
(0.58)
0.0626
(0.54)
Regulatory environment (D9_11)
4.5760
(1.83)*
7.0209
(2.98)***
4.5760
(0.96)
7.0209
(1.63)
Constant term (constant)
-512.0303
(-6.44)***
-354.8648
(-5.46)***
-512.0303
(-1.95)*
-354.8648
(-1.87)*
Number of observations
1078
1078
1078
1078
Number of groups (N)
36
36
36
36
0.0583
0.0631
0.0583
0.0631
1.75{0.122}
11.61{0.170}
Institutional quality (INS)
h
Home-country income (lnY )
f
Overall R
2
F-statistics
Hausman_FE
Breusch-Pagan (B-P) statistics
Source: Author‟s estimation
18.14{0.000}
***
29.49{0.000}
***
132.10{0.000}
***
n/a
n/a
***
++
Robust Random
GLS (RE)
n/a
n/a
n/a
n/a
4910.85{0.000}
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
robust z-statistics in ( ), probabilities in { }, n/a denotes not available or required
++
most efficient and reliable results based on Hausman test and B-P statistics
181
Stellenbosch University http://scholar.sun.ac.za
Table A4.8: Static Panel-Data Modelling of Compensation of Employees to SSA, 1980-2009
Fixed
Effects (FE)
Random GLS
Effects (RE)
Robust FE
Robust Random
++
GLS (RE)
0.7327
(5.59)***
0.6904
(5.50)***
0.7327
(1.15)
0.6904
(1.17)
1.6016
(0.61)
0.8084
(0.37)
1.6016
(0.18)
0.8084
(0.15)
Host-country income (lnY )
-8.5768
(-1.82)*
-4.6367
(-1.19)
-8.5768
(-1.03)
-4.6367
(-0.94)
Rate of inflation (INF)
0.0910
(1.19)
0.0893
(1.18)
0.0910
(1.01)
0.0893
(1.00)
Bank credit to private sector (lnPSC)
0.8000
(0.79)
0.8425
(0.85)
0.8000
(0.46)
0.8425
(0.57)
Real exchange rate (lnRXR)
-0.3033
(-0.23)
-1.2132
(-1.27)
-0.3033
(-0.27)
-1.2132
(-1.49)
Real deposit interest rate (RIR)
0.1266
(1.60)
0.1251
(1.59)
0.1266
(1.01)
0.1251
(1.01)
Regulatory environment (D9_11)
3.2265
(2.31)**
2.7448
(2.09)**
3.2265
(1.19)
2.7448
(1.18)
Constant term (constant)
79.8442
(1.79)*
50.1661
(1.33)
79.8442
(1.00)
50.1661
(0.87)
Number of observations
970
970
970
970
Number of groups (N)
36
36
36
36
0.0202
0.0537
0.0202
0.0537
0.74{0.656}
4.77{0.782}
Institutional quality (INS)
h
Home-country income (lnY )
f
Overall R
2
***
F-statistics
6.66{0.000}
Hausman_FE
2.72{0.951}
Breusch-Pagan (B-P) statistics
Source: Author‟s estimation
54.35{0.000}
***
n/a
n/a
***
n/a
n/a
n/a
n/a
5724.30{0.000}
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
robust z-statistics in ( ), probabilities in { }, n/a denotes not available or required
++
most efficient and reliable results based on Hausman test and B-P statistics
182
Stellenbosch University http://scholar.sun.ac.za
CHAPTER FIVE
THE IMPACT AND CAUSAL EFFECTS OF FINANCIAL LIBERALISATION ON
INTERNATIONAL REMITTANCE INFLOWS IN SUB-SAHARAN AFRICA
5.0 INTRODUCTION
This chapter is devoted to exploring the causal effects and impact of financial liberalisation on
official international remittance inflows in sub-Saharan Africa (SSA). This is motivated by the
proposition that a less restricted and a more inclusive financial system has salutary
consequences for remittance mobilisation through the formal financial system as banks and
allied financial institutions enter into the international remittance market to make it more
competitive, leading to high efficiency and a reduced cost on fund transfers. A sample of 13
SSA countries for which relevant data are available was used. The study begins with
background information in Section 5.1. Section 5.2 presents some selected stylised facts on
international remittance inflows and financial liberalisation in SSA whilst Section 5.3 reviews the
literature on financial liberalisation and financial resource mobilisation. In Section 5.4, the
empirical models, methodology and issues related to data are discussed. The empirical results
are presented and discussed in Section 5.5. Section 5.6, the last, discusses applicable policy
guidelines.
5.1 BACKGROUND
While official remittances to sub-Saharan Africa (SSA) have been increasing steadily over the
past three decades coinciding with the implementation of financial liberalisation programmes,
the sub-region has still remained the region that receives the least migrant remittances by any
good measure (see Figure A3.2 in Chapter Three). But while SSA receives the least
remittances through official channels, Freund and Spatafora (2005) estimate that, in relative
terms, SSA is the highest recipient of informal remittances because of the underdevelopment of
the domestic financial sector. Earlier, Kapur (2004) reported that banks in industrialised
economies79 facilitate the flow of remittances through official transfer channels by competing
with the traditional non-bank remittance service providers. This suggests that a liberalised
financial system which can motivate banks and other formal financial institutions to enter the
global remittance market and compete with other formal remittance service providers such as
79
These are also economies where financial markets are relatively developed with less financial repressive policies.
183
Stellenbosch University http://scholar.sun.ac.za
Money Transfer Operators (MTOs) might be a necessary condition if a country is to receive
higher remittances through official channels. With low participation of banks in the international
remittance market, the existing competition between MTOs and unregulated informal
remittance-transfer service providers hardly enables SSA countries to mobilise optimal formal
remittances.
Although the financial system alone does not represent the entire official remittance channel as
post offices80, mobile telecommunication service providers and international MTOs such as
Western Union, MoneyGram, Vigo, and DolEx are equally recognised as official channels
globally, low participation of domestic banks in the remittance market must be discouraged for
a number of reasons. Firstly, increased competition in the remittance market as a result of
higher bank participation is necessary in mobilising more remittances and in reducing the high
commissions charged by the few existing official transfer agents. The few dominant MTOs that
have market power in the absence of meaningful competition from banks tend to charge above
what would have been the competitive market price on non-competitive remittance-corridors.
Indeed, available data on charges on money transfers by the two leading MTOs in SSA
(Western Union and MoneyGram) show that although there is a high degree of regional
homogeneity in the cost structures, a considerable price variation exists among SSA countries,
depending mainly upon market size and competition on the remittance corridor81,82.
Secondly, because MTOs and other non-bank remittance-transfer institutions do not
intermediate83 in the financial system, the participation of banks and stock exchanges in the
remittance market is vital to ensuring that remittances received are put into optimal use.
80
In most SSA countries, post offices are only directly important in the local remittance market when rural-urban
migrants purchase postal orders. And because typical post offices do not have branches outside their national
borders, they act mainly as agents for MTOs in the international remittance market. And even though some
households may receive remittances in the form of bankers‟ drafts through post offices, in this particular case, post
offices can only be seen as performing their traditional function rather than as direct participants in the international
remittance market.
81
This is based on accessible information on the websites of Western Union and MoneyGram and author‟s personal
inquiries with these MTOs during the period, February 6-22, 2011, on remitting US$500 to selected SSA countries
from France, Spain, Germany, Portugal, UK and the USA using Western Union and MoneyGram. From France, the
least priced countries are Francophone West Africa, especially Mali, Senegal and Togo. From the UK and USA, the
remittance corridors to Botswana, Cape Verde, Ghana, Kenya, Nigeria, Seychelles, South Africa, and Uganda, are
among the least priced. Major price differences among recipient SSA countries are related to the primary minimum
fee rather than the secondary percentage fees charged by the MTOs.
82
A similar observation was made by Suki (2007) on the US-LAC remittance corridor with prices relatively lower on
the US-Mexico corridor largely due to differences in intensity of competition following increased participation of
banks.
83
It must be said that in some SSA countries, for instance, SADC countries, post offices offer post-bank services
where customers can deposit money, but these post offices are not licensed to give loans to customers.
184
Stellenbosch University http://scholar.sun.ac.za
Thirdly, increased bank participation can directly contribute to the elimination of dual exchange
rate regimes as, unlike other agents, banks generally have foreign exchange departments and,
hence, can convert remittances into local currencies using officially approved rates. Fourthly,
increased bank participation can help solve the low degree of „bancarisation‟ and high financial
exclusion syndromes in SSA. Fifthly, given its unique resilient and countercyclical character,
remittances received through banks can be important to the „labour-exporting‟ country through
securitisation of expected remittances as collateral. Ketkar and Ratha (2001; 2004) report that
in the year 2001, the Central Bank of Brazil issued a 5-year bond valued at US$300 million and
secured future remittances from Brazilian migrants in Japan. And, finally, because the 5-year
bond issued by Banco do Brasil in the year 2001 earned an enviable BBB+ rating which is by
far higher than the existing rating of Brazil‟s sovereign bond (Ketkar and Ratha, 2001), it can be
said that receiving higher remittances through banks can improve the international credit rating
of recipient countries.
Thus, there is certainly no doubt that the surest way to attract higher remittances to SSA
through the formal channels and to make the most out of remittance inflows is to liberalise the
domestic financial sector for higher competition so that formal financial institutions in SSA can
be attracted to enter the remittance market and compete with MTOs. The entry of banks and
other formal financial institutions into the remittance market can boost openness, reduce
remittance transmission costs, improve services, including quality standards, increase
efficiency, and expand service options to both senders and receivers of remittances. However,
unless the appropriate sequencing and pace are followed, financial liberalisation can be
associated with a higher risk of financial fragility due to higher speculative behaviour of
economic agents in response to changing trends in capital flows and fluctuations in the rates of
interest and exchange rates (Stiglitz, 1994; Mathieu, 1998). Nevertheless, unlike other forms of
private external capital, international remittances are well-known not only to be risk-free, but
also non-debt creating, non-volatile and possessing the finest shock-mitigating effects
(Bugamelli and Paternò, 2008; Adenutsi, 2011). Notwithstanding the above, it would be
erroneous to think that the mere pursuit of financial liberalisation programmes will cause an
automatic inflow of international remittances. This is because, as noted in Chapter Two, the
concept of financial liberalisation is multidimensional, with Abiad et al. (2010), for example,
185
Stellenbosch University http://scholar.sun.ac.za
identifying nine components84.
Therefore, it can be argued that the implications of financial liberalisation for international
remittance inflows through the formal financial system is contingent upon whether or not the
deregulation of restrictions on capital inflows is accompanied by liberalisation of restrictions on
capital outflows. If there are fewer restrictions on external capital inflows and outflows in a
typical migrant-home country, there is a higher likelihood that international migrants will remit
home using the formal financial system. One reason is that when a migrant finds it convenient
to freely reallocate his/her portfolio investments from his/her home country to other parts of the
world, a typical migrant-home country can receive higher remittances through the formal
financial system as remittances driven by speculative motives increase. Another important
aspect is whether or not, following financial liberalisation, real interest rates and interest rate
spreads in a typical migrant-home developing country are above or below the world market
rates. This is consistent with the self-interest investment theory underlying the flow of
international remittances. According to this theory, when investment in migrant-home countries
is relatively less risky and more rewarding, more non-altruistic remittances are received in the
home country of migrants.
Also, depending upon the extent to which financial liberalisation has led to increased financial
efficiency as measured by reduced cost of international money transfers, increased access of
residents to quality, innovative and reliable financial services and products with wider options
for portfolio diversification, a country could receive more international remittances when
migrants switch from informal remittance service providers to patronise the services of the
formal financial institutions participating in the remittance market. This is also a natural
possibility as the lower transaction cost of remitting alleviates the extra burden on remitters,
especially migrants on low incomes and/or those who remit regularly. And, whether or not, the
implementation of financial liberalisation policies in SSA will cause higher inflows of
international remittances through the formal financial system, is dependent upon the trade-off
between the risk exposure associated with increased financial openness and a host of many
factors including the size of the unofficial remittances received; stock and legal status of
84
These are specific reform policies on directed credit, reserve requirements, interest rate controls, entry barriers,
banking supervision, privatisation, international capital flows, aggregate credit ceilings, and security market
development. Abiad et al. (2010), however, consider directed credit, reserve requirements and aggregate credit
ceilings as complementary policies aimed at achieving the same specific purpose.
186
Stellenbosch University http://scholar.sun.ac.za
migrants; target recipients‟ access to convenient financial services; and the magnitude as well
as the regularity of remittances. This is because even if total remittances (i.e. the sum of official
and unofficial remittances) remain unchanged following the merits of financial liberalisation,
holding all other factors constant, countries which hitherto received higher unofficial
remittances are more likely to receive higher remittances through the formal financial system
than those that traditionally received fewer unofficial remittances. In much the same manner,
countries that traditionally receive more remittances by virtue of having a large stock of legal
and working migrants in the Diaspora are more likely to receive even more remittances through
the formal financial system if these countries adopt financial liberalisation programmes.
As a result, though, in principle, the causal relationship between international remittances and
financial liberalisation may be expected to be uni-directional (running from financial
liberalisation to remittance inflows), because the policies implemented under financial
liberalisation programmes are assumed to be deliberate and independent actions of the
monetary authority of a country; a bi-directional causality cannot be ruled out entirely. Indeed,
the impact of financial liberalisation on international remittance inflows can be positive, zero, or
negative, depending upon the specific aspects of financial liberalisation which received the
most (or domineering) policy reform attention. For instance, one does not expect a strong
causal relationship between remittances and financial liberalisation if the financial reforms
policy implemented is concentrated on reduction of direct reserve requirements and direct
credit control rather than on promoting financial efficiency and competition through privatisation
and relaxation of entry requirements. Although remittances may respond to these financial
policy reforms in the long run, these policies can hardly have a direct short-run impact or causal
effect on attracting remittances through the banking system. Therefore, since the impact and
causal effects of financial liberalisation and international remittances remain theoretically
unresolved, they should be seen as issues of empirical concern.
To fill this research gap, six key research questions are imperative, holding all other factors
constant, and with reference to SSA.
i.
What are the stylised facts about financial liberalisation and international remittance
inflows?
ii. Is there any causal relationship between financial liberalisation and international
remittance inflows?
187
Stellenbosch University http://scholar.sun.ac.za
iii. If yes to (ii), then which specific policy component(s) under the financial liberalisation
programme cause(s) international remittance inflows?
iv. What is the impact of financial liberalisation on international remittance inflows?
v. How does each specific policy component of financial liberalisation impact on
international remittance inflows?
vi. Does the impact of financial liberalisation on international remittances (if any), vary over
time in response to the increasing implementation of the liberalisation policies?
Consequently, the specific objectives of this chapter are six-fold. The first is to provide some
stylised facts on the relationship between international remittances received and financial
liberalisation in SSA. The second is to verify if there is any causal relationship between
international remittance inflows and the broad index of financial liberalisation in SSA. The third
is to trace the direction of any existing line of causality between each specific policy component
of financial liberalisation programmes and international remittance inflows in SSA. Fourthly, it is
to examine the overall impact of financial liberalisation on international remittance inflows in
SSA. The fifth is to determine the policy-specific impact of financial liberalisation on
international remittance inflows. The sixth is to verify if, with the passage of time and as the
financial market became more and more liberalised, the impact of financial liberalisation on
remittances increased.
To achieve these objectives, 13 SSA countries for which relevant data are available from 1980
to 2009 are analysed. This study is essential because, according to the McKinnon-Shaw
hypothesis, financial liberalisation is vital to the mobilisation of domestic resources and various
forms of private external capital and, for that matter, international remittances. It is also
imperative for policy making in SSA being the sub-region that receives the least remittances
through official channels. This is because whilst Orozco (2004) reports that banks do not play
any significant role in the remittance market, (specifically, not more than three per cent on the
US-Mexico corridor85), Alberola and Salvado (2006) develop a notional model to prove that the
presence of banks in the remittance market is crucial to receiving higher official remittances.
85
This is one of the largest remittance corridors in the world. In fact, Mexico, with most of its migrants in the US, is
the largest recipient of remittances in LAC, and the third in the world as at 2009 (see WDI, April 2011).
188
Stellenbosch University http://scholar.sun.ac.za
The motivation behind this study lies in the fact that it addresses some crucial aspects of the
remittance literature, the causal linkages and direct effects of financial liberalisation on
international remittances received through official channels, which have so far been ignored in
empirical studies. In fact, apart from some attempts86 at exploring the impact of remittances on
financial development indicators which are generally considered as measures of financial
reforms outcome, the researcher is not aware of any previous work devoted to exploring the
relationship between financial liberalisation and international remittance inflows.
5.2 SELECTED STYLISED FACTS
Table 5.1 shows the year in which SSA countries embarked upon financial liberalisation
programmes. 19 out of the 38 countries listed in Table 5.1 fully embarked upon financial reform
programmes in the 1980s.
Table 5.1: Implementation of Financial Liberalisation in SSA
Country
Year of FLB
Country
Year of FLB
Benin
1989
Malawi
1987
Botswana⁺
Burkina Faso
1989
1989
Mali
1989
Mauritania
1990
Burundi
Cameroon
1986
1990
Mauritius⁺
Mozambique⁺
1981
1992
Congo, DR
2001
Namibia⁺
1991
Congo Republic
Cote d'Ivoire
1990
1989
Niger
Nigeria⁺
1989
1987
Central Africa Republic
1990
Chad
Ethiopia
1990
1994
Rwanda
Senegal
1994
1989
Seychelles⁺
1984
Gabon
1990
Gambia
Ghana⁺
1985
1987
Sierra Leone
South Africa⁺⁺
1991
1980
Guinea
1996
Sudan
Tanzania⁺
1997
1991
Guinea-Bissau
Kenya⁺
1989
1991
Togo
Uganda⁺
1989
1988
Lesotho
1996
Zambia⁺
1992
Madagascar
1986
Zimbabwe
1993
+ ++
Source: Author, based mainly on various Central Bank reports and website information. ( ) denotes that country is
classified as having frontier (emerging) financial market respectively. Bold means the country is sampled for the
empirical analysis reported in this chapter.
86
See, for example, Aggarwal et al. (2006), Toxopeus and Lensink (2007), Gupta, Pattillo and Wagh (2009), and
Gheeraert et al. (2010).
189
Stellenbosch University http://scholar.sun.ac.za
Republic of South Africa is the first country within the sub-region to liberalise its financial
market, and this was in 1980, whilst Congo DR was the latest to adopt the programme just
about a decade ago. Despite the gradual abandonment of financial repressive policies in favour
of liberalisation policies (see Table A5.7), most SSA still have underdeveloped financial
markets today. According to IMF (2008a), as at 2007, the only emerging financial market in
SSA was that of the Republic of South Africa, with only 12 countries within the sub-region
(those marked + in Table 5.1) having frontier financial markets.
Figure 5.1 reveals that even though countries with emerging and frontier markets have
relatively more liberalised financial markets, they receive far less international remittances (in
per capita or per migrant terms) compared with other countries within the sub-region that have
underdeveloped financial markets (see also Table A5.6 for evidence).
Figure 5.1: Remittances Received and Financial Liberalisation in SSA, 1980-2009
0.800
0.668
1980-1989
0.600
1.400
0.183
0.169
0.000
F&E markets
Other markets
-0.200
Full Sampled
(combined)
Averages
Averages
0.195
0.200
1.200
0.800
0.478
0.442
0.462
0.000
F&E markets
-0.651
2000-2009
1.000
1.400
1.200
1.000
0.752
0.616
0.600
Full Sampled
(combined)
1.065
1.697
1.544
Other markets
1980-2009
1.200
0.689
Averages
1.828
1.600
Averages
0.671
0.600
0.200
-0.600
0.800
1.036
1.000
0.400
-0.400
1.800
1.645
1.600
0.571
0.400
-0.800
2.000
1990-1999
1.800
0.823
0.800
0.600
0.616
0.475
0.407
0.400
0.400
0.443
0.200
0.200
0.000
F&E markets
Other markets
lnREMPC
0.000
Full Sampled
(combined)
F&E markets
FLBI
Other markets
lnREMPC
FLBI
Full Sampled
(combined)
Source: Author‟s estimation. Note: lnREMPC and FLBI represent natural logarithm of remittances per capita (in
US$) and financial liberalisation index respectively. F&E markets stands for frontier and emerging markets.
190
Stellenbosch University http://scholar.sun.ac.za
There can be three possible explanations for this revelation. First, whereas the analysis
stretches over a 30-year period (i.e. 1980-2009), the IMF classification of financial markets was
only done in 2007. This could imply that most of the countries classified as having frontier and
emerging financial markets might have attained this status not too long ago. Second, only six
countries with frontier financial markets (Ghana, Kenya, Mozambique, Nigeria, Tanzania and
Uganda) were included in this analysis because of lack of data on financial liberalisation index
and/or international remittances on the other six frontier countries (Botswana, Cape Verde,
Mauritius, Namibia, Seychelles, and Zambia). Incidentally, whilst Botswana, Cape Verde,
Mauritius and Seychelles are among the traditional leading recipients of remittances per capita,
Ghana, Mozambique and Tanzania are among the very least recipients within the sub-region
(see Figure 3.5 in Chapter Three). Third, it is possible that most of the international remittances
received through official channels in countries with underdeveloped financial markets are not
received through the financial system, which includes banks and the stock exchange, whilst
FLBI is based on policy reforms in the banking sector and the stock market.
One important observation that can be made from Figure 5.1 is that, although SSA countries
with underdeveloped financial markets attracted more officially reported remittances per capita
than their counterparts with frontier and emerging financial markets, over time, it is the latter
group that is increasingly receiving more officially reported remittances per capita in relative
terms. For example, from an average low of natural logarithmic value of -0.651 in the 1980s,
SSA countries with frontier and emerging markets recorded a more than 200 percentage rise
by the end of the 1990s; and a further rise in excess of 270 per cent in the 2000s over the
reported figure in the 1990s. In similar natural logarithmic values, SSA countries with
underdeveloped financial markets recorded a mere 50 per cent rise in remittances per capita in
each successive decade.
The relationship between international remittances received per capita and financial
liberalisation (FLB) including its specific policy components are shown in Figure 5.2. Overall,
international remittances are positively correlated with the degree of financial liberalisation. In
countries with frontier and emerging financial markets, the positive correlation between
international remittances and the various dimensions of FLB excluding policy on directed credit,
reserve requirement and aggregate credit ceilings (DCRR), are stronger than in countries with
underdeveloped financial markets (see also Table A5.5).
191
Stellenbosch University http://scholar.sun.ac.za
Figure 5.2:
Correlation between Remittances Received and Financial Liberalisation in SSA, 1980-2009
0.60000
Correlation coefficient
0.50000
0.40000
0.30000
0.20000
0.10000
0.00000
DCRR
IRC
EBC
Frontier & Emerging markets
Source: Author‟s estimation.
BKS
PVZ
Others (underdeveloped)
ICF
SMK
FLBI
Full Sample (combined)
Refer to Table 5.2 for meaning of IRC, EBC, BKS, PVZ, ICF and SMK.
Correlation, however, only provides evidence of a possible contemporaneous relationship
between a pair of variables, which does not necessarily suggest a direct relationship or
causality between the pair of variables under consideration. For empirical evidence on the
impact or the causal relationship between FLB or the specific FLB policies and international
remittance inflows, a more rigorous statistical analysis, which is the object of this chapter, is
required.
5.3 LITERATURE REVIEW
5.3.1 Theoretical Literature
A survey of the literature suggests that there are three main theories that are central to our
understanding of the role a financial system can play in financial resource mobilisation. These
are the Repressionist theory, the McKinnon-Shaw liberalist theory, and the Neostructuralist
theory.
5.3.1.1 The Financial Repression Theory
The Financial Repression model dominated the financial and monetary policy formulation in
both industrialised and developing economies prior to the mid-1970s (Fry, 1995). Advocates of
financial repression, particularly, Hilferding (1910), Gesell (1911), Keynes (1936), Nicholas
(1974), and Tobin (1965) express misgivings about the efficient role of the financial system in a
capitalist economy and caution that without careful management of money supply, an economy
is bound to suffer severe negative ramifications. This is because capital-intensive investment
which is considered a desideratum in capital-constrained developing countries can only be
192
Stellenbosch University http://scholar.sun.ac.za
promoted at lower cost of capital, hence the need to allow the state to take better control over
money supply. This proposition is based on the assumption that it is the government-regulated
rather than the market-directed financial system that can best determine the type and volume
of savings and investment that are desirable to promote social welfare, especially in developing
countries where institutions are weak (Gerschenkron, 1962). Another justification given for
financial repression is that the government is responsible for protecting borrowers against
usury practices by moderating the free market determination of interest rates. Tobin (1965),
and Giovannini and de Melo (1993) explain that financial repression ensures that interest rates
are maintained below market rates, reduces the cost of servicing debts and enables
governments to finance social development projects.
Repressionists argue that in developing countries, in particular, where institutional constraints
undermine government efforts to mobilise sufficient tax revenue to offset government
expenditure, financial repression is the second best strategy (Fry, 1995). In his liquidity
preference theory, Keynes (1936) demonstrates, among others, that because the relationship
between speculative balances and interest rate is negative, in order to stimulate investment, it
is important for policy makers to keep interest rates low and below the level that interest rates
would have attained at full-employment. According to Keynes (1936), speculative demand for
money increases when people expect the market value of alternative assets such as consols or
government bonds that attract fixed coupon income or dividends to fall. Using his portfolio
allocation model, Tobin (1965) corroborates the findings of the earlier contribution of Gesell
(1911) by showing that welfare can be enhanced by a reduction in deposit interest rates, by
taxing money as proposed by Gesell (1911), or by accelerating the rate of growth in monetary
stock, thereby raising the rate of inflation. Various empirical studies such as those by Drazen
(1981) and Fischer (1979a,b; 1981) confirm the practical validity of the Tobin model. However,
some economists, notably McKinnon (1973) and Shaw (1973), argue against the pursuit of
financial repressive policies because it restricts financial institutions from being competitive in
resource mobilisation and allocation. Repressive financial policies lead to holding large excess
liquidity which limits the capacity of banks to create credit, and even to expand financial
services to the wider population (McKinnon, 1973; Shaw, 1973). Fry (1989; 1993) shows that
financial repression, rather than increasing capital inflows, increases current account deficit and
accelerates accumulation of foreign debts.
193
Stellenbosch University http://scholar.sun.ac.za
The implication is that financial repression policies are by and large restrictive as far as
competition in the financial system is concerned. Therefore, these repressive policies do not
create the ideal competitive environment for financial institutions to adopt efficient, innovative
and aggressive resource mobilisation strategies which could lead to the development of
specialised products and services aimed at international migrants. In the absence of
competition and privatisation, the few existing financial institutions which might offer remittance
services could have the tendency of charging supernormal profits.
5.3.1.2 The McKinnon-Shaw Financial Liberalisation Theory
McKinnon (1973) and Shaw (1973) dismiss the hitherto dominant Keynesian-structuralist
proposition by identifying low private savings and investment, interest rate ceilings, high
reserve requirements, high inflation tax, and direct credit control in favour of the state as the
deleterious consequences of financial repression. Contrary to the repressionists‟ view, financial
liberalists87 propose reliance on less discriminatory market forces for the free determination of
interest rates in order to facilitate an efficient process of financial intermediation and capital
mobility. McKinnon (1973) and Shaw (1973) observe that the savings and investment that take
place under financially repressive regimes are not only inadequate but also of low quality. This
is because the private sector is crowded-out, as the public sector has uncompetitive access to
credit and other services provided by the banks; but, unlike the private sector, the public sector
can save or invest in projects with low or negative rates of return. The McKinnon-Shaw theory
implies the determination of interest rates by market forces; the abandonment of the fixed
exchange rate regime; the relaxation of entry requirements in the formal financial market to
boost competition; the elimination of directed and selective credit controls; the privatisation of
state-owned banks to enhance efficiency, economies of scale and scope, the integration of
informal financial markets into the formal financial system; the removal of artificial barriers to
international capital mobility; and the establishment of vibrant capital markets.
A positive real deposit interest rate is likely to stimulate voluntary private savings because
banks now pay more attractive returns on deposits which can attract savings previously held in
87
Although McKinnon (1973) and Shaw (1973) are recognised as the scholars who formalised the intellectual
debate in favour of financial liberalisation, Fry (1995) traces the call for less government intervention in the financial
th
markets to the 17 century and acknowledges Locke (1695), Smith (1776), Bentham (1787) and Schumpeter (1912)
as the earliest proponents.
194
Stellenbosch University http://scholar.sun.ac.za
non-financial assets88 and outside the banking systems. Furthermore, Schumpeter (1912),
Goldsmith (1969), McKinnon (1973), Shaw (1973), and Bencivenga and Smith (1991) assert
that the extent of financial intermediation in an economy is crucial for efficient resource
mobilisation and allocation, and for dealing with moral hazards, adverse selection and issues
related to transaction costs. Also, because financial liberalisation embraces foreign exchange
deregulation and external account liberalisation, a country can benefit from higher inflows of
external capital in response to improved international integration, interest rate equalisation and
financial openness (Kapur, 1983; Mathieson, 1979; Fry, 1995). Further, under financial reforms,
a country stands a better chance to correct its trade imbalances through foreign exchange
adjustments, given the relative elasticity of imports and exports. Thus, the pursuit of financial
liberalisation enables governments to deploy both monetary and exchange rate policies to
attain external equilibrium. Levine and Zervos (1998) reason that, when the stock market is
liberalised, more risk-sharing between domestic and foreign residents is expected to decrease
equity premium; and increased capital inflows may also improve stock market liquidity whilst
increased liquidity reduces the equity premium.
Yet, using a 3-sector model involving households, firms, and the government, Akyuz (1995),
demonstrates that while the McKinnon-Shaw hypothesis may be applicable to raising financial
savings through higher real deposit interest rates, total savings may decrease due to a shift in
income from firms to renters, and even as lower tax revenue and higher interest payment on
debt reduce government savings. With regard to remittances, Beine et al. (2011) contend that
because remittances impact positively on macroeconomic stability and financial development,
there is a high incentive for migrant-home countries to liberalise their financial sector in order to
receive more remittances as a result of financial openness.
5.3.1.3 The Neostructuralist Financial Theory
Neostructuralists led by van Wijnbergen (1982; 1983), Taylor (1983), Buffie (1984) and
Kohsaka (1984), reject the proposition of the liberalists, arguing that financial liberalisation can
be counter-productive in developing countries because in these economies the role of informal
financial institutions is crucial in financial resource mobilisation and allocation. These
neostructuralists employed various theories including the cost-push inflation, mark-up pricing,
88
For entrepreneurs and corporate institutions, this can take the form of holding foreign currencies, excess raw
materials, inventories or intermediate goods whilst households may prefer the purchase of lands, gold and other
forms of tangible wealth which can appreciate in value.
195
Stellenbosch University http://scholar.sun.ac.za
and the Keynesian-adjustment mechanism models to oppose the predictions of the McKinnonShaw models. Essentially, the neostructuralists argue that because the liberalists failed to take
into account the importance of curb markets in developing countries, their models represent “a
serious lacuna” (Fry, 1995: 130). In the view of the neostructuralists, competitive financial
markets cannot be effective in developing countries because of high market failures due to low
levels of incomes, high fragmentation of the financial markets, and information asymmetry (van
Wijnbergen, 1983; Taylor, 1983; Buffie, 1984; Kohsaka, 1984; Stiglitz, 1994; Ogaki et al. 1996).
Furthermore, in the formal financial sector where commercial banks dominate, the bulk of
mandatory reserve requirements89 constitute an important leakage in the circular flow of funds
during the process of financial intermediation90. These reasons make curb financial markets
essential in the mobilisation of household saving and credit extension, especially in rural
communities where the largest proportion of the population of developing countries resides.
Consequently, Taylor (1983), van Wijnbergen (1983), Buffie (1984) and Khosaka (1984)
conclude that, practically, for developing countries in particular, the pursuit of financial
liberalisation is likely to inhibit the rate of economic growth due to credit constraints because
higher real interest rates increase the costs of production, reduce real wages, and cause
stagflation. In fact, Stiglitz (1994) was very emphatic in suggesting that government intervention
in the financial market should be considered as the better alternative towards achieving Paretooptimality conditions in developing countries, given the high incidence of market failures due to
missing and incomplete information, credit and insurance markets, as well as oligopolistic
tendencies which undermine perfect competition in the financial sector. However, government
intervention in the allocation of critical resources such as finance can only be efficient where
state institutions are strong as reflected in high levels of transparency and accountability. In
SSA, in particular, the application of the neostructuralist model in achieving desirable results is
doubtful because governments lack the capacity, integrity and public confidence to be
entrusted with the leading role of allocating scarce financial resources equitably and without
favour or without seeking egoistic and political interests.
89
Even under financial liberalisation, these requirements may be high in developing countries because in lowincome countries, generally, the propensity to dissave is much higher than the propensity to save.
90
But Courakis (1984; 1986) proves that if demand for loans is highly interest-inelastic relative to the demand for
deposits; equilibrium deposit rate rises as required reserve ratio increases; and the funds mobilised from required
reserves are deposited with specialised development finance institutions other than commercial banks for lending
that would not be undertaken by commercial banks, higher reserve requirements can generate higher deposits.
196
Stellenbosch University http://scholar.sun.ac.za
Additionally, the McKinnon-Shaw hypothesis only calls for the removal of discriminatory
repressive policies that distort financial prices, whilst encouraging governments to implement
prudent policies that allow market forces to lead the efficient determination of resource
allocation. Liberalists acknowledge the fact that perfect markets cannot occur in the real world;
therefore, the McKinnon-Shaw theory should not be taken as advocating for the operation of
perfect financial markets but a more competitive financial system based on the dictates of the
market rather than the government. In fact, McKinnon (1981; 1984) cautions that financial
liberalisation cannot succeed in developing countries without the role of the government
because fiscal discipline is an important pre-requisite since budget deficits in these economies
are often financed by taxing the financial system in a variety of ways. Furthermore, Chang and
Jung (1984) challenge the significance and applicability of the neostructuralists theory on the
grounds that, curb markets in developing countries are non-competitive, less developed and
fragmented, and not as efficient as the neostructuralists might want the world to believe.
Although, theoretically, the impact of financial liberalisation on savings mobilisation may be
ambiguous, the fact remains that if the multidimensional integrated policies of financial
liberalisation are pursued in a sound macroeconomic environment, the policies are likely to
achieve their fundamental objectives. Where a well-implemented financial liberalisation
programme leads to widening the scope of a financial system with higher access of the public
to quality banking services, and creating an attractive saving and investment environment
through higher opportunity for risk diversification and improved information symmetry, higher
official remittances are likely to be received. This is because when the financial sector is
repressed, migrants will be attracted to remit through unofficial channels that are not often
liable to the payment of taxes or meeting other stringent obligations imposed by monetary
authorities on the formal financial institutions91. Also, under a repressive financial environment,
the official remittance market will be uncompetitive with the few dominant MTOs charging
excessively high commissions with the aim of earning monopolistic rents as altruism makes the
supply of remittances highly inelastic to transfer fees. Alberola and Salvado (2006) developed a
2-period financial model to prove, without uncertainty, that when banks enter and compete in
the remittance market, the commissions charged by MTOs on remittance transfers will fall and
more official remittances can be received in migrant-home countries.
91
World Bank (1989:67) observes that financial repression breeds emergence of vibrant informal financial markets
(curb markets) that are not subject to any government regulation – be it “taxes, supervision or otherwise”.
197
Stellenbosch University http://scholar.sun.ac.za
5.3.2 Related Empirical Literature
In a fixed effects panel data analysis of 104 developing countries over the period 1995-2003,
Freund and Spatafora (2005) find that, in most developing countries and particularly those in
SSA, the presence of a dual exchange rate regime and the high transfer fees on remittances
(i.e. the sum of workers‟ remittances, compensation of employees and migrant transfers)
charged by official money transfer institutions are the reasons why SSA countries receive lower
official remittances and relatively higher informal remittances than countries from other regions.
Beine et al. (2011) investigate the relationship between remittances and financial openness in a
sample of 66 mostly developing countries from 1980 to 2005 using a dynamic generalised
ordered logit model and a 2-step process similar to the 2-stage least squares method.
Remittances were measured as the sum of workers‟ remittances, compensation of employees
and migrant transfer as ratio of nominal GDP whilst financial openness (the dependent
variable) was measured as a categorical variable according to the capital account openness
indicator. In the midst of other explanatory variables (political regime, trade openness, and
domestic financial development), it was found out that the impact of remittances on financial
openness was positive and statistically significant, but financial openness had no impact on
remittance inflows.
Based on descriptive statistical indicators, Singh and Hari (2011) conclude that international
remittances increased in India during the post-reforms era because of liberalisation and capital
account convertibility. The study period was 1971-2008; and international remittances were
proxied by the total migrant transfers.
From the literature reviewed, it is clear that none of the underlying theories of financial
liberalisation is self-sufficient for policy design towards attracting higher inflows of international
remittances through official channels. What seems palpable, however, is that under a
liberalised financial environment, there is a higher likelihood that financial institutions will
compete to mobilise resources from internal and international sources. For instance, as
competition in the financial market intensifies, banks are more likely to devise cost-saving
strategies to attract remittances from international migrants, and more especially because
domestic resource mobilisation might be difficult, given the low disposable household incomes
in SSA. Indeed, the implementation of financial liberalisation policies in a stable
macroeconomic environment as a step towards attracting higher international remittances
198
Stellenbosch University http://scholar.sun.ac.za
through official channels has been the most dominant policy recommendation from studies92 on
the sub-region. Moreover, although the role of curb markets in financial intermediation could be
important in developing countries, for which reason policies under financial liberalisation
programmes in SSA should have been taken into account, information on informal financial
sector is at best incomplete due to the fragmentation and low literacy level of participants. Also,
through liberalisation, the domestic financial market is expected to be integrated whereby
informal financial markets will be absorbed into the formal financial system. Furthermore, the
motivation for this study is to encourage the flow of remittances through the formal financial
system. And, because the financial sector policies being implemented in SSA since the 1980s
have been based on the recommendations of the liberalist school (World Bank, 1994; Mathieu,
1998), the analytical framework of this study is in line with the McKinnon-Shaw theory.
Consistent with the McKinnon-Shaw theory, financial liberalisation is expected to cause a
higher inflow of international remittances through official channels. Each specific financial
liberalisation policy, probably with the exception of DCRR (directed credit, reserve
requirements, and aggregated credit ceiling), is also expected to cause a higher inflow of
official international remittances. Similarly, the direct impact of financial liberalisation or each
specific financial liberalisation policy implemented is predicted to have a strong non-negative
contemporaneous impact. Table 5.2 summarises the expected impact of financial liberalisation
on international remittance inflows.
92
Some of these studies are Freund and Spatafora (2005), Giuliano and Ruiz-Arranz (2009), Gupta, Pattillo and
Wagh (2009), and Adenutsi (2011).
199
Stellenbosch University http://scholar.sun.ac.za
Table 5.2:
Expected Impact of Financial Liberalisation Policies on International Remittance Inflows
SPECIFIC
FLB POLICY
Interest rate
liberalisation
(IRC)
Reduction in
reserve
requirements*
Reduction in
directed credit
control*
MAIN MOTIVATION FOR
POLICY REFORM
Ensure real interest rates are
competitive and attractive to
induce higher private savings
and efficient credit allocation
Make more resources available
to banks to create more money
through higher extension of
credit to potential borrowers
especially the private sector
Ensure
efficient
resource
allocation such that projects with
relatively lower risks and higher
return attract the most credit
from lending institutions
Privatisation
of banks
(PVZ)
EXPECTED LONG-RUN IMPLICATIONS FOR OFFICIAL
INTERNATIONAL REMITTANCES
Positive: As real deposit rates in SSA increase relative to the
prevailing rates in migrant-host countries, more investmentoriented remittances are likely to be received in SSA as migrants
decide to save and invest at home rather than abroad.
Ambiguous: No direct effect on remittance inflows but if
permanent migrants from whom the most remittances are
received and who often remit more under sound macroeconomic
conditions in their home countries see this as a sound economic
policy, more remittances can be received. On the other hand,
altruistic-driven remittances may fall when, with a reduction in
reserve requirements, banks can now create more to ameliorate
credit constraints.
Ambiguous: This depends on whether a country receives more
or less altruistic remittances relative to non-altruistic remittances.
Altruistic remittances increase with limited access to private
sector credit, but non-altruistic remittances increase in response
to sound macroeconomic policies which offer improved private
sector investment opportunities.
Positive: With higher professionalism, banks may expand to
reduce the number of unbanked and offer more attractive
products and services to different segments of the market and
deliver services more promptly. Banks, therefore, are likely to
become more competitive in the international remittance market.
Positive: Competition breeds efficiency resulting in lower
average cost in the long run, which enables banks to charge
competitive commissions on international remittances and roll
out innovative products and services to attract migrants to remit
more through the banking system.
Make banks more profit-oriented
through
adherence
to
professional and innovative
practices for efficient service
delivery
Relaxation of
Promote competition among
entry barriers
banks and enhance operational
(EBC)
efficiency often necessitating
expansion
and
strategic
management even as banks
take higher operational risks
Prudential
Prevent predators from taking Positive: With improved confidence in the financial system,
regulation and advantage of the financial migrants will patronise services offered by banks, and cease to
supervision
system to exploit others, boost remit through unofficial money transfer channels.
(BKS)
confidence among agents and
moderate
risk-taking
under
competitive environment
External
Facilitate international capital Positive: International migrants may find it easier to remit more
account
mobility as a pre-condition for funds to SSA at relatively reduced cost; recipients can
liberalisation
mobilising
more
financial conveniently receive remittances in preferred foreign currencies.
(ICF)
resources from overseas.
Development
Create more opportunities for Positive: As opportunity to diversify risk is improved and more
of stock
portfolio risk diversification and investment products are made available, migrants are inclined to
markets
access to longer-term corporate remit more for investment-driven motives.
(SMK)
finance through equity capital
Source: Author. Note: *combined as aggregate credit ceilings, directed credit and reserve requirement (DCRR) in
Figure 5.2 as in Abiad et al. (2010) (see Chapter Two for details).
5.4 EMPIRICAL MODEL, METHODOLOGICAL APPROACH AND DATA ISSUES
A set of two empirical models and analytical approaches were employed to achieve the
objectives of this chapter. Each empirical model and methodological approach is briefly
discussed below.
200
Stellenbosch University http://scholar.sun.ac.za
5.4.1 Empirical Panel Granger Non-Causality Model and Analytical Approach
To test the causal relationship between international migrant remittance inflows and financial
liberalisation in SSA between 1980 and 2009, the Granger non-causality panel-data modelling
with fixed effects as proposed by Hurlin and Venet (2001), and Hurlin (2004) was adopted as
consistent with Equations (5.1) and (5.2).
k 1
k 0
ln REMPCi ,t k ln REMPCi ,t k k Fi ,t k i ,t
k 1
k 0
(5.1)
Fi ,t k Fi ,t k k ln REMPCi ,t k i ,t
(5.2)
where lnREMPC denotes natural logarithm of international remittances per capita, F is FLBI or
any policy component of FLB such as directed credit, reserve requirement and aggregate credit
ceilings (DCRR), interest rate control (IRC), banking sector entry requirements (EBC), banking
supervision (BKS), privatisation of banks (PVZ), international capital flows (ICF), and security
markets development (SMK). is the optimal lag selected according to the Akaike Information
Criterion (AIC), and the Schwarz Information Criterion (SIC) with the log likelihood ratio and
Durbin-Watson statistic playing a „judiciary‟ role in the event of a conflict between AIC and SIC.
The individual countries in the panel are represented by i and i 1, 2,....., N so that N 13
when the full sample analysis is conducted otherwise, either N 7 ; or N 6 when a subsample involving only seven countries with frontier and emerging financial markets; or when the
analysis is restricted to only the six countries with underdeveloped financial markets. The time
period is t and t 1, 2,...., T implying for the period 1980-2009, T 30. 93 Each error term has
two components (Baltagi, 2008), and hence can be decomposed into an unobservable countryand time-specific fixed component and a random disturbance component, where i ,t i ,t i ,t
and i ,t i ,t i ,t in which case
i and i are intercepts whilst i ,t and i ,t are the assumed
independently and normally distributed residuals with
E (i ,t ) 0 ; E ( i ,t ) 0 and finite
2
2
2
2
heterogeneous variances E (i ,t ) ,t ; E ( i ,t ) ,t ; t 1,....., T .
93
Based on the argument that only eight out of the 13 sampled countries initiated FLB programmes in the 1980s and
that most of the countries adopted FLB programmes in the late-1980s (see Table 5.1), T was reduced to 20 in a
second estimation of Equations (5.1) and (5.2). It is acknowledged that doing so in a panel setting of relatively small
T (20) against a small N (13; 7; 6) in the presence of an optimal lag of 2 in a Granger-causality modelling reduced
the degrees of freedom considerably which could adversely affect the reliability of the estimators reported in Table
A5.3.
201
Stellenbosch University http://scholar.sun.ac.za
The F-statistic was used to trace the existence and direction of causality by testing the
following alternative panel Granger non-causality hypotheses (HA) with respect to SSA:
For Equation (5.1), H A1 : k 0, k [1, ]; i [1, N ] and
k 0, k [0, ]; i [1, N ] .
For Equation (5.2), H A2 : k 0, k [1, ]; i [1, N ] and
k 0, k [0, ]; i [1, N ] .
In testing for causality in a panel setting, it is important to pay attention to the question of
heterogeneity which can be caused by permanent cross-sectional disparities especially in this
particular case which involves a small cross-section ( N ) over a large time series (T ) with N
classified into two distinct sub-groups - frontier and emerging, and underdeveloped financial
markets. Therefore, the estimation was carried out for each sub-group and the entire group. In
doing so, it was ensured that the balance within the panel setting remained unchanged just as
the lag order ( k ), so that the critical condition T 5 2k proposed by Hurlin (2004) is not
violated in any estimation.
Prior to the estimation of the empirical models, the tests for panel stationarity and co-integration
were conducted to address the concerns of spurious regression and to ensure that each
estimated regression meets a long-run equilibrium condition. These issues are particularly
relevant in situations where the panel structure is of large T and as is in the case of this study,
N T with
N
T
0 becoming more and more robust in the sub-sample estimations. According
to Kao (1999), it is possible to circumvent spurious regression by using panel data as paneldata estimates give a consistent estimate of the true value of the parameter as both T and N
approach infinity. However, Entorf (1997) proves that spurious regression can still manifest
itself in fixed effects regressions when the true model involves independent random walks, with
or without drifts, and more especially as T and N remains finite. As is the practice in most
panel data studies (Christopoulos and Tsionas, 2004; Hsiao and Hsiao, 20036), this study
relied on more than one panel-data unit root test. The Breitung t-statistic (BT) test developed by
Breitung (2000) and the Hadri Heteroskedasticity Consistent z-statistic (HHC) test by Hadri
(2000) were used in determining the stationarity status of the variables. Where a conflict
between the two aforementioned common-root tests surfaces, the Fisher Phillips-Perron
(Fisher P-P) chi-square test of individual root was conducted to validate the results. BT and
202
Stellenbosch University http://scholar.sun.ac.za
HHC were used rather than the popularly used Im, Pesaran and Shin (IPS) test and Levin, Lin
and Chu (LLC) test because IPS and LLC tests are less efficient as proven by many scholars
including Maddala and Wu (1999), Breitung (2000), Maddala et al. (2000) and Baltagi (2008).
Hlouskova and Wagner (2006) prove that the Breitung test outperforms the Hadri test. Yet, the
HHC test statistic, which is a residual-based Lagrange Multiplier (LM) test, has the speciality of
taking into consideration issues of heteroskedastic consistency. The panel unit root test results
reported in Table A5.1 reveal that all the variables are integrated of order zero.
Since each variable under consideration is I(0), the issue of co-integration is no longer a
serious concern. However, in order to erase any doubts and to reconfirm this principle, the
Engle and Granger (1987) procedure was used to investigate if there is a long-run relationship
between lnREMPC and FLBI; or lnREMPC and each of the components of FLBI in a bivariate
analysis with an intercept as recommended by (Asteriou, 2006). For a decision to be taken on
co-integration, the Johansen unrestricted co-integration rank (JUCR) tests for panel data based
on both the trace statistic and the maximum Eigenvalue statistic would have to ideally validate
the Engle-Granger 2-Step (EG2S) test, but even where JUCR failed to confirm EG2S, the
existence of co-integration cannot be rejected because JUCR is more applicable to nonstationary series (see Asteriou, 2006; Baltagi, 2008). In other words, given the I(0) status of all
the variables under consideration, co-integration is only considered to be non-existent if both
JUCR and EG2S consistently prove this. This is imperative because this study is interested in
only the long-run FLB-impact and causal relationship with international remittances. As was the
case in the EG2S procedure, a constant but no trend was included in the JUCR test with a lag
interval of 1-4, typical of tests involving data of annual frequency. The results of co-integration
tests reported in Table A5.2 suggest that a long-run equilibrium relationship exists between
lnREMPC and FLBI as well as between lnREMPC and each component of FLBI.
5.4.2 Empirical Static Panel Model and Methodological Approach
In order to determine the impact of FLBI as well as the policy-specific impact of FLB on
remittance inflows, a set of bivariate static panel models was analysed 94. The general
mathematical bivariate static panel-data model is of the form:
ln REMPCit f ( Fi ,t ) i 1, 2,..., N
t 1, 2,..., T
94
(5.3)
Severe multicollinearity among the various dimensions of the FLBI as reflected in the close similarity in the
reported pairwise correlation coefficients (see Table A5.5) precludes this study from a multivariate analysis.
203
Stellenbosch University http://scholar.sun.ac.za
where both variables and all notations, except the notation for functional relationship ( f ) , are
as defined in Equation (5.2). Following from Equation (5.3), the specific empirical econometric
model estimated is of the form:
ln REMPCit 0 1Fi ,t it
where, in this case,
(5.4)
0 is a scalar; the two-way composite disturbance term takes the form,
it i t it , so that i is the unobservable individual country-specific heterogeneity, t
captures the unobservable individual-invariant time effect, with the random error term ( it )
accounting for the remaining non-systematic effects, for which reason it
N (0, 2 ). It is
important to note that t takes care of any time-specific effect, such as a structural change or a
shock, that is not included in the estimated regression model (Wooldridge, 2002; Baltagi,
2008). The implication here is that, more explicitly, Equation (5.4) is actually of the form:
ln REMPCit 0 1Fi ,t i t it
(5.5)
Consistent with the McKinnon-Shaw FLB hypothesis, it is expected a priori that when (5.5) is
estimated,
1 0. Unlike Equation (5.2), with Equation (5.5), it is possible to explore the long-
run contemporaneous effects of financial liberalisation on international remittances in SSA
when the dynamic effects of remittances are discounted. Generally, Equation (5.5) can be
estimated by pooled OLS, panel Fixed Effects (FE) or panel Random Effects (RE) models
depending upon the assumptions made about the behaviour of i .
Pooled OLS can only be appropriate with efficient and unbiased estimators if Equation (5.5)
truly has a common constant because there are no differences among the sampled countries
(N), in which case
i 0 , implying:
ln REMPCit 0 1Fi ,t it
where
(5.6)
it t it as the time-variant effects it in a strictly bivariate model is absorbed by the
error term so that the empirical model is now reduced to a “one-way error component
regression model” (Baltagi, 2008: 13). Thus, the common constant approach to estimating
static panel-data models is best applicable under the hypothesis that the panel data under
204
Stellenbosch University http://scholar.sun.ac.za
consideration is a priori strictly homogenous to the extent that i (in Equation 5.5) has no
influence on the intercept ( 0 ) or the disturbance term ( it ) . From experience, the validity of
the assumption
i 0 is a rare possibility, particularly for panel data with large N .
Even though where
i 0 , the i terms can be rewritten as coefficients of a set of dummy
variables designed to account for the belongingness of the cross-sectional unit i and the
modified model can then be estimated with the appropriate dummies to capture the effects of
i ; this approach becomes complicated and impractical in large N and even as N . If i
is correlated with F , failure to account for the effects of i results in heterogeneity bias in the
estimated model due to omitted variable(s). Under this circumstance, the variations in i leads
2
to serial correlation in it , where E (itij ) for t j , pointing to the fact that pooled OLS
estimator would be inefficient with bias standard errors, requiring the adoption of either panel
FE models or panel RE models according to the orthogonality of i .
The panel FE model is appropriate where it is considered that each individual country has a
fixed-effect resulting in parametric shifts of the estimated regression by the fixed value for each
individual country. Mathematically, for an efficient panel FE estimator, Equation (5.6) must be
of the form:
ln REMPCit 1Fi ,t it
where
(5.7)
(0 i ) for i 0 , and where i which is now absorbed into the common
constant ( 0 ) varies according to individual countries (Greene, 2003). But where
i 0 , and
the effects of i is absorbed into a compound error term rather than the intercept, such that
it ( it i ) as in Equation (5.8), panel RE estimator rather FE estimator is more efficient and
less biased in the presence of endogeneity.
ln REMPCit 0 1Fi ,t it
(5.8)
Therefore, whereas in panel FE models i is assumed as constant, in panel RE models, i is
assumed to be drawn independently from some probability distributions. The underlying
205
Stellenbosch University http://scholar.sun.ac.za
principle of panel RE models is that, there is an individual effect, which is random rather than
fixed, and this effect may reflect omitted variables which are not fixed in nature (Maddala, 1971;
Greene, 2003).
A fixed effect can be determined from a selected sample, not a random sample obtained from
experimental design. Hence, inferences are applicable to only the observed effects and not to
the larger population. Another limitation of panel FE estimator is that it cannot be used to
determine the effects of time-constant covariates as those covariates cancel out during the
“within” transformation. A “within” estimator can be obtained from a typical FE model (Equation
5.8) as follows:
(ln REMPCit ln REMPC it ) 1 ( Fit F it ) ( it it )
(5.9)
where the mean of each variable is subtracted from that variable and the constant terms
cancelled out. An alternative way to obtain the FE estimator from Equation (5.8) is to use the
“between” approach, in which case each one lag rather than the mean of each variable is used
as shown in Equation (5.10):
(ln REMPCit ln REMPCit 1 ) 1 ( Fit Fit 1 ) ( it it 1 )
(5.10)
Panel RE is a variant of Generalised Least Squares (GLS) and it is used effectively when the
error term of a given static panel model is heteroskedastic, i.e. E ( ) 2 based on Equation
(5.10). The basic assumptions that must hold for the panel RE estimator to be efficient include
E ( it ) E(i ) 0 , E ( it2 ) 2 , E (i2 ) 2 , E ( iti ) 0 for all t and i , E ( it2 ) 2 2 for
t s , E ( it is ) 2 for t s , and most importantly, E ( Fiti ) 0 for all t and i for RE
estimator to be consistent, which was tested following the popularly used Hausman (1978)
specification test and the Breusch and Pagan (1980) Lagrange Multiplier (LM) test statistic.
Panel RE modelling involves the combination of the panel FE “within” estimator and the FE
“between” estimator, so that the RE estimator is the overall correctly-weighted average
estimator of the FE “within” and the “between” estimators.
Based on the results obtained in Chapter Four and the plausibility of a bi-causal relationship
between remittances and financial liberalisation, it is quite certain that the presence of
endogeneity in the bivariate empirical model (5.4) cannot be disregarded. Endogeneity in
206
Stellenbosch University http://scholar.sun.ac.za
empirical model (5.4) could arise from time effects ( t ) due to a systematic policy shock to F ,
simultaneity emanating from random shocks triggered by variations in F , measurement errors
in reporting F , and very importantly, and unobserved heterogeneity due to model underfitting.
It is well acknowledged in the remittance literature that measurement errors are severe (see
Chapter Two). In Chapter Four, it was found that tougher regulations aimed at clamping down
on using informal channels to remit following the September 11, 2001 terrorist attack on the US
impacted positively and significantly on official remittance inflows in SSA. This implies that by
not specifically accounting for time effects ( t ), the bivariate empirical models (5.6)-(5.10)
cannot be exonerated from obvious endogeneity bias. Evidence that remittances can, under
some circumstances, Granger-cause F (see Table 5.3) is a further indication that the
Equations (5.6)-(5.10) are prone to severe problem of endogeneity.
Considering the high possibility of endogeneity, it might seem obvious that the RE model
should be used for this study. However, the study proceeded to estimate the empirical model
by the three possible methods (pooled OLS, FE and RE) and selected the best model based on
the Haumsan test and the Breusch-Pagan (B-P) test95. The Hausman specification test for the
RE model, which was developed to test orthogonality of the random effects and the regressor,
is based on the notion that under the hypothesis of no correlation, OLS in the Least Squares
Dummy Variable (LSDV) model or FE model and GLS are equally consistent, but OLS
estimator is less efficient96. The B-P Lagrange Multiplier (LM) test designed for evaluating
reliability of random effects estimators derived from GLS is distributed as chi-square with one
degree of freedom on the null hypothesis that 2 0 , the alternative being 2 0 (see
Breusch and Pagan (1980) for further details). Using the overall index of financial liberalisation
(FLBI), the Hausman test as well as the B-P test strongly confirm the RE model as the superior,
hence producing the most efficient estimators (see Table 5.4). Specifically, the high B-P LM
test statistic far exceeds the 99 per cent critical value for chi-square with one degree of
freedom, leading to the rejection of the null hypothesis in favour of RE model (Table 5.4).
Similarly, except for the 1990-1999 model on SSA countries with underdeveloped financial
markets, the Hausman test consistently endorsed the RE estimators as the most efficient
(Table 5.4). Accordingly, the study proceeded to explore the impact of financial liberalisation on
95
2
Because the estimated pooled OLS model could not produce an improved R (see Table A5.4), whilst the basic
diagnostic tests including the F-statistic and Durbin-Watson test failed, it was not preferred over the panel models.
96
For further details on this test, see Hausman (1978) and Greene (2003).
207
Stellenbosch University http://scholar.sun.ac.za
international migrant remittance inflows in SSA based on the panel RE estimators rather than
on the panel FE estimators.
Finally, following the same estimation procedure outlined under 4.5.2.4 in Chapter Four, the
„differential‟ student t-test was used to verify the incidence and degree of decade-based
parameter evolution and instability for each of the relevant estimated decade-based models.
5.4.3 Data Type, Description and Sources
Data on financial liberalisation (see Table A5.8 for summary) was essentially obtained from
Abdul Abiad and Thierry Tressel of the IMF who, together with Enrica Detriagiache, published
“A New Database of Financial Reforms” in IMF Staff Papers, 57(2): 281-302 in the year 201097.
Following Abiad et al. (2010), the author constructed the financial liberalisation index for the
sampled countries for the most recent years, 2006-2009 which are covered by Abiad et al.
(2010). International migrant remittances are the sum of workers‟ remittances and
compensation of employees obtained from the April 2011 e-database and CD-ROM editions of
World Bank‟s WDI and Migration and Remittances Factbook 2011 (MRF-2011), and IMF‟s
BoPS. For Tanzania and Uganda, missing published data on remittances for the period 19801994 was filled in with estimates based on country-specific information obtained from countrydesk officials at the Headquarters of IMF and the World Bank in Washington, DC, USA.
5.5 EMPIRICAL RESULTS AND DISCUSSIONS
5.5.1 The Causal Effects of Financial Liberalisation on Remittance Inflows in SSA
The main empirical panel Granger-causality results of this study are presented in Table 5.3. On
the basis of the overall index (FLBI), the empirical results suggest that, generally, the pursuit of
financial liberalisation programme Granger-causes higher international remittance inflows
through official channels in SSA. This implies that for a typical SSA country to receive higher
inflows of official remittances, the liberalisation of its financial market is a necessary condition.
At 10 per cent level of statistical significance, it can be concluded that there is a reverse
causality between financial liberalisation and international remittance inflows in SSA. The
overall causal effect of financial liberalisation on international remittance inflows is only
statistically significant for the 13 sampled SSA countries and the seven sampled SSA countries
with frontier and emerging financial markets. Thus, in the case of the sampled SSA countries
with underdeveloped financial markets, however, the overall causal effect of financial
97
See Chapter Two for details on the procedures and main sources of raw data for this purpose.
208
Stellenbosch University http://scholar.sun.ac.za
liberalisation on international remittance inflows is statistically insignificant at the conventional
statistical levels.
Table 5.3:
Financial Liberalisation-Remittances Bivariate Panel Granger Non-Causality Results in SSA, 1980-2009
No. of Frontier & Emerging
Lags ‡ Obs
Null Hypothesis:
FLBI does not Granger cause lnREMPC
lnREMPC does not Granger cause FLBI
BKS does not Granger cause lnREMPC
lnREMPC does not Granger cause BKS
DCRR does not Granger cause lnREMPC
lnREMPC does not Granger cause DCRR
EBC does not Granger cause lnREMPC
lnREMPC does not Granger cause EBC
ICF does not Granger cause lnREMPC
lnREMPC does not Granger cause ICF
IRC does not Granger cause lnREMPC
lnREMPC does not Granger cause IRC
PVZ does not Granger cause lnREMPC
lnREMPC does not Granger cause PVZ
SMK does not Granger cause lnREMPC
lnREMPC does not Granger cause SMK
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
196
196
196
196
196
196
196
196
196
196
196
196
196
196
196
196
Others (underdeveloped) Full Sample (combined)
F-statistic(F-prob.) Obs
F-statistic(F-prob.)
Obs
F-statistic(F-prob.)
3.3233(0.04)**
2.6842(0.07)*
1.7021(0.19)
3.0983(0.05)**
0.9431(0.39)
0.0277(0.97)
2.0624(0.13)
0.3700(0.69)
1.0574(0.35)
2.5167(0.08)*
2.9737(0.05)*
1.2062(0.30)
0.4190(0.66)
1.3669(0.26)
4.9066(0.01)***
0.2817(0.75)
1.0733(0.34)
0.1451(0.87)
0.0249(0.98)
0.2006(0.82)
0.6127(0.54)
1.5260(0.22)
0.9141(0.40)
0.1869(0.83)
1.8263(0.16)
6.4845(0.00)***
4.3704(0.01)**
0.5226(0.59)
0.7238(0.49)
0.6084(0.55)
1.4194(0.24)
0.8156(0.44)
364
364
364
364
364
364
364
364
364
364
364
364
364
364
364
364
5.0715(0.01)***
2.3481(0.10)*
1.6752(0.19)
2.5310(0.08)*
0.8491(0.43)
0.5416(0.58)
3.0587(0.05)**
0.4457(0.64)
2.2426(0.11)
5.2644(0.01)***
5.9720(0.00)***
0.4452(0.64)
1.3058(0.27)
1.8524(0.16)
6.7773(0.00)***
0.8173(0.44)
168
168
168
168
168
168
168
168
168
168
168
168
168
168
168
168
‡
Source: Author. Note: is a model-specific optimal lag selected according to SIC, AIC and log likelihood.
REMPC here refers to migrant remittances per capita
The specific financial liberalisation policies implemented in SSA which have so far Grangercaused higher inflows of international remittances through official channels between 1980 and
2009, in order of magnitude and statistical relevance, are policy reforms on stock market
development (6.7773), interest rate deregulation (5.9720), and relaxation of entry barriers to
promote competition in the banking industry (3.0587). In each of these cases, the causality
direction is only one way – from the specific policy reform to remittance inflows. The effect of
interest rate deregulation on international remittance inflows in SSA countries with
underdeveloped financial markets is more robust compared to countries with frontier and
emerging markets. The causal effect of relaxation of barriers to entry into the banking industry
to allow for greater competition (EBC), although significant for the entire sample, is statistically
insignificant for each specific sub-sample.
Quite strikingly, although external account liberalisation does not Granger-cause increased
remittance inflows in SSA, remittances received in the sub-region were necessary to cause
elimination of policy constraints to international capital inflows. This holds for SSA countries
with frontier and emerging financial markets but more importantly, for countries with relatively
209
Stellenbosch University http://scholar.sun.ac.za
less developed financial markets. This finding seems to suggest that the more remittances
received, the more policy makers in SSA are encouraged to relax restrictions of international
capital inflows, probably, as a strategy to receive more remittances from abroad.
The results imply that other FLB policies such as reforms on directed credit and reserve
requirements, international capital flows, and privatisation have not caused increased flow of
remittances through official channels to SSA. Whereas the composite index of financial
liberalisation and specific FLB policies on stock market development and interest rate
deregulation Granger-cause international remittance inflows in SSA frontier and emerging
financial markets, so far it is only interest rate deregulation that Granger-causes higher inflow of
remittances in SSA countries with underdeveloped financial markets. As far as financial
liberalisation is concerned in SSA, FLB policy on stock market development is the main cause
of remittance inflows in countries with frontier and emerging financial markets, whilst interest
rate liberalisation is the most important cause of remittance inflows in countries with
underdeveloped financial markets. These results seem to lend support to the self-interest
economic motive and imply that higher returns on interest-bearing banking sector financial
assets or reduced interest rate related charges on remittances are the first and foremost
influencing factors that cause higher inflow of international remittances in SSA countries with
relatively underdeveloped financial markets. Similarly, there is strong evidence for the selfinterest investment motive in the case of SSA countries with frontier and emerging financial
markets where stock markets are relatively vibrant and with higher returns. There is also
evidence of international remittances causing improved banking supervision in SSA; and more
particularly in countries with frontier and emerging financial markets.
5.5.2 Empirical Results on the Impact of FLB on International Remittances in SSA
In Table 5.4, the empirical results of the overall impact of financial liberalisation as measured
by FLBI, as well as the FLB policy-specific impact on international remittance inflows in SSA
are presented. For the full sample of 13 SSA countries, the overall impact of financial
liberalisation on international remittance inflows was zero in the 1980s, but this turned positive
in the 1990s and the positive impact became even more robust, both statistically and
economically in the 2000s. Specifically, when all other factors are held constant, a one unit
increase in FLBI resulted in a 1.7241 and 5.2825 percentage increases in remittance inflows at
10 and 5 per cent levels of statistical significance in the 1990s and 2000s respectively. As a
sub-group, neither SSA countries with frontier and emerging markets nor other SSA countries
210
Stellenbosch University http://scholar.sun.ac.za
with underdeveloped financial markets recorded a statistically significant impact of FLBI on
remittance inflows in the 1980s and 1990s. In the 2000s, the impact of FLBI on remittance
inflows in SSA countries with frontier and emerging markets, and those with underdeveloped
financial markets was positive and significant for both sub-groups. Whilst the positive impact in
the 2000s of FLBI on SSA was economically more significant for countries with
underdeveloped financial markets, for the entire post-reforms era, 1980-2009, the impact of
FLBI on remittance inflows in SSA countries with frontier and emerging markets was relatively
higher 3.8050 compared to 2.0804 for countries with underdeveloped financial markets (Table
5.4). Taking into consideration the economic and the statistical significance of FLBI, it is
apparent that between 1980 and 2009, there was a consistent increasing impact of financial
liberalisation on remittance inflows in SSA; and this trend of impact is more consistent in
countries with frontier and emerging financial markets than in other SSA countries with
underdeveloped financial markets.
Turning to the specific FLB policies implemented in SSA over the past three decades, it can be
argued that with the exception of policies on entry requirements into the banking industry for
competition (EBC), generally, in the 1980s, FLB impacted negatively on remittance inflows.
More specifically, policies of banking supervision and regulations (BKS), and to some extent
international capital flow deregulation (ICF) in the case of underdeveloped financial markets, as
well as the privatisation of state-owned banks (PVZ) and stock market development (SMK) in
the case of frontier and emerging markets, were deterrent to remittance inflows in SSA in the
1980s. Thus, during the initial years of implementing FLB programmes in SSA, as far as
remittance inflows were concerned, it was the countries with underdeveloped financial markets
that, ironically, benefitted the most, and where reform policies on EBC, IRC, and SMK were the
most productive. This might be due to the fact that most of the sampled countries are at
similarly low levels of liberalisation in the 1980s, as these countries generally embarked upon
FLB programmes in the late 1980s (see Table 5.1).
Apart from an improved statistical significance on the impact of policies on directed credit,
aggregate
credit
ceilings
and
reserve
requirements
(DCRR),
SSA
countries
with
underdeveloped financial markets failed to build on the initial gains made on the effects of FLB
on remittance inflows during the 1990s. In the 1990s, besides DCRR, SSA countries with
frontiers and emerging financial markets can only boast of having enjoyed higher positive
impacts of two FLB-specific policies (BKS, and SMK) on remittance inflows when compared
211
Stellenbosch University http://scholar.sun.ac.za
with other SSA countries that have underdeveloped financial markets (Table 5.4).
In the 2000s, it was countries with underdeveloped financial markets that received a more
significant positive impact from FLBI and all the specific FLB-policy components except DCRR,
on officially reported remittance inflows. Unlike in the 1980s and the 1990s, during the 2000s,
FLB had a consistent positive impact on both categories of SSA countries. Apart from ICF in
countries with frontier and emerging financial markets, each FLB-specific policy had a positive
and more significant impact on remittance inflows in both categories of countries in the 20002009 decade. During the 2000s, for the entire sample of 13 countries, in descending order of
economic value, the FLB-specific policy effects are: IRC (0.8022), EBC (0.6580), SMK
(0.6096), PVZ (0.6002), DCRR (0.5999), BKS (0.3780), and ICF (0.2005) (see Table 5.4).
Table 5.4:
Results of the Impact of Financial Liberalisation on International Remittance Inflows in SSA 1980-09
Frontier and Emerging Markets
Full Sample (Both Markets)
Underdeveloped Financial Markets
1980-89 1990-99 2000-09 1980-09 1980-89 1990-99 2000-09 1980-09 1980-89 1990-99 2000-09 1980-09
FLBI
BKS
DCRR
EBC
0.0536
1.9421
3.7140
3.8050
7.5113
0.9393
9.8180
2.0804
2.5143
1.7241
5.2825
3.2078
(0.03)
(1.55)
(4.16)***
(3.11)***
(1.48)
(0.87)
(7.15)***
(1.96)**
(1.00)
(1.90)*
(3.51)***
(3.76)***
0.6277
-0.3657
0.5760
0.3800
1.0140
(2.48)** (-35.75)***
(2.14)**
(2.52)**
(4.49)***
-0.3598
0.9644
0.3278
1.1975
(-30.77)***
(2.72)***
(2.06)**
(3.91)***
IRC
PVZ
SMK
Overall R²
Obs
Groups
Hausman_FE
B-P stat_RE
0.1002
0.6067
(0.74)
(2.66)**
-0.2120
0.6280
0.7638
0.9939
0.9579
0.3908
0.6236
0.5484
0.2983
0.5378
0.5999
0.7572
(-1.23)
(2.12)**
(2.12)**
(2.35)**
(1.38)
(3.85)***
(2.03)**
(2.59)***
(0.67)
(2.90)***
(2.05)**
(3.15)***
0.4367
0.1461
0.4566
0.7679
0.9111
0.1513
1.1286
0.4385
0.6545
0.1548
0.6580
0.6290
(1.58) (31.73)***
(2.97)***
(8.02)***
(1.10)
(9.90)***
(2.22)**
(1.77)*
(2.00)**
(3.08)***
(3.71)***
0.6508
-2.4975
0.0155
0.5181
0.4130
-0.0341
0.1014
0.2005
0.6118
(3.50)*** (-17.49)***
(0.06)
(2.42)**
(0.78)
(-0.09)
(0.65)
(1.98)**
(3.58)***
(0.68)
ICF
?
n/a
0.2837
0.1063
0.1578
(1.25)
(0.58)
(1.36)
0.0190
0.2487
0.6093
0.5044
0.9042
0.2159
0.9025
0.4847
0.34
(0.10)
(1.38)
(3.26)***
(2.94)***
(1.88)*
(0.71)
(5.15)***
(1.60)
(1.14)
1.60
0.25
0.8022
0.6086
(4.45)***
(3.46)***
-1.1064
0.2554
0.4325
0.7583
0.2436
-0.0592
0.7171
0.2949
-0.6792
0.1954
0.6002
0.5448
(-2.72)***
(1.00)
(7.96)***
(2.09)**
(1.47)
(-0.24)
(5.93)***
(1.68)*
(-1.29)
(1.03)
(5.66)***
(2.77)***
1.4552
1.7788
0.0241
1.3075
0.7616
0.2926
0.2651
0.6096
1.2227
(-0.12) (34.06)***
(2.12)**
(0.38)
(1.32)
(3.48)***
(5.17)***
-0.6466
0.5240
0.4547
(-2.73)***
(1.66)*
(5.45)***
0.1265
0.0887
0.3920
0.3075
0.0505
0.3304
0.0950
0.1469
0.0693
0.1554
0.0681
0.2189
70
70
70
210
60
60
60
180
130
130
130
390
7
7
7
7
6
6
6
6
13
13
13
13
0.99
0.10
0.01
0.05
0.29
61.46
1.21
0.09
0.11
1.71
0.5
0.06
{0.32}
{0.75}
{0.94}
{0.82}
{0.59}
{0.00}***
{0.27}
{0.77}
{0.74}
{0.19}
{0.48}
{0.81}
(5.30)*** (20.26)***
235.57
181.01
262.16
1290.65
201.70
172.75
211.25
1501.22
456.16
358.17
469.69
2786.03
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
{0.00}***
Source: Author‟s estimation.
Note: ***(**)* represent statistical significance at 1%(5%)10% respectively.
Constant term included in each estimation.
z-statistics in ( ); 2 -statistics in { }
All statistics based on robust standard errors.
? means nonexistent (absorbed by constant term), n/a means not applicable
2
R , Hausman‟s specification tests, and B-P tests are based on estimated
models involving FLBI
212
Stellenbosch University http://scholar.sun.ac.za
For the group of 13 sampled SSA countries, since the adoption of FLB in the 1980s, holding all
other determinants of remittance inflows constant, policy reforms on developing the stock
market emerged as the most important (Table 5.4.1). Among the bank-based liberalisation
policies, banking supervision and prudential regulation emerged the most economically
significant. Other factors in descending order of economic significance are: DCRR (0.7572),
EBC (0.6290), IFC (0.6118), IRC (0.6086) and PVZ (0.5448) (see Table 5.4).
Table 5.4.1:
Financial Liberalisation-Remittance Impact by Rank of Economic Significance in SSA, 1980-09
BKS
Frontier & Emerging
DCRR
EBC
3
4
ICF
IRC
6
7ⁿ
5
1
3.20776
5
1
2.08042
6
7
1
3.80504
Others (Underdeveloped)
2
3
4
7
6ⁿ
Full Sample
2
3
4
5
Source: Author based on Table 5.4.
2
PVZ
SMK
FLBI-IMPACT
n
Note: means not statistically significant.
1=first (most significant),……….., 7=seventh (least significant)
The economic significance order of the specific FLB-policy impact on remittance inflows is
virtually the same in SSA countries irrespective of the level of financial market development
(Table 5.4.1). For example, the first five economically most important FLB policies (SMK, BKS,
DCRR, EBC, and PVZ) that impact on the inflow of remittances in both categories of SSA
countries are the same. And, while IRC and ICF are the sixth and seventh most important
specific policies in SSA countries with frontier and emerging financial markets, these two FLBspecific policies did not impact on the inflow of international migrant remittances in SSA
countries with underdeveloped financial markets when the entire 30-year period is taken into
consideration. This, notwithstanding, is the positive and significant effect of these two on
migrant remittance inflows in the 13 SSA countries during the 2000s (see Table 5.4).
Table 5.4.2 presents the results that investigate the statistical justification for the apparent
changing impact of financial liberalisation on official international remittance inflows in SSA
countries having frontier and emerging markets between 1980 and 2009. The empirical results
suggest that there is consistent evidence that the estimated decade-based parameters are
statistically different from one another lending further support to the hypothesis favouring a
pattern of evolution across the three decades in SSA countries with frontier and emerging
financial markets. The few cases in which this hypothesis could not hold consistently across
decades concern some of the components of the financial liberalisation index, namely: entry
barriers and pro-competition (EBC), international capital flows (ICF), privatisation of banks
(PVZ) and stock market development (SMK).
213
Stellenbosch University http://scholar.sun.ac.za
Table 5.4.2:
Parameter Evolution and Instability Test Results in Frontier
and DecadeEmerging SSA Financial Markets
Non-Overlapping
Estimated Decade-Based Results
FLBI
BKS
DCRR
EBC
ICF
IRC
A
B
C
1980-89
1990-99
2000-09
Decade-Based Rolling Based Coefficient Stability Test
Results
Estimated Results
D
Overlapping Decade-Based Coefficient
Stability Test Results
E
1985-1994 1995-2004
A-B
B-C
A-C
A-D
B-D
B-E
0.0536
1.9421
3.7140
2.8033
1.9890
-1.8885
-1.7719
-3.6604
-2.7497
-0.8612
-0.0469
C-E
1.7250
[1.7853]
[1.2529]
[0.8928]
[0.9014]
[0.6812]
[0.5324]
[0.3602]
[0.8925]
[0.8840]
[0.3516]
[0.5718]
[0.2116]
{8.15}***
{0.03}
{1.55}
{4.16}***
{3.11}***
{2.92}***
{-3.55}***
{-4.92}***
{-4.10}***
{-3.11}***
{-2.45}**
{-0.08}
-0.3598
0.9644
0.3278
1.9437
0.4003
-1.3241
0.6366
-0.6876
-2.3034
-0.9793
0.5641
-0.0725
[0.0117]
[0.3545]
[0.1591]
[0.3202]
[0.1067]
[0.3428]
[0.1954]
[0.1474]
[0.3085]
[0.0343]
[0.2478]
[0.0524]
{-30.77}***
{2.72}***
{2.06}**
{6.07}***
{3.75}***
{-3.86}***
{3.26}***
{-4.66}***
{-7.47}*** {-28.52}***
{2.28}**
{-1.38}
-0.2120
0.6280
0.7638
0.4528
0.0039
-0.8400
-0.1358
-0.9758
0.6240
0.7599
[0.1724]
[0.2962]
[0.3603]
[0.2461]
[0.1965]
[0.1238]
[0.0641]
[0.1879]
[0.0737]
[0.0501]
[0.0997]
[0.1638]
{-1.23}
{2.12}**
{2.12}**
{1.84}*
{0.02}
{-6.78}***
{-2.12}**
{-5.19}***
{-9.02}***
{3.49}***
(6.26}***
{4.64}***
-0.6648
0.1752
0.4367
0.1461
0.4566
0.2361
0.1465
0.2906
-0.3105
-0.0199
0.2006
-0.0900
-0.0003
0.3102
[0.6422]
[0.0925]
[0.0144]
[0.1276]
[0.2441]
[0.5497]
[0.0781]
[0.6278]
[0.5146]
[0.0351]
[0.1516]
[0.2297]
{0.68}
{1.58}
{31.73}***
{1.85}*
{0.60}
{0.53}
{-3.98}***
{-0.03}
{0.39}
{-2.56}**
{-0.00}
{1.35}
0.2837
0.1063
0.1578
0.3994
0.1651
0.1775
-0.0515
0.1260
-0.1157
-0.2931
-0.0588
-0.0073
[0.2270]
[0.1832]
[0.1160]
[0.2481]
[0.1162]
[0.0438]
[0.0672]
[0.1110]
[0.0211]
[0.0649]
[0.0670]
[0.0002]
{1.25}
{0.58}
{1.36}
{1.61}
{1.42}
{4.06}***
{-0.77}
{1.14}
{-5.48}***
{-4.52}***
{-0.88}
{-30.45}***
0.0190
0.2487
0.6093
0.1497
0.0585
-0.2297
-0.3606
-0.5903
-0.1307
0.0990
0.1902
0.5508
[0.1900]
[0.1802]
[0.1869]
[0.1062]
[0.2249]
[0.0098]
[0.0067]
[0.0031]
[0.0838]
[0.0741]
[0.0447]
[0.0380]
{14.50}***
{0.10}
{1.38}
{3.26}***
{1.41}
{-1.56}
{1.34}
{4.26}***
-1.1064
0.2554
0.4325
0.4600
0.2118
-1.3618
-0.1771
-1.5389
-1.5664
-0.2047
0.0436
0.2207
[0.4077]
[0.2553]
[0.0543]
[0.5823]
[0.0861]
[0.1514]
[0.2010]
[0.3524]
[0.1755]
[0.3269]
[0.1693]
[0.0317]
{-2.72}***
{1.00}
{7.96}***
{0.79}
{2.46}**
{-8.99}***
{-0.88}
{-4.37}***
{-8.92}***
{-0.63}
{0.26}
{6.95}***
-0.6466
0.5240
0.4547
0.9876
0.1718
-1.1706
0.0694
-1.1013
-1.6342
-0.4636
0.3522
0.2829
[0.2368]
[0.3157]
[0.0834]
[0.2896]
[0.1468]
[0.0788]
[0.2322]
[0.1534]
[0.0528]
[0.0261]
[0.1689]
[0.0634]
{-2.73}***
{1.66)*
{5.45}***
{3.41}***
{-7.18}*** {-30.97}*** {-17.79}***
{2.09}**
{4.46}***
0.1265
0.0887
0.3920
0.0576
0.0539
0.1076
0.2404
0.2592
0.0921
0.0732
0.0713
0.2230
70
70
70
70
70
70
70
70
70
70
70
70
7
7
7
7
7
7
7
7
7
7
7
Hausman_FE
0.99(0.32) 0.10(0.75)
0.01(0.94)
0.00(0.98)
0.02(0.89)
-
-
-
-
-
-
-
B-P Stat_RE
235.57***
262.16***
214.90***
250.18***
-
-
-
-
-
-
-
PVZ
SMK
Overall R²
Obs
Groups
181.01***
Source: Author‟s estimation
{0.26} {-23.48}*** {-54.02}*** {-189.96}***
{1.17} {-14.85}***
{0.30}
*/**/*** denotes significant at 10/5/1 per cent statistical levels respectively.
Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
With reference to column A-B, the estimated coefficients of the 1980-89 differ statistically from
the corresponding estimated coefficients of 1990-99 for the overall financial liberalisation index
(FLBI) and each of the components of FLBI, except EBC. Similarly, with the exception of ICF,
PVZ and SMK, the computed t -statistics reported in column B-C of Table 5.4.2 point to the fact
that the estimated coefficients of the 1990-99 decade are statistically different from the
corresponding estimates of the 2000-09 decade. In a similar vein, the estimated parameters of
the 1980-89 decade are statistically different from the corresponding parameter estimates of
the 2000-09 decade as reported in column A-C of Table 5.4.2.
214
7
Stellenbosch University http://scholar.sun.ac.za
In columns A-D, B-D, B-E and C-E of Table 5.4.2, the inter-temporal evolution of the
differences in the estimated decade-based coefficients involving the impact of financial
liberalisation on international remittance inflows in SSA countries with frontier and emerging
financial markets are reported. The only consistent statistical evidence of coefficient instability
in the evolution throughout the study period, with reference to SSA countries with frontier and
emerging financial markets, relates to DCRR (directed credit, reserve requirement and
aggregate credit ceilings) and stock market development (SMK) components of the overall
financial liberalisation index (FLBI). There is also an appreciably strong statistical evidence of
consistency in the instability of the estimated decade-based coefficients associated with the
impact of the overall financial liberalisation index (FLBI), banking supervision (BKS) and
international capital flows (ICF) on international remittance inflows in SSA countries with
frontier and emerging financial markets. In the case of FLBI and ICF, however, the coefficient
instability test fails when the respective estimated decade-based coefficients of the 1990-99
decade are compared with the corresponding estimates of the 1995-2004 overlapping decade.
Finally, there is widespread evidence in favour of decade-based coefficient instability when the
estimated parameters of the 2000-09 decade are compared with those of the 1995-2004
decade as reported in column C-E of Table 5.4.2.
Table 5.4.3 reports the results of the statistical inquiry into the estimated decade-based
changing impact of financial liberalisation index (FLBI) and its components on international
remittance inflows in SSA countries with underdeveloped financial markets. As in previous
related cases, columns A-B, B-C and A-C report the results of the statistical differences of the
estimated decade-based coefficients of 1980-89 and 1990-99, 1990-99 and 2000-09, as well
as 1980-89 and 2000-09 respectively. The estimated results suggest that in the case of SSA
countries with underdeveloped financial markets, when the corresponding estimated decadebased coefficients of 1980-89 and 1990-99 are compared, the various estimated decade-based
coefficients are statistically different from each other at the conventional statistical levels. A
similar result was obtained when the respective estimated coefficients of 1990-99 and 2000-09
decades are compared. The results, however, show that the estimated decade-based
coefficients of DCRR (directed credited, reserve requirements, and aggregate credit ceilings)
are statistically the same, implying that with reference to SSA countries having underdeveloped
financial markets, the decade-based impact of DCRR on international migrant remittance
inflows is statistically the same in the 1980s, the 1990s and the 2000s. The empirical results of
the statistical test on comparing the estimated decade-based coefficients of the 1980-89
215
Stellenbosch University http://scholar.sun.ac.za
decade to those of the 2000-09 decade show that besides FLBI, DCRR and interest rate
control (IRC), the various components of FLBI actually impact on international remittance
inflows differently in the 1980s and the 2000s as reported in Table 5.4.
Table 5.4.3:
Parameter Evolution and Instability Test Results in Underdeveloped SSA Financial Markets
Decade-Based Rolling
Estimated Results
Estimated Decade-Based Results
FLBI
A
B
C
D
E
1980-89
1990-99
2000-09
1985-1994
1995-2004
EBC
ICF
IRC
PVZ
SMK
Overall R²
Obs
Groups
B-C
A-C
A-D
B-D
B-E
C-E
0.9393
9.8180
0.8444
2.8612
6.5720
-8.8787
-2.3067
6.6669
0.5040
-1.9219
6.9568
[5.0752]
[1.0796]
[1.3731]
[0.4352]
[0.7569]
[3.9956]
[0.2935]
[3.7021]
[4.6400]
[0.6444]
[0.3227]
[0.6162]
{1.64}*
{-30.25}***
{11.29}***
{0.87}
{7.15}***
{1.94}**
{3.78}***
{0.78}
{-5.96}***
?
0.1002
0.6067
0.1639
0.4466
?
-0.5065
?
?
-0.0387
-0.3464
0.1601
?
[0.1354]
[0.2281]
[0.1389]
[0.1861]
?
[0.0927]
?
?
[0.1939]
[0.0507]
[0.0420]
{-0.62}
{1.44}
{-5.46}***
{3.81}***
{0.74}
{2.66}***
{1.18}
{2.40}**
?
?
?
{-0.20}
{-6.83}***
0.9579
0.3908
0.6236
0.1515
0.5358
0.5671
-0.2328
0.3344
0.8065
0.3106
-0.1450
0.0878
[0.6942]
[0.1015]
[0.3072]
[0.0802]
[0.1728]
[0.5926]
[0.2057]
[0.3870]
[0.6140]
[0.0214]
[0.0713]
[0.1344]
?
DCRR
A-B
Overlapping Decade-Based Coefficient Stability
Test Results
7.5113
{1.48}
BKS
Non-Overlapping Decade-Based
Coefficient Stability Test Results
{1.38}
{3.85}***
{2.03}**
{1.89}*
{3.10}***
{0.96}
{-1.13}
{0.86}
{1.31}
{14.55}***
{-2.03}**
{0.65}
0.9115
0.1513
1.1286
0.2101
0.2640
0.7602
-0.9773
-0.2171
0.7014
0.0779
-0.1127
0.8646
[0.1136]
[0.1375]
[0.1140]
[0.0735]
[0.1941]
[0.0240]
[0.0236]
[0.0004]
[0.0402]
[0.0641]
[0.0565]
[0.0801]
{8.02}***
{1.10}
{9.90}***
{2.86}**
{1.36}
{31.75}***
{-41.48}***
{-562.38}***
{17.47}***
{1.21}
{-1.99}**
{10.79}***
-2.4975
0.0155
0.5181
0.4328
0.3631
-2.5130
-0.5027
-3.0156
-2.9303
-0.1967
-0.3477
0.1550
[0.1428]
[0.2578]
[0.2141]
[0.2121]
[0.1737]
[0.1150]
[0.0437]
[0.0713]
[0.0694]
[0.0457]
[0.0841]
[0.0404]
{-17.49}***
{0.06}
{2.42}**
{2.04}**
{2.09}**
{-21.84}***
{-11.50}***
{-42.29}*** {-42.25}***
{-4.30}***
{-4.13}***
{3.84}***
0.9042
0.2159
0.9025
0.0042
0.7666
0.6884
-0.6867
0.0017
0.9000
0.1311
-0.5507
0.1360
[0.4810]
[0.3040]
[0.1752]
[0.0848]
[0.2808]
[0.1769]
[0.1288]
[0.3057]
[0.3962]
[0.2192]
[0.0232]
[0.1055]
{1.88}*
{0.71}
{5.15}***
{0.05}
{2.73}***
{3.89}***
{-5.33}***
{0.01}
{2.27}**
{0.60}
{-23.69}***
{1.29}
0.2436
-0.0592
0.7171
0.1538
0.1729
0.3028
-0.7763
-0.4735
0.0898
-0.1590
-0.2321
0.5442
[0.1657]
[0.2465]
[0.1209]
[0.0999]
[0.1081]
[0.0808]
[0.1256]
[0.0448]
[0.0659]
[0.1467]
[0.1385]
[0.0128]
{1.47}
{-0.24}
{5.93}***
{1.54}
{1.60}*
{3.75}***
{-6.18}***
{-10.57}***
{1.36}
{-1.08}
{-1.68}*
{42.37}***
1.7788
-0.0241
1.3075
-0.0736
0.5163
1.8029
-1.3316
0.4713
1.8524
-0.2345
-0.5404
0.7912
[0.0878]
[0.2008]
[0.0384]
[0.2104]
[0.1660]
[0.1129]
[0.1624]
[0.0494]
[0.1226]
[0.0097]
[0.0347]
[0.1276]
{20.26}***
{-0.12}
{34.06}***
{-0.35}
{3.11}***
{15.96}***
{-8.20}***
{9.54}***
{15.11}***
{-24.30}***
{-15.55}***
{6.20}***
0.0505
0.3304
0.0950
0.1646
0.2625
0.1905
0.2127
0.0728
0.1076
0.2475
0.2965
0.1788
60
60
60
60
60
60
60
60
60
60
60
60
6
6
6
6
6
6
6
6
6
6
6
61.46(0.00)*** 1.21(0.27)
0.65(0.42)
0.30(0.58)
-
-
-
-
-
-
-
213.17***
203.41***
-
-
-
-
-
-
-
Hausman_FE
0.29(0.59)
B-P STAT-RE
201.70***
172.75***
Source: Author‟s estimation
211.25***
*/**/*** denotes significant at 10/5/1 per cent statistical levels respectively.
Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
Again, with reference to the computed „differential‟ t-statistics reported in Table 5.4.3, it can be
concluded that generally, the evolution of the estimated decade-based coefficients are not
statistically stable over time except in the case of SMK and ICF, and to some extent entry
barriers and pro-competition (EBC) and FLBI. The remaining decade-based coefficients exhibit
inconsistent and isolated cases of stability in two out of the possible four estimations. The
statistical evidence of relatively low and inconsistent parameter stability over time in SSA
countries with underdeveloped financial markets might be due to the relative slow pace of
financial liberalisation process in these countries especially during the early stages of economic
216
6
Stellenbosch University http://scholar.sun.ac.za
reforms (see Abiad et al. 2010). It is also possible that a significant proportion of the officially
reported remittances received in SSA countries with underdeveloped financial markets during
the 1980s and the 1990s might have passed through other officially recognised channels such
as the MTOs and post offices hence outside the banking system and the stock markets upon
which the financial liberalisation index was developed.
The results of the estimated decade-based coefficients evolution and stability tests on the
impact of financial liberalisation on international remittances in 13 sampled SSA countries for
which relevant data are available for a bivariate panel-data analysis are presented in Table
5.4.4. More specifically, within the context of a bivariate panel-data analysis, the computed tstatistics providing the statistical evidence for the changing impact of financial liberalisation on
international remittance inflows in the 1980-89, 1990-99 and 2000-09 decades are reported in
columns A-B, B-C and A-C respectively of Table 5.4.4.
The results reported in column A-B show that the impact of the overall index of financial
liberalisation (FLBI), directed credit, reserve requirements and aggregate credit ceilings
(DCRR), international capital flows (ICF), interest rate control (IRC) and stock market
development (SMK) on international remittance inflows in the 1980-89 decade is not
statistically different from the corresponding estimated coefficients reported for the 1990-99
decade reported in Table 5.4. This might be due to the fact that most of the sampled SSA
countries actually began the implementation of financial reform policies in the latter part of the
1980s. Secondly, the pace of financial deregulation and reforms during the initial stages of
implementation might be relatively slow. Apart from DCRR, in SSA, FLBI as well as each of its
component indicators had a statistically different impact on international remittance inflows in
1990-99 decade when compared with the corresponding estimated coefficients for the 2000-09
decade as suggested by the computed „differential‟ t-statistics reported in columns B-C of Table
5.4.4. There is substantial evidence that, both at the integrated and disaggregated levels,
financial liberalisation impact on international remittance inflows in SSA differs statistically
across the 1980-89 decade and the 2000-09 decade as shown in column A-C.
217
Stellenbosch University http://scholar.sun.ac.za
Table 5.4.4:
Financial Liberalisation-Remittance
Parameter
Evolution and Instability Test Results in SSA
Decade-Based
Estimated Decade-Based Results
FLBI
BKS
DCRR
EBC
ICF
IRC
PVZ
SMK
Overall R²
Obs
Groups
A
B
C
1980-89
1990-99
2000-09
Rolling Estimated
Results
D
Non-Overlapping Decade-Based Overlapping Decade-Based Coefficient Stability
Coefficient Stability Test Results
Test Results
E
1985-1994 1995-2004
B-C
A-C
2.5143
1.7241
5.2825
1.9261
2.3444
A-B
0.7902
-3.5584
-2.7683
A-D
0.5882
B-D
-0.2020
B-E
-0.6203
C-E
2.9381
[2.5142]
[0.9074]
[1.5050]
[0.5335]
[0.5042]
[1.6068]
[0.5976]
[1.0093]
[1.9807]
[0.3739]
[0.4032]
[1.0008]
{2.94)***
{1.00}
{1.90}**
{3.51}***
{3.61}***
{4.65}***
{0.49}
{-5.95}***
{-2.74}**
{0.30}
{-0.54}
{-1.54}
-0.3657
0.5760
0.3800
0.8911
0.4102
-0.9416
0.1960
-0.7456
-1.2567
-0.3151
0.1658
-0.0302
[0.0102]
[0.2692]
[0.1508]
[0.1853]
[0.0926]
[0.2589]
[0.1184]
[0.1406]
[0.1750]
[0.0839]
[0.1765]
[0.0582]
{-35.75}***
{2.14}**
{2.52}**
{4.81}***
{4.43}***
{-3.64}***
{1.66}*
{-5.30}***
{-7.18}***
{-3.75}***
{0.94}
{-0.52}
0.2983
0.5378
0.5999
0.2561
0.2716
-0.2395
-0.0621
-0.3016
0.0422
0.2817
0.2662
0.3283
[0.4452]
[0.1854]
[0.2926]
[0.1133]
[0.1275]
[0.2598]
[0.1072]
[0.1526]
[0.3319]
[0.0721]
[0.0579]
[0.1651]
{0.67}
{2.90}***
{2.05}**
{2.26}**
{2.13}**
{-0.92}
{-0.58}
{-1.98}**
{0.13}
{3.91}***
{4.59}***
{1.99}**
0.6545
0.1548
0.6580
0.2272
0.2206
0.4996
-0.5032
-0.0036
0.4273
-0.0724
-0.0658
0.4374
[0.3697]
[0.0774]
[0.2136]
[0.0817]
[0.1511]
[0.2923]
[0.1362]
[0.1561]
[0.2880]
[0.0043]
[0.0737]
[0.0625]
{6.99}***
{1.77}*
{2.00}**
{3.08}***
{2.78}**
{1.46}
{1.71}*
{-3.69}***
{-0.02}
{1.48}
{-16.81}***
{-0.89}
-0.0341
0.1014
0.2005
0.4295
0.2237
-0.1356
-0.0991
-0.2346
-0.4637
-0.3281
-0.1222
-0.0232
[0.3791]
[0.1561]
[0.1013]
[0.1746]
[0.0956]
[0.2230]
[0.0548]
[0.2778]
[0.2045]
[0.0185]
[0.0605]
[0.0057]
{-4.08}***
{-0.09}
{0.65}
{1.98}**
{2.46}**
{2.34}**
{-0.61}
{-1.81}*
{-0.84}
{-2.27}**
{-17.69}***
{-2.02}**
0.3379
0.2467
0.8022
0.1141
0.2949
0.0912
-0.5555
-0.4643
0.2238
0.1326
-0.0483
0.5073
[0.2964]
[0.1542]
[0.1803]
[0.0731]
[0.1705]
[0.1422]
[0.0261]
[0.1161]
[0.2232]
[0.0810]
[0.0163]
[0.0098]
{0.64} {-21.28}***
{51.79}***
{1.14}
{1.60}
{4.45}***
{1.56}
{1.73}*
{-4.00}***
{1.00}
{1.64}
{-2.96}**
-0.6792
0.1954
0.6002
0.1995
0.2006
-0.8746
-0.4048
-1.2794
-0.8787
-0.0041
-0.0052
0.3996
[0.5265]
[0.1897]
[0.1060]
[0.1750]
[0.0671]
[0.3368]
[0.0837]
[0.4205]
[0.3515]
[0.0147]
[0.1226]
[0.0389]
{10.26}***
{-1.29}
{1.03}
{5.66}***
{1.14}
{2.99}***
{-2.60}**
{-4.84}***
{-3.04}***
{-2.50}**
{-0.27}
{-0.04}
0.2926
0.2651
0.6096
0.7117
0.3056
0.0275
-0.3445
-0.3170
-0.4192
-0.4466
-0.0406
0.3039
[0.7699]
[0.2008]
[0.1752]
[0.2039]
[0.1092]
[0.5691]
[0.0257]
[0.5947]
[0.5660]
[0.0031]
[0.1951]
[0.1845]
{0.38}
{1.32}
{3.48}***
{3.49}***
{2.80}**
{-0.53}
{-0.74} {-143.60}***
{-0.21}
{1.65}*
0.0693
0.1554
0.0681
0.0792
0.1353
0.1124
{0.05} {-13.43}***
0.1118
0.0687
0.0743
0.1173
0.1454
0.1017
130
130
130
130
130
130
130
130
130
130
130
130
13
13
13
13
13
13
13
13
13
13
13
13
Hausman-FE
0.11(0.74) 1.71(0.19)
0.50(0.48) 0.18(0.67) 0.29(0.59)
-
-
-
-
-
-
-
B-P STAT-RE
456.16***
469.69***
-
-
-
-
-
-
-
358.17***
Source: Author‟s estimation
438.26***
458.71***
*/**/*** denotes significant at 10/5/1 per cent statistical levels respectively.
Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
Considering the results of coefficient instability test reported in columns A-D, B-D, B-E and C-E
of Table 5.4.4, it is quite obvious that apart from the estimated coefficients of ICF, none of the
estimated decade-based coefficients exhibits a complete and consistent pattern of instability
over the three decades. The overriding implication of this result is that although, generally, the
impact of financial liberalisation on remittance inflows seems to vary from decade to decade,
there is very little evidence in favour of instability among the estimated decade-based
coefficients over the period 1980-2009. This might be due to the general slow pace of the
liberalisation process in the early stages of the economic reforms and the fact that most of the
officially reported remittances received in SSA during the 1980s and the 1990s might have
been transferred through MTOs and post offices rather than the formal financial institutions
218
Stellenbosch University http://scholar.sun.ac.za
such as banks and stock markets.
5.6 CONCLUSIONS AND POLICY RECOMMENDATIONS
As SSA countries with underdeveloped financial markets received more official remittances
than the SSA countries with frontier and emerging financial markets, it is implied that most of
the officially reported remittances received in SSA between 1980 and 2009 were transferred
through other officially approved channels other than the banking system and the stock market.
This is evidence that SSA banks are not active participants in the international remittance
market; leaving MTOs mainly Western Union and MoneyGram to be enjoying monopolistic
advantages. Under this circumstance, given the altruistic nature of remittances and the demand
for MTO services on official remittances, international money transfer might be highly priceinelastic enabling these MTOs to charge higher fees on remittance transfers in a bid to earn
supernormal profits on the major SSA remittance-corridors. The low participation of banks in
the remittance market might be due to mistrust of banks and the relative superior efficiency of
MTOs, at least, in connection with the speed, coverage, precision and reliability of the service
offered. The low participation of SSA banks may also be due to the existence of structural and
systemic constraints to competitive conditions such as poor financial infrastructure,
inappropriate legal and regulatory framework, and lack of convenient access to banking
services.
In contrast to MTOs, the few SSA banks that have offshore facilities to facilitate cross-border
payment systems can be described not only as generally inefficient and unreliable, but also
costly due to lack of automated direct links for retail transfers. This is why effecting a single
remittance transfer to a typical SSA country through the banking system often involves a multistage series of network correspondent bank transactions that do not only increase the total cost
of remitting but also delay the time taken for the target recipient to receive the funds
transferred. The forefront position of SSA bourses in attracting official remittances should not
be taken for granted as, unlike SSA banks, the majority of these stock exchanges are part of
the global capital markets and have online facilities. Clearly, SSA banks are not taking
advantage of financial liberalisation policies pursued in the sub-region to attract more
remittances strategically. In other words, even though the implementation of FLB programme is
necessary, it is not in itself sufficient to attract more remittances through the formal financial
system.
219
Stellenbosch University http://scholar.sun.ac.za
The results of this study imply that the underlying objectives have been achieved and the
answers to the specific research questions posed have been provided. The main answers are
that, by the stylised facts, SSA countries with frontier and emerging financial markets are, in
relative terms, more likely to receive more remittances through the formal financial system
when compared with other SSA countries with underdeveloped financial markets. Generally,
financial liberalisation Granger-causes international remittance inflows with a low statistical
evidence for a reverse causal effect. Policies on the stock market development Grangercaused higher inflows of remittances than the financial liberalisation policies on the banking
system. On the impact of financial liberalisation, and each of the specific reform policies
implemented under the programme, the findings of this study provide the affirmation that the
overall and specific policy impact of financial liberalisation on remittance inflows is positive and
that this impact is more significant in countries with frontier and emerging financial markets
than in other SSA countries where financial markets are underdeveloped. The statistical and
economic significance of the effects of financial liberalisation and each of its specific polices on
international remittance inflows have been generally increasing over time.
It is, therefore, recommended that, in order to attract more official remittances to SSA through
the formal financial system, domestic banks should devise strategies to enter into the lucrative
remittance market and compete away supernormal profits earned by MTOs. This will reduce
the cost of remitting on the official remittance-corridors of SSA, and encourage migrants to
patronise the services offered by banks in the international remittance market. More
specifically, by the evidence of the empirical results and best practices elsewhere, it is
suggested that SSA banks should:
i.
Introduce differentiated services and develop remittance-products through technological
innovation. Innovative products, online and automation in payment systems are
necessary to reduce the cost of handling small cross-border money transfers. Although
this may require huge capital investment, banks must appreciate the substantial longterm benefits from such an investment project mainly because remittance inflows, which
are less negatively affected by economic downturns, can cushion banks in periods of
recession. Consequently, SSA banks should develop the payment systems that can
directly communicate across borders or even develop products that can use existing
payment systems such as credit cards and/or ATM products.
220
Stellenbosch University http://scholar.sun.ac.za
ii. Open overseas branches and offer more offshore services to residents at home to
facilitate payments and receipts of remittances internationally. With an effective network
system, it should be possible to develop and extend the internal electronic proprietary
payment systems to all branches including overseas branches, and complement this
payment system with an account-to-account collection and delivery system. Through
this, remittances deposited in overseas branches should be easily and readily
accessible to target recipients at home at relatively low transaction costs.
iii. Through further liberalisation of the financial market, promote competition that breeds
financial innovation both in terms of products and services. One aspect of the
competitive landscape of which banks can take advantage is to target migrants as
customers by offering relatively low costs on remittance services whilst aiming at
making reasonable gains from other supporting services offered, using remittances as
leverage. Examples include providing relevant information on investment opportunities
at home, offering mortgages and housing loans, and assisting migrants in planning for
their retirement and in insuring their valuable assets.
iv. Establish bilateral and multilateral partnerships and networks with one another not
excluding rural and community banks, and with post offices and foreign banks, in order
to build an efficient and reliable national and international payment systems among
collaborating banks and institutions, in a manner that will make banks appear more
visible and conveniently accessible in the remittance market at home and overseas. In
order to succeed in the long run, SSA banks must focus on strategic partnerships,
networks, and negotiated alliances and franchises similar to the models used by large
multinationals, so as to enable them to overcome the challenges of high operational
costs and the geographic fragmentation of the remittance markets.
Meanwhile, because stock markets tend to be the most important channel through which SSA
migrants remit (with reference to the impact on official remittances received), it is hereby
suggested that policy makers should design policies aimed at further enhancing electronic
trading via stock market automation, and improving financial literacy on the role of stock
markets as well as the regional and international integration of SSA stock markets among
migrants and potential recipients of remittances. In SSA countries where stock markets do not
currently exist or are underdeveloped, regionalisation of capital markets could be of
tremendous benefit in this regard.
221
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 5
Table A5.1: Panel Unit Root Test Results
PANEL UNIT ROOT TEST STATISTICS
Breitung t-stat
Hadri HC z-stat
Fisher P-P chi-square
Conclusion
st
At Level
1 Difference
At Level
At Level
BKS
-1.4719*
4.5862***
I(0)
{0.0705}
(0.0000)
DCRR
-1.5026*
4.8241***
I(0)
{0.0665}
(0.0000)
EBC
-2.5113***
6.6481***
I(0)
{0.0060}
(0.0000)
FLBI
-2.5523***
3.5286***
I(0)
{0.0054}
(0.0002)
ICF
-2.7046***
5.4839***
I(0)
{0.0034}
(0.0000)
IRC
-0.2735
-3.8695***
5.6366***
-2.5925***
I(0)
{0.3922}
{0.0001}
(0.0000)
[0.0048]
lnREMPC
-0.3146
-7.2297***
6.0585***
46.2448***
I(0)
{0.3765}
{0.0000}
(0.0000)
[0.0086]
PVZ
-2.9646***
7.0261***
I(0)
{0.0015}
(0.0000)
SMK
-2.9011***
5.8985***
I(0)
{0.0019}
(0.0000)
*** ** *
Source: Author‟s computations
Note: Figures in brackets are respective probability values. / / significant
at 1/5/10 level statistical level respectively. Constant and trend included.
Table A5.2: Results of Panel Co-integration Tests
Engel-Granger 2-Step (EG2S)
Explanatory Variable Resid_Mean
BKS
DCRR
EBC
FLBI
ICF
IRC
PVZ
SMK
1.76E-16
6.38E-17
3.60E-16
1.48E-16
1.40E-16
1.84E-16
1.21E-16
2.32E-16
Source: Author‟s estimation
t-stat(t-prob.)
6.0950(0.00)***
6.9910(0.00)***
6.7017(0.00)***
4.5307(0.00)***
5.5711(0.00)***
6.8417(0.00)***
7.3991(0.00)***
7.6376(0.00)***
Johansen Unrestricted Cointegrating Rank (JUCR) Test
Hypothesis
Trace stat
5% Critical Value Max-Eigen stat
5% Critical Value
None
7.824
15.4947
6.8273
14.2646
At most 1
0.9967
3.8415
0.9967
3.8415
None
22.9069***
15.4947
17.2493**
14.2646
At most 1
5.6576**
3.8415
5.6576**
3.8415
None
30.5906***
15.4947
22.1948***
14.2646
At most 1
8.3957***
3.8415
8.3957***
3.8415
None
19.5899**
15.4947
12.6661*
14.2646
At most 1
6.9238***
3.8415
6.9238***
3.8415
None
20.9139***
15.4947
15.7397**
14.2646
At most 1
5.1742**
3.8415
5.1742**
3.8415
None
29.1684***
15.4947
23.2101***
14.2646
At most 1
5.9582**
3.8415
5.9582**
3.8415
None
17.1967**
15.4947
14.3087**
14.2646
At most 1
2.8880*
3.8415
2.8880*
3.8415
None
16.3152**
15.4947
13.5167**
14.2646
At most 1
2.7985*
3.8415
2.7985*
3.8415
Note: ***/**/* represent statistical significance at 1%/5%/10% respectively.
Constant no trend included the HHC test for the EG2S and the JUCR tests.
222
Stellenbosch University http://scholar.sun.ac.za
Table A5.3:
Financial Liberalisation-Remittances Bivariate Panel Granger Non-Causality Results in SSA, 1990-2009
Null Hypothesis:
FLBI does not Granger cause lnREMPC
lnREMPC does not Grangercause FLBI
BKS does not Granger cause lnREMPC
lnREMPC does not Granger cause BKS
DCRR does not Granger cause lnREMPC
lnREMPC does not Granger cause DCRR
EBC does not Granger cause lnREMPC
lnREMPC does not Granger cause EBC
ICF does not Granger cause lnREMPC
lnREMPC does not Granger cause ICF
IRC does not Granger cause lnREMPC
lnREMPC does not Granger cause IRC
PVZ does not Granger cause lnREMPC
lnREMPC does not Granger cause PVZ
SMK does not Granger cause lnREMPC
lnREMPC does not Granger cause SMK
Source: Author.
No. of Frontier & Emerging
Others (underdeveloped) Full Sample (combined)
Lags ‡ Obs F-statistic(F-prob.)
Obs
F-statistic(F-prob.)
Obs
F-statistic(F-prob.)
108
108
108
108
108
108
108
108
108
108
108
108
108
108
108
108
0.3249(0.72)
0.6783(0.51)
0.3499(0.71)
0.0039(0.99)
0.1765(0.84)
2.7280(0.07)*
0.2385(0.78)
0.1116(0.45)
0.2937(0.75)
0.8024(0.45)
2.1224(0.13)
2.0724(0.13)
0.0971(0.91)
0.4279(0.65)
2.4403(0.09)*
1.7109(0.19)
234
234
234
234
234
234
234
234
234
234
234
234
234
234
234
234
0.9785(0.38)
1.5404(0.22)
0.0353(0.97)
2.6058(0.08)*
0.2054(0.81)
0.5508(0.58)
0.7132(0.49)
0.1287(0.88)
0.3627(0.70)
1.7068(0.18)
0.1639(0.85)
1.1212(0.33)
0.5578(0.57)
1.8102(0.17)
8.1991(0.00)***
1.2681(0.28)
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
126
126
126
126
126
126
126
126
126
126
126
126
126
126
126
126
0.6587(0.52)
1.3263(0.27)
0.9363(0.40)
2.8622(0.06)*
0.0587(0.94)
3.5036(0.03)**
1.0642(0.35)
0.6269(0.54)
0.2254(0.80)
0.7919(0.46)
0.2709(0.76)
0.6033(0.55)
0.0687(0.93)
1.5406(0.22)
6.7021(0.00)***
0.2721(0.76)
‡
Note: is a model-specific optimal lag selected according to SIC, AIC and log likelihood.
REMPC here refers to migrant remittances per capita
223
Stellenbosch University http://scholar.sun.ac.za
Table A5.4: Empirical Modelling Robustness Test for Impact of Financial Liberalisation on International Remittances in SSA, 1980-2009
Pooled Ordinary Least Squares
FLBI
BKS
DCRR
EBC
ICF
Panel Fixed Effects
Panel Random Effects
1980-89
1990-99
2000-09
1980-09
1980-89
1990-99
2000-09
1980-09
1980-89
1990-99
2000-09
1980-09
3.6820
3.5006
3.1291
3.4454
2.3926
1.6491
5.4055
3.2042
2.5143
1.7241
5.2825
3.2078
[3.68]***
[4.90]***
[4.05]***
[9.38]***
[0.89]
[1.83]*
[3.35]***
[3.75]***
(1.00)
(1.90)*
(3.51)***
(3.76)***
0.4944
0.3764
0.0116
0.7655
-0.3805
0.5811
0.3903
1.0170
-0.3657
0.5760
0.3800
1.0140
[1.99]**
[1.37]
[-0.06]
[6.67]*** [-32.89]***
[2.15]**
[2.44]**
[4.50]*** (-35.75)***
(2.14)**
(2.52)**
(4.49)***
1.7848
0.6545
-0.2647
0.8952
0.2239
0.5347
0.7274
0.7547
0.2983
0.5378
0.5999
0.7572
[5.21]***
[3.10]***
[-1.24]
[8.39]***
[0.51]
[2.86]***
[2.11]*
[3.13]***
(0.67)
(2.90)***
(2.05)**
(3.15)***
0.6418
0.3831
0.6431
0.7135
0.6572
0.1426
0.6597
0.6267
0.6545
0.1548
0.6580
0.6290
[5.15]***
[2.66]***
[4.80]***
[8.58]***
[1.55]
[1.88]*
[2.99]***
[3.67]***
(1.77)*
(2.00)**
(3.08)***
(3.71)***
0.8364
0.6059
0.4618
0.8612
-0.0597
0.0819
0.1845
0.6048
-0.0341
0.1014
0.2005
0.6118
[2.35]**
[4.03]***
[3.35]***
[8.63]***
[-0.15]
[0.53]
[1.65]
[3.60]***
(-0.09)
(0.65)
(1.98)**
(3.58)***
IRC
0.0155
0.2669
0.0928
0.4512
0.3481
0.2459
1.0198
0.6111
0.3379
0.2467
0.8022
0.6086
[0.11]
[1.89]*
[0.54]
[5.66]***
[1.14]
[1.56] [5.9e04]***
PVZ
0.2089
0.8137
0.5567
0.6942
-1.0075
[1.46]
[5.94]***
[4.35]***
[9.09]***
SMK
-0.0362
0.2963
0.2516
0.6928
Overall R²
Obs
Groups
[3.34]***
(1.14)
(1.60)
(4.45)***
(3.46)***
0.1333
0.6023
0.5407
-0.6792
0.1954
0.6002
0.5448
[-2.07]**
[0.78]
[5.67]***
[2.74]***
(-1.29)
(1.03)
(5.66)***
(2.76)***
0.3141
0.2636
0.6379
1.2312
0.2926
0.2651
0.6096
1.2227
[-0.10]
[1.22]
[1.46]
[5.74]***
[0.40]
[1.32]
[3.48]***
[5.16]***
(0.38)
(1.32)
(3.48)***
(5.17)***
0.0693
390
13
0.1554
390
13
0.0681
390
13
0.2189
390
13
0.0693
390
13
0.1554
390
13
0.0681
390
13
0.2189
390
13
0.0693
390
13
0.1554
390
13
0.0681
390
13
0.2189
390
13
Source: Author‟s estimation.
Note: ***/**/* represent statistical significance at 1%/5%/10% respectively.
Constant term included for each estimation.
z-statistics in ( ); t-statistics in [ ]; with both statistics based on robust
2
standard errors. R , Hausman‟s specification tests, and B-P tests are based on estimated models involving FLBI
224
Stellenbosch University http://scholar.sun.ac.za
Table A5.5: Pairwise Correlation Coefficients of Financial Liberalisation Indicators and
Remittances in SSA, 1980-2009
Frontier & Emerging Markets
lnREMPC
DCRR
IRC
EBC
BKS
PVZ
lnREMPC
1.0000
DCRR
0.3313
1.0000
IRC
0.4095
0.7183
1.0000
EBC
0.4864
0.4909
0.3691
1.0000
BKS
0.4534
0.6472
0.6308
0.5485
1.0000
PVZ
0.5165
0.6644
0.6412
0.4420
0.6769
1.0000
ICF
0.4202
0.3280
0.5639
0.5069
0.5040
0.4928
ICF
SMK
FLBI
1.0000
SMK
0.4242
0.6776
0.5697
0.5084
0.6616
0.6686
0.4541
1.0000
FLBI
0.5546
0.8132
0.8319
0.6922
0.8344
0.8377
0.7042
0.8058
1.0000
DCRR
IRC
EBC
BKS
PVZ
ICF
SMK
FLBI
Others (underdeveloped)
lnREMPC
lnREMPC
1.0000
DCRR
0.5318
1.0000
IRC
0.2008
0.6158
1.0000
EBC
0.3761
0.5886
0.5952
1.0000
BKS
0.1142
0.6124
0.7271
0.5423
1.0000
PVZ
0.3636
0.5853
0.5811
0.8485
0.5939
1.0000
ICF
0.3303
0.4288
0.6677
0.5953
0.5723
0.5568
1.0000
SMK
0.1319
0.5765
0.4762
0.5959
0.6570
0.5401
0.4020
1.0000
FLBI
0.3833
0.7831
0.8237
0.8760
0.8059
0.8681
0.7330
0.7207
1.0000
PVZ
ICF
SMK
FLBI
Full Sample (frontier & emerging plus underdeveloped)
lnREMPC
DCRR
IRC
EBC
BKS
lnREMPC
1.0000
DCRR
0.4226
1.0000
IRC
0.2932
0.5965
1.0000
EBC
0.4178
0.5116
0.4573
1.0000
BKS
0.3188
0.6005
0.6608
0.5439
1.0000
PVZ
0.4347
0.5991
0.6193
0.6309
0.6471
1.0000
ICF
0.3939
0.3592
0.5496
0.5102
0.5102
0.4920
1.0000
SMK
0.2817
0.5516
0.5786
0.5316
0.6581
0.6203
0.4023
1.0000
FLBI
0.4679
0.7620
0.8236
0.7698
0.8255
0.8513
0.6898
0.7717
Source: Author‟s estimation
225
1.0000
Stellenbosch University http://scholar.sun.ac.za
Table A5.6: Descriptive Statistics of Financial Liberalisation Indicators and Remittances Data
Frontier & Emerging Markets
LNREMPC
DCRR
IRC
EBC
BKS
PVZ
ICF
SMK
FLBI
Mean
0.6158
1.3119
1.9619
2.1381
0.8095
1.5333
1.1095
1.1048
0.4747
Median
1.2544
1.0000
3.0000
3.0000
1.0000
2.0000
1.0000
1.0000
0.5238
Maximum
4.1897
3.0000
3.0000
3.0000
3.0000
3.0000
3.0000
3.0000
0.9643
Minimum
-4.4361
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
Std. Dev.
2.0608
0.8917
1.3725
1.0871
0.8869
1.2301
1.0317
0.8741
0.2772
Skewness
-0.8828
0.0418
-0.6326
-0.7686
0.7521
-0.0770
0.6709
0.3990
-0.2940
Kurtosis
Jarque-Bera
Probability
3.0062
1.6321
1.4831
2.0667
2.5435
1.4172
2.3320
2.4388
1.8857
27.2761
16.4347
34.1403
28.2972
21.6219
22.1298
19.6563
8.3283
13.8896
0.0000
0.0003
0.0000
0.0000
0.0000
0.0000
0.0001
0.0155
0.0010
Sum
129.3189
275.5000
412.0000
449.0000
170.0000
322.0000
233.0000
232.0000
99.6905
Sum Sq. Dev.
887.5945
166.1952
393.6952
246.9952
164.3810
316.2667
222.4810
159.6952
16.0577
210
210
210
210
210
210
210
210
210
Observations
Others (underdeveloped)
lnREMPC
DCRR
IRC
EBC
BKS
PVZ
ICF
SMK
FLBI
Mean
1.0645
1.5833
1.2889
1.8944
0.6389
1.2444
1.2278
0.6611
0.4066
Median
1.2571
1.8750
1.0000
2.0000
1.0000
1.0000
1.0000
1.0000
0.4226
Maximum
4.7948
2.5000
3.0000
3.0000
2.0000
3.0000
2.0000
2.0000
0.7500
Minimum
-4.3811
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
Std. Dev.
1.6961
0.8954
0.9542
1.1458
0.6666
1.1415
0.6415
0.5610
0.2330
Skewness
-0.4526
-0.3459
-0.0231
-0.3291
0.5605
0.1443
-0.2432
0.1038
-0.2380
Kurtosis
3.2633
1.5871
1.8926
1.5384
2.2915
1.5162
2.3231
2.2833
1.6592
Jarque-Bera
6.6659
18.5619
9.2136
19.2728
13.1909
17.1370
5.2114
4.1758
15.1825
Probability
0.0357
0.0001
0.0100
0.0001
0.0014
0.0002
0.0739
0.1239
0.0005
Sum
191.6162
285.0000
232.0000
341.0000
115.0000
224.0000
221.0000
119.0000
73.1905
Sum Sq. Dev.
514.9460
143.5000
162.9778
234.9944
79.5278
233.2444
73.6611
56.3278
9.7199
180
180
180
180
180
180
180
180
180
Observations
Full Sample (frontier & emerging plus underdeveloped markets)
lnREMPC
DCRR
IRC
EBC
BKS
PVZ
ICF
SMK
FLBI
Mean
0.8229
1.4372
1.6513
2.0256
0.7308
1.4000
1.1641
0.9000
0.4433
Median
1.2554
1.0000
2.0000
3.0000
1.0000
2.0000
1.0000
1.0000
0.4881
Maximum
4.7948
3.0000
3.0000
3.0000
3.0000
3.0000
3.0000
3.0000
0.9643
Minimum
-4.4361
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
Std. Dev.
1.9120
0.9025
1.2425
1.1198
0.7964
1.1973
0.8745
0.7774
0.2597
Skewness
-0.8038
-0.1314
-0.2109
-0.5565
0.7944
0.0389
0.4611
0.6351
-0.2059
Kurtosis
Jarque-Bera
Probability
Sum
Sum Sq. Dev.
Observations
3.3650
1.5594
1.4238
1.7528
2.8286
1.4566
2.5955
3.0992
1.8569
44.1603
34.8452
43.2616
45.4029
41.4942
38.8065
16.4811
26.3740
23.9900
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0003
0.0000
0.0000
320.9352
560.5000
644.0000
790.0000
285.0000
546.0000
454.0000
351.0000
172.8810
1422.0570
316.8359
600.5744
487.7436
246.7308
557.6000
297.4974
235.1000
26.2271
390
390
390
390
390
390
390
390
390
Source: Author‟s computation
226
Stellenbosch University http://scholar.sun.ac.za
Table A5.7: Degree of Financial Liberalisation in Contemporary SSA, 2005-2009
Ccode Country Name
Year DCRR
IRC
EBC
BKS
PVZ
ICF
SMK
FLBI
1
Burkina Faso
2005
2.50
2.00
2.00
1.00
2.00
2.00
1.00
0.595238
1
Burkina Faso
2007
2.50
2.00
2.00
1.00
2.00
2.00
1.00
0.595238
1
Burkina Faso
2009
2.50
2.00
2.00
2.00
2.00
2.00
1.00
0.642857
2
Cameroon
2005
2.50
1.00
3.00
1.00
3.00
1.00
1.00
0.595238
2
Cameroon
2007
2.50
2.00
3.00
1.00
3.00
1.00
1.00
0.642857
2
Cameroon
2009
2.50
2.00
3.00
1.00
3.00
1.00
1.00
0.642857
3
Côte d'Ivoire
2005
2.50
2.00
3.00
1.00
2.00
2.00
1.00
0.642857
3
Côte d'Ivoire
2007
2.50
2.00
3.00
1.00
2.00
2.00
1.00
0.642857
3
Côte d'Ivoire
2009
2.50
2.00
3.00
1.00
3.00
2.00
1.00
0.690476
4
Ethiopia
2005
2.00
1.00
1.00
1.00
0.00
1.00
2.00
0.380952
4
Ethiopia
2007
2.00
1.00
1.00
1.00
1.00
1.00
2.00
0.428571
4
Ethiopia
2009
2.00
1.00
2.00
1.00
2.00
1.00
2.00
0.523810
5
Ghana
2005
2.00
3.00
2.00
2.00
2.00
1.00
2.00
0.666667
5
Ghana
2007
2.00
3.00
2.00
2.00
3.00
2.00
2.00
0.761905
5
Ghana
2009
3.00
3.00
3.00
2.00
3.00
2.00
3.00
0.904762
6
Kenya
2005
1.00
3.00
3.00
1.00
2.00
3.00
2.00
0.714286
6
Kenya
2007
1.00
3.00
3.00
1.00
2.00
3.00
2.00
0.714286
6
Kenya
2009
1.50
3.00
3.00
2.00
2.00
2.00
2.00
0.738095
7
Madagascar
2005
1.75
3.00
3.00
2.00
3.00
2.00
1.00
0.750000
7
Madagascar
2007
1.75
3.00
3.00
2.00
3.00
2.00
1.00
0.750000
7
Madagascar
2009
1.75
3.00
3.00
2.00
3.00
2.00
1.00
0.750000
8
Mozambique
2005
2.50
3.00
3.00
1.00
3.00
1.00
1.00
0.6904762
8
Mozambique
2007
2.50
3.00
3.00
1.00
3.00
2.00
1.00
0.738095
8
Mozambique
2009
2.50
3.00
3.00
2.00
3.00
2.00
1.00
0.785714
9
Nigeria
2005
2.50
3.00
3.00
2.00
3.00
0.00
3.00
0.785714
9
Nigeria
2007
2.50
3.00
3.00
2.00
3.00
1.00
3.00
0.833333
9
Nigeria
2009
2.50
3.00
3.00
2.00
3.00
1.00
3.00
0.833333
10
Senegal
2005
2.50
2.00
3.00
1.00
3.00
2.00
1.00
0.690476
10
Senegal
2007
2.50
2.00
3.00
2.00
3.00
2.00
1.00
0.738095
10
Senegal
2009
2.50
2.00
3.00
2.00
3.00
2.00
1.00
0.738095
11
South Africa
2005
2.25
3.00
3.00
2.00
3.00
3.00
2.00
0.869048
11
South Africa
2007
2.25
3.00
3.00
3.00
3.00
3.00
2.00
0.916667
11
South Africa
2009
2.25
3.00
3.00
3.00
3.00
3.00
3.00
0.964286
12
Tanzania
2005
2.50
3.00
3.00
2.00
2.00
2.00
2.00
0.785714
12
Tanzania
2007
2.50
3.00
3.00
2.00
2.00
2.00
2.00
0.785714
12
Tanzania
2009
2.50
3.00
3.00
2.00
2.00
2.00
2.00
0.785714
13
Uganda
2005
1.00
3.00
3.00
3.00
3.00
2.00
1.00
0.761905
13
Uganda
2007
1.00
3.00
3.00
3.00
3.00
2.00
1.00
0.761905
13
Uganda
2009
1.00
3.00
3.00
3.00
3.00
3.00
1.00
0.809524
Source: Abiad et al. (2010) and author‟s own computation based on various Central Bank reports and IMF‟s World
Economic Financial Survey Reports
227
Stellenbosch University http://scholar.sun.ac.za
Table A5.8: Data Description, Measurement and Sources
Variable
Dependent Variable
Remittances
per
capita
Notation
Description, Measurement and Main Source(s)
REMPC
Migrant remittances defined as the sum of workers‟ remittances plus
compensation of employees received divided by total population. It was the
natural logarithmic form of REMPC denoted as lnREMPC that was used.
Source: Author‟s computation based mainly on WDI, BoPS and MRF-2011.
Explanatory Variables
+
FLBI
Financial
liberalisation index
Financial liberalisation comprising the normalised index of BKS, DCRR,
EBC, ICF, IRC, PVZ and SMK on a zero to one scale. Source: Abiad et al.
(2010) and author.
Bank supervision and
Policies on the adoption of a capital adequacy ratio based on the Basel
+
BKS
prudential regulation
standard, independence of monetary authorities, and the degree of effective
bank supervision through on-site and off-site supervision, and coverage of
supervision by monetary authorities. Source: Abiad et al. (2010) and author.
Reduction in directed
Policies on degree of restrictions on reserve requirements, minimum
credit,
aggregate
amount of credit that must be channelled to certain state priority sectors,
+/0/credit ceilings, and DCRR
mandatory credit supply to certain sectors at subsidised rates, and
reserve requirements
aggregate credit ceilings including bank-specific credit ceilings imposed by
the Central Bank. Source Abiad et al. (2010) and author.
+
Relaxation of entry EBC
Policies on the extent to which government allows new banks and foreign
barriers
banks to enter into the domestic financial market, restrictions on branching,
and the freedom banks have to engage in a wide range of activities.
Source: Abiad et al. (2010) and author.
International account
Policies on exchange rate system unification, restrictions on capital inflows,
+
and restrictions on capital outflows. Source: Abiad et al. (2010) and author.
ICF
liberalisation
Interest
rate
Degree to which deposit rates and lending rates are separately free from
+
government control set or subject to a binding ceiling or floor. Source:
IRC
deregulation
Abiad et al. (2010) and author.
Privatisation of banks
Degree of privatisation of banks coded according to the proportion of
+
privately-owned bank assets to state-owned bank assets. Source: Abiad et
PVZ
al. (2010) and author.
Policies such as introduction of auctioning of treasury bills or the
+
Securities
market SMK
establishment of a security commission towards the development of
reforms
securities markets; and the extent to which the equity market is open to
participation of foreign investors. Source: Abiad et al. (2010) and author.
Source: Author‟s compilation. Note: The a priori sign is indicated by +/- by the notation column of each variable.
228
Stellenbosch University http://scholar.sun.ac.za
CHAPTER SIX
REMITTANCES AND ECONOMIC GROWTH IN SUB-SAHARAN AFRICA,
6.0 INTRODUCTION
A dynamic panel-data model covering 36 SSA countries was estimated following the system
GMM estimation procedure to explore the direct long-run impact of remittances on economic
growth. Under the assumption that the effects of remittances on economic growth could vary
over time in response to the macroeconomic policy environment in remittance-recipient
countries, a decade-based analysis was undertaken. This chapter further explores the possible
size-effects of international migrant remittances on economic growth in SSA over the past three
decades, 1980-2009. The section that follows presents the background whilst Section 6.2
presents selected stylised facts. Section 6.3 outlines the theoretical framework and reviews the
empirical literature on remittance inflows and economic growth. In Section 6.4, the empirical
model, methodological approach, and data issues are presented. The estimated results are
presented and analysed in 6.5 before the concluding remarks and policy recommendations are
outlined in Section 6.7.
6.1 BACKGROUND
Consistent with increasing international migration of active labour from the developing world to
the advanced economies in recent years, there has been an upsurge and a continuous flow of
migrant remittances to the developing world of which sub-Saharan Africa (SSA) is a part.
Without any immediate aspirations for a narrowing income gap between the advanced world
and the developing world in this era of increasing elimination of trade barriers under the tenets
of globalisation, the South-North trend in international migration is set to continue unabated.
From a mere US$2.05 billion in 1970, global migrant remittance inflows increased to US$36.69
billion in 1980, US$68.38 billion in 1990, US$131.49 billion in 2000, US$274.54 billion in 2005
and to US$416.12 billion in 2009. During this same period, SSA received US$0.02 billion,
US$1.40 billion, US$1.88 billion, US$4.64 billion, US$9.42 billion and US$20.75 billion
respectively. As at 2009, the global official inflows of migrant remittances had emerged as the
A paper based on this chapter and titled “The Impact of International Remittances on Economic Growth in SubSaharan Africa,” was presented at ESSA bi-annual conference, September 5-7, 2011, Cape Town, South Africa.
A related published paper by the author is: “Financial Development, International Migrant Remittances and
Endogenous Growth in Ghana,” Studies in Economics and Finance, 28(1): 68-89.
229
Stellenbosch University http://scholar.sun.ac.za
second largest source of external capital (second only to foreign direct investment (FDI)). By
the end of 2009, global remittance inflows accounted for 3.39 per cent of goods exports, 35.76
per cent of FDI and as much as 326.30 per cent, which implied more than thrice the volume of
overseas development assistance (ODA). This trend can also be discerned in the case of SSA,
as officially reported remittances received by the sub-region represents 8.03 per cent of goods
exports, 71.31 per cent of FDI and 46.62 per cent of ODA as of 2009. Globally, from 36
countries in 1980, the number of countries that received migrant remittances representing at
least one per cent of their GDP increased to 58 in 1990, and 81 in the year 2000, with a further
rise to 96 countries by the end of 2009. Of this figure, 25 countries, including four from SSA98,
received remittances representing more than 10 per cent of their GDP by the end of 2009. This
could just be one of the key reasons why the implications of international remittances in
recipient countries have become increasingly important as far as economic policy research,
design and implementation are concerned in recent years.
Apart from the persistent positive growth trend, migrant remittance inflows are known to exhibit
a unique feature which clearly distinguishes it from other forms of external capital received by
developing countries. Remittances are clearly the least volatile form of external capital (see
Figure 3.1). It can be inferred from the remittance literature99 that because the flow of
remittances is largely influenced by the altruistic feelings of migrants and, for this reason
altruism underlies all other motives behind remittance inflows, these private transfers, unlike
other forms of capital, often increase in response to harsh economic conditions and crises
afflicted by shocks in migrant-home countries. Another distinguishing feature of altruistic
remittances is that they do not often involve the recipient in any financial risk or cost as they are
often directly associated with smoothing the consumption of the target recipient.
The theoretical role of remittances in enhancing long-run economic growth in migrant-home
developing countries is not straightforward. From a theoretical standpoint, it can be argued that
because remittances are used mainly for consumption smoothing and investment in land and
other non-tradable assets such as construction or redevelopment of private residential
apartments100 which do not directly generate income, remittance-recipient countries could be at
risk of suffering from the Dutch disease. In this case, the inflows of remittances can actually
98
SSA countries that received 10 per cent or more of remittances as a percentage of GDP in year 2009 are Lesotho
(26.23 per cent), Gambia (10.88 per cent), Togo (10.75 per cent) and Senegal (10.64).
99
See, for instance, Johnson and Whitelaw (1974), Stark and Lucas (1988), Giuliano and Ruiz-Arranz (2009), World
Bank (2006a,b), Acosta et al. (2008a), Barajas et al. (2009), Mundaca (2009), and Fayissa and Nsiah (2011).
100
See Chapter Four for cross-country evidence on the uses of migrant remittances.
230
Stellenbosch University http://scholar.sun.ac.za
inhibit long-run economic growth as export earnings fall due to a significant reduction in
international competitiveness of small-open and import-dependent remittance-recipient
countries, which are traditionally exporters of primary products. The World Bank (2006a),
however, downplays the Dutch disease effects associated with remittance inflows as a serious
concern on the basis that increases in remittances are gradual. As remittance inflows can
exacerbate international dependency, intensify emigration syndrome and reduce labour
productivity through moral hazards in recipient countries, some scholars including Wiest
(1984), Chami et al. (2005), and Kapur (2004) contend that, theoretically, remittances can
impair long-run growth in developing countries.
On the reverse side of the argument is the positive role of remittances in enhancing long-run
economic growth as these funds are considered as additional income to boost household
consumption, and private investment and thereby create job opportunities through increased
output expansion in capital-constrained migrant-home countries. In this context, by financing
private
consumption
and
entrepreneurial
activities,
remittances
can
help
increase
manufacturing output through increased aggregate demand and, hence, higher private
investment resulting in increased demand for labour for industrial output expansion. This
suggests that remittances carry along with them some positive multiplier effects and optimistic
externalities, so that by helping to reduce income volatility, minimise credit market failures and
smooth consumption in low-income migrant-home countries, they also help stabilise
macroeconomy which is favourable for attracting private investment. Bugamelli and Paternò
(2008) and Chami et al. (2009) find evidence for an automatic output volatility mitigating the
element of remittances in remittance-recipient countries. Another important positive effect of
remittances is its potential to ameliorate BoP problems which can improve the international
credit rating of remittance-recipient countries that can, in the long run, affect both the
magnitude and trend in government spending on the provision of public goods and the type of
infrastructure that can crowd-in private investment.
The arguments above imply that the effect of remittances on long-run growth is purely an
empirical issue in the absence of a theoretical consensus. Unfortunately, however, conclusions
from various empirical findings buttress the theoretical controversy surrounding the long-run
growth impact of remittances as findings range widely from negative, zero, positive and to
conditional effects. For instance, Stark and Lucas (1988), Chami et al. (2005), Lee (2008) and
Karagöz (2009) conclude from various studies that the impact of remittances on economic
231
Stellenbosch University http://scholar.sun.ac.za
growth is negative. IMF (2005), Baldé (2009) and Barajas et al. (2009) find a zero effect of
remittances on economic growth. Other studies including those by Stark and Lucas (1988),
Faini (2006a), Catrinescu et al. (2006), Ahortor and Adenutsi (2009), Ziesemer (2008; 2009),
and Adenutsi (2011), however, find a direct positive impact of remittances on growth. In
connection with indirect or conditional effects of remittances on economic growth, Giuliano and
Ruiz-Arranz (2009), Fajnzylber and López (2007) find a substitutability relationship between
remittances and financial development, such that, remittances promote long-run growth in
countries with poorly developed financial markets and where liquidity constraints are most
severe.
Previous studies, however, suffer from various technical deficiencies. One fundamental
limitation associated with previous studies relates to the poor definition and arbitrary
measurement of international migrant remittances (see Table A6.11 in the Appendix). In this
study, migrant remittances are measured to include only the two relevant current account
components (workers‟ remittances and compensation of employees) because migrants‟
transfers do not flow frequently, and even when they do flow, they are mostly received by the
returnee migrants themselves101. Furthermore, remittances as a ratio to GDP is not likely to
yield reliable results as the issue of factor productivity is brought into question, especially in a
typical cross-country study. Another important problem with previous studies is that they fail to
provide an insight into the possible changing impact of remittances on economic growth in
remittance-recipient countries. Some previous studies also try to model the impact of
remittances on growth through an ad hoc indirect mechanism. This study recognises the fact
that the channels through which remittances can affect growth could be many102 and cannot be
adequately addressed in one particular empirical study; hence, the need to rather concentrate
on how remittances can directly affect growth either contemporaneously or asynchronously as
the macroeconomic environment evolves in response to the implementation of financial
liberalisation policies. These problems are addressed in this study using 36 SSA countries.
101
For further clarifications, see Chapter Two of this dissertation which has been particularly devoted to definition
and measurement of key concepts, including remittances.
102
For example, remittances can indirectly affect growth through human capital development in terms of improved
access to healthcare or higher skills acquisition which are essential for higher labour productivity, financial
development by augmenting domestic savings to improve credit extension, increased aggregate demand through
consumption of manufactured goods, increased government expenditure on provision of critical infrastructure as
government revenue increases from import tariffs and consumption tax such as the Value Added Tax (VAT).
232
Stellenbosch University http://scholar.sun.ac.za
Consequently, the principal research questions addressed in this chapter are:
i.
Do international remittances have a consistent and direct impact on long-run growth
since the inception of financial liberalisation programmes in SSA? Overall, is the impact
of remittances on economic growth contemporaneous or asynchronous?
ii.
Are there any direct impact variations of remittances on long-run growth based on the
rate of economic growth in remittance-recipient SSA countries? If yes, which category
of SSA countries is more likely to benefit from international remittances in this regard?
Accordingly, with reference to SSA, the related specific objectives of this chapter include:
i.
To estimate the direct impact of international migrant remittance inflows on economic
growth in the long run.
ii.
To examine if the direct impact of international migrant remittance inflows on economic
growth is contemporaneous or asynchronous.
iii.
To determine if the impact of migrant remittance inflows on economic growth evolves
over time in response to macroeconomic environment since the implementation of
financial liberalisation programme three decades ago.
iv.
To verify the existence and impact of economic growth size-effects of international
migrant remittances in remittance-recipient countries.
The achievement of the above-stated research objectives is very important in a number of
ways. First and foremost, in terms of scope (both time span and country inclusiveness), it
represents the most comprehensive empirical study on the remittance-growth relationship in
SSA. Besides, the decade-based impact analysis alongside the overall period analysis makes
this study not only a novelty but also the most detailed in examining the effects of remittances
on economic growth. Thus, the findings of this study, among other things, reveal the timevarying effects of remittances on economic growth in SSA; and this plays a crucial role in the
quest for the appropriate policy formulation for contemporary „labour-exporting‟ SSA countries.
Furthermore, an insight into the economic growth size-effects of remittances is crucial in
designing specific relevant pro-growth policies for SSA countries with similar growth features.
233
Stellenbosch University http://scholar.sun.ac.za
6.2 SELECTED STYLISED FACTS
One of the underlying reasons why the impact of migrant remittances on economic growth
should not be expected to be consistent across the developing world is that these developing
regions differ widely regarding net remittance inflows, but given the problems related to data,
the implications of remittances are often analysed based on what has been received rather
than on the net inflows. In Figure 6.1, it is shown that, notwithstanding the fact that developing
economies are net recipients of remittances (see Figure A6.1), these developing economies
differ widely in terms of the proportion of remittances paid relative to remittances received (see
Figure 6.1). For example, between 1980 and 2009, the amount of remittances paid by SSA to
the rest of the world constituted as much as 53.36 per cent of the total remittances received by
the sub-region compared to only 4.30 per cent and 8.16 per cent in the case of South Asia
(SAS) and LAC. Over the same period, the Middle East and North Africa (MNA) paid 15.83 per
cent of the remittances received to the rest of the world, East Asia and the Pacific (EAP) paid
17.21 per cent, whilst Europe and Central Asia paid as much as 47.44 per cent of their
remittances. Thus, ideally, if country-based data were readily available, it would have been
more relevant and appropriate to use net remittance inflows when analysing the effects of
remittances on economic growth. Indeed, in a comparative study between LAC and SSA
countries, Ahortor and Adenutsi (2009) show that although remittances have positive effects on
economic growth in both regions, the impact is more robust in LAC.
Figure 6.1: Total Inflows and Outflows of Remittances in Developing Economies, 1980-2009
700
Remittances (US$ billions)
600
589.92
587.28
552.71
541.79
500
528.91
486.19
439.02
400
369.53
321.12
300
200
168.79
152.33
100
156.14
101.09
83.31
69.49
48.13
72.83
23.79
0
EAP
ECA
LAC
Paid
Received
Source: Author‟s estimation based on WDI (April 2011)
234
MNA
Net Received
SAS
SSA
Stellenbosch University http://scholar.sun.ac.za
The cyclicality of remittance inflows in SSA is shown in Figure 4.1, with the evidence pointing to
the fact that remittances were pro-cyclical in the 1980s and the 2000s but countercyclical in the
1990s. The cyclicality of remittance inflows shows the likely use to which remittances might
have been put and, hence, the potential changing impact of migrant remittances on growth in
SSA over the past three decades. For example, during the periods when migrant remittances
were pro-cyclical, it is most likely that the self-interest motive might have dominated the
motives behind remittance inflows, implying a higher likelihood of using remittances to finance
income-generating projects, which have a higher positive multiplier effect on an economy than
when remittances were countercyclical with higher likelihood of being spent on consumer
goods. Chami et al. (2005) and Barajas et al. (2009) show that when remittances are
countercyclical, they are also counterproductive and, hence, cannot have a direct positive
impact on growth. In Table 3.1, it is shown that, actually, SSA as a sub-region recorded its
lowest real per capita income and worst general macroeconomic performance in the 1990s
over the past three decades.
Figure 6.2:
Correlation between Remittances and Key Macroeconomic Indicators in SSA, 1980-2009
1.000
REMGDP
0.914
0.841
Private Capital
0.758
0.572
Portfolio Equity
-0.900
-0.748
0.453
Population growth
0.899
FDI
Investment
0.725
0.018
0.282
0.524
RGDPPC,PPP
0.763
0.381
RGDPPC
-1.000
-0.500
0.647
0.000
0.500
Cor_REMGDP
Source: Author‟s based on WDI (April 2011).
1.000
1.500
Cor_REMPC
Note: Cor_ denotes correlation, RGDPPC denotes real GDP per
capita. PPP stands for purchasing power parity
235
Stellenbosch University http://scholar.sun.ac.za
The unique cyclicality of migrant remittance inflows brings to the fore the likely endogeneity103
problems that are associated with estimating growth models involving remittances.
Unfortunately, many previous studies (see Table A6.11) failed to account for this endogeneity
problem whilst modelling remittances on economic growth and this has obvious negative
implications for the results obtained. Remittances received between 1980 and 2009 had a
significant positive correlation with other non-debt capital inflows and core macroeconomic
performance indicators such as investment rate and the level of real per capita income as
shown in Figure 6.2. Here, it is also shown that it is migrant remittances per capita, which is the
best available proxy for remittances per migrant that reasonably correlate with investment and
general economic performance measured by real GDP per capita. This suggests that the
conclusions from studies that used remittances as a ratio of GDP rather than remittances per
migrant could be misleading, especially because in terms of remittances as percentage of GDP
or exports, SSA emerges as one of the leading recipients in the world today, although the subregion is, in fact, the least recipient in terms of actual volume (Table A6.1) and per migrant (see
Figure 3.3). Table A6.1 also shows that, as a percentage of official development assistance
(ODA), SSA was the least recipient as at 2009, implying that the sub-region, compared to other
developing economies, remains the sole dependant on foreign aid.
6.3 THEORETICAL FRAMEWORK AND LITERATURE REVIEW
The theoretical framework and the empirical literature of the effects of international migrant
remittances on economic growth are presented in this section.
6.3.1 Theoretical Framework
In line with the theoretical contributions of Rapoport and Docquier (2006) and Barajas et al.
(2009), this study appeals to the endogenous growth model to evaluate the impact of
international migrant remittances on economic growth in SSA. The inspiration for operating
within this theoretical framework is based on the emphasis of the endogenous growth model on
the role of knowledge which is measurable by the stock of the quality of human capital
available rather than mere population size often determined by the quantity of human capital
available. The application of the endogenous growth model to verify the potential effects of
remittances on long-run growth is logical in view of the fact that, in per capita terms, it is the
103
Given the altruistic dominance behind remittance inflows, countries with unfavourable economic conditions and
negative external shocks often receive more remittances than those with sound governance and higher growth
prospects. See Clarke and Wallsten (2004), Kapur (2004) and Yang (2005) for evidence.
236
Stellenbosch University http://scholar.sun.ac.za
quality rather than quantity of human capital that is more relevant since the country with the
largest number of migrants is not necessarily the leading recipient of remittances per capita or
remittances per migrant.
Furthermore, the endogenous growth framework has been adopted because it is the best
known model that adequately addresses the shortcomings of the famous neoclassical growth
model proposed by Solow (1956). Prior to the famous contributions of Solow (1956; 1957), it
was the classical aggregate production function proposed by Cobb and Douglas (1928) that
dominated the analytical framework of models on economic growth. Although Solow‟s
neoclassical growth model made a significant contribution to empirical analysis of economic
growth and development through its emphasis on the direct link between investment in tangible
assets and growth, it has some limitations. First, it has been argued that the steady-state
growth in per capita income envisioned by Solow (1956) will only remain an illusion without
exogenous technical progress as capital accumulation is subject to diminishing returns (Romer,
1986; Barro, 1990; Grossman and Helpman, 1991; Aghion and Howitt, 1992). Secondly, the
neoclassical growth model as proposed by Solow (1956) failed to explain what it means by
technical progress and how it will be achieved and sustained in the long run (Romer, 1986;
Barro, 1990). Another widely cited criticism of the neoclassical growth model is the narrow
definition of capital accumulation to include investment in tangible assets only without any
value placed on intangible assets (Romer, 1986; 1990; Lucas 1988; Barro, 1990; Grossman
and Helpman, 1991).
In addressing the shortcomings in the neoclassical growth model, the endogenous growth
school makes room for technological knowledge that emphasises the incentives driving
innovation, invention and creativity as the main pillars around which sustainable economic
growth evolves. For instance, Romer (1986; 1990), Lucas (1988), Barro (1990), and Rebelo
(1991) argue for economic growth models wherein the rate of growth of an economy is
endogenously determined because it is related to the elements of total factor productivity. The
endogenous growth models predict self-sustaining growth with exogenous technical progress in
an economy in the long run. This growth rate may occur because, in the long run, tastes and
preferences, state of technology104, income distribution, governance and institutional
104
Technological advancement often endogenously generates positive externalities such that the production function
exhibits increasing returns to scale due to the presence of spill-over effects associated with knowledge generation
and/or education (Romer, 1986; Lucas, 1988; Stokey, 1991).
237
Stellenbosch University http://scholar.sun.ac.za
arrangements are not likely to be static in the real world. Therefore, as long as international
migrant remittances received in developing countries can affect any of these factors, (for
instance, income distribution, technological advancement, and preferences), in the long run,
they can have not only level-effects but also growth-size effects on the economies of migranthome countries.
The potential impact of international remittance inflows on long-run economic growth can be
determined within the context of the simple generic endogenous AK-technology model
proposed by Rebelo (1991) in which aggregate output is a constant-returns function of the
aggregate capital stock. The aggregate capital stock is physical capital stock plus human
capital stock including current state of knowledge, so that the aggregate output (Y ) of a closed
economy is assumed to be dependent on the stock of capital (K) which includes physical
assets accumulated ( K k ) and human capital as well as the stock of knowledge ( K h ).
Y f ( K ) f ( Kk , Kh )
where Y
N
N
Y , K K
i 1
(6.1)
i
i 1
i
, and N is the number of firms in the economy under consideration.
Following Cobb and Douglas (1928), the mathematical expression of the AK production
function in Equation (6.1) takes the form:
Y A0 K A0 Kk Kh
(6.2)
where A0 is the total factor productivity (TFP) which is a function of the stock of knowledge105; K
is the investment in both physical assets and human capital; whilst K k and K h are components
of K representing the investment in physical assets and human capital accumulation
respectively.
Each component of K (i.e. physical and human capital stock including knowledge) is
reproducible with identical technologies (Pagano, 1993). This is why the simple endogenous
AK model does not assign any productive role to labour (L) and other non-reproducible factor
105
According to the learning-by-investing hypothesis proposed by Arrow (1962), it is assumed that at any time t,
technology is endogenously generated by
At M t (0 1) where
Mt, the stock of experience at time t is a
function of previous investment undertaken by various firms in the productive sector of an economy in which it is
assumed that, for convenience, the rate of depreciation of physical assets is zero.
238
Stellenbosch University http://scholar.sun.ac.za
inputs106 because what is assumed relevant to the production process is the quality of adjusted
labour, which is the human capital that is accumulated as each new successive generation is
assumed to be more knowledgeable than the one before (Romer, 1986; Lucas, 1988; Rebelo,
1991; Pagano, 1993).
Equation (6.2) can be specified in its intensive form when both sides are divided by the labour
force L under the assumption of constant returns to scale, hence Equation (6.3)
y Ak
(6.3)
where A is the average or marginal productivity of capital; y Y L representing output per
labour; and k K L
Kk
L
Kh L measuring capital-labour ratio.
A typical closed economy has both a demand side (consumption) and a supply side
(production), since the goods produced in this economy are either consumed or saved which
then augments the existing stock of capital. Also, as in the Keynesian national income
determination, capital market equilibrium condition requires that gross savings (S) equates
gross investment ( I ) . However, it is known from real world experience that a proportion of the
savings given as a leakage of 1 is incurred in the process of converting savings into
investment (Pagano, 1993). Accordingly, it can be shown, as in Equation (6.4) that
St equals
I t at any particular time t:
St (1 )St It St St St It St It
(6.4)
Thus, in excess of consumption (C), the evolution of capital stock due to production efficiency
is given as:
It Kt 1 (1 d ) Kt
where d
(6.5)
is the rate of depreciation of physical capital, which if equated to zero for
convenience sake, It Kt 1 Kt .
106
In
fact,
even
if
the
production
Y f ( K , L) f ( Kk , Kh , L) so
that Y
function
were
1
AKk Kh L
assumed
to
take
which implies that
L
, where L denotes labour input,
L 1 , hence Y f ( K , L) A0 KL)
0
239
expanded
106
Rebelo (1991) endogenous AK production function assumes constant returns to scale,
1
an
is actually
form
of
because the
1 as 1
Y f ( K ) A0 K .
Stellenbosch University http://scholar.sun.ac.za
The Demand Side of the Product Market - the Household Sector
According to Bond and Wang (1996), and Barro and Sala-i-Martin (1997; 2003), consumption is
analysed with households maximising an inter-temporal utility function (U) that has a constant
elasticity of substitution and, hence, takes the form of Equation (6.6):
U t (ct ) ct1 1
(6.6)
where ct is the per capita consumption, whilst 0 1 is the factor to consider in making
decisions on inter-temporal substitution in consumption at time t.
Sena and Fontenele (2004) specify the dynamic optimisation problem of a representative agent
as:
Maximise U t (ct )
(c
0
1
t
) /(1 ) e rt dt
(6.7)
subject to the budget constraint kt Akt ct
where k is the change in the capital-output ratio; r is the discount rate that connotes the
constant rate of time preference considered by the household sector in making decisions
between present consumption and future consumption of wealth which includes remittances
received. represents a relative risk aversion coefficient such that a rise implies a faster
proportionate rate of decline in the utility derived from consuming remittances at the present
time rather than saving them.
To set up the stage for solving Equation (6.7), the present-value Hamiltonian maximisation
problem required is given as:
H C {ct(1 ) /(1 )} ut ( Akt ct )
(6.8)
which, according to Sena and Fontenele (2004: 6), yields Equations (6.9 - 6.11) as, “the three
maximum principle conditions”, obtained by taking the first-order conditions of Equation (6.8) as
follows:
H C
ct
0
ut ut r H
(6.9)
C
(6.10)
Kt
kt Akt ct
where
(6.11)
t denotes the present-value shadow price of per capita household income.
240
Stellenbosch University http://scholar.sun.ac.za
Subsequent to Barro and Sala-i-Martin (1997), Sena and Fontenele (2004) derived the optimal
steady-state per capita consumption growth rate (Equation (6.12)) by taking the logarithm and
the time derivative of Equation (6.9) and using the result of Equation (6.10).
gct ct / ct ( A r ) /
(6.12)
where gct 0 as long as A r since 0 1 .
The Supply Side of the Product Market - the Business Sector
It is in the business sector of an economy that investment actually takes place through the
inflow of non-altruistic remittances (here, the remittances received in excess of present
consumption). Taking the logarithm and the time derivative of both sides of the third
Hamiltonian optimality condition represented by Equation (6.11), Barro and Sala-i-Martin
(1997), and Sena and Fontenele (2004) show that, in the long run, per capita capital growth
equals the long-run per capita consumption; and this is positive and constant along the
discount rate r as specified in Equation (6.13).
gkt kt / kt ct / ct
(6.13)
This implies that in the long run, all positive rates of growth are equal and constant over time in
a closed economy where leakages are equal to injections; given that firms are rational and,
hence, only spend on good quality projects even as consumers spend on locally produced
goods and services.
The Equilibrium – the „Complete Economy‟ Balanced Growth Rate
For a steady-state growth rate of an autarky economy that is in a long-run equilibrium, the
demand side (consumption) and the supply side (production) must equal each other. This
requires that:
g yt gkt gct g yt gkt or g yt gct since gkt gct 0
(6.14)
when the market of each product in this economy is cleared. Here, g yt represents the long-run
growth rate in per capita output of the economy which is directly commensurate to the rate of
growth in „quality‟ investment by firms or the rate of growth in household consumption of locally
produced goods and services in a two-sector closed economy setting.
From Equation (6.3), it can be shown that, taking logarithm of both sides,
g yt yt / yt kt / kt
(6.15)
241
Stellenbosch University http://scholar.sun.ac.za
hence, the optimal steady-state growth rates107 of this economy are:
g yt yt / yt kt / kt ct / ct ( A r ) /
(6.16)
Likewise, from Equations (6.15) and (6.16), the long-run steady-state growth rate of this
economy is:
g yt A YItt A st
where s S
Y
(6.17)
is the gross saving rate also the long-run marginal or average propensity to
save, which is inversely related to the discount rate r ; and where it is assumed that d 0 .
Equation (6.16) is, thus, essentially the same as Equation (6.17). From Equation (6.17) it is
apparent that remittances received can influence the long-run economic growth rate through an
increase in the saving rate s (or the rate of investment), the proportion of remittances saved or
invested , and the social marginal productivity of capital, A . Besides, as shown in Equation
(6.16), altruistic remittances consumed by the household sector can equally stimulate
economic growth in the long run through increased aggregate demand for locally produced
goods. It is also important to re-emphasise the fact that the extent to which non-altruistic
remittances can contribute to long-run growth is dependent upon the quality of the investment
to which these funds are put by the business sector.
The underlying assumptions for the application of this model in the context of this study are:
i.
International migrant remittances are received as „additional income‟ by households and
firms in developing countries.
ii.
Remittances received are either to fulfil altruistic objective or self-interest investment
motive, and altruistic remittances in excess of settlement of „contractual‟ family debt108,
are spent on consumables such as food, shelter, clothing, transportation, potable water,
electricity, telecommunications and other basic necessities of modern life. Self-interest
investment-driven remittances are spent on income-generating goods in the form of
savings and mainly investment which include stocks, bonds, fixed deposit, treasury bills,
107
This is closely related to the well-known Keynesian macroeconomic accounting approach to national income
determination in which output, income and expenditure equal one another in a closed-economy setting. The
implication of Equation (6.16) is that irrespective of the use to which remittances are put they can stimulate long-run
growth in a migrant-home country that has no foreign sector and government.
108
Especially for migrants under implicit social contract with members of their family or “sponsors” who either
financed their education / training or their travelling abroad, or both.
242
Stellenbosch University http://scholar.sun.ac.za
working capital, education, vocational training, and healthcare. Therefore, the
household sector is assumed to be recipients of altruistic remittances whereas the
business sector is assumed to be the recipients of non-altruistic remittances. This is
notwithstanding the fact that some altruistic remittances can be saved or invested for
future consumption, at least, in a bid to realise the precautionary motive for holding
money.
iii.
Altruistic migrant remittances are received essentially for the sake of smoothing the
consumption pattern of the recipient household, whilst non-altruistic remittances are, to
all intents and purposes, meant for financial gains through saving and investment.
iv.
The opportunity cost associated with using international migrant remittances to finance
private sector led investment projects is not higher than the real cost of borrowing from
a financial institution in the migrant-home country where capital is relatively scarce.
v.
Any saved remittances end up in the formal financial system where financial
intermediaries in the migrant-home countries are efficient enough to swiftly convert
short-term liabilities to medium and long-term financial assets, such that there is no time
lag for this conversion. Thus, in other words, saved remittances behave just like other
forms of non-altruistic remittances received outside the formal financial system because
international migrants or their representatives at home decide to invest directly.
vi.
The necessary condition for a country to receive remittances from abroad is to move out
of autarky to an open economy with the government playing a key role in formulating
policies and regulations on international migration based on the notion that migrants
with the appropriate travel documents are more likely to get decent jobs and remit
home.
One important implication of the fundamental assumptions of the theoretical framework is the
necessity to make crucial modifications to the simple AK endogenous growth model proposed
by Rebelo (1991). Thus, in this study, following the theoretical contributions of Barro (1990),
and Grossman and Helpman (1991), government spending and international trade are
introduced to augment the model specified in Equation (6.17). Another implication of these
assumptions is that because an open-economy case with the role of a government is assumed
243
Stellenbosch University http://scholar.sun.ac.za
in this study, the potential effect of remittances on economic growth rate can no longer be seen
as definitely positive, but as dependent upon whether or not altruistic remittances are spent on
locally produced consumables rather than on imported consumables; and, whether or not
government expenditure finances critical public infrastructure that crowds-in the private sector
rather than crowd-out the private sector. All in all, from a theoretical viewpoint, the impact of
remittances on economic growth can be dependent upon the crucial fundamental features of
the migrant-home country as can be reflected in the rate of economic growth. In other words,
theoretically, the impact of remittance inflows on economic growth can either be negative, zero,
or positive in a typical migrant-home developing country possibly depending on the unique
characteristics of the recipient economy.
The ultimate impact of remittances on economic growth is not straightforward because in as
much as remittances have the potential of spurring long-run growth, they can equally exert a
negative impact on productivity through the problem of moral hazards in developing countries
(Stark and Levhari, 1982; Lipton, 1980; Chami et al. 2005; 2009). In a contribution to the
formulation of remittances-growth theory, Barajas et al. (2009) posit that the effects of
remittances on economic growth are transmitted through three main channels – capital
accumulation, labour force growth and TFP growth - none of which has a one-directional
potential impact on long-run growth in remittance-recipient countries. The contributions of
Barajas et al. (2009) are not novel as they are essentially parallel to an earlier contribution by
Rapoport and Docquier (2006) who identify two broad channels109 through which remittances
can affect economic growth in remittance-recipient countries. The theoretical contributions of
both Rapoport and Docquier (2006) and Barajas et al. (2009) are consistent and juxtaposed to
the tenets of the endogenous growth model.
On the potential positive effects of migrant remittances on long-run growth through capital
accumulation, Barajas et al. (2009) do not disagree with earlier propositions by Stahl and
Arnold (1986), Massey et al. (1998), and de Haas (2003) that remittances can contribute to
growth by reducing macroeconomic volatility, liquidity and productive investment constraints;
raising real income levels, and minimising balance of payments problems in developing
countries. Besides, remittance inflows help to narrow the trade gap, control external debt,
109
These are the „liquidity constraint 1: entrepreneurship‟ and liquidity constraint 2: human capital‟. Each of these
theories has various cases under it that shows how remittances can potentially affect economic growth in the long
run (see Rapoport and Docquier, 2006 for further details). The recent work by Barajas et al. (2009) is, however, an
adequate representation of Rapoport and Docquier (2006).
244
Stellenbosch University http://scholar.sun.ac.za
facilitate debt servicing, increase credit worthiness, and increase the supply of foreign
exchange in remittance-recipient countries (Adenutsi, 2011). Remittances may also reduce the
cost of borrowing in capital constrained remittance-recipient countries as demand for credit
from the private sector reduces and has a high potential for reducing profit margin and the
default risk component of quoted lending rates by banks and other financial intermediaries. All
these can enhance long-run growth in remittance-recipient countries.
However, the potential positive impact of remittances on economic growth in recipient countries
can only manifest itself if remittances are less altruistic hence saved or spent mainly on „quality‟
investment goods rather than on imported consumables. One important fact is that many
developing countries are import dependent and, therefore, spending remittances on consumer
goods is likely to result in a leakage rather than an injection of funds into the income flow of
remittance-recipient countries. Indeed, it is conceivable that when remittances become
permanent income transfers, they are very likely to be spent mainly on the consumption of
leisure and imported consumer goods110 rather than being spent on investment goods, though
this is an unlikely event in the long run if migrants behave rationally.111 However, if altruistic
remittances are spent on locally made consumables, they can engender long-run growth
through a higher demand for manufactured goods leading to an increased demand for factor
inputs by local industries as firms expand production to meet the increased domestic demand
and even target the export market. This can also culminate in higher wages and deposit
interest rates, with the potential of reducing further migration and boosting private sector
savings and investment, which can ultimately entice migrants and recipients of remittances to
save or to invest in migrant-home countries thereby increasing the positive multiplier effects of
remittances. Barajas et al. (2009) again argue that the remittances-growth channel through
capital
accumulation
could
suffer
undesirable
consequences
by
destabilising
the
macroeconomy of a financially developed remittance-receiving economy, where remittances
act as substitutes rather than complements of credit allocation by the financial sector.
110
Even when remittances are spent on imported goods and leisure, the receipt of remittances can expand the tax
net in migrant-home countries, enabling governments of these low-income countries to mobilise more resources for
redistribution especially through provision of critical social infrastructure that can crowd-in the private sector to boost
growth.
111
A rational migrant will not continue to remit without expecting a positive impact of remittances on the lives of the
recipients since the continuous inflow of remittances, whether, altruistic or self-interest, is based on mutual trust that
guarantees the interest and satisfaction of both parties through strong social ties. No rational migrant who derives
utility from remitting will continue to remit large funds home when social ties between him/her and the family back
home become weak or the trust and confidence that the remitter has in the target recipient have diminished.
245
Stellenbosch University http://scholar.sun.ac.za
As a country cannot receive international migrant remittances without losing the services of a
proportion of its labour force to the outside world, it is argued that international migration drains
developing countries of highly trained and skilled labour and capital by crowding-out the
domestic production of tradable goods in the brain-drained underdeveloped economy (Lipton,
1980; Taylor, 1984; Rubenstein, 1992; Ahlburg; 1991). Thus, migrant remittances, apart from
deepening the foreign dependency mentality of developing countries112, can also cause higher
inequality among households and macroeconomic instability in the form of inflation through
excess demand for consumables and relative deficit in the domestic production capacity of
developing countries (Adenutsi, 2011). In a contribution, Barajas et al. (2009) re-emphasise the
remittances-growth nexus through labour force participation in economic activities in
remittance-recipient countries where remittances may act as substitutes for wages earned from
being engaged in economic activities, through the moral hazard problem. Chami et al. (2005;
2008) assert that there is a high possibility of this moral hazards problem occurring because
migrants remit under asymmetric information having been separated from their family by
distance, and with limited chances of monitoring and enforcing compliance of how remittances
should be used. Therefore, recipients of remittances can divert these funds to spending on
leisure and unproductive activities, thereby reducing labour participation in remittance-recipient
countries in the long run. Nevertheless, since social ties and trust underlie the motivation of a
migrant to remit, a rational remitting migrant is not likely to continue remitting if there is lack of
reliable information on the uses of remittances because migrant remittances are often sent for
specific known purposes.
According to Rapoport and Docquier (2006) and Barajas et al. (2009:7), the effects of
remittances on growth through TFP in a remittance-recipient economy are dependent upon a
variety of factors as this channel is transmitted through the efficiency of domestic investment as
well as through the effects on the size of the domestic productive sectors that generate a set of
„dynamic productive externalities‟. If remittances are invested rather than spent on consumer
goods, then these funds may affect the efficiency of investment in recipient countries based on
the informational advantage or disadvantage of the migrant or the person acting on his/her
behalf in this capacity (Barajas et al., 2009). Therefore, if the migrant or his/her investor agent
does not have more adequate financial literacy and relevant investment information than the
domestic financial intermediaries, then the altruistic remittances rather than a capital inflow
112
In countries where remittances form a significant proportion of national output, governments may become
complacent and less aggressive in implementing pro-growth economic policies by over-relying on remittances.
246
Stellenbosch University http://scholar.sun.ac.za
intermediated by domestic financial intermediaries will reduce the efficiency of domestic
investment (Barajas et al. 2009). Again, remittances can reduce capital productivity when the
agents of migrants or the direct beneficiaries of remittances decide to invest in riskier projects
than they would have normally done if these funds were not considered as risk-free transfers.
Furthermore, remittances have the potential of affecting the formal financial system of the
recipient economies in financial resource allocation since remittance inflows most often
increase the volume of funds that flow through the formal financial system (Aggarwal et al.
2006). This can promote financial market development and, hence, higher economic growth
through increased economies of scale in financial intermediation (Barajas et al. 2009). But
substantial inflows of international remittances can also result in equilibrium real exchange rate
appreciation, a recipe for Dutch disease infestation which implies less international
competitiveness of export commodities of the remittance-recipient country which can impact
negatively on long-run growth in the migrant-home developing economies.
From the theoretical viewpoint, therefore, it is apparent that the impact of remittances on the
long-run economic growth is indeterminate and likely to depend on the amount of non-altruistic
remittances received, the unique features and the macroeconomic policy environment of a
migrant-home developing country. These distinctive features of the migrant-home country can
be very many113 but, often, they collectively reflect in the long-run growth rate of an economy
and level of economic development. For instance, the underdevelopment of financial markets,
low private sector savings and investment, high income inequality, protracted fiscal deficit due
to excessive government spending, and import dependency, which result in economic
stagnation, are common characteristics of developing countries (Todaro and Smith, 2002;
Thirlwall, 2011).
Accordingly, the focus of this chapter is to examine the direct impact of international migrant
remittances on long-run economic growth, and to investigate the possible presence and impact
of the economic growth rate size-effect of international remittance inflows in SSA.
113
For example, the degree of financial development, income inequality, dependency on imports, quality of labour
force, and government policy can influence the pace of economic growth.
247
Stellenbosch University http://scholar.sun.ac.za
6.3.2 Empirical Literature on International Remittance Inflows and Economic Growth
Recent years have seen a growing interest in investigating the effects of international
remittance inflows on the economic growth of developing countries. Between the year 2003
and 2011 alone, this author has come across as many as 30 empirical studies (see Table
A6.11 in the Appendix) on the relationship between remittances and economic growth in
migrant-home countries. The majority of these empirical studies (24 out of 30) involved crosscountry/panel studies whilst the remaining six were devoted to country-specific studies. Of the
24 cross-country/panel studies, 12 of which only four114 (Baldé, 2009; Kagochi et al., 2010;
Lartey, 2010; Singh et al., 2010) were strictly devoted to SSA, were regional-based. Thus,
cross-country/panel studies on the impact of remittances on the economic growth in SSA are
not only relatively scarce, but also relatively recent.
Apart from Ziesemer (2008; 2009) who went as far back as 1960 and covered the period, 19602003, hence 44 years, cross-country/panel studies that covered the most time period normally
ranged between 1970 and early 2000s (see Table A6.11). Most of these studies115 were
generally on developing countries and not restricted to regional studies. With reference to
regional cross-country/panel studies on SSA countries, the study period covered ranged
between 17 years (as in Kagochi et al., 2010) and 25 years (as in Baldé, 2009). So far, only the
study by Baldé (2009) covered the 1980s (specifically, 1980-2004) as all other related studies
exclusive to SSA as a sub-region covered 1990/91-2007/08. Of the empirical literature
reviewed, Lartey (2010), and Singh et al. (2010) who analysed 36 SSA countries in their
various studies, compared to six by Kagochi et al. (2010) and 29 by Baldé (2009), represent
the most inclusive cross-country/panel studies on SSA as a sub-region in terms of the number
of sampled countries. Therefore, with 36 SSA countries over the period, 1980-2009, the
empirical findings from this study on the effects of remittances on economic growth are the
most comprehensive in terms of coverage (both time and country) on SSA as a sub-region.
From Table A6.11, it is observed that 23 out of the 30 empirical studies reviewed used
international remittances as a ratio of nominal GDP (REMGDP), whilst in three other studies,
the logarithm of gross international remittances was used. In the remaining four studies (those
by Fayissa and Nsiah, 2008; 2010; 2011; Siddique et al., 2010) remittances per capita
114
This is exclusive of the studies by Fayissa and Nsiah (2008) on 37 African countries, and Ahortor and Adenutsi
(2009) in a comparative study involving 31 small-open developing countries from LAC (16) and SSA (15).
115
This is exclusive of Garcia-Fuentes and Kennedy (2009) on 14 LAC countries from 1975 to 2000; and Mundaca
(2009) on 25 LAC countries between 1970 and 2002.
248
Stellenbosch University http://scholar.sun.ac.za
(REMPC) were used, but, unlike this study, none of these authors explained the motivation
behind the choice of REMPC over REMGDP. The justification for using remittances per capita
rather than remittances as a ratio of nominal GDP is provided in Chapter Three.
With regard to the composition of what constitutes international remittances, 13 of the studies
summed up the three commonly used components – workers‟ remittances, compensation of
employees, and migrant transfers; while in eight other studies, remittances were defined as
comprising only workers‟ remittances and compensation of employees. Also, seven of the 30
empirical studies summarised in Table A6.11 defined international remittances as gross private
capital transfers comparable to the broadest definition of international remittances (see Chapter
Two); while in the four remaining studies, workers‟ remittances were used to represent
international migrant remittances. With the exception of Barajas et al. (2009), practically no
previous study gave academically justifiable explanations for the choice or inclusion of the
specific components of the remittances data, as decisions on what to include or to exclude from
the determination of migrant remittances were based merely on easy access to data on
remittances and the manner of reporting by the source institution concerned. For example, the
World Bank reports only workers‟ remittances and compensation of employees as components
of remittances in its WDI, but the IMF also reports the third component (migrant transfers) in its
BoPS. With the exception of Adenutsi (2011), all previous authors who measured remittances
as the sum of workers‟ remittances and compensation of employees were those who
coincidentally used the remittance data from the WDI and not the BoPS. In this study, the
narrow definition of international remittances as the sum of workers‟ remittances and
compensation of employees is used based on the explanations provided in Chapter Two.
The conclusions from various empirical studies suggest that the direct impact of remittances on
long-run economic growth is mixed even though the majority of the results favour a positive
impact (see Table A6.11). The obvious controversy in the literature on remittances has been
whether or not remittances have a direct or an indirect positive impact on long-run economic
growth. In various empirical studies, Amuedo-Dorantes and Pozo (2004), Chami et al. (2005),
Fajnzylber and López (2007), Jongwanich (2007), Barajas et al. (2009), and Singh et al.
(2010), it was found out that remittances do not have a direct positive growth-impact, or as in
some cases, directly retard economic growth. However, the conclusions from many other
studies including those by León-Ledesma and Piracha (2004), Glytsos (2005), Lucas (2005),
World Bank (2006b), Calderón et al. (2008), Ahortor and Adenutsi (2009), Catrinescu et al.
249
Stellenbosch University http://scholar.sun.ac.za
(2009), Jayaraman et al. (2009), Mundaca (2009), Rao and Hassan (2009) and Ziesemer
(2009) show that remittances do have a direct positive impact on long-run growth, except that
the direct impact is marginal in most cases. Other studies that found a direct positive impact of
remittances on economic growth include those by Lartey (2010), Adenutsi (2011), Ahmed et al.
(2011), and Fayissa and Nsiah (2011).
One thing that appears quite clear from the studies that fail to find a direct (positive) impact of
remittances on economic growth is that even where remittances failed to have direct (positive)
impact on growth, they do have a significant positive impact on most of the factors in the
neoclassical and endogenous growth models. For instance, IMF (2005), Fajnzylber and López
(2007), Jongwanich (2007), Fayissa and Nsiah (2008), Le (2008) and Chami et al. (2009) show
that even where remittances are injurious to growth, they do impact positively on other growthenhancing factors like investment in physical assets and human capital accumulation, improved
institutions, macroeconomic stability, and financial development through higher savings and in
making more funds available for credit expansion.
The general conclusion from the various empirical studies on SSA as a sub-region point to the
fact that the direct effect of remittances on economic growth is mixed. While Baldé (2009)
found no direct impact, Kagochi et al. (2010) found a positive impact on SSA countries with
relatively higher GDP per capita but no impact on SSA countries with lower GDP per capita.
Lartey (2010) found direct positive impact of remittances on economic growth in SSA, but
Singh et al. (2010) found a significant negative impact. These previous studies on SSA as a
sub-region, however, suffer several defects especially with regard to the appropriate contextual
definition and measurement of international remittances discussed in Chapter Two, and sample
representativeness. For example, the results obtained by Kagochi et al. (2010) can be
misleading in view of the fact that only six SSA countries with relatively vibrant and advanced
financial markets116 and relatively higher levels of income were analysed. Again, given the
fundamental macroeconomic disparity across the sampled countries, as shown in Chapter
Five, parameter estimates from pooled OLS may be inefficient, biased and unreliable.
Moreover, none of these previous authors defined and measured international remittances in
per capita terms or as the sum of workers‟ remittances and compensation of employees. Unlike
this study, previous studies explored only the contemporaneous effects of international
116
See Table 5.1 and Adenutsi et al. (2012) for evidence.
250
Stellenbosch University http://scholar.sun.ac.za
remittances on economic growth in SSA. However, as shown in this study, 117 at least in the
case of SSA, restricting empirical analysis of the potential remittance impact on growth to
contemporary effects of remittances is, at best, an underestimation.
Apart from Lartey (2010), previous regional studies on SSA measured economic growth as
logarithm of real per capita GDP and not the growth rate in real per capita GDP. It is important
to emphasise that, in a cross-country/panel study, using logarithm of real per capita GDP as a
measure of economic growth cannot be as appropriate as using the real per capita GDP growth
for two main reasons. First, since the former invariably measures income level which can
remain fairly high (in comparison with what is recorded in low-income countries) during periods
of global recession as was evident in the 2007-2009 credit crunch, even though the
industrialised countries with higher real per capita GDP suffered the heaviest recession and the
lowest growth rates, the real per capita GDP of these countries was still far higher than the real
per capita GDP of developing countries. Second, there are instances when a low-income
country can record a higher growth rate above the real per capita growth recorded in a highincome country due to a number of reasons given by proponents118 for the catch-up effect.
It can be seen from the literature reviewed that remittances can affect economic growth either
directly or indirectly through a variety of mechanisms including quality of institutions,
macroeconomic stability, human capital accumulation, investment physical capital, and
financial development. However, this indirect influential behaviour is not unique to international
remittances. For instance, when remittances, just like domestic financial resources, are saved
but these savings are not translated into quality investment, remittances cannot be blamed for
undermining economic growth. In many developing countries where default risks are high,
financial institutions try to avoid extending credit to Small and Medium-Scale Enterprises
(SMEs), self-employees and informal sector workers. Under this scenario, even if financial
development improves by way of higher credit extension to the private sector, and yet the
beneficiaries are mostly formal sector employees who are not entrepreneurs, improved private
sector access to bank credit may not impact positively on economic growth as a result of low
private investment. Similarly, investment in human capital development is theoretically progrowth, but the effect of human capital accumulation on long-run growth could only be positive
when measures are put in place to employ the educated to perform skill-related jobs in the
117
118
Compare the results reported in Table A6.8.1 and Table A6.9 with those reported in Table 6.1 for evidence.
These include Ramsey (1928), Solow (1956), Lucas (1990), and Barro and Sala-i-Martin (1997).
251
Stellenbosch University http://scholar.sun.ac.za
domestic economy. In situations where remittances were found to have no direct
contemporaneous impact but an indirect impact on long-run economic growth, it is plausible
that remittances could have a direct asynchronous effect on long-run growth. It is for this
reason that this study does not restrict itself to exploring only the direct contemporaneous
contribution of remittances to economic growth in SSA.
6.3.3 A Brief Literature Review on Other Potential Determinants of Economic Growth
Thus, far, available studies show limited and often contradictory evidence of the impact of
remittances on economic growth. Theoretically, the effect of the traditional sources of growth
has remained purely an empirical concern given the somewhat contentious theoretical
prediction of the potential impact of each of these determinants of economic growth. Therefore,
in order to provide both theoretical and empirical foundation for the empirical results of this
study that under some circumstances might be considered as counter-intuitive, a brief review of
the literature on the potential impact of the traditional sources of economic growth is presented
below.
6.3.3.1 Investment (INV)
Classical and neoclassical economists led by Smith (1776), Domar (1947), and Harrod (1948)
identify capital accumulation and productive investment as an important factor in the process of
long-run economic growth. In an apparent support of this proposition, proponents of
endogenous economic growth theories including Barro (1990), and Grossman and Helpman
(1991) argue that capital (broadly defined as ideas (or knowledge), learning by doing and
human capital accumulation) is a sine qua non for long-run growth through the steady-state
growth rate (Lucas, 1988; Romer, 1986; 1990). The endogenous growth theory further predicts
that despite the reality of the law of diminishing returns, marginal factor productivity can be
increased through investment. A typical instance is where massive commercialisation,
diversification, industrialisation, and technological progress financed through productive capital
investments increase total factor productivity and long-run economic growth. Nonetheless, it is
not theoretically illogical or counter-intuitive to predict a non-positive relationship between
capital investment and economic growth especially in the developing world. The reason is that
the extent to which capital investment can contribute positively to economic growth is
dependent upon the quality of the investment which is easily undermined by information
252
Stellenbosch University http://scholar.sun.ac.za
asymmetry, corruption119, low quality human resources, and unfavourable political and
macroeconomic environments in low-income countries. In other words, theoretically, high
investment ratio does not necessarily guarantee economic growth since the magnitude, the
quality and the productivity of investment in a stable socio-political and ideal macroeconomic
policy environment are necessary pre-conditions. Thus, to a very large extent, the existence of
appropriate policy, political and social infrastructure is crucial determinants of the effectiveness
of investment in enhancing growth (Hall and Jones, 1999; Artadi and Sala-i-Martin, 2003;
Fafchamps and Schündeln, 2013). In addition, because macroeconomic risk and geopolitics
can influence the performance of private investment, it is conceivable that the potential
contribution of investment to economic growth can be influenced by the relative dominance of
private over public ownership strategic assets and firms and the quality of relevant investment
information available.
Contrary to popular views and conclusions from majority of related previous studies, empirical
evidence from studies conducted by Klenow and Rodriguez-Clare (1997), and Hall and Jones
(1999) suggests that capital accumulation is not a primary source economic growth.
6.3.3.2 Government Expenditure (GXP)
As the largest consumer of final goods and services in a money economy, a government has
the option to implement its fiscal policy through a deliberate action on its level of expenditure. In
a typical endogenous growth framework, Rebelo (1991) demonstrates how economic policy
including government expenditure, a key fiscal policy instrument, can affect the rate of long-run
economic growth. According to the endogenous growth economists especially Barro (1990),
King and Rebelo (1990), Lucas (1990) and Stokey and Rebelo (1995), the share of public
expenditure in output or the composition of expenditure and taxation affects the steady-state
growth rate. As has been the tradition in most import-dependent developing countries,
however, unwarranted government expenditure on final goods and services leads to low public
sector saving and investment, low exports and high imports with undesirable implications for
price hikes and exchange rate depreciation culminating in less growth, at least, in the short run.
In fact, excessive government consumption can distort market outcomes and ultimately
depress economic growth. It is for this reason that governments in developing countries have
been under pressure from the Bretton Woods institutions and the international donor
119
In the developing world like SSA, the undesirable impact of corruption on the quality of capital investment is most
common and severe within the public sector.
253
Stellenbosch University http://scholar.sun.ac.za
community to implement austerity measures as a means of stabilising the macroeconomic
environment for sustained accelerated long-run growth. However, higher government
expenditure devoid of corrupt practices on locally produced final goods and services can
expand the market size of the domestic economy thereby accelerating long-run growth. In
various empirical studies, Landau (1983), Grier and Tullock (1987), Barro (1991), Ghura
(1995), and Fölster and Henrekson (1999; 2001) observe that government expenditure is
negatively correlated with economic growth, whilst Kormendi and Meguire (1985) find no
empirical relationship. In sharp contrast, Aschauer (1990), Engen and Skinner (1992), Kelly
(1997), Knoop (1999), and Alexious (2007) find out that government expenditure has a robust
positive impact on growth.
6.3.3.3 Openness to International Trade (OPN)
The fundamental liberalist argument is that openness to international trade has a potential
positive implication for long-run growth, since openness enables countries to allocate resources
more efficiently by promoting large-scale production, industrial research, innovative product
development and entrepreneurial activities arising from international competition and easier
access to larger product and capital markets globally. Additionally, openness to world trade can
enhance international capital flows thereby increasing financial openness and financial depth
which in turn facilitate financial development necessary for economic growth (Obstfeld, 1994).
In the opinion of Grossman and Helpman (1991), Romer (1993), and Barro and Sala-í-Martin
(1995), countries that are more open have a greater opportunity to catch up with leading
technologies of the rest of the world as market size for both finished and intermediate goods
expands. Enlarged market size raises research and development which is subject to increasing
returns to scale, hence economic growth.
On the other hand, openness to international trade compels local industries to face higher
foreign competition usually resulting in domestic industries of capital-constrained developing
countries losing their market share at home leading to capacity underutilisation and
retrenchment thereby impeding economic growth. Indeed, Alesina et al. (2000; 2005) develop a
theoretical model whereby there is an inverse relationship between openness to trade and
country size. Feenstra (1996) asserts that if intermediate goods are not traded, the integration
effect of trade openness can hardly be beneficial to smaller countries. Chang et al. (2009) point
out that openness promotes the efficient allocation of resources through comparative
advantage, allows dissemination of knowledge and technological progress, and encourages
254
Stellenbosch University http://scholar.sun.ac.za
competition in domestic and international markets. In contrast, Krugman (1994), and Rodrik
and Rodriguez (2001) argue that the effect of trade openness on growth is doubtful.
Depending upon which indicator is used to proxy trade openness, some empirical studies,
including Levine and Renelt (1992), Harrison (1996), Sala-í-Martin (1997), and Rodrik and
Rodriguez (2001) inter alia conclude that a negative relationship exists between trade
restrictions and economic growth. Yanikkaya (2003), however, finds that although trade
openness is positively correlated with growth, contrary to popular view, trade restrictions
positively and significantly affect economic growth in most developing countries. Thus, the
trade openness-growth nexus is basically an empirical question and has been extensively
interrogated in both theoretical and especially empirical studies with majority of the studies
finding a strong and statistically significant positive impact of trade openness on economic
growth. Notable among these studies are Dollar (1992), Lee (1993), Islam (1995), Sachs and
Warner (1995), Harrison (1996), Vamvakidis (1999), Frankel and Romer (1999), Greenaway et
al. (2002), Lee et al. (2004), Salinas and Aksoy (2006), Foster (2008), Kneller et al. (2008),
Wacziarg and Welch (2008), Chang et al. (2009), Kim (2011), Ulaşan (2012), Mercan et al.
(2013).
6.3.3.4 Human Capital Accumulation (HCA)
Despite the microeconomic theoretical consensus on the crucial contributory role of human
capital on growth (Barro, 1990; 1991; Mankiw et al. 1992), macroeconomic empirical findings of
the impact of human capital accumulation on economic growth are mixed. This may be due to
the fact that the availability of institutions and socially accountable governance, job-related
skills, access to job, and social infrastructure play a crucial role in determining the quality of
human capital which subsequently affects long-run growth. Besides, the quality of education,
retention of the educated workforce in the domestic economy and work ethics within the formal
sector have a direct impact on labour productivity, hence the contribution of human capital to
economic growth. Indeed, there are a host of macro-level studies that found weak, no, or
negative impact of human capital on growth as reported by Benhabib and Spiegel (1994), Islam
(1995), and Caselli et al. (1996). Quite recently, Bils and Klenow (2000), Bond et al. (2001),
Pritchett (2001), Easterly (2001), Easterly and Levine (2001), and Kumar (2006) reaffirm the
non-positive contribution of human capital to growth. Fedderke (2005) provides empirical
evidence in favour of positive impact of the quality but not the quantity of human capital
accumulation on total factor productivity growth. Meanwhile, Temple (1999a) and Krueger and
255
Stellenbosch University http://scholar.sun.ac.za
Lindahl (2001) argue that educational stock is only positively associated with economic growth
if the initial educational endowment of a country is relatively low.
What some may perceive as a counter-intuitive result of the impact of human capital in terms of
education on economic growth might be due to the fact that education could have a lag rather
than contemporaneous effect. Also, human capital proxies, including education stock is, to a
large extent, subject to measurement errors (de la Fuente and Domenech, 2002).
6.3.3.5 Rate of Inflation (INF)
Inflation is the proxy for economic uncertainty and investment risk and, therefore, has the
potential of discouraging private investment in the form of non-altruistic remittances. Romer
(2006) argues that, among other things, a high variability of inflation can suppress long-term
investment since this can be regarded as a signal of government malfunctioning that is capable
of eroding potential capital/investment gains. According to Temple (1999b), high inflation is
accompanied by exchange rate volatility, political instability and other undesirable factors that
impede economic growth. Cukierman et al. (1993), Fischer (1993) and Gillman et al. (2002)
obtain a negative relationship between inflation and growth. Conclusions from several other
studies, including Ghosh and Phillips (1998), and Nell (2000) show that whether or not inflation
promotes or undermines economic growth depends upon the rate of inflation as admittedly, an
inflation rate beyond a certain threshold can jeopardise growth.
More specifically, Fischer (1993) in a cross-country study comprising both developing and
industrialised countries established a negative non-linear relationship between inflation and
economic growth, noting that inflation only hampers economic growth after 40 per cent
threshold. Hasanov (2010) based on 2001-2009 annual data established a non-linear
relationship between inflation and growth rate at a 13 per cent threshold above which inflation
became injurious to the economic growth prospects in the Azerbaijani economy. Having
controlled for unit roots and co-integration, Umaru and Zubairu (2012) revealed that in the case
of Nigeria it is GDP growth rate that Granger-caused inflation and not inflation Granger-causing
economic growth. Mallik and Chowdhury (2001) find out that the relationship between
economic growth and the rate of inflation is positive and statistically significant for Bangladesh,
Pakistan, India and Sri Lanka. In this study, it was further established that the sensitivity of
growth to changes in inflation rates was smaller than that of inflation to changes in growth
rates. By these results, it was suggested that although moderate inflation promotes growth,
256
Stellenbosch University http://scholar.sun.ac.za
faster growth rate absorbs into inflation by overheating the economy. In a panel of 124
countries comprising advanced and developing countries, Bick et al. (2009) find a threshold of
two per cent for the industrialised countries and 17 per cent for developing countries. In the
case of Brazil over the 1980-1995 period, Faria and Carneiro (2001) find that the impact on
inflation on economic growth was negative in the short run but in the long run, inflation does not
affect growth, a results which seems to validate the super-neutrality theory of money.
Meanwhile, Wai (1959), Bhatia (1960), Dorrance (1963), and Sidrauski (1967) found no
evidence for a relationship between inflation and economic growth, whilst De Gregorio (1993
and Saaed (2007) find a negative relationship in their respective studies.
6.3.3.6 Bank Credit to the Private Sector (PSC)
There seems to be not much controversy that a relationship exists between finance and
economic growth. Both theoretically and empirically, what seems to be the debate has been the
direction of the causality, and the degree and type of impact of this relationship under different
macroeconomic conditions. Some scholars including (Bagehot, 1873; Schumpeter, 1912;
Hicks, 1969; Miller, 1998) argue that finance is a major contributor to growth while others such
as Robinson (1981) suggest that growth leads financial development and Lucas (1988) shows
that finance is over-stressed in explaining growth. In a theoretical contribution, Patrick (1966)
identifies a contrasting two-way hypothesis suggesting that the finance-growth causality is
either supply-leading or demand-following. In line with this postulation, a causal relationship
that runs from the indicators of financial development such as private sector credit allocation to
economic growth is described as supply-leading because it is believed that the activities of the
financial institutions increase the supply of financial products and services that stimulates
economic growth. In a similar fashion, when higher growth of an economy results in an
increase in the demand for financial products and services, often necessitating higher
competition and innovation within the financial sector hence financial development, then the
demand-following hypothesis is said to have prevailed. In the presence of high degree of
financial repression and weak financial sector via high-level of non-performing loans, high
information and transactions costs that inhibit financial deepening and the lagging behind of
financial reforms, financial development may not necessarily promote long-run growth, at least,
contemporaneously. In other words, bank credit to the private sector may not enhance the
prospects of economic growth in economies with high level of macroeconomic imbalances and
uncertainty, limited participation of the private sector in the productive sectors of the economy,
weak institutional environment, low labour productivity and widespread rent-seeking activities.
257
Stellenbosch University http://scholar.sun.ac.za
In effect, bank credit to the private sector can have a conflicting effect on economic growth
depending essentially upon the cost, volume and how the credit facility is used. In the financegrowth literature, it is hypothesised that if credit expansion to the private sector is enhanced in
a competitive financial market environment this can lead to higher investment in economically
viable projects thereby stimulating private sector-led growth in the long run (McKinnon, 1973;
Shaw, 1973; King and Levine, 1993). Rajan and Zingales (1998) observe that business
enterprises receiving the majority of their operational funding from financial institutions do not
expand normally in the economies which are financially developed. Similarly, financial
development resulting in credit reallocation (or redistribution of finances) is a driving force
behind high industrial growth rates (Fisman and Love, 2003; 2004; Hartmann et al. 2007).
Indeed, Hsieh and Klenow (2009) attribute the robust economic breakthroughs and the
enviable achievements of high performers and newly-emerging industrialised countries such as
China and India of the 21st Century to reallocation of quality financial resources from lesser to
higher productive sectors of their economies by financial intermediaries.
The aforementioned prospects of finance enhancing growth notwithstanding, the potential
capacity of private sector credit to stimulate the desired economic growth is dependent upon a
variety of factors including the amount, terms and the cost of the credit, the quality of the
investment, and the policy environment. When the private sector is given adequate credit
facility at relatively low cost under favourable terms of repayment in a stable and an
investment-friendly political and macroeconomic environment, there is a higher likelihood that
bank credit to the private sector will boost growth rather than when the contrasting conditions
prevail. In fact, recent developments in the global financial front seem to suggest that there is a
high conditional probability that private sector lending boom can lead to financial crises.
Habibullah and Eng (2006) following the Blundell and Bond (1998) GMM estimation technique
analysed the causal relationship between financial development and economic growth in a
panel of 13 Asian developing countries. The results, which is consistent with previous causality
studies by Calderón and Liu (2003), Fase and Abma (2003), and Christopoulos and Tsionas
(2004) confirm the Schumpeterian hypothesis of financial development promoting growth. In a
related study, the IMF (2008) in its Global Financial Stability Report noted a statistically
significant impact on credit growth on GDP growth. Specifically, the IMF reports that a credit
squeeze and a credit spread evenly over three quarters in USA will reduce growth by about 0.8
258
Stellenbosch University http://scholar.sun.ac.za
per cent and 1.4 per cent points year-on-year respectively assuming no other supply shocks to
the system.
In a panel study, Favara (2003) finds that the relationship between financial development and
economic growth is at best weak; and revealing further that there is no indication that finance
spurs growth and in some specifications, the relationship is puzzlingly negative. The results
obtained by Favara (2003) further suggest that financial development does not have a firstorder effect on economic growth; the link between them is not linear and if the dynamic
specification and slope heterogeneity across countries are taken into account, the effect is
substantially negative.
In the case of Turkey, Kar and Pentecost (2000) observe that when bank deposit, private
sector credit or domestic credit ratios are alternatively used as a proxy for financial
development, causality runs from economic growth to financial development, suggesting that
economic growth leads financial development. Meanwhile, Demetriades and Hussein (1996)
observe that bi-directional causality is possible. From a study involving 16 less developed
countries, they find a bi-directional causality in six countries, reversal causality in six countries,
while there was no evidence of a causal relationship in the case of South Africa. Shan and
Jianhong (2006) also find a bi-directional relationship in the case of China. Abu-Bader and AbuQarn (2008) for Egypt for the period 1960-2001 within a trivariate VAR framework, used four
different measures of financial development (ratio of M2/GDP, (M2-M1)/GDP ratio, ratio of bank
credit to the private sector to GDP, and the ratio of credit issued to private sector to total
domestic credit). The empirical results show that a bi-directional exists between financial
development and economic growth, and the indicators of financial development stimulate
growth through investment and economic efficiency. Arcand et al. (2012) reveal that there is a
threshold effect of 80-90 per cent above which the contribution of financial development to
economic growth is negative.
In an empirical study on the relationship between financial development and economic growth
in Egypt based on 1974-2002 data, Bolbol et al. (2005) report that the effect of bank-based
indicators on total factor productivity was negative unless these indicators are interacted with
per capita income. Ayadi et al. (2013) using data from 1985-2009 find that credit to the private
sector and bank deposits are negatively associated with economic growth, which affirms
deficiencies in credit allocation in the north or south Mediterranean countries. After controlling
259
Stellenbosch University http://scholar.sun.ac.za
for stock market development, Naceur and Ghanzouani (2007) report that banking sector
development indicators negatively impact on economic growth in 11 selected Middle East and
North African countries based on 1979-2003 annual panel data following dynamic modelling.
Other examples of studies that find a positive impact of financial development on economic
growth include Goldsmith (1969) in a cross-country study involving countries where there are
no appropriate financial structure and/or infrastructure. For 77 countries, King and Levine
(1993) confirm the results of Goldsmith (1969). Also, Beck et al. (2000), Rioja and Valev
(2004a,b) and Rousseau and Wachtel (2000) for a sample of countries in a panel data
framework. These conclusions seem to be the case especially in middle and high-income
countries.
6.3.3.7 Broad Money to GDP Ratio (M2/GDP)
Financial market development is an important ingredient for economic growth and development
(Hicks, 1969; Fry, 1995). Developed financial systems have efficient financial markets that
provide a better platform for monitoring operations and projects undertaken by financial
intermediaries, information and the safety net necessary for lowering transaction costs, and
channelling savings towards new and quality investments, thus leading to economic growth
(Greenwood and Jovanovic, 1990; Levine, 1991; Bencivenga and Smith, 1991, Blackburn and
Hung, 1996). Nevertheless, a negative impact of financial development proxied by M2/GDP is
not impossible under some circumstances. De Gregorio and Guidotti (1995) observe that
where financial repression exists or where financial liberalisation process is too fast and
characterised by a poor regulatory environment, the development of the banking sector can
lead to lower savings and investment rate, thereby impeding growth. In fact, Lartey (2010)
obtained a similar result.
6.3.3.8 Foreign Direct Investment (FDI)
This is one of the traditional sources of private external capital in the developing world. It
usually comprises the transfer of modern technology and (new) knowledge to enable the
recipient country to exploit the experience for an accelerated growth and sustainable
development. The macroeconomic impact of FDI is not automatically positive but primarily
dependent upon the nature and scope of FDI in terms of the scale, beneficiary sector,
concentration of local firms in the sector, duration of business and many other secondary
conditions (Manning and Shea, 1989). In much the same manner, Lipsey et al. (1994), Epstein
260
Stellenbosch University http://scholar.sun.ac.za
(1999), and Vo (2004) assert that FDI is favourable to economic prosperity only if the
appropriate conditions such as adequate absorptive capacity and quality human capital exist in
the beneficiary target country. Also of importance in determining the impact of FDI in the host
country are the capacities of domestic enterprises to face and hold out to foreign competition,
abundance of projects and market gaps that cannot be filled up by domestic entrepreneurs in
the host country (ibid). A host of empirical studies including Blomström et al. (1996),
Borensztein et al. (1998), and de Mello (1999) found evidence of a positive impact of FDI on
economic growth.
From various empirical studies, Bengelsdijk et al. (2008) find that unlike in the case of
developed countries, FDI has no significant impact on developing countries. Jackman (1982),
Akinlo (2004), Carkovic and Levine (2005), and Schneider (2005) find no relationship between
FDI and economic growth in LDCs. In related empirical studies, Bos et al. (1974), Rothgeb
(1984), Saltz (1992), Kholdy (1995), Mencinger (2003), Durham (2004), and Darrat et al.
(2005), find a negative effect of FDI on growth in host developing or non-OECD countries.
6.3.3.9 Institutional Quality (INS)
Hall and Jones (1999), Kaufmann et al. (2003) and Acemoglu et al. (2005) argue that good
institutions stimulate economic growth and development whilst poor institutions impede growth
and development. Weak institutions breed corruption, and corruption creates an unstable and
unsafe business environment (Gray et al. 2004) because of increased selectivity, inequality and
the opacity of the rules for the outsiders. Lack of confidence in the domestic political system,
weak institutions and bad governance normally breed macroeconomic instability arising from
economic mismanagement, direct unproductive rent-seeking economic activities, and public
sector corruption especially in the form of public funds (Acemoglu et al. 2001; Hermes and
Lensink, 2001; Lensink et al. 2000). Similarly, Owens (1987), North (1981; 1990), Sen (1999),
and Todaro and Smith (2002) posit that quality institutions arising from political, economic and
social rights and freedom coupled with transparent governance and security are a necessary
condition for long-run economic growth and development. The contribution of the state of
institutions on growth may not be significant in the short run because Acemoglu et al. (2001;
2002; 2005) argue that the influence of institutions on growth is more of long term rather than
short term. Resnick and Birner (2006) express uncertainty about the role of institutions in the
growth and the development process of an economy.
261
Stellenbosch University http://scholar.sun.ac.za
Persson et al. (2000) developed a theoretical model that shows how strong institutions in terms
of good parliamentary system, a proxy for participatory democracy enhance effective public
spending, and invariably good economic performance. This theory has been corroborated in
empirical studies undertaken by Knack and Keefer (1995), Hall and Jones (1999), Knack
(2000), Acemoglu et al. (2001; 2005), Dollar and Kraay (2003), Persson and Tabellini (2003;
2004), Glaeser et al. (2004), and Rodrik et al. (2004).
6.3.3.10 Official Development Assistance (ODA)
Given the high dependency of SSA countries on the outside world for donor support (see
Chapter Three), it is of crucial relevance to include overseas development assistance or aid
(ODA) in a growth model of a typical developing country to capture the impact of official
external assistance on growth (Burnside and Dollar, 2000; Easterly, 2003). Proponents of
foreign aid, notably Chenery and Strout (1966), Papanek (1973), Levy (1988) and Islam (1995),
argue that ODA is crucial to the growth process of developing countries. However, Heller
(1975) and Boone (1994) argue that foreign aid cannot be a propeller of domestic savings and
economic growth in developing countries.
Theoretical arguments against foreign aid include dependency mentality (Kanbur, 2000),
privates sector crowding-out effects (Bauer, 1976; Krauss, 1983), worsening bureaucratic
quality (Knack and Rahman, 2007), weakening governance (Knack, 2000; Rajan and
Subramanian, 2007), and lowering international trade competitiveness through the Dutch
disease effects (Rajan and Subramanian, 2005). Chenery and Strout (1966) caution against
over-reliance on foreign aid by arguing that foreign aid can be detrimental and counterproductive to growth mainly because the potential contribution of aid to production and
investment, hence growth, is dependent upon the absorption capacity of the aid-recipient
country to make good use of the aid. Factors that enhance the absorptive capacity of foreign
aid include existing infrastructure, availability of skilled labour, the institutional and
administrative capacity of national and local governments.
On the reverse side of the argument, Sachs et al. (2004) and Sachs (2005a,b) maintain that
foreign aid is beneficial to low-income countries and actually advocate for more aid to the
developing world because aid is the most surest means by which escaping poverty traps in
low-income countries is possible. In another contribution to the literature, Easterly (2007a,b)
262
Stellenbosch University http://scholar.sun.ac.za
argue that history does not favour the effectiveness of aid in promoting economic growth,
hence more aid carry undesirable consequences for growth in aid-recipient countries.
While microeconometric studies on the impact of foreign aid on economic growth are clear and
generally positive, macroeconometric studies are generally inconclusive; perhaps, due to the
paradoxical micro-macro conflicting outcomes in empirical studies, which can be attributed to
well-known measurement errors and aggregation problems.
Some macro-level studies, including Guillaumont and Chauvet (2001), Clemens et al. (2004),
Dalgaard et al. (2004) and Moreira (2005) find foreign aid as a positive determinant of
economic growth in aid-recipient countries. In an empirical contribution, Burnside and Dollar
(2000: 847) observe that foreign aid has “a positive impact on economic growth in developing
countries with good fiscal, monetary and trade policies, but has little effect in the presence of
poor policies.” The findings by Burnside and Dollar (2000) have been validated by Collier and
Dollar (2001; 2002),and Collier and Hoeffler (2004) that foreign aid only works effectively on
growth in aid-receiving countries with “good policies.” In many ways, these results have since
influenced some donor countries to be increasingly and conditionally allocating aid to
developing countries that perform well and/or have good policies. Having investigated the
impact of foreign aid on economic growth in 71 aid-receiving countries using annual data from
1960 to 1997, Karras (2006) concludes that the impact of foreign aid on economic growth is not
only positive, but also significant, permanent and sizeable. Similarly, Hansen and Tarp (2000;
2001) point out that a positive aid-growth link exists even under favourable economic
conditions in aid-receiving countries.
In other related studies, Boone (1994; 1996), Easterly et al. (2004), and Easterly (2005)
conclude that the foreign aid-growth relationship is neutral in various empirical contexts. Bobba
and Powell 2007) find a negative impact of foreign aid on economic growth.
6.4 EMPIRICAL MODEL, METHODOLOGY AND DATA ISSUES
6.4.1 The Empirical Model and Methodology
From the literature reviewed, stylised facts and the theoretical framework presented above, it is
obvious that remittances are likely to correlate with many traditional determinants of growth in
different ways. This is a recipe for a severe endogeneity problem that poses a challenge when
there is an attempt to analyse the impact of remittances on economic growth using a
263
Stellenbosch University http://scholar.sun.ac.za
macroeconometric technique. Among the possible panel-data estimation techniques involving
large cross-sections ( N ) and small time period (T ) , a dynamic model following the
Generalised Method of Moments (GMM) procedure is recommended (see Chapter Four). This
study, therefore, used the system GMM procedure suggested by Blundell and Bond (1998) to
estimate an empirical dynamic panel-data model, with the dimension ( N 36) and (T 10) for
decade-based analysis, and N T 36 30 for overall period analysis. The empirical model in
its general form is specified as Equation (6.18).
growthi ,t 1 growthi ,t 1 2 ln INVi ,t 3 ln GXPi ,t 4 ln OPNi ,t 5 ln REMPCi ,t 6Zi ,t
t i i ,t
(6.18)
where economic growth is measured by an annual percentage change in real per capita GDP
in US$ which signifies improvements in productivity120; investment (INV) measured as the ratio
of gross fixed capital formation to GDP; trade openness (OPN), government expenditure
(GXP), and REMPC represents migrant remittances per capita proxied by the sum of workers‟
remittances and compensation of employees121 as a ratio of population. Central government
final consumption expenditure (GXP) as a percentage of nominal GDP was introduced to
capture the role and size of government (Barro, 1990). Considering the important role of the
trade sector in endogenous growth (Grossman and Helpman, 1991), openness to international
trade (OPN) was proxied by the sum of exports and imports as a percentage of nominal GDP
was included in the model. The elements of the matrix Z are the set of principal control
variables. The original elements of Z are foreign direct investment, official development
assistance, bank credit to the private sector, broad money ratio, human capital accumulation,
inflation, real exchange rate, and institutional quality. Initial growth rate is included so as to
capture the possibility of the absolute common convergence phenomenon122 among the
sampled countries.
The subscripts i and t are the country and time identities respectively whilst
t and i are the
time-specific and country-fixed effects respectively, and is the specific significant lag
120
This also denotes economic efficiency and the long-run output expansion rate of material production. It captures
the extent to which economic production rises in relation to increases in population size of a country.
121
Workers‟ remittances are funds transferred back home by permanent migrants whilst compensation of employees
are the funds sent home by migrants who are temporarily resident (less than 12 months) abroad.
122
See Ramsey (1928), Solow (1956), Swan (1956), Koopmans (1965), Baumol (1986), and Barro and Sala-i-Martin
(1997) for justification; even though the unconditional convergence hypothesis may not necessarily hold in a panel
study involving countries with wide structural disparities in terms of technological advancement, saving rate, rate of
depreciation, and population growth as shown in Aghion and Howitt (1997), Gaulier et al. (1999), and Barro (2003).
264
Stellenbosch University http://scholar.sun.ac.za
operator on INV and REMPC following the accelerator principle123, such that 0 4 . The
idiosyncratic disturbance term ( ) which is assumed to be normally distributed with a constant
variance and zero mean, takes into account the unobserved time-variant factors that can
influence economic growth. Equation (6.18) is a semi-log endogenous growth model. Thus, the
empirical model of this study states that economic growth in country i at year t is determined
by initial growth, government expenditure ratio, openness to international trade, current and
past rates of investment and remittances and other orthodox growth determinants contained in
Z . In line with the endogenous growth theory, the a priori signs are 2 , 3 , 4 0 ; but for the
other coefficients, their signs are indeterminate a priori although the expected sign of
5 is
skewed towards positive, because for the most part of the period under study, migrant
remittance inflows in SSA were pro-cyclical.
The system GMM estimation procedure adopted in this study yields more efficient, precise and
reliable estimators than the first-difference GMM proposed by Arellano and Bond (1991) and
deviations GMM proposed by Arellano and Bover (1995) as noted in Chapter Four. The merits
of the system GMM (sys-GMM) over the other alternative estimation techniques for a panel
setting of this nature are thoroughly discussed in Chapter Four (see also Blundell and Bond,
1998; Behr, 2003; Baltagi, 2008).
In order to obliterate any such misgivings concerning the reliability of the results for the entire
period when N can be thought of as reasonably large, the time series properties of each of the
variables were investigated, using the Fisher Phillips-Perron (Fisher P-P) chi-square test and
the Hadri Heteroskedasticity Consistent (HHC) z -test. In the event of a conflict between the
Fisher P-P statistic and the HHC statistic, the study employed the Levin-Lin-Chu (LLC)
adjusted t test124. The panel unit root test results presented in Table A6.5 show that each
variable is integrated into order zero, hence the estimated model is co-integrated125, and the
123
An implication of this theory is that the effect of (private) investment and, for that matter, self-interest investmentdriven remittances on economic growth is less likely to be instantaneous.
124
A concise discussion of these panel unit root tests is presented in Chapter Four.
Note that the problem of co-integration relates to trended time series of a particular non-zero order that can
potentially create problems in empirical econometrics if the linear combination of these variables does not yield a
stationary residual. Therefore, in models exclusively involving I(0) time series, the problem of co-integration does not
arise (Engle and Granger, 1987; Gujarati, 1995; Asteriou, 2006).
125
265
Stellenbosch University http://scholar.sun.ac.za
empirical results obtained from the long-run equilibrium relationship are not econometrically
spurious126.
To verify if migrant remittances received have a long-run size-effect on economic growth,
according to the growth rates among the 36 sampled SSA countries, an expanded Equation
(6.18) was re-estimated127 to include a median-dummy variable. This median-dummy variable
(MDV) is a dichotomous variable that takes the value of one if in a particular time period t , the
growth rate of a country i exceeds the median growth rate of the sampled countries; otherwise
it takes the value of zero. In other words,
MDV 1, if growthi ,t growthmedian , and
MDV 0, if otherwise. A statistical significance of the estimated parameter corresponding to
MDV indicates evidence of size-effect of remittances within the sample requiring a further
investigation of the nature of this bias. Therefore, where MDV is statistically significant, the
study proceeds to the third-stage of estimation, in which MDV-remittance interactive effect was
explored to evaluate how remittances received impacted on economic growth in SSA countries
where the economic growth rate exceeds the median-level of the sampled group of countries.
The corresponding expanded second-stage and third-stage estimated models are specified as
Equation (6.19) and Equation (6.20) respectively:
growthi ,t 1 growthi ,t 1 2 ln INVi ,t 3 ln GXPi ,t 4 ln OPNi ,t 5 ln REMPCi ,t 6i ,t
7 MDVi ,t t i i ,t
(6.19)
growthi ,t 1 growthi ,t 1 2 ln INVi ,t 3 ln GXPi ,t 4 ln OPNi ,t 5 ln REMPCi ,t 6i ,t
7 MDVi ,t t i i ,t
(6.20)
where the elements in matrix are the „final‟ set of explanatory variables that emerged from
the original set of control variables contained in matrix Z in Equation (6.18); MDV is the
median-dummy variable defined above, and MDV is the interaction of MDV and migrant
remittances per capita in the exact empirical context as in Equation 6.18. The computed
median values of economic growth are 0.414481 for 1980-89; 0.881176 for 1990-99; 2.181857
126
If a stationary or an I(0) combination exists, time series is said to be co-integrated and a long-run equilibrium
relationship of the variables exists.
127
Augmenting Equation (6.18) with the introduction of MDV does not affect the original panel structure in which
N T .
266
Stellenbosch University http://scholar.sun.ac.za
for 2000-09; and 1.315342 for the overall study period, 1980-09. All other variables in
Equations (6.19) and (6.20) are defined under Equation (6.18) and Table A6.2 in the Appendix.
Finally, for the overall study period, 1980-2009, static panel-data models were estimated to
provide further empirical evidence on the relationship between migrant remittance inflows and
economic growth in SSA. Both the conventional and the robust static panel-data Fixed (within)
Effects (FE) and Random GLS Effects (RE) models were estimated. In this econometric
exploration, decade-based estimations were not carried out essentially because as explained
under 4.5.2 in Chapter Four, static panel-data modelling of the issue at stake cannot be the
most appropriate in terms of the efficiency and reliability of the estimators within the confines of
this particular analysis. Certainly, the estimates from the static panel-data models are not
expected to necessarily confirm those obtained from system dynamic GMM estimations.
Clearly, in the event of the anticipated conflict in results, it is the robust two-step sys-GMM
results that are relied upon to inform policy imperatives.
6.4.2 Data Sources and Description
Unless otherwise specified in Table A6.2, the annual panel data used in this study were
collated from the April 2011 Edition of World Development Indicators (WDI) published by the
World Bank and World Economic Outlook (WEO) published by the IMF. The list of countries
included in the analysis is presented in Chapter One. Economic growth is measured as an
annual growth rate in real GDP per capita in a typical migrant-home SSA country. In the
absence of available data on capital stock, investment in physical capital (INV) measured by
gross fixed capital formation as a ratio to GDP was used. Gross fixed capital formation
comprises the monetary value of land improvements, plant, machinery and equipment
purchases, construction of roads, railways, and other infrastructure like schools, hospitals,
offices, private residential dwellings, and commercial and industrial buildings. Government
expenditure (GXP) was proxied by central government final consumption spending as a
percentage of nominal GDP. Openness to international trade (OPN) was proxied by the sum of
exports and imports as a percentage of nominal GDP. Openness measures based on trade
flows and trade dependent ratio are by far the most commonly used in empirical studies as
shown in the works of Frankel and Romer (1999), Irwin and Tervio (2002), Frankel and Rose
(2002), Dollar and Kraay (2004), and Squalli and Wilson (2011).
267
Stellenbosch University http://scholar.sun.ac.za
Secondary school enrolment used as a proxy for human capital development (HCA) was
introduced into the model as a key determinant of growth in order to be consistent with the
models that appeal to endogenous growth theory (see Romer, 1986, 1990; Lucas, 1988; Barro,
1990, 1991; World Bank, 2006b; Calderón et al. 2008). Private sector credit (PSC) is the stock
of claims by deposit money banks and other financial institutions on the private sector as a
percentage of nominal GDP. Broad money ratio (M2/GDP) was measured as broad money as a
percentage of nominal GDP.
Foreign direct investment (FDI) is the net inflows of investment, being the sum of equity capital,
reinvestment of profits, other long-term capital, and short-term capital, to acquire long-term
management interest in an enterprise operating in an economy other than that of the investor,
expressed as a percentage of nominal GDP128. Rate of inflation (INF) is the annual percentage
change in the cost to the average consumer acquiring a basket of basic essential goods and
services in an economy. Official development assistance (ODA) is the disbursement flows (net
of repayments) from official donors to a country as a percentage of nominal GDP. Institutional
quality was included among the control variables so as to assess the effects of governance on
economic growth. This variable was proxied by polity2 index (which ranges from -10 for low
democratic governance to +10 for high democratic governance and strong institutions) was
obtained from Marshall and Jaggers (2011) who developed this index. Real exchange rate was
also included in the initial set of control variables. In the final estimation, however, institutional
quality and real exchange rate were excluded, based on the efficiency test of the empirical
model. The set of explanatory variables129 included in the empirical model has been the most
popularly used in empirical growth modelling involving remittances (see Table A6.11). The
statistical description and the bivariate correlation coefficients of the dataset are presented in
Table A6.3 and Table A6.4 respectively.
6.5 EMPIRICAL RESULTS AND DISCUSSIONS
First and foremost, in Table A6.10, the estimated results of the impact of migrant remittance
inflows on economic growth within the context of static panel-data modelling of 36 SSA
countries over the period 1980-2009 are presented. Based on the Hausman specification test
conducted on the empirical Fixed (within) Effects (FE) and the Random GLS Effects (RE)
models, the empirical results of the FE model were relied upon. The Breusch-Pagan statistic,
128
129
This is how FDI (net inflows) reported in WDI is defined by the World Bank.
See Table A6.2 for the „final‟ set of the explanatory variables used in the estimated model.
268
Stellenbosch University http://scholar.sun.ac.za
however, shows that the reported standard errors of the conventional FE and RE empirical
models are not homoscedastic. Accordingly, alongside the estimated conventional FE and RE
models, robust heteroskedasticity-corrected versions of the empirical static FE and RE models
were estimated and reported in Table A6.10 as robust FE and robust RE respectively.
Therefore, with reference to the reported static panel-data modelling of the impact of migrant
remittances on economic growth, the estimated results of the robust FE are deemed the most
efficient and reliable in the context of this study. The estimated empirical robust FE results
show that, consistent with the dynamic panel-data estimation by sys-GMM reported in Table
6.1, migrant remittances have a significant positive asynchronous impact on economic growth
in SSA between 1980 and 2009. Although, in comparison with dynamic panel-data modelling,
the empirical results from the robust static panel-data model estimations are less reliable
because of the omission of dynamic effects and the presence of endogeneity bias, the results
suggest that for the overall period 1980-2009, human capital accumulation had a significant
positive impact on economic growth, whereas bank credit to the private sector, and broad
money to GDP ratio inhibited growth in SSA. The computed R 2 of 0.0531 suggests that the
explanatory power of the estimated robust static FE panel-data model is merely 5.31 per cent
leaving as much as 94.69 per cent of the total variations in economic growth in SSA
unexplained. Invariably, this gives further evidence of the inappropriateness of static panel-data
estimation in the context of this particular analysis. More importantly, the main empirical results
of this chapter which are based on dynamic panel-data estimations following sys-GMM in
determining the impact of migrant remittances on economic growth in SSA are presented in
Table 6.1.
The Sargan test for the hypothesis that the estimated growth model and the over-identifying
conditions are correctly specified and valid is upheld at one per cent level of statistical
significance. The Arellano-Bond test statistic suggests that there are no second-order serial
correlations in the first-differenced disturbances at the conventional levels of statistical
significance. The Wald statistic confirms that, for each of the estimated sys-GMM, the
explanatory variables jointly explained the variations in the rate of economic growth in SSA
over the past three decades. Based on the robustness of the model performance diagnostic
269
Stellenbosch University http://scholar.sun.ac.za
tests130, the findings of this study as presented in Table 6.1 (cf. Table A6.8.1 in the Appendix)
also show that the direct impact of migrant remittances on long-run growth is not
contemporaneous, but with a one-year lag131.
Table 6.1: Estimated Impact of Remittances on Economic Growth in SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-89
1990-99
2000-09
1980-2009
Initial economic growth (growth_1)
-0.1027
(-4.83)***
-0.2159
(-30.41)***
1.3386
(4.54)***
0.1000
(1.24)
Investment (lnINV_1)
-5.5122
(-3.90)***
1.0946
(1.68)*
-2.8670
(-2.33)**
0.9692
(1.81)*
Government expenditure (lnGXP)
-1.2861
(-0.88)
2.0337
(1.77)*
-0.8103
(-1.54)
-11.1898
(-2.46)**
Trade openness (lnOPN)
6.3826
(3.44)***
1.1837
(0.74)
0.8309
(0.67)
13.0927
(3.02)***
Migrant Remittances (lnREMPC_1)
1.0378
(3.64)***
0.6717
(3.91)***
-0.1917
(-0.68)
1.9733
(2.50)**
Human capital accumulation (lnHCA)
1.8855
(1.64)*
3.2674
(2.63)***
-1.9897
(-2.23)**
-5.0397
(-1.50)
Rate of inflation (INF)
-0.0053
(-0.92)
0.0723
(4.74)***
-0.0959
(-5.72)***
-0.0023
(-0.40)
Bank credit to private sector (lnPSC)
0.1995
(0.24)
-2.5885
(-3.04)***
1.6250
(1.71)*
-2.4258
(-1.36)
Broad money to GDP ratio (lnM2/GDP)
-1.5905
(-4.72)***
-9.3526
(-8.35)***
-2.4518
(-2.28)**
-11.9825
(-2.80)***
Foreign direct investment (FDI)
0.3246
(6.20)***
-0.0899
(-4.51)***
0.0819
(2.43)**
-0.1072
(-1.99)**
Constant term
2.6847
(1.00)
11.0032
(1.78)*
20.2703
(6.12)***
30.6042
(2.07)**
Number of observations
319
322
324
1037
Number of groups (N)
36
36
36
36
Number of instruments
2
Wald [10],
54
54
54
444
1337.39***
22377.75***
403.15***
79.89***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
0.3336{0.739}
-1.2508{0.211}
0.9028{0.367}
0.2574{0.797}
Sargan test of over-identifying restrictions:
[2],
Source: Author‟s estimation
130
[43], 29.6174
[43], 23.4327
[43], 24.7495
[433], 21.7956
*/**/*** denotes statistical significance at 10%, 5% and 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
Generally, the reported statistics and probabilities of the Wald , Arellano-Bond order-2 tests and Sargan tests
2
are more robust for the estimation involving the asynchronous impact of remittances on economic growth (Table 6.1)
than the results involving the contemporaneous impact of remittances on growth (Table A6.8.1).
131
In fact, comparing the results in Table 6.1 to those reported in Table A6.9, both remittances and investment are
more asynchronous than contemporaneous.
270
Stellenbosch University http://scholar.sun.ac.za
The empirical results show that between 1980 and 2009 migrant remittances received had a
significant positive impact of 1.9733 on long-run economic growth in SSA, but the size of this
impact varies over time in response to some macroeconomic fundamentals. In the 1980s and
the 1990s, previous year remittances impacted positively on long-run growth, whilst in the
2000s, the remittances had a negative but statistically insignificant impact on growth in SSA.
These findings are in spite of the facts that since the implementation of financial liberalisation
programmes in SSA in the 1980s, international remittance inflows have been increasing
steadily (see Figure 3.2) and remittance inflows in SSA were somewhat countercyclical in the
1990s compared to the pro-cyclical trend of the 1980s and the 2000s (see Figure 4.1).
More specifically, the results show that a one percentage rise in per capita international migrant
remittances received between 1980 and 1989 had a 1.0378 positive impact on the real per
capita GDP growth rate in SSA. Between 1990 and 1999, a similar rise in the receipt of per
capita international migrant remittances had a 0.67171 impact on economic growth rate in
migrant-home SSA countries. In more recent years (i.e. 2000-2009), however, the ordinary
effect of international remittances on economic growth rate in SSA countries is statistically
zero. The results of the decade-by-decade analysis show that, even though international
migrant remittances had a significant positive impact on economic growth in SSA between
1980 and 2009, the impact of remittances on SSA countries as a group has been declining
over the years132. These results seem to suggest that a mere policy switch towards the
implementation of financial liberalisation programmes alone may not be adequate to maximise
the impact of international remittances on economic growth in SSA.
Overall, in descending order of the magnitude in terms of economic significance, other
important macroeconomic factors that contributed to economic growth in SSA between 1980
and 2009 are international trade openness, broad money to GDP ratio, government
expenditure, investment in physical assets and FDI. By implication, the general contributions of
human capital accumulation, private sector credit and the rate of inflation to economic growth in
SSA between 1980 and 2009 were statistically insignificant.
132
This declining trend is similar to the contemporaneous impact of remittances on economic growth in SSA
reported in Table A6.8.1.
271
Stellenbosch University http://scholar.sun.ac.za
Specifically, the empirical results show that trade openness and investment impacted positively
on economic growth over the 30-year period. On the average, a one per cent increase in
openness to international trade and investment to GDP ratio resulted in approximately 13.09
per cent and 0.97 per cent rise in economic growth in SSA. Therefore, holding all other factors
constant, over the past three decades, migrant remittances emerge as the second most
important contributor to economic growth in SSA. Over the same period, however, increases in
broad money, government expenditure and FDI were harmful to economic growth in SSA. In
the case of M 2 / GDP ratio, this could be attributed to the inherent limitations associated with
its measurement, a measurement problem more pronounced in the developing world because
of the excessive dominance of the currency in circulation ( M 1 ) over quasi money, hence
measuring degree of monetisation rather than financial development (see 7.2.2.4 for further
details).
The finding that government expenditure to GDP ratio undermines economic growth in SSA
should not be considered counter-intuitive since majority of sampled SSA countries are importdependent, implying that the governments of these SSA countries might have spent more on
imported consumables rather than locally-produced goods and services. Again, high level of
public sector corruption could lead to over-invoicing and financial loss to the state by way of
over-payment of bills for the purchase of final goods and services by the government.
The negative contribution of FDI could be a reflection in the high participation of foreigners in
the retail markets of most SSA countries rather than investment in the valued-added productive
sectors of the FDI-receiving economy. In other words, on the average, net FDI inflows have not
been effective in contributing to economic growth in SSA over the past three decades, probably
due to the low absorptive capacity of the recipient SSA countries.
The results of the statistical test verifying the extent to which the estimated decade-based
coefficients of the empirical growth model evolve over time across the three decades are
presented in columns A-B, B-C and A-C of Table 6.2. Column A-B of Table 6.2 reports the
results on how the estimated coefficients of the 1980-89 decade differ statistically from the
corresponding estimated coefficients of the 1990-99 decade. The results on the extent to which
the estimated decade-based coefficients of 1990-99 and 2000-09 are statistically different, are
reported in column B-C. In column A-C of Table 6.2, the results of the extent to which how the
estimated coefficients of 1980-89 decade are statistically different from the corresponding
272
Stellenbosch University http://scholar.sun.ac.za
estimated coefficients of the 2000-09 decade are reported. The null hypothesis that there is no
statistical difference between any pair of the corresponding estimated decade-based
coefficients to signify evolution in the empirical growth model across the three decades is
rejected with 95 per cent statistical confidence. In other words, statistically, international
migrant remittances just like the other explanatory variables in the empirical growth model had
statistically significant decade-based changing impact on economic growth in SSA between
1980 and 2009.
With reference to columns A-D, B-D, B-E and C-E of Table 6.2, there is also a robust statistical
evidence that the reported differences among the estimated decade-based coefficients of
international migrant remittances are consistent providing further evidence for parameter
instability over time between 1980 and 2009. Furthermore, the instability of the estimated
decade-based coefficients of the other explanatory variables are statistically significant and
generally consistent except in the case of foreign direct investment (FDI) as reported in
columns A-D, B-D, B-E and C-E of Table 6.2.
273
Stellenbosch University http://scholar.sun.ac.za
Table 6.2: Results of Decade-Based Parameter Evolution and Instability Tests for Impact of Migrant Remittances on Growth in SSA
Initial economic growth (growth_1)
Investment (InINV_1)
Government expenditure (lnGXP)
Trade openness (lnOPN)
Migrant Remittances (lnREMPC_1)
Human capital accoumulation (lnHCA)
Rate of Inflation (INF)
Bank credit to private sector (lnPSC)
Broad money to GDP ratio (lnM₂/GDP)
Foreign direct investment (FDI)
Constant term
Number of observations
Number of groups
Number of instruments
Wald statistic
A-B 2ⁿᵈ-order autocorrelation test
Sargan over-identifying restrictions
Source: Author‟s estimation
Estimated
A
1980-89
-0.1027
[0.0213]
{-4.83}***
-5.5122
[1.4134]
{-3.90}***
-1.2861
[1.4614]
{-0.88}
6.3826
[1.8554]
{3.44}***
1.0378
[0.2851]
{3.64}***
1.8855
[1.1497]
{1.64}*
-0.0053
[0.0057]
{-0.92}
0.1995
[0.8313]
{0.24}
-1.5905
[0.3370]
{-4.72}***
0.3246
[0.0524]
{6.20}***
2.6847
[2.6847]
{1.00}
319
36
54
1337.39***
0.334(0.74)
29.617(0.93)
Decade-Based Rolling
Non-Overlapping Decade-Based Overlapping Decade-Based Coefficient Stability
Estimated Results
Coefficient Stability Test Results
Test Results
Decade-Based Results
B
C
D
E
1990-99
2000-09
1985-1994
1995-2004
A-B
B-C
A-C
A-D
B-D
B-E
C-E
-0.2159
1.3386
-0.2428
0.0559
0.1132
-1.5544
-1.4413
0.1401
0.0269
-0.2717
1.2827
[0.0071]
[0.2948]
[0.0208]
[0.0248]
[0.0142]
[0.2877]
[0.2736]
[0.0005]
[0.0137]
[0.0177]
[0.2700]
{-30.41)***
{4.54}***
(-11.70)***
{2.25}**
{7.99}***
{-5.40}***
{-5.27}*** {274.61}***
{1.97}***
{-15.33}***
{4.75}***
1.0946
-2.8670
1.0883
-0.0833
-6.6067
3.9616
-2.6451
-6.6005
0.0063
1.1779
-2.7837
[0.6515]
[1.2305]
[0.9804]
[0.8332]
[0.7618]
[0.5789]
[0.1829]
[0.4329]
[0.3289]
[0.1817]
[0.3973]
{1.68}*
{-2.33}**
{1.11}
{-0.10}
{-8.67}***
{6.84}*** {-14.46}*** {-15.25}***
{0.02}
{6.48}***
{-7.01}***
2.0337
-0.8103
-6.1741
3.8095
-3.3198
2.8440
-0.4758
4.8881
8.2078
-1.7758
-4.6198
[1.1490]
[0.5262]
[1.3751]
[1.1440]
[0.3124]
[0.6228]
[0.9353]
[0.0863]
[0.2261]
[0.0050]
[0.6178]
{1.77}*
{-1.54}
{-4.49}***
{3.33}*** {-10.62}***
{4.56}***
{-0.51}
{56.61}***
{36.30}*** {-355.87}***
{-7.48}***
1.1837
0.8309
1.6063
5.6055
5.1989
0.3529
5.5517
4.7763
-0.4226
-4.4218
-4.7746
[1.5997]
[1.2401]
[1.3613]
[0.9389]
[0.2558]
[0.3595]
[0.6153]
[0.4941]
[0.2384]
[0.6607]
[0.3012]
{0.74}
{0.67}
{1.18}
{5.97}***
{20.33}***
{0.98}
{9.02}***
{9.67}***
{-1.77}**
{-6.69}***
{-15.85}***
0.6717
-0.1917
-0.0862
0.3459
0.3661
0.8634
1.2295
1.1240
0.7579
0.3258
-0.5376
[0.1718]
[0.2819]
[0.3916]
[0.2402]
[0.1133]
[0.1101]
[0.0032]
[0.1065]
[0.2198]
[0.0684]
[0.0417]
{3.91}***
{-0.68}
{-0.22}
{1.44}
{3.23}***
{7.84}*** {380.65}***
{10.55}***
{3.45}***
{4.76}***
{-12.89}***
3.2674
-1.9897
-0.2258
-3.9252
-1.3819
5.2571
3.8752
1.8475
3.2294
3.2662
-1.9909
[1.2424]
[0.8922]
[2.2578]
[1.6919]
[0.0927]
[0.3501]
[0.2575]
[1.1081]
[1.0154]
[0.4495]
[0.7997]
{2.63}***
{-2.23}**
{-0.10}
{-2.32}** {-14.91}***
{15.01}***
{15.05}***
{1.67}*
{3.18}***
{7.27}***
{-2.49}**
0.0723
-0.0959
0.0380
0.0012
-0.0775
0.1681
0.0906
-2.0235
-1.9460
1.5548
1.3867
[0.0152]
[0.0168]
[0.0115]
[0.0110]
[0.0095]
[0.0015]
[0.0110]
[0.0057]
[0.0038]
[0.0042]
[0.0058]
{4.74}***
{-5.72}***
{3.32}***
{0.11}
{-8.16}*** {110.61}***
{8.22}*** {-354.07}*** {-514.13}*** {366.69}***
{240.74}***
-2.5885
1.6250
2.0182
-1.4825
2.7880
-4.2135
-1.4255
0.4253
-2.3628
1.3367
5.5502
[0.8515]
[0.9503]
[0.7106]
[0.5573]
[0.0202]
[0.0988]
[0.1190]
[0.1207]
[0.1408]
[0.2942]
[0.3929]
{-3.04}***
{1.71}*
{2.84}***
{-2.66}** {138.02}*** {-42.65}*** {-11.98}***
{3.53}*** {-16.77}***
{4.54}***
{14.12}***
-9.3526
-2.4518
-0.9019
-4.1919
7.7622
-6.9009
0.8613
-0.6886
-8.4508
-5.1608
1.7401
[1.1201]
[1.0753]
[1.2355]
[0.9570]
[0.7831]
[0.0447]
[0.7384]
[0.8985]
[0.1154]
[0.1630]
[0.1183]
{-8.35}***
{-2.28}**
{-0.73}
{-4.38}***
{9.91}*** {-154.24}***
{1.17}
{-0.77} {-73.25}***
{-31.65}***
{14.71}***
-0.0899
0.0819
0.1707
-0.0307
0.4145
-0.1718
0.2427
0.1539
-0.2606
-0.0592
0.1126
[0.0199]
[0.0337]
[-5.5682]
[-0.0740]
[0.0324]
[0.0138]
[0.0187]
[5.6205]
[5.5881]
[0.0939]
[0.1077]
{-4.51}***
{2.43}**
{3.72}***
{-0.67}
{12.78}*** {-12.47}***
{13.01}***
{0.03}
{-0.05}
{-0.63}
{1.05}
11.0032
20.2703
4.7133
-2.4543
-8.3185
-9.2671
-17.5856
-2.0286
6.2900
13.4575
22.7246
[6.1816]
[3.3121]
[5.2369]
[4.4088]
[3.4969]
[2.8695]
[0.6274]
[2.5523]
[0.9446}
[1.7989]
[1.0705]
{1.78}*
{6.12}***
{0.90}
{-0.56}
{-2.38}**
{-3.23}*** {-28.03}***
{-0.79}
{6.66}***
{7.48}***
{21.23}***
322
324
324
322
320.5
323
321.5
321.5
323
322
323
36
36
36
36
36
36
36
36
36
36
36
54
54
54
54
54
54
54
54
54
54
54
22377.75***
403.15***
2448.41***
3832.23*** 11857.57*** 11390.45***
870.27*** 1892.90*** 12413.08*** 13104.99***
2117.69***
-1.251(0.21) 0.903(0.37) -0.044(0.96) -0.626(0.53)
23.433(0.99) 24.750(0.99) 19.542(0.99) 22.211(0.99)
-
*/**/*** denotes significant at 10/5/1 per cent statistical levels respectively.
Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
274
Stellenbosch University http://scholar.sun.ac.za
In Table 6.3, the results as to whether international migrant remittances received have a sizeeffect on long-run growth in SSA migrant-home countries with varying rates of economic growth
are presented133. The results show that, indeed, between 1980 and 2009, international migrant
remittances had a significant economic growth rate size-effect on SSA. This is based on the
one per cent level statistical significance of MDV reported in Table 6.3. The results further show
that although in the 1980s it was SSA countries with relatively lower growth rates that
benefitted more from the receipt of international remittances as far as the impact on economic
growth was concerned, this trend has since the 1990s switched in favour of SSA countries with
relatively higher economic growth rates. Yet, the overall positive impact of international
remittances on economic growth in SSA between 1980 and 2009 was about 1.9733 (Table
6.1), but for SSA countries with relatively higher growth rates, the impact was 0.8897 (Table
6.3).
Table 6.3: Estimated Results of Remittance-Growth Size-Effect on SSA, 1980-2009
Type of Dummy Effect
Independent Dummy
1980-1989
1990-1999
1980-2009
4.8045 (23.99)***
6.5931 (22.11)***
-0.0459 (-0.13)
0.4973 (3.56)***
0.9131 (9.50)***
0.8897 (4.51)***
Number of observations
319
322
324
1037
Number of groups
36
36
36
36
55
487.15 ***
55
13680.52 ***
55
2918.34***
445
98.70 ***
0.3896{0.6968}
[43], 30.9076
-1.1889{0.2345}
[43], 22.5484
0.6677{0.5043}
[43], 23.0167
0.3944{0.6933}
[433], 27.9407
MDV-Remittance Interactive
Instruments
Wald ( χ²₁ ₁)
Arellano-Bond Test
Sargan Test (χ² ₍₀₎ )
Source: Author‟s estimation
8.2954 (12.25)*** 6.0315 (16.62)***
2000-2009
Note: *** denotes significant at 1 per cent. Diagnostic tests in italics apply to
estimated MDV-Remittance Interactive model only.
The results in Table 6.3 reveal that between 1980 and 1989, international migrant remittances
received in SSA had no significant impact on economic growth in migrant-home SSA countries
with relatively higher growth rates. Therefore, given that, for the entire sample of 36 SSA
countries, the impact of international migrant remittances on economic growth was most
statistically significant during the 1980-1989 period (see Table 6.1), it can be concluded that it
was SSA countries with relatively lower growth rate that benefited from the positive impact of
remittances on economic growth during that era (i.e. the early years of financial liberalisation).
In other words, between 1980 and 1989, „labour-exporting‟ SSA countries with relatively higher
growth rates did not benefit from the impact of migrant remittances on economic growth.
133
The complete estimated results on the existence of size-effect are reported in Table A6.6 whilst the impact of this
size-effect on countries with growth rates above the median growth rate of the sub-region is presented in Table A6.7.
275
Stellenbosch University http://scholar.sun.ac.za
Between 1990 and 1999, remittance-receiving SSA countries with relatively higher growth rates
had a significant share amounting to 0.4973 (Table 6.3) of the entire sample impact of 0.6717
(Table 6.1). Thus, contrary to the 1980s, it is a typically „labour-exporting‟ SSA country with a
growth rate above the median growth rate of the sub-region in the 1990s that profited more
from maximising the growth-impact of remittances during that recessionary era. This implies
that migrant remittances can still stimulate economic growth during hard times such as periods
of regional economic recession. This is possible in a less import-dependent country when
remittance-recipient households become less luxurious and patronise more locally produced
goods even as the propensity of saving or directly investing remittances fall due to the higher
cost of living.
An important finding in this chapter is that, in very recent years (the 2000s), it is only
remittance-receiving SSA countries with relatively higher economic growth rates that benefitted
from maximising the positive impact of migrant remittances on economic growth. Whereas the
overall impact of international remittances on economic growth in SSA was statistically
insignificant with some potential of inhibiting growth during the 2000s (Table 6.1), in migranthome SSA countries with relatively higher growth rates, remittances impacted positively and
significantly on growth, the magnitude of which is 0.9131 (Table 6.3). One possible explanation
for this finding is that, unlike in the past, in the 2000s, migrant remittance flows to SSA
countries have become relatively more pro-cyclical and this trend emanates more from the less
altruistic workers‟ remittances than from compensation of employees (see Figure 4.2). Further,
in determining the cyclical behaviour of migrant remittances as shown in this study, it is the
growth in remittance inflows instead of the actual receipts that exhibits the factual cyclicality in
remittance inflows in SSA.
On the other hand, for the 30 sampled SSA countries, international remittances had no
contemporaneous effect on economic growth over the entire study period; and the
contemporaneous effect of remittances on economic growth was positive, negative, and zero in
the 1980s, 1990s, and the 2000s respectively (see Table A6.8.1). Consistent with the estimates
from the asynchronous model, the results in Table A6.8.2 show that the contemporaneous
growth rate size-effect of remittances has since the 1990s favoured SSA countries with
relatively higher growth rates. Even in the 1990s when the general contemporaneous impact of
international remittances on growth was -0.5616 (Table A6.8.1), remittances had a 0.9085
276
Stellenbosch University http://scholar.sun.ac.za
(Table A6.8.2) contemporaneous impact on remittance-receiving SSA countries with relatively
higher rates of economic growth. Similarly, during the 2000s, the contemporaneous impact of
remittances on economic growth in migrant-home SSA countries with relatively higher growth
rates was 1.0131 (Table A6.8.2) even when the contemporaneous impact on the entire sample
was statistically zero (A6.8.1). Besides, between 1980 and 2009, at one per cent level of
statistical significance, the contemporaneous impact of international remittances on economic
growth in remittance-receiving SSA countries with relatively higher growth rates in any
particular year was 1.1934 (Table A6.8.2), although for both categories of the sampled
countries, the impact was statistically zero (Table A6.8.1).
Evidently, the long-run impact of international migrant remittances on economic growth in SSA
is not automatic. Migrant-sending SSA countries with relatively higher rates of growth at any
point in time have a better chance of benefiting more from remittances in terms of growth
prospects. In other words, although generally during the 2000-2009 decade international
remittances may not necessarily impact significantly on economic growth in the 36 sampled
countries, they contributed to substantial growth in remittance-receiving SSA countries with
relatively higher growth rates. It is, thus, noted that between 1980 and 2009, at worst,
international remittances had no impact on economic growth in SSA. Unquestionably, a close
examination of the trend in the changing impact of remittance inflows on economic growth
shows their continuous rising positive impact on growth in high-growth „labour-exporting‟ SSA
countries since the implementation of financial liberalisation programme in the 1980s.
Overall, the findings of this study invalidate the widely held view that remittances are antigrowth in low-income countries where the propensity to consume is high. To the extent that, in
more recent years, the positive impact of remittances on long-run growth is more pronounced
in remittance-receiving SSA countries with relatively higher real per capita GDP growth rates,
this finding contradicts earlier conclusions drawn by Chami et al. (2005), Jongwanich (2007),
Barajas et al. (2009), and Karogöz (2009) that remittances are deterrent to economic growth in
developing countries. An important source of this contradiction is the fact that unlike in previous
studies, this study analysed the impact of remittances on real per capita GDP growth rate, and
not the logarithm of real per capita GDP (as in Barajas et al. (2009) and Karagöz (2009)), which
is commensurate with income status, or a group of countries classified as low-income countries
(as in Chami et al. (2005) and Jongwanich (2007)). Thus, unlike in those previous related
studies, this study took recognition of the fact that, in line with the cyclicality character of
277
Stellenbosch University http://scholar.sun.ac.za
international remittance inflows, it is the actual growth rate rather than mere income level of
remittance-receiving country that is more relevant in understanding the implications of
remittances for economic growth in an empirical context. The reason is that while developing
(or low-income) countries are the major recipients of international remittances (see Figure A6.1
and Table A6.1), economic growth can occur in any economy at one time or the other,
irrespective of its income level. For example, a developing country like Ghana recorded a real
GDP per capita growth 12.42 per cent in 2011 which is far higher than the growth rate of OECD
countries in that year even though the logarithm of Ghana‟s real per capita GDP is far lower
than that of the OECD countries for that same year. Again, these aforementioned previous
studies just like Baldé (2009) and Singh et al. (2010) that found negative or no direct impact of
remittances on economic growth failed to take cognisance of the fact that the direct effects of
international remittances on economic growth could be asynchronous and not merely
contemporaneous as proven by this study.
The findings of this study, however, confirm the results obtained from a host of previous studies
including the World Bank (2006b), Ramirez and Sharma (2008), Acosta, Baerg and
Mandelman (2009), Ahortor and Adenutsi (2009), Catrinescu et al. (2009), Mundaca (2009),
Kagochi et al. (2010), Lartey (2010), Cooray (2012), and Shera and Meyer (2013). Thus, with
the exception of Baldé (2009) and Singh et al. (2010) the findings of this study are generally
consistent with all known previous studies on SSA and other regions of the developing world.
6.6 CONCLUSIONS AND POLICY RECOMMENDATIONS
This study explored the impact of international migrant remittance inflows on long-run economic
growth in SSA over the period, 1980-2009. The analysis was carried out for the overall study
period and for each of the three past decades, 1980-1989, 1990-1999, and 2000-2009.
Consistent with the underlying objectives, both the contemporaneous and asynchronous effects
of remittances were explored. Finally, the implications of international remittance inflows for
economic growth based on the economic growth size-effects of SSA since the inception of
financial liberalisation programmes in the sub-region were analysed.
The general conclusion from this study is that when international migrant remittances are
properly measured and the economic growth model is correctly specified and instrumented to
reduce the well-known endogeneity problems associated with remittance inflows, the overall
impact of international remittances on long-run growth in SSA is significantly positive and
278
Stellenbosch University http://scholar.sun.ac.za
consistently more asynchronous. Thus, the impact of remittances on economic growth in SSA
is less contemporaneous and varies over time in consistency with the macroeconomic policy
environment during the liberalised regime. Generally, international migrant remittances had a
higher positive impact on economic growth in SSA in periods when remittances were procyclical as was the case in the 1980s and the 2000s than during the 1990s when remittances
were countercyclical to economic growth. This implies there is no significant evidence that
remittances retard economic growth in SSA. At worst, the impact of remittances on economic
growth was found to be statistically insignificant. Given the general decreasing positive impact
of remittances on economic growth in the 36 sampled countries but, at the same, the
increasing positive impact of remittances on the sampled countries with relatively higher growth
rates, it is important to note that although the pursuit of financial liberalisation programme in
SSA is indispensable to attracting more official remittances into the sub-region, this is not a
sufficient condition for remittances to have a consistent positive impact on economic growth in
migrant remittance-receiving SSA countries. It is a must for remittance-receiving SSA countries
to implement other pro-growth strategies in order to optimise the remittance-growth potential.
Conceivably, the most persuasive evidence revealed by this study is the fact that, as far as the
implications for long-run economic growth are concerned, international migrant remittances
have considerable size-effects in SSA countries. The results have shown that in remittancereceiving SSA countries with relatively higher growth rates at any point in time, the impact of
international remittances on economic growth was both statistically and economically robust,
and this positive impact has been increasing steadily over time. The results imply that, with
reference to SSA, the automatic economic growth and macroeconomic stabilisation effects of
international remittances that existed in the past, have gradually faded away. Consequently, in
contemporary SSA, it is only migrant remittance-receiving countries with relatively higher
growth rates that have the prospects of maximising the potential contribution of international
migrant remittances on economic growth in the long run, provided that these countries can
mobilise substantial remittances through official channels. Therefore, in spite of the broad
positive impact of international remittances on economic growth in SSA, it can still be
economically suicidal for policy makers in SSA to live under unguided optimism of remittances
serving as an automatic macroeconomic stabiliser and a growth stimulant in recipient countries
irrespective of macroeconomic fundamentals.
279
Stellenbosch University http://scholar.sun.ac.za
The findings from this study imply that even though it is imperative for remittance-receiving
SSA countries to implement well-coordinated and effective policies geared towards attracting
optimal remittances through official routes to stimulate economic growth, this may not be
sufficient unless policy makers support these efforts with the implementation of other progrowth policies. In view of this, in addition to the remittance attraction strategies espoused in
the preceding chapters, the following policy recommendations are made towards maximizing
the economic growth potentials of international remittance inflows in SSA.
i.
Strategic policy measures must be put in place to stabilise the macroeconomy, and
create an investment climate that encourages the private sector to invest more in high
profit-yielding projects. Under this condition, more international migrants are likely to
invest directly in profitable projects at home while, at the same time, recipients of
remittances may now find it more lucrative to save and invest a higher proportion of the
remittances received, rather than consume these funds instantly. In this regard,
monetary and fiscal policy coordination aimed at low and stable rates of inflation,
exchange rate stability, and financial policy transparency are highly recommended.
ii.
Complementary policies such as stronger institutions for good governance and reduced
public sector corruption which can reduce „wasteful‟ government spending are required.
These policies are imperative because excessive government spending can crowd-out
the private sector and subsequently retard economic growth whilst government
spending on critical social infrastructure and services such as the construction of roads,
extension of electricity and telecommunication services can crowd-in the private sector
to boost investment in the productive sectors of the economy. This means that it is
remittance-receiving SSA countries that can boost the confidence of international
migrants through good governance, stronger institutions and reasonable government
spending that are likely to maximise the growth potentials of international remittances
received.
iii.
There is the need for trade and industrial policies that are aimed at reducing overdependency on imports of basic necessities in the form of food and clothing that can be
locally produced. This also requires modernisation of production techniques to uplift the
quality of locally produced goods. This is because the spending of international
remittances on imported consumables rather than on locally produced goods tends to
280
Stellenbosch University http://scholar.sun.ac.za
reduce the growth potential of remittances in migrant-home countries. Specific policies
that can be implemented for this purpose include the introduction of special subsidies or
tax holidays for local industries; the advancement of pro-competition policies; privatepublic sector ownership of strategic capital-intensive industries; and the establishment
of venture capital funds to provide easy access to medium-term enterprise credit. Other
relevant policies include the establishment of special funds and institutions for the
conduct of industrial innovatory research towards a permanent transformation and
revitalisation of local industries and SMEs as well as regionalisation and globalisation.
iv.
Policies on financial development towards higher efficiency of financial institutions when
playing their intermediation role can stimulate long-run economic growth through the
saving rate and the social marginal productivity of invested international remittances in
SSA. Therefore, it is important to provide an enabling environment through improved
information symmetry and financial market deregulation that will give financial
institutions the confidence and the incentive to switch private sector credit allocation
from the consuming household sector to the investing business sector.
v.
Finally, to maximize the growth potentials of international remittance inflows in SSA, it is
imperative for policy makers to put measures in place to make use of the available
human capital and to mitigate the tendency of international remittances to become an
incentive for further migration of skilled labour. This is important because remittances
received in youthful and aged populated countries are more likely to be spent on
imported consumables when the remittance-recipient countries are not capable of
producing adequate basic necessities locally. Excessive spending of international
remittances on imported consumables will increase the dependency of the migrantsending SSA country and worsen its macroeconomic vulnerability to adverse external
shocks.
281
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 6
Figure A6.1: Global Outlook of Migrant Remittances Received and Paid, 1980-2009
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Paid
Received
World
Net Received
Developing Countries
Source: Author based on WDI
Table A6.1:
Global Inflows of Migrant Remittances and Major Forms of External Capital (as of 2009)
Remittances
(US$ billions)
Remittances as a percentage of:
GDP
Export of Goods
FDI
ODA
World
416.12
0.75
3.39
35.76
326.30
Developing Economies
307.65
1.88
0.08
0.86
2.42
East Asia and Pacific
85.79
1.36
4.91
84.58
834.64
Europe and Central Asia
36.02
1.41
5.38
41.81
444.67
Latin America and the Caribbean
56.59
1.43
8.12
73.85
621.62
Middle East and North Africa
33.44
3.22
n/a
120.44
246.10
South Asia
75.06
4.45
35.63
195.40
523.72
Sub-Saharan Africa
20.75
2.49
8.03
71.31
46.62
Source: Author based on WDI (April 2011 Edition)
282
Stellenbosch University http://scholar.sun.ac.za
Table A6.2: Data Description, Measurement and Sources
VARIABLE
NOTATION
Dependent Variable
growth
Economic growth
DESCRIPTION, MEASUREMENT AND MAIN SOURCES
Annual percentage change rate in real per capita GDP (in constant US$).
Source: WDI.
Explanatory Variables*
+/Investment
in lnINV
physical assets
+/lnGXP
Government
expenditure
+/Openness
to lnOPN
international trade
+/Remittances
per lnREMPC_1
capita lag one
Gross fixed capital formation as a ratio to nominal GDP. Source: WDI and
WEO.
Central government final consumption expenditure as a percentage of
nominal GDP. Source: Author based on WDI, IFS and WEO.
Sum of exports and imports as a percentage of nominal GDP. Source:
Author‟s computation based on WDI and WEO.
Lag one of the sum of workers‟ remittances and compensation of
employees as ratio of population. Source: WDI, BoPS, MRF-2011 CDROMs and e-databases and estimates based on country-specific
information obtained from country-desk officials of the IMF and the World
Bank.
Human
capital lnHCA
Net enrolment ratio of children of official school age based on International
accumulation
Standard Classification of Education 1997 who enrolled in post-primary
school relative to the population of the corresponding official school age.
Source: WDI.
+/Rate of growth in annual average of consumer price index. Source: WDI
INF
Inflation rate
and author based on IFS and WEO.
+/Broad money to lnM2/GDP
Sum of currency outside banks, demand deposit other than those of the
GDP ratio
central government, and time, savings and foreign currency deposits of
resident sector other than the central bank as ratio of GDP. Source: WDI
and author based on IFS and WEO.
/Domestic credit to lnPSC+
Total domestic credit to the private sector by the financial system as a ratio
of nominal GDP. Source: WDI, Africa Development Indicators (ADI), and
private sector
the Central Bank website of selected sampled countries.
+/Foreign
direct FDI
Net inflows of investment, being the sum of equity capital, reinvestment of
investment
profits, other long-term capital, and short-term capital, to acquire long-term
management interest in an enterprise operating in an economy other than
that of the investor, expressed as a percentage of nominal GDP. Source:
Author based on WDI and WEO.
Source: Author based on April 2011 editions of WDI, ADI, WEO and IFS were primarily used. Note: The a priori sign
+/is indicated by
by the notation column of each variable. Only final explanatory variables are included in Table
A6.2, thus excluding control variables that were dropped from the final estimated model. ln preceding a variable
means that variable is in its natural logarithmic value.
283
Stellenbosch University http://scholar.sun.ac.za
Table A6.3: Descriptive Statistics of Dataset
VARIABLE
Obs
Mean
Std. Dev.
Min
Max
growth
1080
0.9829
5.4418
-52.2019
37.8386
lnINV_1
1044
2.9077
0.4786
-3.1397
4.3398
lnGXP
1080
2.6623
0.4265
0.2683
3.9985
lnOPN
1080
4.1373
0.5579
1.8954
5.5232
lnREMPC_1 1044
1.4404
2.2497
-6.1374
5.7400
lnHCA
1080
3.1584
0.7503
0.8774
4.7694
INF
lnM2GDP
1078
13.5133
22.2021
-100.0000
200.0260
1080
3.1085
0.6503
-2.7186
4.7652
lnPSC
1080
2.6018
0.8090
-1.2469
5.1301
FDI
1074
2.3102
4.3904
-28.6243
36.1138
Source: Author‟s estimation
Table A6.4: Bivariate Correlation of Variables
growth lnINV_1 lnGXP lnOPN lnREMPC_1
lnHCA
INF lnM2GDP
lnPSC
growth
1.0000
lnINV_1
0.1258
1.0000
lnGXP
0.0083
0.3226
1.0000
lnOPN
0.0905
0.4137
0.3952
1.0000
lnREMPC_1
0.1826
0.2796
0.2652
0.4579
1.0000
lnHCA
0.1060
0.2136
0.2512
0.5300
0.3789
1.0000
INF
lnM2GDP
0.0022
-0.1641
-0.2470
-0.2696
-0.2756
-0.2804
1.0000
0.0159
0.2035
0.2370
0.3143
0.3938
0.4615
-0.1674
1.0000
lnPSC
-0.0092
0.3103
0.3254
0.3435
0.3849
0.3827
-0.3039
0.5291
1.0000
FDI
0.1340
0.2765
0.0994
0.3633
0.2601
0.2667
-0.1180
0.1764
0.0489
Source: Author‟s estimation
284
FDI
1.0000
Stellenbosch University http://scholar.sun.ac.za
Table A6.5: Results of Panel Unit Root Tests
VARIABLES
PANEL UNIT ROOT TEST STATISTICS
Fisher P-P chi-square
Hadri HC z-stat
LLC Adjusted t-stat
Conclusion
At Level
At Level
At Level
growth
59.1524***
3.7901***
I(0)
{0.0000}
[0.0001]
lnINV
5.4517***
32.2079***
I(0)
{0.0000}
[0.0000]
lnGXP
5.8323***
33.3928***
I(0)
{0.0000}
[0.0000]
lnOPN
9.0605***
23.7910***
I(0)
{0.0000}
[0.0000]
lnREMPC
2.7415***
37.9585***
I(0)
{0.0031}
[0.0000]
lnHCA
-0.7845
44.0794***
-4.4781***
I(0)
{0.7836}
[0.0000]
(0.000)
INF
n/a
n/a
25.8682***
I(0)
n/a
n/a
{0.0000}
lnM2/GDP
0.2765
36.2907***
-3.5126
I(0)
{0.3911}
[0.0000]
(0.0002)***
lnPSC
0.6143
38.0580***
-1.5210*
I(0)
{0.2695}
[0.0000]
(0.0641)
FDI
n/a
n/a
18.8927***
I(0)
n/a
n/a
{0.0000}
Source: Author‟s computations
Note: Figures in brackets { }, [ ], and ( ) are probability values of chi-square, z*** *
statistics, and t-statistics respectively.
( ) significant at 1%(10%) level
respectively. Constant and trend included except LLC. Where applicable (in
Fisher P-P and LLC tests), optimal lag 2 was included. n/a means not applicable
due to missing data.
285
Stellenbosch University http://scholar.sun.ac.za
Table A6.6:
Estimated Impact of Median-Dummy Variable (MDV) on Growth in SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-89
1990-99
2000-09
1980-2009
Initial economic growth (growth_1)
-0.1118
(-5.01)***
-0.2259
(-18.79)***
0.1031
(3.98)***
-0.0573
(-1.24)
Investment (lnINV_1)
-5.8496
(-6.55)***
-3.3227
(-0.53)
-1.8533
(-1.11)
0.3504
(0.50)
Government expenditure (lnGXP)
0.3866
(0.28)
3.5794
(1.84)*
-0.1456
(-0.31)
1.1172
(0.52)
Trade openness (lnOPN)
2.2973
(3.37)***
-4.3048
(-3.17)***
-0.2780
(-0.34)
0.4241
(0.20)
Migrant Remittances (lnREMPC_1)
0.0609
(0.34)
0.8970
(2.72)***
-0.0551
(-0.15)
-0.1670
(-0.27)
Human capital accumulation (lnHCA)
1.5868
(1.91)**
2.2190
(1.73)*
-0.5793
(-0.55)
1.6318
(0.71)
Rate of inflation (INF)
0.0001
(0.01)
0.0927
(7.40)***
-0.0925
(-6.44)***
0.0211
(4.15)***
Bank credit to private sector (lnPSC)
2.2357
(1.88)*
-2.3794
(-1.44)
0.1257
(0.15)
-1.8981
(-0.98)
Broad money to GDP ratio (M2/GDP)
-1.7864
(-3.16)***
-6.0046
(-2.79)***
-3.1372
(-4.26)***
-1.2747
(-2.05)**
Foreign direct investment (FDI)
-0.0042
(-0.12)
0.0681
(2.38)**
0.0135
(0.45)
-0.0416
(-1.68)*
Median Dummy (MDV)
8.2954
(12.25)***
6.0315
(16.62)***
4.8045
(23.99)***
6.5931
(22.11)***
Constant term
-2.7333
(-0.44)
22.2861
(4.02)***
19.3791
(6.67)***
-4.1941
(-0.72)
Number of observations
319
322
324
1037
Number of groups (N)
36
36
36
36
Number of instruments
2
Wald [11],
55
55
55
445
1285.81***
6103.88***
6057.29***
1026.37***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
0.7679{0.443}
1.2097{0.226}
1.9954{0.046}**
0.7918{0.429}
Sargan test of over-identifying restrictions:
[2],
Source: Author‟s estimation
[43],
17.7359
[43],
21.2452
[43],
27.0407
[433],
28.0675
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
286
Stellenbosch University http://scholar.sun.ac.za
Table A6.7:
Median Dummy Variable-Remittances Interactive Effect on Economic Growth in SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-89
1990-99
2000-09
1980-2009
Initial economic growth (growth_1)
-0.1135
(-3.50)***
-0.2181
(-25.35)***
0.1252
(4.27)***
-0.0313
(-0.85)
Investment (lnINV_1)
-5.9107
(-3.73)***
2.1314
(3.57)***
-2.0687
(-2.82)***
1.2177
(2.69)***
Government expenditure (lnGXP)
-1.5255
(-0.73)
-0.4817
(-0.22)
-0.0918
(-0.26)
1.1936
(0.50)
Trade openness (lnOPN)
5.5104
(3.00)***
0.0932
(0.07)
1.2389
(1.11)
4.4187
(1.65)*
Migrant Remittances (lnREMPC_1)
0.8088
(2.35)**
0.1352
(0.57)
-0.5921
(-2.03)**
0.3572
(0.49)
Human capital accumulation (lnHCA)
-1.5177
(-1.07)
4.8085
(3.47)***
-1.5728
(-1.28)
-0.2881
(-0.18)
Rate of inflation (INF)
-0.0022
(-0.30)
0.0884
(6.24)***
-0.1001
(-6.81)***
0.0123
(2.54)**
Bank credit to private sector (lnPSC)
1.3965
(1.02)
-1.5441
(-1.61)*
0.8751
(0.96)
-1.9582
(-1.28)
Broad money to GDP ratio (M2/GDP)
-1.5213
(-3.09)***
-12.3836
(-8.03)***
-2.9902
(-3.52)***
-1.1578
(-0.50)
Foreign direct investment (FDI)
0.2657
(3.03)***
-0.0808
(-2.72)***
0.0258
(1.27)
-0.0165
(-0.45)
MDV-Remittance Interactive Effect
-0.0459
(-0.13)
0.4973
(3.56)***
0.9131
(9.50)***
0.8897
(4.51)***
Constant term
3.8045
(0.76)
20.9344
(3.91)***
16.9125
(4.25)***
-15.6067
(-1.78)*
Number of observations
319
322
324
1037
Number of groups (N)
36
36
36
36
Number of instruments
2
Wald [11],
55
55
55
445
487.15***
13680.52***
2918.34***
98.70***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
0.3896{0.697}
-1.1889{0.235}
0.6677{0.504}
0.3944{0.693}
Sargan test of over-identifying restrictions:
[2],
Source: Author‟s estimation
[43],
30.9076
[43],
22.5484
[43],
23.0167
[433],
27.9407
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
287
Stellenbosch University http://scholar.sun.ac.za
Table A6.8.1: The Contemporaneous Impact of Remittances on Growth in SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-89
1990-99
2000-09
1980-2009
Initial economic growth (growth_1)
-0.0874
(-3.95)***
-0.1901
(-15.16)***
0.1150
(3.89)***
-0.0720
(-1.25)
Investment (lnINV_1)
-4.2215
(-3.11)***
1.7371
(3.40)***
-2.6489
(-2.01)**
1.3823
(2.49)**
Government expenditure (lnGXP)
-1.4266
(-1.10)
3.0085
(1.65)*
-0.4708
(-0.88)
-1.9675
(-0.54)
Trade openness (lnOPN)
5.6675
(3.57)***
0.7993
(0.81)
1.4759
(1.05)
3.5264
(1.22)
Migrant Remittances (lnREMPC)
1.1778
(4.48)***
-0.5616
(-2.65)***
0.0559
(0.19)
-0.3161
(-0.44)
Human capital accumulation (lnHCA)
-1.8469
(-1.79)*
8.0193
(4.63)***
-2.2491
(-2.39)**
3.0523
(0.91)
Rate of inflation (INF)
-0.0037
(-0.66)
0.0797
(6.13)***
-0.0816
(-4.41)***
0.0057
(1.16)
Bank credit to private sector (lnPSC)
0.5825
(0.60)
-3.0861
(-3.39)***
1.3957
(1.16)
-2.2319
(-2.30)**
Broad money to GDP ratio (lnM2/GDP)
-1.4102
(-2.49)**
-10.1077
(-4.41)***
-2.6217
(-2.31)**
-5.2225
(-2.62)***
Foreign direct investment (FDI)
0.2780
(8.08)***
-0.0841
(-3.99)***
0.0597
(1.68)*
-0.0197
(-0.72)
Constant term
0.3041
(0.09)
-1.3404
(-0.16)
17.2455
(4.62)***
0.5389
(0.05)
Number of observations
319
322
324
1037
Number of groups (N)
36
36
36
36
Number of instruments
2
Wald [10],
54
54
54
444
408.59***
14059.59***
543.44***
54.99***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
0.4235{0.672}
-0.9508{0.342}
0.7934{0.428}
-0.1369{0.891}
Sargan test of over-identifying restrictions:
[2],
Source: Author‟s estimation
[43],
24.3549
[43],
22.0945
[43],
22.4493
[433],
23.9006
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
288
Stellenbosch University http://scholar.sun.ac.za
Table A6.8.2: Contemporaneous Size-Effect of Remittances on Growth in SSA, 1980-2009
Type of Dummy Effect
1980-1989
1990-1999
2000-2009
1980-2009
Independent MDV
9.2427 (16.14)***
MDV-Remittance Interactive
0.5061 (1.78)*
6.4127 (18.96)***
4.8690 (4.15)***
6.1316 (17.91)***
0.9085 (5.46)***
1.0131 (9.76)***
1.1934 (7.28)***
Number of observations
319
322
324
1037
Number of groups
36
36
36
36
55
681.36***
55
8901.23***
55
1551.25***
445
169.96***
0.4941{0.6212}
-0.8531{0.3936}
0.6988{0.4847}
0.2100{0.8336}
₍₄₃₎, 25.1545
₍₄₃₎, 21.0542
₍₄₃₎, 24.0050
₍₄₃₃₎, 19.4533
Instruments
Wald (χ²₁₁)
Arellano-Bond Test
Sargan Test (χ²₍₀₎)
Source: Author‟s estimation
Note: *(***) denotes significant at 10(1) per cent. Diagnostic tests in italics apply
to estimated MDV-Remittance Interactive model only.
289
Stellenbosch University http://scholar.sun.ac.za
Table A6.9:
Robustness Test Results of Contemporaneous Investment and Remittances on Growth in SSA
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-89
1990-99
2000-09
1980-2009
Initial economic growth (growth_1)
-0.0529
(-1.79)*
-0.1923
(-11.86)***
0.1351
(4.36)***
0.0275
(0.64)
Investment (lnINV)
2.2876
(1.16)
0.3664
(0.52)
0.8885
(0.67)
-0.6420
(-0.40)
Government expenditure (lnGXP)
-3.0070
(-1.52)
2.0886
(1.01)
-0.3621
(-0.66)
-4.0264
(-1.76)*
Trade openness (lnOPN)
3.8383
(1.48)
-0.7285
(-0.62)
0.8229
(0.72)
7.7684
(2.47)**
Migrant Remittances (lnREMPC)
1.1634
(3.14)***
-0.5997
(-3.29)***
-0.1911
(-0.94)
-0.2465
(-0.30)
Human capital accumulation (lnHCA)
2.3937
(1.54)
8.2896
(7.06)***
-2.5539
(-2.26)**
-1.4554
(-0.57)
Rate of inflation (INF)
-0.0086
(-1.32)
0.0705
(5.58)***
-0.0981
(-5.49)***
0.0047
(1.19)
Bank credit to private sector (lnPSC)
-2.8983
(-2.49)**
-3.5998
(-2.25)**
1.1131
(0.92)
-2.2459
(-1.65)*
Broad money to GDP ratio (lnM2/GDP)
-1.7036
(-4.61)***
-9.8037
(-4.43)***
-3.2262
(-3.11)**
-3.2760
(-2.79)***
Foreign direct investment (FDI)
0.2175
(4.74)***
-0.0625
(-2.66)***
0.0717
(1.96)*
-0.0612
(-1.74)*
Constant term
-4.0003
(-0.51)
10.1952
(1.05)
12.6526
(2.90)***
2.5215
(0.24)
Number of observations
319
322
324
1037
Number of groups (N)
36
36
36
36
Number of instruments
2
Wald [10],
54
54
54
444
732.83***
13071.66***
996.58***
190.85***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
0.2166{0.829}
-1.1533{0.249}
0.8176{0.414}
0.3800{0.704}
Sargan test of over-identifying restrictions:
[2],
Source: Author‟s estimation
[43],
22.0898
[43],
23.4782
[43],
24.5138
[433],
26.4248
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
290
Stellenbosch University http://scholar.sun.ac.za
Table A6.10:
Static Panel-Data Modelling of Remittances on Economic Growth in SSA, 1980-2009
Fixed
Effects (FE)
Random
Effects (RE)
Robust FE
Investment (lnINV_1)
1.8083
(3.93)***
1.8498
(4.51)***
1.8081
(1.65)
1.8489
(1.93)**
Government expenditure (lnGXP)
-0.7649
(-1.25)
-0.7251
(-1.56)
-0.7649
(-1.06)
-0.7251
(-1.35)
Trade openness (lnOPN)
1.6326
(2.53)**
-0.3114
(-0.71)
1.6326
(1.69)*
-0.3114
(-0.53)
Migrant Remittances (lnREMPC_1)
0.4521
(2.91)***
0.4718
(4.87)***
0.4521
(2.04)**
0.4718
(5.03)***
Human capital accumulation (lnHCA)
1.3334
(2.49)**
0.7129
(2.28)**
1.3334
(2.34)**
0.7129
(2.34)**
Rate of inflation (INF)
0.0023
(0.26)
0.0058
(0.71)
0.0023
(0.12)
0.0058
(0.41)
Bank credit to private sector (lnPSC)
-0.9280
(-2.46)**
-0.6566
(-2.38)**
-0.9280
(-2.35)**
-0.6566
(-2.09)**
Broad money to GDP ratio (lnM2/GDP)
-1.0436
(-2.77)***
-0.5993
(-1.79)*
-1.0436
(-4.41)***
-0.5993
(-2.44)**
Foreign direct investment (FDI)
0.0342
(0.74)
0.0655
(1.52)
0.0342
(0.70)
0.0655
(1.27)
Constant term
-8.3100
(-2.51)***
-0.7598
(-0.39)
-8.3100
(-2.47)**
-0.7598
(-0.39)
Number of observations
1037
1037
1037
1037
Number of groups (N)
36
36
36
36
0.0531
0.0740
0.0531
0.0740
8.23{0.000}***
123.26{0.000}***
Overall R
2
F-statistics
7.74{0.000}***
25.49{0.003}***
Hausman_FE
25.00{0.003}***
n/a
Breusch-Pagan (B-P) statistics
Source: Author‟s estimation
n/a
++
Robust Random
GLS (RE)
n/a
n/a
n/a
n/a
2.09{0.074}*
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
robust z-statistics in ( ), probabilities in { }, n/a denotes not available or required
++
most efficient and reliable results based on Hausman test and B-P statistics
291
Stellenbosch University http://scholar.sun.ac.za
Table A6.11: Summary of Empirical Studies on the Impact of Remittances on Economic Growth
Author(s), Year
Case Study
Solimano (2003)
LAC countries
with
countrybased evidence
from Bolivia and
Ecuador
Chami
(2005)
et
al.
IMF (2005)
World
(2006a)
Bank
Study
Period
1987-2002
Model & Estimation
Method
Single equation country
specific modelling by
OLS
113 developing
countries
1970-1998
101 developing
countries
1970-2000
(nonoverlapping
5-year
annual
average)
1991-2005
Panel Fixed Effects
(FE)
and
panel
Random Effects (RE)
instrumental
variable
modelling
Pooled single equation
bivariate
model
estimated by OLS
67
developing
countries
Unspecified
Jongwanich
(2007)
17
developing
Asia-Pacific
countries
1993-2003
Panel Fixed Effects
model and dynamic
model by system GMM
Fayissa
and
Nsiah (2008)
37
African
countries
1980-2004
Simple log-log linear
using dynamic paneldata model following
GMM. Robustness FE
and RE model
Variables Included
Key Finding(s)
Dependent: GDP per capita growth rate
Explanatory: Logarithm of GDP, investment proxied by
gross fixed capital formation (GFCF)/GDP, logarithm of
terms of trade (ToT) change lag 1, government
consumption/GDP,
logarithm
of
ratio
of
remittances/GDP lag 1.
Dependent: Annual growth in real GDP per capita
Explanatory: Remittances (WR+CE)/GDP, changes in
remittances/GDP ratio. Control variables: Investment
proxied by (GFCF)/GDP, inflation, net private capital
inflows/GDP, regional dummies
Dependent: Real GDP per capita
Explanatory: Remittances (WR+CE+MT)/GDP ratio
Remittances promote long-run growth in
both Bolivia and Ecuador
Dependent: Logarithm of real GDP per capita
Explanatory: Logarithm of: Initial GDP per capita,
remittances
(WR+CE)/GDP,
secondary
school
enrolment ratio capturing human capital, private sector
credit/GDP ratio, political risk, openness, inflation, real
exchange
rate
overvaluation,
government
consumption/GDP, time dummies
Dependent: Annual growth of real GDP per capita
Explanatory: Initial real GDP per capita growth,
logarithm of remittances (WR+CE+MT), human capital
development, logarithm of investment (GFCF)/GDP at
time
and
logarithm
of
government
t
t-1,
consumption/GDP, logarithm of openness, CPI-based
inflation.
Dependent: Natural logarithm of real GDP per capita
Explanatory: (Natural logarithm of) remittances
(WR+CE) per capita, GFCF/GDP, secondary school
enrolment, foreign aid (AID/GDP), foreign direct
investment (FDI/GDP), terms of trade, political rights,
initial level of real GDP per capita
Consistent positive relationship between
remittances and economic growth, both
when investment was present and absent
from the model. But in the absence of
investment,
the
contribution
of
remittances to economic growth became
small.
Remittances have direct negative impact
on economic growth, but it impacts
positively on growth indirectly through
investment in physical assets and human
capital accumulation.
292
Remittances are countercyclical in nature
and with a negative impact on economic
growth.
Impact of remittances on long-run growth
is not statistically significant.
Remittances promote growth in countries
where
the
financial
sector
is
underdeveloped as they serves as an
alternative source of investment finance
and
helping
overcome
liquidity
constraints
Stellenbosch University http://scholar.sun.ac.za
Le (2008)
49
selected
countries
1970-2005
(5-year
period)
Dynamic
panel-data
model. Single equation
OLS with pooled data
and Panel Fixed Effects
2-Stage Least Squares
(FE2SLS) instrumental
variable (IV) models
robustness check
Ramirez
and
Sharma (2008)
23
American
countries
1990-2005
Panel Unit Root and
Panel
Co-integration
test
using
FullyModified
OLS
approach.
Latin
Two main estimations:
With
and
without
financial development
Dynamic
panel-data
models and system
GMM
Ziesemer (2008)
50
poor
developing
countries
(i.e.
countries
with GDP per
capita less than
US$1200
(in
2000 prices)
1960-2003
Acosta,
Baerg
and Mandelman
(2009)
10
countries
Longitudinal
survey data
(average
period: 20002004)
Dynamic
panel-data
model using GMM
Ahortor
and
Adenutsi (2009)
31 small-open
developing
countries from
LAC (16) and
SSA (15)
1986-2006
Dynamic
panel-data
model using system
GMM
LAC
Dependent: Logarithm of real GDP per capita 5-year
annual average. Also, average annual 5-year real GDP
per capita growth
Explanatory: Initial real GDP per capita growth, trade
openness measured as exports (X) plus imports (M) as
a ratio of GDP, remittances (WR+CE+MT) as a ratio of
GDP, quality institutions from polity IV project, and
vector of other variables including religious affiliation,
and education.
Dependent: Changes in logarithm of real GDP per
capita
Explanatory: Logarithm of remittances/GDP ratio, and a
set of control variables that include fixed capital
formation/GDP, openness, labour force, M2/GDP and
domestic credit/GDP.
Institutions foster growth but remittances
hamper economic growth.
Dependent: Logarithm of GDP per capita
Explanatory: (Logarithm of) GDPt-5, literacy rate,
ODA/GDP, logGFCF/GDP(-x), interest rate, remittances
(WR)/GDP, labour force, world GDP proxied by GDP of
USA
Remittances enhance savings, public
expenditure on education and growth, but
reduce tax revenues and emigration.
Taking into account direct and indirect
effects of remittances on levels and
growth rates of GDP per capita, it was
found that remittances impact positively
on economic growth, investment and
literacy rates.
Remittances promote long-run growth.
Dependent: Logarithm of per capita income
Explanatory: Logarithm of initial per capita income,
remittances (WR+CE+MT) but with some exceptions
where two or less components are used. Control
variables include average years of secondary school
education for male population, and for the female
population, price of investment goods relative to that of
the USA. All explanatory variables used are of one lag.
Dependent: Natural logarithm of real GDP per capita
Explanatory: Initial growth, remittances (WR+CE+MT+
other current transfers)/GDP, investment (GFCF)/GDP,
human capital measured as secondary school
enrolment, openness (X+M)/GDP, logarithm of CPI as
293
With financial development, remittances
have higher positive impact on growth
than without the presence of financial
development. In both cases (i.e. with or
without financial development), the
impact of remittances on upper-middle
income group is more positive than it is
the case of lower income group.
Generally remittances have positive
impact on long-run growth in small-open
developing countries. The impact is more
robust
in
LAC
than
SSA.
Contemporaneously,
remittances
Stellenbosch University http://scholar.sun.ac.za
proxy for inflation, government spending/GDP
Baldé (2009)
Barajas
(2009)
29
countries
SSA
1980-2004
(3-year
moving
average
data)
Unbalanced
panel
Panel
2SLS
IV
estimation technique
al.
84
developing
and
emerging
countries
receiving
remittances
1970-2004
(5-year
period
average)
Pooled OLS IV and FEIV
Catrinescu et al.
(2009)
162 developing
countries
1970-2003
(unbalanced
panel data)
Dynamic
panel-data
modelling in the context
of GMM
Garcia-Fuentes
and
Kennedy
(2009)
14
countries
1975-2000
(overlapping
5-year
moving
average)
Panel Random Effects
(RE) 2SLS with pooled
OLS and RE for
robustness
Giuliano
Ruiz-Arranz
(2009)
100 developing
countries
1975-2002
(5-year
annual
average
data)
System
GMM
with
Pooled OLS and FE
models for robustness
test
et
and
LAC
Dependent: Natural logarithm of average of 3-year GDP
per capita
Explanatory: Natural logarithm of initial GDP per capita,
remittances (WR+CE+MT)/GDP, ODA/GDP, population
growth rate, trade openness (X+M)/GDP, secondary
school enrolment for human capital formation,
government consumption/GDP, inflation, investment
(GFCF)/GDP and political stability
Dependent: Logarithm of real GDP per capita
Explanatory: (Logarithm of) initial growth, remittances
2
(WR)/GDP denoted as REMGDP, REMGDP ,
REMGDP*M2/GDP interaction, and average growth rate
in top-20 trading partners. Control variables: logarithm
of trade/GDP, FDI/GDP, fiscal balance/GDP, population
growth rate, and M2/GDP; political risk
Dependent: Logarithm of real GDP per capita
Explanatory: (Logarithm of) real GDP per capita lag 1,
remittances (WR+CE)/GDP with control variables as
gross
capital
formation/GDP,
gross
domestic
savings/GDP, net private capital inflows/GDP, inflation
rate and regional dummies
Dependent: Growth of output per worker
Explanatory: Human capital stock (HCAP), human
capital growth, remittances (WR+CE+MT)/GDP defined
as (REMGDP), HCAP*REMGDP interaction, growth
rates of HCAP and physical capital plus control
variables including time dummies, investment/GDP,
government consumption/GDP, and inflation
Dependent: Logarithm of per capita GDP
Explanatory: Logarithm of Initial level of GDP per
capita, remittances (measured according to countryspecific reporting data)/GDP ratio defined as
(REMGDP), financial development proxied by M2/GDP,
domestic credit/GDP), bank deposits/GDP, and bank
loans/GDP. Control variables include trade openness,
294
positively affect growth with higher
impact in LAC. In dynamic terms,
remittances retard growth, but with
overall positive impact
Remittances do not have a direct positive
impact on economic growth
At best, remittances have no effect on
economic growth in the long run,
probably because poor institutions do not
make remittances to be channelled to
growth-enhancing projects
Remittances have a weak impact positive
impact on long-run growth, but the
positive impact improves in the presence
of sound macroeconomic policies and
institutions.
Remittances positively impact on human
capital development but directly deter
economic growth. Also, there is
significant positive effect of the
interaction between human capital and
economic growth. Thus, the impact of
remittances on growth is dependent upon
the level of human capital development
Remittances impact positively on longrun growth in countries with less
developed financial systems by serving
as an alternative finance of investment
and
entrepreneurial
activities
to
overcome credit constraints. In the
absence of financial development,
Stellenbosch University http://scholar.sun.ac.za
human capital growth rate denoted by secondary
school enrolment, government fiscal balance/GDP,
investment/GDP rate, inflation, and population growth
rate
Jayaraman et al.
(2009)
Samoa
1981-2008
Karagöz (2009)
Turkey
1970-2005
Mundaca (2009)
25
countries.
Rao and Hassan
(2009)
LAC
40
highest
remittancerecipients as of
1970-2002
1970-2006
(unbalanced
panel)
Single
equation
Autoregressive
Distributed Lag (ARDL)
bounds testing model
Single equation double
logarithmic model using
OLS
estimation
procedure
Dynamic panel data
following first-difference
GMM.
Full
sample
as
estimated
alongside
three
sub-samples
categorised as: (i) large
recipients relative to
GDP; (ii) low, lower
middle & upper middle
income but with large
receipts of remittances
but poorest; and (iii)
Central
American
countries.
Dynamic
panel-data
modelling
following
system GMM
Dependent: Logarithm of real GDP
Explanatory:
(Logarithm
of)
remittances
(WR+CE+MT)/GDP,
private
sector
credit/GDP,
exports/GDP
Dependent: Logarithm of GDP per capita
Explanatory: (Logarithm of) initial GDP per capita,
remittances (all private transfers implying WR+CE+MT+
other current transfers)/GDP, FDI/GDP, exports/GDP
Dependent: Annual growth of output per capita
Explanatory: Initial output growth rate, logarithm of
investment proxied by (GFCF) per capita, remittances
measured as WR/GDP ratio at time t-1, indicators of
financial development at time t-1 (here main emphasis is
on bank private sector credit (PSC)/GDP. Human
capital development measured as literacy rate among
adults aged 15 years and above.
remittances alone do not have a positive
impact on economic growth. Remittances
have a positive impact on growth at both
the median and the mean level of
financial development, but their impact
becomes zero and eventually turns
negative in countries with developed
th
financial systems above the 75 per
centile of the sample distribution
Remittances have a direct significant
positive impact on economic growth
Remittances impact negatively on
economic growth whilst exports and
domestic
investment
are
positive
determinants of economic growth.
The long-run impact of remittances on
economic growth is positive and
significant in all four groups (full sample,
and three sub-samples) analysed.
Expansion of financial services to citizens
of remittance recipient countries should
lead to better use of remittances and
boost long-run growth.
Initial estimation involved on three explanatory
variables: investment, initial growth and remittances.
Dependent: Growth of GDP per worker proxied by GDP
divided by labour force
Explanatory: Financial development proxied by
295
Remittances have positive growth effects
although the impact is small.
Stellenbosch University http://scholar.sun.ac.za
2007 of which 9
134
are from SSA
Ziesemer (2009)
96 countries that
received
remittances of at
least US$1 in
2003
1960-2003
Dynamic
panel-data
model using GMM
Fayissa
and
Nsiah (2010)
18
American
countries
1980-2005
(unbalanced
panel data)
Dynamic
panel-data
model one-step GMM.
Pooled OLS, FE and
RE for robustness
Kagochi
(2010)
6 SSA countries
(Botswana,
Ghana, Kenya,
Nigeria, South
Africa,
Swaziland)
36
SSA
countries
1991-2007
Pooled OLS
1990-2008
Dynamic
panel-data
models within 2-step
system
GMM
framework
Largest
20
remittance
recipients as of
2008
1980-2008
Descriptive
statistics
and
trend
analysis
alongside
correlation
coefficient
et
al.
Lartey (2010)
Morton
(2010)
134
et
al.
Latin
M2/GDP, PSC/GDP), government expenditure/GDP,
investment/GDP rate, remittance (WR+CE+MT)/GDP,
inflation (GDP deflator), real effective exchange rate
(REER), and human capital
Dependent: 5-year logarithm differences in GDP per
capita (i.e. logGDPPCt-logGDPPCt-5)
Explanatory: Logarithm of lagged dependent variables,
literacy rate (-5), logGFCF/GDP, logGFCF/GDP(-5),
remittances (WR)/GDP, logarithm of labour force
Dependent: Natural logarithm of real GDP per capita
Explanatory: (Natural logarithm of) remittances
(WR+CE) per capita, tertiary school enrolment,
GFCF/GDP, FDI/GDP, ODA/GDP, other official
flows/GDP, trade openness, economic reform index,
exchange rate fluctuations
Dependent: Logarithm of real GDP per capita
Explanatory: (Logarithm of) Remittance (WR+CE+MT),
GFCF per capita, population growth rate, human capital
(proxied by life expectancy and education)
Dependent: Annual growth of GDP per capita
Explanatory: Remittance (WR+CE+MT)/GDP, private
sector credit/GDP), deposit money bank assets/GDP,
government expenditure/GDP, inflation (GDP deflator
based), FDI, trade openness, terms of trade, population
growth rate (all in logs except GDP growth and inflation)
Dependent: Real GDP per capita growth rate. Also,
absolute real GDP per capita and annual GDP growth
Explanatory: Remittances (WR+CE)/GDP ratio, poverty
headcount ratio (US$2 per day PPP as percentage of
population), income share lowest 20 per cent of
population, gross domestic savings, final consumption
expenditure/GDP, capital formation growth rate, CPI,
Poorer countries (those with less than
US$1200 (2000) GDP per capita have
greater positive impact of remittances on
long-run growth. Savings react much
more strongly than investment, with
remittances reducing amounts of debts
incurred and debt service paid.
Remittances have significant positive
effects on economic growth in Latin
America where the financial system is
less developed.
Remittances are a positive determinant
of economic growth in countries where
GDP per capita is high, but in low GDP
per capita countries, their effect is zero.
Remittances have positive impact on
economic growth just as the interaction
effect of remittances and financial depth
Remittances
reduced
poverty
but
aggravated income inequality. Traditional
factors such as physical capital
formation, human capital formation and
good governance are found to be crucial
determinants of growth.
These SSA countries are Ethiopia (2 per cent), Kenya (5.4 per cent), Mali (3.3 per cent), Mauritius (2.9 per cent), Mozambique (1.3 per cent), Nigeria (6.7per cent), Rwanda (1.9
per cent), Senegal (8.5 per cent), Sierra Leone (9.4 per cent), and Uganda (7.2 per cent). Figures in brackets are the remittances ratio to GDP in each sampled country in 2007 cited
by authors. Conspicuously missing from the list of SSA countries are traditionally well-known largest remittance-recipients like Lesotho, Cape Verde, Gambia and Sudan.
296
Stellenbosch University http://scholar.sun.ac.za
Siddique
(2010)
Singh
(2010)
et
et
al.
Bangladesh,
India and Sri
Lanka
1976-2006
Granger-causality
under VAR framework
al.
36
countries
1990-2008
Double log panel FE
and panel FE 2SLS
SSA
Adenutsi (2011)
Ghana
1987(3)2007(4)
Dynamic
equilibriumcorrection mechanism
model, unrestricted cointegration model and
Granger-causality test
Ahmed
(2011)
Pakistan
1976-2009
ARDL
OLS
64
countries
from Africa (29),
Asia (14) and
LAC (21)
1985-2007
Panel Unit-Root tests,
co-integration
model
and
Panel
FullyModified
OLS
(PFMOLS)
Dynamic
panel-data
modelling by sys-GMM.
Also, OLS and FE
Model
et
al.
Fayissa
and
Nsiah (2011)
modelling
by
literacy rate and population growth rate.
Dependent: Annual GDP per capita growth
Explanatory: Remittances (WR+CE+MT) per capita
Dependent: Logarithm difference of real GDP per capita
Explanatory: (Logarithm of) Initial growth, remittances
(WR+CE+MT)/GDP, M2/GDP, domestic credit/GDP,
population growth, government expenditure/GDP,
openness, terms of trade, political risk, real exchange
rate, REMGDP*institutions and REMGDP*financial
development
Dependent: Natural logarithm of real GDP
Explanatory: Natural logarithm of initial real GDP,
secondary school enrolment as proxy for human capital
formation, investment (GFCF/GDP), remittances
(WR+CE)/GDP, and financial development indicators
(M2/GDP and bank credit to private sector as ratio of
total bank credit). Control variables include government
expenditure/GDP, openness to trade (X+M)/GDP,
exchange rate, CPI-based inflation, AID and FDI
Dependent: Logarithm of real GDP
Explanatory:
(Logarithm
of)
remittances
(WR+CE+MT)/GDP,
M2/GDP,
government
expenditure/GDP, dummy for natural calamity
(earthquake)
Dependent: Logarithm of real GDP per capita
Explanatory: Remittances (WR+CE) per capita,
economic freedom, capita-labour ratio (GFCF/labour
force), economic openness
Remittances Granger-cause economic
growth in Bangladesh. In Sri Lanka, the
causal relation is bi-directional, whereas
there exists no causal relationship in the
case of India
Remittances have direct negative impact
on economic growth, but countries with
higher quality institutions have better
potential for harnessing the contribution
of remittances to growth
Although remittances generally promote
economic growth in the short run and in
the long run, their impact is low and lower
in the long run. In the short run, there is a
stronger lagged impact of remittances on
growth. There is no causality between
credit to the private sector remittance
inflows, but a bi-directional causality
exists between remittances and M2/GDP
Remittances have both short-run and
long-run significant positive impacts on
economic growth.
Remittances have significant positive
impact on growth in all three regions as
well as in the full sample as a group
Dependent: Natural log of output per capita
Six South Asian 1970-2008
Remittances are found to have a direct
Explanatory: Remittances (WR+CE+MT)/GDP ratio, significant positive impact on economic
countries (India,
Bangladesh,
human capital development proxied by secondary growth. Remittances also have significant
Nepal, Maldives,
school enrolment, government expenditure as ratio of positive interactive effects on growth
Pakistan,
Sri
GDP, openness proxied by (X/GDP), FDI/GDP, polity through educational levels and financial
Lanka)
index by Marshall & Jaggers for institutional quality
sector development
Source: Author‟s compilation from various sources. Note: WR, CE and MT denote workers‟ remittances, compensation of employees and migrants‟ transfers respectively as defined in
Chapter Two.
Cooray (2012)
297
Stellenbosch University http://scholar.sun.ac.za
CHAPTER SEVEN
THE DEVELOPMENTAL-IMPACT OF REMITTANCES IN SUB-SAHARAN AFRICA
7.0 INTRODUCTION
This chapter verifies whether or not international remittance inflows contribute to various
aspects of economic development in sub-Saharan Africa (SSA). The aspects of economic
development outcomes covered in this study are those related to poverty, income inequality,
labour market, human welfare and development, and financial development. In the case of
financial market development, an attempt was made to investigate whether remittances had
varying impact on SSA countries over time as the pursuit of financial liberalisation programme
progressed. Consequently, the background information on the relevance of this chapter is
presented in Section 7.1. This is followed by the review of the theoretical and the empirical
literature on remittance inflows and economic development in Section 7.2. Section 7.3
discusses the econometric issues as related to the analytical framework and the empirical
model. This section also outlines issues related to the data used for the empirical analysis. In
Section 7.4, the empirical results are presented and discussed, whilst Section 7.5 concludes
with policy implications and recommendations.
7.1 BACKGROUND
Arguably, international migration and its consequential effects on the economic transformation
of migrant-home countries have received the most attention from academics, policy makers
and researchers in the area of development economics and finance in this era of globalisation.
The upsurge of research interest in international migrant remittances, in particular, is not too
surprising given the magnitude and stability in the positive growth trend. There is one other
important reason why a lot of policy research on the implications of remittances for economic
development might have overtaken related studies on other forms of development finance
since the recognition of remittances as an alternative source of development finance. Unlike all
other forms of development finance, remittances can have a direct impact on the disintegrated
levels of a remittance-receiving economy. Consequently, from the theoretical perspective, the
direct linkages of international remittances and economic development can be explored at
three possible levels – micro, meso, and macro.
298
Stellenbosch University http://scholar.sun.ac.za
At the micro level, remittances are a major source of additional income for sustenance and
capital for financing small and medium-scale enterprises (SMEs) in remittance-receiving
households. Unlike aid, remittances flow directly to individual households and institutions; and
unlike loans they attract no direct interest and financial repayment obligations. Besides
contributing to increased consumption in the short run by empowering recipients to settle food,
clothing, shelter, healthcare, funerals and festival bills and so on, remittances can engineer
longer term development processes through investment in education, skills training, land,
housing, and SMEs. At the meso level, local communities can benefit from social development
projects such as the construction of roads, schools, and hospitals as well as the supply of
educational materials and healthcare equipment, initiated and funded by overseas-based
associations of native migrants. Associations of migrants can also mobilise funding through
non-governmental organisations (NGOs) and other development-oriented organisations in
support of important social projects such as vaccination against communicable diseases and
the provision of potable water in their local communities back home. Accordingly, besides the
indirect trickling down effects, families without international migrants can also benefit directly
from international migration at the meso level.
At the macro level, international remittances are an essential source of foreign exchange, as
they inject substantial foreign capital into an economy which may help remittance-receiving
countries to stabilise the macroeconomy through reduction in balance of payments (BoP)
problems and budget deficits. On the reverse side, remittances may contribute to destabilising
the macroeconomy of the receiving countries by sparking inflation through excess demand and
worsening BoP problems in import-dependent small-open economies. Meanwhile, remittance
inflows are also generally countercyclical as they increase during economic downturns; hence,
they contribute significantly to accommodating various forms of negative natural and
macroeconomic shocks in migrant-home countries. For instance, global evidence has
consistently shown that remittance inflows have always increased in disaster and conflict
inflicted countries (Clarke and Wallsten, 2004; Yang, 2007; Yang and Choi, 2007). Besides,
remittance inflows have consistently remained the most resilient form of private external capital
during global financial crises and violent conflicts in migrant-home developing countries.135 In
this sense, international remittances represent a more stable source of poverty reduction than
other forms of capital inflows, at least, at the macro level. And as available statistics suggests,
135
The World Bank estimates that in disaster-afflicted Haiti remittances represent about 17 per cent GDP in 2001,
while in some areas of war-torn Somalia, they accounted for up to 40 per cent of GDP in the late 1990s.
299
Stellenbosch University http://scholar.sun.ac.za
international remittances are more equally spread across developing countries than other forms
of foreign capital.
It can, thus, be said that depending upon the structure of the economy of a migrant-home
country, remittances can play a direct role in the economic development of a country through
increased consumption of the basic needs of life, job creation, financial market development,
human capital development, poverty alleviation, and economic empowerment. Remittances do
not only directly affect the various units of the economy of the remittance-recipient economy,
but they also have the potential to influence migrant-home countries indirectly in a number of
ways. Remittances received in excess of present consumption can be saved which could then
put financial institutions in a better position to expand credit at a relatively reduced cost. This, in
turn, can lead to higher job creation and poverty reduction. In fact, such private investments
can even attract additional investment, either by decreasing the risks of specific projects for
private investors, or by establishing business networks and openings that promise new
business opportunities for private financial institutions and multinational companies. Despite
these, from theoretical perspective, the implications of migrant remittances on labour
productivity and income inequality are far from being universally conclusive when the issue of
moral hazards, further migration of highly-trained labour, and the socioeconomic background of
migrants are broadly considered. In this respect, at the macro level, the extent to which
migrants remittances can impact on any specific developmental outcomes in a migrant-home
economy can be seen as being dependent upon some macroeconomic fundamentals.
Following from the above, an analysis of the effects of remittances at each level – micro, meso,
and macro – should provide the best and the most comprehensive insight into the actual direct
effects of migrant remittances on economic development in migrant-home countries. However,
achieving such an objective seems impossible across countries in the absence of accurate and
reliable micro- and meso-level data in the sampled countries. Accordingly, the focus of this
study is to explore the implications of remittances for economic development at the macro
level, based on the principle that the micro and meso effects of remittances on economic
development will, in the long run, reflect at the macro level. Meanwhile, empirical studies on the
impact of remittances on developmental outcomes have been far from being conclusive,
irrespective of the level of the analyses (see Table A7.1), justifying the theoretical stance that,
at any level of empirical analysis, migrant remittance inflows can have both forward and
backward linkages to the development prospects of migrant-home countries.
300
Stellenbosch University http://scholar.sun.ac.za
This study, therefore, seeks to contribute to filling this research gap by providing answers to the
following research questions within the context of SSA:
i.
Do migrant remittances reduce poverty headcount, poverty gap and poverty severity?
ii.
What is the impact of migrant remittance inflows on income inequality?
iii.
What is the impact of migrant remittances on human welfare and development
outcomes such as educational attainment and life expectancy?
iv.
How do migrant remittance inflows affect labour market outcomes with reference to
labour productivity, labour participation and unemployment?
v.
To what extent do migrant remittances promote financial market development? And,
does the impact of remittances on financial market development change over time as
the pursuit of policies under the financial liberalisation programme progresses?
vi.
Do migrant remittances have a universal impact on various aspects of economic
development in all categories of countries? Otherwise, which category of countries
benefit the most from receiving international migrant remittance inflows as far as
economic development is concerned?
Finding the appropriate answers to each of the above-stated questions constitutes the
underlying objective of this chapter. Nevertheless, with respect to SSA, the specific objectives
that this chapter seeks to achieve are:
i.
To determine the impact of international migrant remittance inflows on poverty.
ii. To examine the impact of international migrant remittance inflows on income inequality.
iii. To evaluate the effects of international migrant remittance inflows on human welfare
and development outcomes viz. human development indicators, educational attainment
and life expectancy.
301
Stellenbosch University http://scholar.sun.ac.za
iv. To determine the impact of migrant remittance inflows on labour market outcomes such
as labour productivity, labour participation, and unemployment.
v. To evaluate the impact of migrant remittance inflows on financial development.
vi. To examine if, in each case (i.e. objectives i-v), the developmental-impact of migrant
remittance inflows change over time as financial liberalisation policies are implemented.
vii. To verify if, in each particular case (i.e. objectives i-v), the developmental-impact of
migrant remittance inflows has a size-effect and, if so, estimate the impact of the sizeeffect on economic development with particular reference to each specific indicator of
economic development.
This study is important because the recent euphoria concerning the upsurge of international
migrant remittances has attracted an appreciable number of empirical studies on the possible
implications of remittances for economic development across the developing world of which
sub-Saharan Africa (SSA) is a part. To date, comprehensive macroeconomic policy options for
the developmental-impact of migrant remittances have not been studied in any systematic way.
For instance, the negative effects of remittances on one developmental outcome (say, (lower)
labour participation) could be due to the positive effects of remittances on another
developmental outcome (say, (higher) educational attainment)136.
7.2 THE LITERATURE ON REMITTANCE INFLOWS AND ECONOMIC DEVELOPMENT
7.2.1 Theories of the Developmental-Impact of International Migrant Remittances
Theoretically, three main schools of thought can be identified concerning the possible long-run
impact of international remittances on „labour-exporting‟ developing countries. These schools of
thought are the remittance-optimistic developmental, the remittance-pessimistic structural
dependence and the transnational-migrant remittance based on the theories of international
migrant migration137. The theories of international migration are being applied to international
remittance flows because of the absence of an existing specific theory on international migrant
136
This can happen when remittances are used to finance the education and training of children of school going age
who were hitherto out of school due to poverty, and are compelled to work for income in support of their families.
137
The main theories of international migration are the migration optimism (also known as the developmentalist and
neoclassical school of the 1950s and 1960s), the migration pessimism (or the historical structural dependency
school of the 1970s and 1980s) and the migration pluralism (or the New Economics of Labour Migration and
Livelihood School) which has been dominating the approaches to analysing the effects of international migration
since the 1990s (de Haas, 2007).
302
Stellenbosch University http://scholar.sun.ac.za
remittances in the context of economic development from a macroeconomic perspective. In
connection with this, it is important to note that even though it is generally known that migration
is a pre-requisite for receiving migrant remittances, it is also possible that the desire for
receiving remittances can influence international migration. Actually, there is evidence138 to
show that the receipt of migrant remittances in developing countries can engender further
migration139 to the industrialised world.
From the viewpoint of the developmentalist school, international remittances have a strong
potential to accelerate economic development processes in both industrialised „labourimporting‟ countries and non-industrialised „labour-exporting‟ countries, as large scale SouthNorth migration is adequately compensated for by large scale North-South migrant remittances
leading to international factor price equalisation. In other words, labour is transferred from
capital-constrained developing countries where labour is abundant and, often in excess supply,
hence relatively cheaper, to labour-constrained industrialised countries where capital is in
abundant and, often in excess supply hence relatively cheaper. Proponents, notably,
Kindleberger (1965), Beijer (1970), Penninx (1982), and Stark et al. (1997) of this neoclassicalinclined doctrine argue that, all other things remaining equal, international migration can,
therefore, lead to an increase in global production of goods and services, especially as
technological knowledge, attitudes, modernisation, information, rational and democratic ideas
are also transferred to developing countries. For instance, international remittances can
contribute positively to the removal of production and investment constraints, raising real
income levels, and lessening, if not solving the perennial BoP problems of developing
countries. In addition, remittances can help to narrow the trade gap, control external debt,
facilitate debt servicing, reduce exchange rate volatility and accumulate foreign exchange. The
developmentalist school also argues that the emergence of migration on the global scene is
aiding the industrialised countries in increasing production at a faster rate than it would have
been possible without access to cheap labour from developing countries. Therefore,
international migration has a two-sided positive impact on the global economy. On the side of
the „labour-importing‟ industrialised countries, increased supply of labour as a result of
immigration reduces the cost of hiring labour, whilst on the side of the „labour-exporting‟
138
See, for example, Cox and Jimenez (1992) for Peru, and Ilahi and Jafarey (1999) for Pakistan.
This can happen in two possible ways: (i) When sponsored migrants under implicit social contract with their
sponsors (often their families) are obliged to remit in order to finance another family member (see Poirine, 1997;
Brown and Poirine, 2005). (ii) When non-migrant families, upon seeing the life-transforming impact of remittances on
the families of a migrant, decide to sponsor a family member to go abroad for the purpose of receiving remittances.
303
139
Stellenbosch University http://scholar.sun.ac.za
developing countries, the emigration of „excess‟ labour reduces the supply of labour, thereby
increasing the cost of hiring labour towards equilibrium in the international labour market.
Besides the direct positive effects of international remittances on socioeconomic development,
the developmentalist school also contends that even when migrants fail to return home they
often contribute to financing development projects in their native communities either personally
or through their involvement in the activities of charitable organisations in countries where they
are permanently resident (Massey et al. 1998). Furthermore, from the experiences gained from
abroad, returnee migrants often act as agents of positive social change in governance,
innovation and entrepreneurship in their home countries. Consequently, in the long run,
developing countries stand to gain from the migration of their nationals who would otherwise
have been unemployed or underemployed and lowly paid at home. This benefit comes directly
through remittances, and indirectly through other afore-stated channels such as donations,
social work, skills and knowledge transfers.
The remittance-pessimistic structural dependence school that emerged in the 1970s following
the global economic decline with industrial restructuring and increasing unemployment as a
result of the 1973 oil crisis contends that international migration drains underdeveloped
migrant-home countries of skilled labour. The remittance-pessimist school further argues that
international migration crowds-out domestic production of tradable goods in the brain-drained
underdeveloped economy. This school of thought does not see how the negative effects of
brain drain can adequately be compensated for by the receipt of international remittances. The
remittance-pessimistic theorists (Almeida, 1973; Bhagwati, 1976; Lipton, 1980; Reichert, 1981;
Taylor, 1984; Rubenstein, 1992) argue that international migration only reinforces the
underdevelopment syndrome of developing countries through lower production capacity and
over-dependency, as remittances received are not adequate compensation for the lost labour
efforts in developing countries.
More specifically, remittance-pessimists contend that it is the industrialised countries that stand
to gain more in international migration through access to cheap labour, and high taxation on
migrant earnings and even commissions on transferring remittances. In effect, the low wages
paid to migrants in industrialised economies are not sufficient to help in narrowing the
development gap between the North and the South. Worst of all, even when remittances are
received in large amounts, there are very good reasons to predict that, given the abysmally low
304
Stellenbosch University http://scholar.sun.ac.za
incomes and widespread poverty in developing countries, it is difficult or impossible to avoid
conspicuous consumption and put remittances into productive use (Lipton, 1980; Entzinger,
1985; Lewis, 1986). This can include remittances being “wasted” on housing, family debt
settlement, land purchase, land and chieftaincy litigations, transportation, funerals, festivals,
financing conflicts, leisure, and other non-productive goods and services. In this case,
remittances can destabilise the macroeconomy by way of demand-pull inflation (Russell,
1986a,b; Appleyard, 1989; Rubenstein, 1992), with higher trade deficit in developing countries
which are predominantly net importers of essential goods. Furthermore, it is argued that higher
remittance inflows may aggravate higher income inequality as the very poorest cannot afford to
send a family member abroad (Lipton, 1980; Stahl, 1982). The tendency for further migration is
also high when more remittances are received in low-income countries. Another possible
negative consequence of higher inflow of remittances in the household is the moral hazard
problem when recipients reduce work efforts (Chami et al. 2005); and at the national level,
governments of developing countries may also over rely on these funds rather than
implementing sustainable pro-growth economic policies.
Subsequent to the more recent emergence of the New Economics of Labour Migration (NELM)
paradigm as proposed by Stark, (1978; 1991), Stark and Bloom (1985), Taylor (1999), Bracking
(2003), Carling (2004) and Robinson (2004) it is possible to identify a third school of thought,
the transnational-migrant remittance school based on the concept of pluralism. This school
sought to reconcile the two strictly divergent perceptions on the outcome of international
migration by focusing on how remittances together with socioeconomic networks, link local and
global development processes (Levitt, 2001). This approach does not restrict itself to
considering financial remittance flows alone, but also takes into account the flow of goods,
services and new ideas that impact on the broader social fabric and structures of the
economies of both „labour-importing‟ and „labour-exporting‟ countries (Datta et al. 2006). By
taking a balanced view of the implications of international migration, the transnational migrantremittance school focuses on how remittances are embedded within an emerging structure
where various economic, social, institutional and even political transactions occur. This neoliberalist functionalist ideology relates migration decisions with the impact of migration to
collective household survival and the pursuit of income and/or initial capital for productive
investment as a means of insuring against both income and production risks at the household
level (Stark, 1978; 1991; Taylor, 1999; Stark and Levhari, 1982). This is the fundamental
reason why remittances are seen as being beneficial at the household level with positive spill305
Stellenbosch University http://scholar.sun.ac.za
overs to the national economy; as with increased disposable incomes, effective demand for
industrial goods and services increase, and this, in turn, stimulates domestic production
(Skeldon, 2002; Ratha, 2003). Higher remittance inflows may also result in the increased
vibrancy of domestic capital markets and the expansion of productive infrastructure (Ballard,
2004).
According to Vertovec (1999), the transnationally adopted identities and connections between
international migrants and migrant households in home countries can lead to radical
modernisation of telecommunication infrastructure such as cellular networks, telephone,
internet, and satellite, as migrants remit through globalised banking systems. Therefore, the
transnational-migrant remittance school hypothesis is essential to the understanding of the
framework within which migrants remit as it does not only take cognisance of how remittances
reinforce and create inequality and differentiation, but it also recognises the fact that these
private transfers have various degrees of positive social effects and, hence, have a huge
potential to contribute to poverty alleviation and socioeconomic transformation (Ballard, 2004;
Carling, 2004). It is probably due to the broad inclusiveness of the transnational-migrant
remittance theory inferred from the NELM doctrine that explains its dominance in analytical
studies over the past two or three decades (see Table A7.1). The quest to follow the pluralist
dimension of the popular transnational-migrant remittance theory explains the underlying
analytical macroeconometric framework of this chapter.
7.2.2 Literature Review on Effects of Remittances and Developmental Outcomes
In addition to the summary of reviews of the impact of remittances on economic development
reported in Table A7.1, in this section, a classified approach has been adopted to review
empirical studies on the effects of international remittances on poverty, income inequality,
human capital development and financial market development.
7.2.2.1 Effects of International Remittances on Poverty and Income Inequality
From both theoretical and empirical literature, it seems that there is less controversy
concerning the positive effects of remittances on poverty in migrant-home countries than the
possible conflicting effects on income inequality. The main theoretical debate centres on the
fact that it is only households with relatively higher incomes that can afford to finance the cost
of international migration. Therefore, international migrant remittances can widen income
inequality in migrant-home countries (Lipton, 1980; Stahl, 1982). This poses a serious
306
Stellenbosch University http://scholar.sun.ac.za
challenge to policy makers, given that inequality is often a determinant of poverty140 as it
indirectly undermines long-run growth by reducing motivation for optimal labour productivity
and, hence, can result in the perpetuation of the poverty cycle.
Adams (1991), in a micro-level study based on a survey of 1000 households in rural Egypt,
using income data from households with and without migrants to determine the effects of
remittances on poverty, income distribution and rural development, observes that migrant
remittances were important in alleviating poverty. For a sample of 77 developing countries over
the period 1980-2008, the UN (2011) obtained a similar poverty-alleviating impact of
remittances. However, Adams (1991) concludes that despite the direct poverty-mitigating
effects of international remittances, they also contributed to inequality in the distribution of
income. Chimhowu et al. (2004) provide evidence in support of the view that remittances do
increase inequality at a national level, but internationally they transfer resources from
developed to developing countries, thereby contributing to reducing income inequality across
countries. Analogous to these inequality-aggravation findings is the result obtained by
Rodriguez (1998) on Philippines.
In contrast, inequality-reducing effects of remittances were found by Barham and Boucher
(1998) for Nicaragua; Adams (2006) in the case of Ghana; and the World Bank (2007) for
households in East European and former Soviet Union countries. Gustafsson and Makonnen
(1993) reveal that in Lesotho, migrant remittances do not only reduce poverty but they actually
decrease income inequality. For Mexico, Esquivel and Huerta-Pineda (2007) find that
remittance-recipient households are less likely to be poor, based on the National Household
Survey Data on income and expenditure for year 2002. Evidence from various cross-country
studies including those by Adams and Page (2005), Spatafora (2005), Acosta et al. (2008b),
Shafiq et al. (2012) and Orzell (2013) lend support to the fact that remittances directly reduce
poverty; whilst many more studies including those by Stark et al. (1986), Taylor (1992),
McKenzie and Rapoport (2007), and Unger (2005) show that remittances directly reduce
inequality.
7.2.2.2 Effects of International Remittances on Labour Market Outcomes
The question as to whether remittances affect labour market outcomes is very important
because migrant remittances are received at the cost of losing the participation of the emigrant
140
See Table A7.1 for evidence on empirical models on poverty.
307
Stellenbosch University http://scholar.sun.ac.za
in the home-country labour market. It is also known that remittances can directly affect the
labour market in a number of ways including: (i) job creation when remittances are driven by
the investment motive; (ii) job creation through market expansion when remittances are spent
on locally produced goods and services; and (iii) higher economic inefficiency which increases
the unemployment rate (especially of the other factors of production) through reduced output
level due to brain drain. International remittances can also affect the labour market indirectly,
through moral hazard effects, because when they are received in „satisfactory‟ amounts and
become permanent incomes, they can reduce productivity and participation of labour.
Theoretically, the extent to which remittances (as compensation for losing the services of a
migrant at home) can affect the labour market of the migrant-home country is dependent upon
the degree of friction in the domestic labour market in question.
According to Pond and McPake (2006), almost a quarter of the new overseas trained
physicians that registered with the National Health Services of the United Kingdom between the
years 2002 and 2003 came from SSA alone. The large-scale emigration of skilled professionals
has created high job vacancies in some key sectors in many SSA countries, but it is common
knowledge that developing countries like those in SSA have high rates of graduate
unemployment and underemployment141,142. Based on survey data on the city of Managua in
Nicaragua, Funkhouser (1992) finds that international remittances lead to about five per cent
reduction in the labour force participation of women, as well as by 2.1 per cent of men. At the
same time, however, remittances increase the probability of self-employment by 1.2 per cent
among men and 1.1 per cent among women. Hanson (2007), based on the 2000 population
census survey, obtains similar results for Mexico where remittances reduce female labour
supply relatively more than in the case of male remittance recipients.
In Zambia and Zimbabwe, just like in Ghana, Bach (2006) finds that the annual rate of attrition
in public health employment due to emigration ranges between 15 per cent and 40 per cent. In
another empirical work, Gupta, Pattillo and Wagh (2009) find that, on the average, 20 per cent
of SSA tertiary educated population above 15 years of age are employed in OECD countries
compared with less than 10 per cent for South Asia. And within SSA, Angola, Guinea-Bissau
and Mozambique have expatriation rates in excess of 50 per cent of their tertiary educated
141
Bhagwati (1976) argues that brain drain can have a detrimental effect on economic development of migrantsending countries because even where skilled labour is unemployed, their social marginal impact is not necessarily
zero as they could move inland the countryside, where they would have been employed productively.
142
However, the issue of underemployment and rural unemployment as is the common case in contemporary SSA
can neutralise Bhagwati‟s argument.
308
Stellenbosch University http://scholar.sun.ac.za
population. Of the top-10 countries with the highest emigration rate of tertiary educated
population, six are from SSA alone.143 Strangely, with the exception of Mauritius, none of these
six SSA countries mentioned as having exported the most educated migrants is a major
recipient of remittances in terms of actual volume or in relative terms. This seems to confirm
earlier conclusions by Steiner and Velling (1994) and Rodriguez and Horton (1995), that the
educational level of migrants has no impact on the amount of funds transferred in the form of
remittances144. However, even if remittances are spent on domestically produced consumables,
they are expected to carry substantial positive multiplier effects on employment that can
manifest in the labour market. Increased demand for domestically produced goods and
services, increased retail activities and small-scale industrialisation, hence, higher demand for
factor inputs, are some of these positive effects. Also, Ratha (2003) finds that the negative
effects of brain drain are largely offset by inward migrant remittances. It is, therefore, important
to explore the extent to which remittances have been able to impact on labour market
outcomes in SSA as a sub-region.
7.2.2.3 Effects of International Remittances on Human Development and Welfare
From the typology of the uses of remittances (see Chapter 4), it is clear that migrant
remittances in excess of daily consumption expenditure are spent on financing education,
vocational training and improved access to quality healthcare services, each of which promotes
the development of human capital. This implies that remittances could contribute directly to
reducing income constraints that limit maximum human capital investment for optimal labour
productivity. Human capital accumulation is central to the economic development prospects of
a country through higher labour productivity and greater prospects of reducing dependency
ratios and breaking the seemingly perpetuating cycle of poverty. However, the effects of
remittances on human capital development seem ambiguous in the face of international
migration because: (i) it is active labour with relevant skills that can be attracted to compete
more favourably in the international labour market, and are therefore the most likely to jump
onto the exodus wagon; (ii) skilled labour can only contribute meaningfully to economic
development of their native countries if they are gainfully employed and retained in a skilled
143
These are Guinea-Bissau (70.4 per cent), Angola (53.8 per cent), Mozambique (52.3 per cent), Mauritius (50.3
per cent), Gambia (42.4 per cent), and Burundi (35.0 per cent). And among the top-20 countries, 75 per cent are
SSA countries (Gupta, Pattillo and Wagh, 2009 based on OECD, Trends in International Migration Database, 2006).
144
Possible explanations for this are: (i) migrants with higher levels of education might not necessarily be coming
from poor homes where remittances are much more needed to augment meagre family incomes; and (ii) educated
migrants are more likely to have residential status, given the quality of their skills and, hence, are more likely to
reunite with their families in the foreign country when compared with their illiterate counterparts.
309
Stellenbosch University http://scholar.sun.ac.za
related profession in the home country; and (iii) even where highly skilled citizens migrate into
the Diaspora, they are not necessarily the highest remitters (Rodriguez and Horton, 1995).
Conclusions from empirical studies have been largely unanimous on the fact that migrant
remittance inflows directly promote human capital development in migrant-home countries. For
instance, in Zimbabwe, although households with migrants abroad tend to have less cultivated
lands, these remittance beneficiary households also tend to have a higher level education than
their non-remittance receiving counterparts (de Haan, 2000). Duryea et al. (2005) find
significant evidence for lower incidence of infant mortality when female participation in the
labour market was reduced upon receipt of remittances. This is likely to result from a higher
time allocation to maternal care. Using the case of Philippines, Yang (2004) shows that
reduced labour force participation is associated with increased school attainment among
Pilipino children aged between 17 and 21 years in remittance-recipient households. CoxEdwards and Ureta (2003) also find that remittances directly and instantaneously reduce
school dropout rate in El Salvador. For Mexico, López-Córdova (2005) confirms earlier results
by Hanson and Woodruff (2003) and McKenzie and Rapoport (2006) that in remittancerecipient homes, illiteracy rates are lower among boys and girls of school going age and
teenagers. For a group of Latin American countries, Acosta et al. (2008b) obtain a similar result
whereby migrant remittances enhance educational attainment even when counterfactual
scenarios of migration without remittances, and, no migration and, hence, no remittances were
taken into account.
With regard to the direct role of remittances in promoting human capital accumulation through
higher access to improved healthcare system in developing migrant-sending countries where
public healthcare is inefficient and pro-rich in the absence of an effective universal health
insurance system, international evidence has shown that remittances have been most useful.
In Mexico, for instance, Amuedo-Dorantes and Pozo (2006) report that remittances received
directly increase healthcare expenditure by households, and that these expenditures are more
responsive to increases in remittances than non-remittance incomes. Duryea et al. (2005)
conclude from an empirical study on Mexico that migrant remittances have a direct positive
impact on reducing infant mortality through higher mother-child time allocation, increased
access to improved housing conditions and potable water. Hildebrandt and McKenzie (2005)
also find international remittances to have reduced child mortality rate and increased birth
weight of infants from remittance-receiving households in Mexico. Similarly, from a study
310
Stellenbosch University http://scholar.sun.ac.za
carried out on Mexican municipalities, López-Córdova (2005) reports that as more remittances
are received, infant mortality declines. In the case of 11 Latin American countries, Acosta et al.
(2008b) also discover that remittances contribute substantially to improvements in health
indicators. To provide an insight into the empirics of the effects of migrant remittances on
human development and welfare in SSA, this chapter explores the implications of international
remittances received for both dimensions – educational attainment and life expectancy as well
as integrated human welfare.
7.2.2.4 Effects of International Remittances on Financial Market Development
The large volume and strong stability in the inflow of migrant remittances offer remittancereceiving developing countries a good opportunity to develop their financial sector in order to
attract more official inflow of these funds from their citizens residing abroad. A financial sector
is considered as developed if financial intermediaries can more freely and efficiently provide
quality and reliable payments mechanism, facilities for financial resource mobilisation and
credit allocation, information symmetry, liquidity and risk mitigation (Pagano, 1993; World Bank,
2005). Essentially, McKinnon (1973), Shaw (1973), Fry (1995), Kar and Pentecost (2000), and
the World Bank (2005) consider financial development to be a consequence of financial
liberalisation since the pursuit of financial repressive policies undermines the scope and pace
of financial development.
Theoretically, migrant remittances can either substitute or complement the role of the financial
sector in resource mobilisation and allocation depending upon the level of financial
development in migrant-home countries. According to the substitutability hypothesis of
remittances, the restricted access of the private sector to the formal sector credit in low-income
countries where credit markets are imperfect and the financial sector is underdeveloped can be
partially offset by higher inflows of remittances (Giuliano and Ruiz-Arranz, 2009). The inflow of
high international remittances allows recipients to invest in high return investment projects
despite the difficulties in accessing bank credit. Besides, it is this credit constraint awareness
that compels migrants to remit more funds home in excess of family consumption and to
potential investors who lack the collateral assets to access credit from the formal financial
market. Therefore, in the case of the substitutability hypothesis, there is an inverse relationship
between financial development (FDV) and international remittance inflows, such that, if
D
V
R
E
M
P
C
F
D
V
fR
(E
M
P
C
,
) and F
, then FDV
0 1
2
311
REMPC
1 , but
Stellenbosch University http://scholar.sun.ac.za
ˆ1 0 ; where ˆ1 and denote estimated 1 and other macroeconomic determinants of FDV
respectively; and when any possible reverse causality between FDV and REMPC is ignored.
On the contrary, the complementarity hypothesis of remittances holds when higher international
remittance inflows and a higher degree of financial development foster each other. According to
Giuliano and Ruiz-Arranz (2009), in economies where capital market imperfections are limited
and access to credit is readily available, and where potential investors can rely on the financial
sector; remittances can be counter-productive and have moral hazard effects. On the one
hand, a country receives more international remittances because of a higher degree of financial
development, which results in higher financial sophistication and reduced transaction costs
associated with remittance inflows. In other words, the propensity to remit through the formal
financial system increases as the financial sector of the migrant-home country develops and
there is higher access to quality financial services and innovative financial products at
competitive prices. On the other hand, higher inflows of international remittances stimulate the
incentive of formal financial institutions, including the monetary authorities, to implement
prudent legal and institutional reforms to boost remittance inflows as well as enhance the
productive uses of remittances received. Accordingly, higher levels of financial development
help migrants to remit more, and in turn, a significant inflow of remittances contributes to the
development of the domestic financial system in many ways, but, in particular, financial
inclusion (Terry and Wilson, 2005). For the complementarity hypothesis of remittances to be
valid with respect to how remittances contribute to financial development in migrant-home
D
V
fR
(E
M
P
C
,
),
countries, it is expected that, once it is established that, in general, F
ˆ
F
D
V
D
V
R
E
M
P
C
0
.
and, specifically, F
, then
0 1
2
R
E
M
P
C1
The World Bank (2005) asserts that financial development can be determined from five main
dimensions. These are the ability of financial intermediaries to provide savings facilities for
resource mobilisation, credit allocation and the monitoring of borrowers, payment mechanisms,
risk mitigation, and liquidity services (World Bank, 2005). There are a set of indicators for each
of these aspects of financial development, as shown in Table 7.1.
Of the numerous indicators of financial development, this study adopts only two measures –
broad money to GDP ratio and private bank credit as ratio of GDP – due mainly to data
312
Stellenbosch University http://scholar.sun.ac.za
limitations on the other indicators in the sampled countries; and for the sake of comparability
with a majority of previous related studies.
Table 7.1: Functions of Financial System and Financial Sector Development Indicators
Function
Key Indicators
Broad money (M2) to GDP
Ratio of bank deposits to GDP
Proportion of population with bank accounts
Total number of bank branches
Population per bank branch
Distribution of branches and other outlets
Household and corporate holdings of non-bank financial assets
Credit allocation and
Private sector bank credit as ratio of GDP
monitoring of
Ratio of bank loans to bank deposits
borrowers
Volume of finance raised from the issuance of bonds and money market instruments
Provision of
Proportion of payments (volume and value) made with different payment instruments
payments
Number of days for clearing cheques
mechanism
Number and distribution of clearing centres
Risk mitigation
Ratio of insurance premiums to GDP
Number of insurance and derivative products and services available
Insurance and derivative products held as a ratio of population
Provision of liquidity
Interest rate spread
services
Interest rate structure
Prices of basic financial services
Source: Author based on World Bank (2005)
Provision of savings
facilities for resource
mobilisation
Broad money to GDP: According to the World Bank (2005: 20), “the overall extent of financial
savings can be ascertained by examining the level and trends in the ratio of broad money to
GDP”. Broad money is recorded as M2 or M3 in the standing of money supply by monetary
authorities, although data on M3 is relatively scarce in many developing countries. This
indicator, specifically ( M 2 / GDP) , which is the standard and most commonly used indicator of
financial sector development (World Bank, 1989; Kar and Pentecost, 2000) may, however,
inflate the real size and depth of the financial sector if currency (M1) constitutes a high
proportion of broad money (De Gregorio and Guidotti, 1995; World Bank, 2005). When
currency outside the banking system constitutes a larger proportion of broad money, then the
use of broad money as a ratio to nominal GDP merely measures the degree of monetisation
rather
than financial
development.
Cash-based
economies,
a common feature
of
underdeveloped financial systems, automatically have a higher degree of monetisation in the
absence of other sophisticated financial instruments. Accordingly, De Gregorio and Guidotti
(1995), suggest the use of less liquid forms of monetary aggregates (i.e. M2 or M3) as a proxy
for financial sector development. However, the problem of monetisation could still be present if
M2 or M3 is measured as M1 plus quasi money, which indeed is the case, taking into
313
Stellenbosch University http://scholar.sun.ac.za
consideration the vertical composition of money supply as recorded in the Balance Sheets of
Central Banks. In this case, it would have been more appropriate to use (M2 or M3 minus M1) as
a ratio of GDP to measure FDV in highly monetised SSA countries. Yet, data on M1 over the
study period, 1980-2009, is lacking in most of the sampled SSA countries. In spite of the
limitation of M 2 / GDP as an indicator of FDV, it continues to enjoy a popular patronage in
empirical studies. Some of the recent studies that used M 2 / GDP in the remittance literature
include those of de Leon-Manlagnit (2006), Drinkwater et al. (2006), Shahbaz et al. (2007),
Ebeke and Le Goff (2009), Giuliano and Ruiz-Arranz (2009), Gupta, Pattillo and Wagh (2009),
and Adenutsi (2011).
Bank credit to private sector as a ratio to GDP: According to Kar and Pentecost (2000: 6), this
“is one of the five most commonly used proxies for financial development” to evaluate the
extent of financial intermediation by banks. Bank credit to private as a ratio to GDP is
considered as a more direct measure of financial intermediation when compared with domestic
credit/GDP ratio because the former directly captures the proportion of credit extended by
banks to finance productive private-sector investment projects. In other words, bank credit to
government agencies and state institutions are excluded from the computation of this indicator.
The underlying theory is that the private sector, unlike the public sector, is more efficient in
utilising debt capital because the private sector is confronted with more stringent loan
repayment obligations, a higher quest for entrepreneurial success and an intrinsic desire to
avoid perpetual dependency on debt capital. Notwithstanding the fact that this indicator
exclusively measures credit directed at the private sector, one major limitation145 of this
indicator is that it does not suggest how bank loans to the private sector are actually utilised.
Financial development is expected to culminate in raising returns on investment and reducing
the cost of capital and the risk of investment by ameliorating information symmetry, reducing
information and transactions cost, and facilitating risk management (Rajan and Zingales, 1998;
Wurgler, 2000). However, this indicator does not provide information related to these aspects of
financial development (Levine et al. 2000). All the same, private sector credit to GDP enjoys
extensive patronage in empirical studies as shown in Table A7.1 in the Appendix.
145
Another critical drawback for using this indicator of financial development for countries in SSA is that many of the
countries in the sub-region included credit to public enterprises as part of private sector claims, especially before as
well as in the earlier years of economic reforms. Meanwhile, during those years, it was state-owned enterprises
(SOEs) that received the majority of the credits extended by banks, the majority of the banks which were also state
owned (various SSA Central Bank reports).
314
Stellenbosch University http://scholar.sun.ac.za
So far, conclusions from available empirical studies have been quite unanimous on the direct
positive impact of migrant remittance inflows on financial development in remittance-recipient
countries (see Drinkwater, 2006; Acosta et al., 2008a; Shahbaz et al., 2007; Gheeraert et al.
2010; Gani and Sharma, 2013), with this impact often turning more robust in countries with
lowly-developed financial markets as implied by Giuliano and Ruiz-Arranz (2009).
The financial development strategies that developing countries can adopt to attract a higher
inflow of official remittances include lowering the cost of international funds transfers, widening
financial services to advance financial inclusion; providing offshore banking facilities; and rolling
out innovative financial products with diversified risks. Meanwhile, Acosta, Baerg and
Mandelman (2009) show that well-developed financial markets of remittance-recipient countries
can be important in channelling remittances into productive uses in migrant-home countries. It
is for this reason that examining the impact of remittances on financial development should be
seen as vital to the understanding of the development prospects of migrant remittance inflows
in SSA as a region where the majority of the countries have underdeveloped financial markets.
7.3 ANALYTICAL FRAMEWORK, EMPIRICAL MODEL AND DATA ISSUES
7.3.1 Analytical Framework and Empirical Model
An important methodological challenge related to modelling the effects of migrant remittances
on economic development outcomes is endogeneity bias that could arise from reverse
causality, omitted variable bias and migrants‟ self-selection bias of target recipients. In addition,
remittance inflows do not only affect the socioeconomic welfare of direct recipients but also
non-migrant households, the business sector, the local community, and the nation as a whole.
To circumvent this problem, it is important to adopt an econometric approach where it is
possible to overcome endogeneity in the empirical model. Analysts who take serious
cognisance of this problem often use either instrumental variable techniques or dynamic paneldata modelling especially where the data dimension is of a larger cross-section over time
series146. Of these two approaches, dynamic panel-data modelling by Generalised Method of
Moment (GMM) dominates the empirical studies of recent years and even where the two
approaches are used for robustness tests, conclusions have been based mainly on results from
GMM estimators147. Therefore, to estimate the macroeconomic impact of remittance inflows on
146
See Table A7.1 for details.
See, for example, Aggarwal et al. (2006), Giuliano and Ruiz-Arranz (2006), Acosta et al. (2008a,b), Acosta,
Baerg and Mandelman (2009), Jongwanich (2007), Gyimah-Brempong and Asiedu (2009), and Adenutsi and Ahortor
(2010).
315
147
Stellenbosch University http://scholar.sun.ac.za
economic development in SSA, this study relies on a dynamic panel-data modelling by system
GMM148.
Each estimated model has remittances incorporated into an otherwise standard endogenous
growth-type economic development model149 with motivation from the transnational-migrant
remittance paradigm. This is because, as noted in Section 7.2, within the context of the
transnational-migrant remittance theory, it is possible to explore the effects of remittances on a
wide array of developmental outcomes from both the migrant-pessimist and the migrantoptimist perspectives. Also, because economic development is a multi-dimensional concept150,
there can be many developmental outcomes, but consistent with the afore-stated objectives,
the study restricts itself to analysing the impact of remittances on poverty (headcount, gap,
severity), income inequality, three indicators of labour market outcomes, three indicators of
human capital development and welfare, and two indicators of financial development. The
choice of each indicator was based essentially on the popularity in empirical studies and data
availability.
The general empirical dynamic panel-data model is specified as Equation (7.1), which states
that any measure of economic development outcome ( Ei ,t ) in country i at time t is explained
by the initial level of the specific measure of the economic development outcome in question
( Ei ,t _1 ), current remittances per capita which also connotes remittances per capita151 received
in a sampled in country ( Ri ,t ) plus a set of other possible macroeconomic determinants of E .
Mathematically, it is specified that:
e
c
o
n
o
m
i
c
d
e
v
e
l
o
p
m
e
n
t
l
n
E
l
nl
E
n
R
Z
i
,
t
i
,
t
1
i
,
t
1
2
i
,
t
3
i
,
t
t
i
i
,
t
(7.1)
where the regressand, E , denotes a measure of economic development outcome of interest;
R, Z , t , i and i ,t are as previously defined in Chapter Six. However, Z now contains
additional control variables such as annual GDP growth rate as a proxy for business cycle,
literacy rate, real lending rate, and real GDP per capita at purchasing power parity (PPP).
148
1,
The relative superior qualities of system GMM over alternative GMM estimation techniques of dynamic paneldata models are well discussed in Chapter Four of this dissertation.
149
See Chapter Six for details of endogenous growth model.
150
See Todaro and Smith (2002), and Thirlwall (2011).
151
For the justification of this analogy, see Chapter Three.
316
Stellenbosch University http://scholar.sun.ac.za
2 and 3 being the corresponding parameter estimates of E , R and Z . The notation ln
preceding R and Z signifies natural logarithm; and all control variables previously used in
Chapter Six remain as defined and as to whether they are in natural logarithmic or algorithmic
forms, specific details are provided in Tables A7.2 and A7.6 in the Appendix. Although, both
intuitively and by anecdotal evidence, Z may contain a wide array of potential explanatory
variables, Perotti (1996), Acosta et al. (2008a,b) and Acosta, Baerg and Mandelman (2009)
offer a reasonable justification for selecting regressors within the context of economic growth
model152,153. Beyond the underlying theoretical relevance, a further justification for adopting this
approach of choosing the regressors is to make room for comparability of results with previous
related studies, and at the same time satisfying the condition of parsimonious approach to the
empirical modelling. This is important because apart from estimations involving financial
development indicators, for all other estimations, the number of observations reduced
drastically particularly due to unavailability of annual data154 in the 36 sampled countries. It is
expected a priori that migrant remittances have poverty-mitigating effects and impact positively
on human welfare, school attainment and life expectancy, but with regard to income inequality,
financial development and labour market outcomes, the a priori effects are indeterminate.
Similar to the approach used in Chapter Six, a three-step estimation procedure was used to
determine the impact of migrant remittances on a given economic development outcome at
step one; investigate the presence or absence of discriminatory impact of remittances on the
given developmental outcome at step two; and, given that this size-effect exists, estimate the
impact on the relevant group at step three. In doing so, a median-dummy variable (MDV) was
introduced as an additional variable in the „final‟ parsimonious empirical model at the secondstep estimation. In step three, MDV was replaced with the MDV-remittance interactive variable
in the empirical model for re-estimation. With the exception of the empirical unemployment
model, MDV takes the value of one if in a particular time period t , E of a country i exceeds
the median E ; otherwise it takes the value of zero (see Table A7.3).
Consistent with previous chapters, for the empirical models involving annual panel data over
the entire study period, 1980-2009 and for the 36 sampled SSA countries, static panel-data
152
This has been the norm in macro level cross-country panel-data studies. For examples, see Table A7.1.
Here, the study choice of regressors is those of the endogenous growth model as espoused in Chapter Six.
154
Only 5-year average data are available for socioeconomic development variables such as indicators of poverty,
inequality, human welfare and development indicators (excluding school enrolment) and labour market outcomes.
153
317
Stellenbosch University http://scholar.sun.ac.za
models were estimated to, among other things, demonstrate that, indeed, dynamic panel-data
modelling outperforms the former. In pursuit of this objective, it is only the empirical models
determining the impact of migrant remittance inflows on the two selected financial development
indicators namely bank credit to the private sector and broad money to GDP ratio that were
subjected to this compelling exercise. For each of these financial development indicators, both
the conventional and heterokesdasticity-corrected robust Fixed (within) Effects (FE) and GLS
Random Effects (RE) models were estimated. Based on the econometric issues discussed
under 4.5.2 in Chapter 4, the results of the estimated robust static panel-data parameters are
not expected to confirm the parameter estimates from the two-step sys-GMM estimators from
the empirical dynamic panel-data models in either case.
7.3.2 Data Issues
The empirical analysis of this chapter encompasses 36 sampled SSA countries listed in
Chapter One of this dissertation. However, in some specific cases, the sample size was
reduced due to constraint on relevant data. For instance, in the case of empirical poverty and
inequality models, only 34 countries (excluding Mauritius and Sudan) were analysed. For this
same reason, the number of countries was further reduced to 27 (excluding Congo Republic,
Côte d‟Ivoire, Ethiopia, Gambia, Guinea, Guinea-Bissau, Niger, Senegal and Togo) in the
estimation involving the rate of unemployment.
With the exception of educational attainment proxied by secondary school enrolment and
financial development indicators, annual panel-data on most of the developmental outcome
variables at the centre of analysis in this chapter are scarce over the study period, 1980-2009.
In this respect, in estimating the impact of remittances on human welfare and human capital
development outcomes, poverty, inequality, and labour market outcomes, rather than using
annual panel data as was done in the case of financial development, a 5-year non-overlapping
average data was used. Using non-overlapping 5-year average data reduces the time
dimension of the panel data from 30 to 6 for the overall study period as there are only two
observations per decade. With the panel-data dimension still having the structure of N T ,
when the sys-GMM estimation technique is executed on the empirical dynamic model in which
N 36 and T 6 as a result of the 5-year non-overlapping averaging of data, the efficiency of
system estimators was not compromised155.
155
Even where N was reduced to 27 in the unemployment model due to data constraint, the system GMM was still
applicable because T=6.
318
Stellenbosch University http://scholar.sun.ac.za
In line with previous related studies (see Table A7.1), poverty headcount ratio and poverty gap
index which measure the incidence and depth of poverty respectively, and squared poverty gap
index as a proxy for poverty severity are the indicators of poverty used in this study whilst the
Gini index was used to measure income inequality. Human development index (HDI)
representing the geometric mean of three normalised indices of life expectancy, knowledge and
education156, and living standards as measured in natural logarithm of gross national income at
PPP was used as a proxy for general human welfare status. Secondary school enrolment was
used as a measure for educational attainment, whilst life expectancy was used as a narrow
measure for human welfare. Unemployment rate, labour force participation rate and labour
productivity rate were used to measure labour market outcomes. The definition, specific
measurement and main source of the dependent variables and explanatory variables not
previously used in this study as explanatory variables are outlined in Table A7.6 in the
Appendix. Unless otherwise specified, each variable is in its natural logarithmic form.
7.4 EMPIRICAL RESULTS AND DISCUSSIONS
7.4.1: The Impact of Remittances on Poverty and Income Inequality in SSA
The empirical results on the impact of international migrant remittances on poverty and income
inequality in SSA are reported in Table 7.2. As can be seen in Table 7.2.1, the study further
shows that the poverty-alleviating effects of remittances differ across SSA countries, using the
group median-level indicators of poverty as a reference point. From Table 7.2, it is shown that a
one percentage increase in remittances per migrant received in SSA reduces poverty in terms
of incidence, gap and severity by 0.0217, 0.0292 and 0.0584 respectively.
With statistically significant estimated coefficients of 0.0452, 0.0750 and 0.1500 for poverty
headcount, poverty gap and poverty severity respectively reported in Table 7.2.1, this study
reveals that when the incidence of poverty by any of the three measures is above the median
level (see Table A7.3), official remittances received aggravate poverty in migrant-receiving
countries, at least, internationally157. Thus, although generally, remittances alleviate poverty in
SSA, in migrant remittance-receiving SSA countries with relatively high probability incidence of
poverty, remittances actually aggravate poverty.
156
Knowledge is proxied by adult literacy rate with two-thirds weighting, whilst primary, secondary and tertiary gross
school enrolment which captures education, takes one-third weighting.
157
Note that the poverty line used in this study is based on international caloric requirements based on PPP.
Therefore remittances can actually reduce poverty at national levels in both categories of countries (which is outside
the scope of this study), but not in terms of comparative international landscape (as revealed by this study).
319
Stellenbosch University http://scholar.sun.ac.za
Table 7.2: Impact of Remittances on Poverty and Inequality in SSA, 1980-2009
Group variable: Code
Time variable: Year (5-year average)
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
Poverty
Poverty
Poverty
Headcount
Gap
Severity
Income
Inequality
Initial Dependent variable (•t-1)
1.0506
(31.51)***
0.8323
(7.21)***
0.8323
(7.21)***
0.9112
(11.22)***
Migrant remittances (lnREMPC)
-0.0217
(-3.04)***
-0.0292
(-2.07)**
-0.0584
(-2.07)**
-0.0014
(-0.27)
Human capital accumulation (lnHCA)
………..
………..
-0.3216
(-5.14)***
-0.6432
(-5.14)***
-0.0334
(-2.06)**
Real GDP per capita (lnY_PPP)
-0.0305
(-0.55)
-0.0748
(-2.18)***
-0.1496
(-2.18)***
0.0742
(2.85)
Investment in physical assets (lnINV)
0.0862
(1.73)*
-0.1702
(-2.20)**
-0.3405
(-2.20)**
0.0680
(2.77)***
Foreign direct investment (FDI)
-0.0178
(-3.30)**
………..
………..
………..
………..
………..
………..
Official development assistance (lnODA)
0.0195
(0.82)
0.0671
(1.93)*
0.1342
(1.93)*
-0.0200
(-2.97)***
Trade openness (lnOPN)
-0.1351
(-1.80)*
0.0417
(0.42)
0.0834
(0.42)
-0.1359
(-2.88)***
Rate of inflation (INF)
-0.0012
(-2.22)**
0.0005
(0.71)
0.0092
(0.71)
………..
………..
Government expenditure (lnGXP)
0.2993
(4.10)***
0.6133
(5.49)***
1.2266
(5.49)***
………..
………..
Real exchange rate (lnRXR)
0.0159
(0.53)
0.0354
(0.61)
0.0707
(0.61)
-0.0290
(-1.76)*
Business cycle (BZC)
………..
………..
………..
………..
………..
………..
-0.0090
(-5.75)***
Institutional quality (INS)
………..
………..
………..
………..
………..
………..
0.0035
(1.23)
Constant term
-0.5860
(-1.03)
0.2443
(0.30)
0.4885
(0.30)
0.4926
(1.39)
Number of observations
169
169
169
175
Number of groups
34
34
34
36
Number of instruments
25
25
25
25
[11],17667.07***
[11],2518.00***
[11],2518.00***
[10],596.03***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-5.186{0.604}
0.7467{0.455}
0.7467{0.455}
-0.0169{0.987}
Wald
[2 ],
Sargan test of over-identifying restrictions:
2
[13],
Source: Author‟s estimation
16.0871
20.0526*
20.0526*
11.4428
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics are in ( ), z-probabilities in { }
320
Stellenbosch University http://scholar.sun.ac.za
This may be due to the fact that, in SSA, the poorest countries in terms of real GDP per capita,
such as Burundi, Congo Republic, Ethiopia, Malawi and Rwanda, receive the least official
remittances per capita (see Figure 3.5). Besides, even in the 2000-2009 decade when SSA
received its highest amount of migrant remittances, no country could receive even up to US$1
per day, as Cape Verde (the highest recipient) received only US$249.88 per annum. This
seems to justify the apprehension of remittance-pessimists that severely poor families would
normally not have the means to sponsor migrants to high-income countries in order to receive
migrant remittances.
To the extent that remittances generally have a direct and instantaneous poverty-mitigating
effect on SSA, the findings of this study are consistent with the results obtained in related
previous studies including those of Adams and Page (2005), López-Córdova (2005), Acosta et
al. (2006), Gupta, Pattillo and Wagh (2009), Kalim and Shahbaz (2009), Gubert et al. (2010),
UN (2011), Shafiq et al. (2012) and Orzell (2013).
Table 7.2.1: Comparative Analysis of Remittance Effects on Poverty and Inequality in SSA
Type of Dummy Effect
Poverty Headcount
Poverty Gap
Poverty Severity
Income Inequality
Independent Median
0.1415 (2.83)***
0.2620 (4.73)***
0.5240 (4.73)***
0.1388 (6.27)***
MDV-Remittance Interactive
0.0452 (2.67)**
0.0750 (4.40)***
0.1500 (4.40)***
0.0357 (4.09)***
169
169
169
175
Number of observations
Number of groups
34
34
34
36
Instruments
26
26
26
25
Wald (χ²₍₀₎)
Arellano-Bond Test
Sargan Test (χ² ₁₃)
Source: Author‟s estimation
[12],
9448.46***
[12],
1240.28***
[12],
1240.28***
[11],
486.67***
-0.8852{0.3760}
0.7114{0.4769}
0.7114{0.4769}
0.1813{0.8561}
12.5386
18.8135
18.8135
11.5240
Note: **(***) denote statistical significance at 5(1) per cent respectively
2-step robust z-statistics in ( ); z-probabilities in { }
At the conventional levels of statistical significance, it can be concluded that remittances did not
contribute significantly to equalising incomes in „labour-exporting‟ SSA countries during the
1980-2009 period (Table 7.2). From Table 7.2.1, it is shown that migrant remittances received
actually aggravate poverty and income inequality in SSA countries with relatively higher levels
of poverty and income inequality. Thus, the potential poverty and income-equalising effects of
migrant remittance inflows are a preserve advantage for only remittance-receiving SSA
countries with relatively lower poverty and income inequality.
To a large extent, this result is consistent with the findings of Nguyen (2008) and Ekebe and Le
Goff (2009) that although remittances reduce poverty, they are less inequality-mitigating. This
321
Stellenbosch University http://scholar.sun.ac.za
finding, however, is in contrast with the findings by Acosta et al. (2008a) for 10 LAC countries
and that of Gubert et al. (2010) for Mali; that, remittances either reduce or have no effects on
income equality. Differences in sampled countries, period of study and the level of analysis
could be the source of this variation. For example, this is a purely macro-level study unlike in
the case of Acosta et al. (2008a) and Gubert et al. (2010) that are micro-level studies.
7.4.2 The Impact of Remittances on Labour Market Outcomes in SSA
The empirical results on the implications of remittance inflows for labour market outcomes in
SSA are presented in Table 7.3. Contrary to the trepidation of the remittance-pessimistic
structural dependence view, the findings of this study show that in SSA between 1980 and
2009, international migrant remittance inflows did not impair labour market outcomes, as they
contributed directly to reducing the unemployment rate. The estimated results (see Table 7.3)
show that a 100 per cent increase in international migrant remittances per capita had an
approximately -1.67 per cent impact on unemployment rate in the 27 sampled SSA countries.
Although this finding may seem to suggest that the direct unemployment-reducing impact of
international migrant remittances is economically low, it is important to note that international
remittances are the second most important only after trade openness in reducing
unemployment rate in SSA between 1980 and 2009.
Even more striking is the fact that between 1980 and 2009, the direct effect of international
remittance inflows was even more important than government expenditure when it comes to
solving the perennial unemployment problem in SSA. One possible explanation for this finding
is that whereas international remittance inflows are directly and instantaneously beneficial to
the private sector, as a result of bad governance and weak institutions, a colossal amount of
government spending over the period was not pro-poor or was due to public sector corruption
and debt servicing which culminated in crowding-out the private sector.
For the 36 sampled SSA countries, the impact of international remittance inflows on labour
market participation was both economically and statistically insignificant. Similarly, between
1980 and 2009, international migrant remittances did not have any significant impact on labour
productivity in SSA (Table 7.3). More specifically, in Table 7.3, it is shown that, on the average,
international remittances had no moral hazard effects with reference to labour force
participation and productivity, given that, the estimated coefficients of -0.0002 and 0.0046
respectively, are not only low but also statistically insignificant even at 10 per cent.
322
Stellenbosch University http://scholar.sun.ac.za
Table 7.3: Impact of Remittances on Labour Market Outcomes in SSA, 1980-2009
Group variable: Country Code
Time variable: Year (5-year average)
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
Labour
Unemployment
Participation
Labour
Productivity
Initial Dependent variable (•t-1)
0.4472
***
(17.62)
0.7878
***
(18.12)
0.7579
***
(13.24)
Migrant remittances (lnREMPC)
-0.0167
**
(-2.02)
-0.0002
(-0.48)
0.0046
(0.54)
Human capital accumulation (lnHCA)
0.0082
(0.13)
0.0047
**
(2.08)
0.0640
**
(2.44)
Real GDP per capita (lnY_PPP)
0.0364
(1.03)
0.0076
**
(2.26)
0.1046
***
(2.77)
Investment in physical capital (lnINV)
………..
………..
………..
………..
0.1248
***
(4.92)
Institutional quality (INS)
0.0071
(0.81)
0.00199
**
(2.41)
-0.0023
(-1.50)
Trade openness (lnOPN)
-0.1313
***
(-3.64)
………..
………..
-0.1117
**
(-2.50)
Rate of inflation (INF)
………..
………..
-0.0002
(-3.10)
-0.0016
***
(-2.80)
Government expenditure (lnGXP)
0.1977
**
(2.33)
………..
………..
-0.2170
***
(-6.65)
Real exchange rate (lnRXR)
………..
………..
0.0057
***
(2.99)
-0.0636
***
(-4.03)
Business cycle (BZC)
0.0270
***
(6.98)
……….
……….
……….
……….
Official development assistance (lnODA)
………..
………..
………..
………..
0.0194
(1.55)
Foreign direct investment (FDI)
………..
………..
………..
………..
0.0035
*
(1.80)
Constant term
0.8263
**
(2.61)
0.8578
***
(4.52)
1.6949
***
(4.29)
Number of observations
132
179
179
Number of groups
27
36
36
Number of instruments
22
Wald
[2 ],
21
[8],1093.41
***
[7],1310.25
26
***
[12],3193.04
***
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-1.1477{0.251}
0.1721{0.863}
0.6324{0.527}
Sargan test of over-identifying restrictions:
2
[13],
Source: Author‟s estimation
12.5546
18.9955
14.3350
0.483
0.123
0.351
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics in ( ), z-probabilities in { }
The results in Table 7.3.1 show that the unemployment-solving effect of remittances did not
prevail when the unemployment rate of an SSA country in any particular year fell below the
323
Stellenbosch University http://scholar.sun.ac.za
median level for the sampled SSA countries during the period under study. There is statistical
evidence of a size-effect in relation to unemployment rate. The implication here is that, over the
30-year period, the unemployment-reducing effects of migrant remittances received are merely
to the benefit of SSA countries with relatively higher unemployment rates.
Table 7.3.1: Comparative Analysis of Remittance Effects on Labour Market Outcomes
Type of Dummy Effect
Unemployment
Labour Participation
Labour Productivity
Independent Median
-0.4691 (-3.70)***
0.0105 (5.26)***
0.1211 (3.25)***
-0.0133 (-0.49)
0.0006 (0.77)
0.0176 (2.77)
Number of observations
132
180
179
Number of groups
27
36
36
Instruments
23
22
27
MDV-Remittance Interactive
Wald (χ²₍₀₎)
Arellano-Bond Test
Sargan Test (χ² ₁₃)
Source: Author‟s estimation
[11],
1609.80***
[8],
1287.00***
[13],
2683.23***
-1.1557{0.2478}
0.2109{0.8329}
0.7514{0.4524}**
12.6260
18.3883
14.6549
Note: **(***) denotes statistical significance at 5(1) per cent respectively
2-step robust z-statistics in ( ); z-probabilities in { }
The statistical significance of the estimated coefficients of the „autonomous‟ effects of MDV on
the rates of labour participation and labour productivity, (0.0105 and 0.1211 respectively),
provide evidence of size-effect of remittances on labour market outcomes (Table 7.3.1). In this
case, if received in larger amount, migrant remittances can have the potential of inducing
higher rates of productivity in countries with higher rates of labour productivity, but for countries
with lower rates of labour productivity the impact of remittances is zero. And, whereas in
migrant-receiving SSA countries with labour market participation rates above the median level
of the group, remittances have positive but statistically insignificant impact, in the case of
countries with lower rates of participation, the impact is negative but statistically insignificant. In
effect, it is only SSA countries with higher rates of labour participation that stand the chance of
benefitting more from international remittances received as the amount received increases. The
result that migrant remittances do not negate labour market outcomes confirms those obtained
by Drinkwater et al. (2006) for 19 developing countries and Orrenius et al. (2010) for Mexican
states.
324
Stellenbosch University http://scholar.sun.ac.za
7.4.3 The Impact of Remittances on Human Welfare and Development in SSA
The empirical results on the impact of international remittance inflows on human welfare and
human capital development in SSA are reported in Table 7.4.
Table 7.4: Human Development and Welfare Impact of Remittances in SSA, 1980-2009
Group variable: Country Code
Time variable: Year (5-year average)
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
Educational
Human
Attainment
Welfare
Life
Expectancy
Initial Dependent variable (•t-1)
0.8624
(19.96)***
0.8686
(13.43)***
0.7291
(32.89)***
Migrant remittances (lnREMPC)
0.0203
(4.43)***
0.0427
(2.81)***
0.0060
(2.24)**
Real GDP per capita (lnY_PPP)
0.0384
(3.13)***
0.0525
(2.02)**
………..
………..
Investment in physical capital (lnINV)
0.0173
(1.11)
………..
………..
0.0167
(1.73)*
Institutional quality (INS)
0.0034
(2.71)***
0.0157
(3.52)***
………..
………..
Trade openness (lnOPN)
-0.0343
(-1.67)*
0.0027
(0.07)
………..
………..
Rate of inflation (INF)
0.0010
(5.66)***
-0.0049
(-2.83)***
………..
………..
Government expenditure (lnGXP)
0.0191
(1.05)
………..
………..
-0.0023
(-0.15)
Real exchange rate (lnRXR)
0.0100
(2.32)**
0.0547
(2.88)***
0.0031
(1.52)
Business cycle (BZC)
0.0014
(1.96)**
……….
……….
0.0098
(16.42)***
Real lending rate (RLR)
………..
………..
-0.0038
(-2.69)***
………..
………..
Official development assistance (lnODA)
………..
………..
………..
………..
0.0194
(2.71)***
Foreign direct investment (FDI)
………..
………..
0.0002
(0.08)
-0.0015
(-1.45)
Constant term
-0.4601
(-5.25)***
-0.1282
(-0.51)
1.0163
(11.07)***
Number of observations
177
171
176
Number of groups
36
36
36
Number of instruments
24
23
22
[10],17893.81***
[9],4225.10***
[8],2299.28***
Wald
2
[ ],
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-1.3212{0.186}
-2.3683{0.018}
-1.3154{0.188}
Sargan test of over-identifying restrictions:
[2 ],
Source: Author‟s estimation
21.1125*
18.4159
16.8539
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics are in ( ), z-probabilities in { }
325
Stellenbosch University http://scholar.sun.ac.za
The findings of this study suggest that, in SSA, international remittance inflows have significant
positive effects on overall human welfare, educational attainment and life expectancy; with an
impact magnitude of about 0.0203 per cent, 0.0427 per cent and 0.0060 per cent respectively,
in response to a one percentage rise in remittances per capita (Table 7.4). The empirical
results show that, generally, international remittances received in SSA between 1980 and 2009
contributed positively to promoting human welfare (or socioeconomic wellbeing), educational
attainment, and life expectancy. These remittance-developmental effects, however, vary
according to the level of development in remittance-recipient countries (see Table 7.4.1).
As far as the positive effect of remittances on socioeconomic development is concerned, the
findings of this study are consistent with related previous studies by Adenutsi and Ahortor
(2010) for 31 developing countries from Latin America and the Caribbean (LAC) and SSA and
Adenutsi (2010a,b) for selected SSA countries. On the positive direct impact on schooling, the
results of this study confirm earlier results obtained by Cox-Edwards and Ureta (2003) for El
Salvador, and Ponce (2008) for Ecuador, in various household survey studies. Again, the
results of the direct positive impact of remittances on life expectancy are consistent with those
obtained by Ajayi et al. (2009) for 38 SSA countries based on 2007 data.
In Table 7.4.1, it is shown that the positive impact of remittances on human welfare is more
beneficial to SSA countries with relatively higher indices of human development (HDI). In the
case of educational attainment, the positive impact of remittances was only beneficial to SSA
countries with relatively lower level of educational attainment, as the impact of remittances on
educational attainment is actually negative (-0.0159) in migrant-sending SSA countries with
higher levels of attainment. With reference to life expectancy, the MDV-remittance interactive
effect suggests that over the past three decades, it is remittance-receiving SSA countries with
relatively higher life expectancy that benefited more from the direct positive impact of migrant
remittances.
326
Stellenbosch University http://scholar.sun.ac.za
Table 7.4.1: Comparative Analysis of Remittance Effects on Human Development and Welfare
Type of Dummy Effect
Human Welfare
Educational Attainment
Life Expectancy
Independent Median
0.1045 (6.34)***
0.0577 (6.23)***
0.0537 (4.79)***
MDV-Remittance Interactive
0.0126 (2.52)**
-0.0159 (-2.42)**
0.0146 (3.66)***
Number of observations
177
171
176
Number of groups
36
35
36
Instruments
Wald (χ²₍₀₎)
Arellano-Bond Test
Sargan Test (χ² ₁₃)
Source: Author‟s estimation
25
[11],
24
20805.06***
[10],
4287.92***
23
[9],
1381.19***
-1.3035{0.1924}
-2.4975{0.0125}**
-1.3199 {0.1869}
20.9558
18.8594
21.1165
Note: **(***) denotes statistical significance at 5(1) per cent
2-step robust z-statistics are in ( ); z-probabilities are in { }
7.4.4: The Impact of Remittances on Financial Development in SSA
7.4.4.1: The Impact of Remittances on Bank Credit to the Private Sector
Overall, migrant remittances did not impact on private sector credit allocation in SSA between
1980 and 2009. However, the results from the decade-by-decade analysis suggest that migrant
remittances exert a direct and a significant positive impact on bank credit allocation to the
private sector under sound macroeconomic policy environment, with this impact increasing
over time as SSA countries pursued financial liberalisation programme over the past three
decades. The results in Table 7.5.1 suggest that, in the 1980s and 2000s, a percentage rise in
international remittances per capita led to increased private sector credit allocation by banks by
0.0429 per cent and 0.0684 per cent respectively in SSA.
327
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.1: Impact of Remittances on Private Sector Bank Credit in SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-1989
1990-1999
2000-2009
1980-2009
Initial private sector credit (lnPSC_1)
0.5577
(26.13)***
0.7830
(13.65)***
0.9892
(47.66)***
0.7850
(28.20)***
Migrant remittances (lnREMPC)
0.0429
(7.14)***
-0.0350
(-6.50)***
0.0684
(8.68)***
0.0085
(1.09)
Real GDP per capita (lnY_PPP)
0.2062
(6.64)***
0.2643
(3.98)***
-0.0161
(-1.00)
0.3153
(5.94)***
Rate of inflation (INF)
-0.0015
(-10.69)***
-0.0020
(-6.91)***
0.0035
(4.17)***
-0.0010
(-2.08)**
Government expenditure (lnGXP)
-0.0232
(-1.22)
0.2611
(5.83)***
0.0407
(7.74)***
0.0183
(1.04)
Real lending rate (RLR)
-0.0004
(-0.10)
0.0045
(13.26)***
0.0037
(6.59)***
0.0024
(8.95)***
Trade openness (lnOPN)
0.0742
(1.92)*
-0.1762
(-4.30)***
-0.1105
(-5.37)***
-0.0704
(-3.99)**
Constant term
-0.5467
(-3.41)***
-1.3059
(-3.53)***
0.3440
(2.31)**
-1.5033
(-3.86)***
Number of observations
275
298
312
952
Number of groups
33
35
35
35
Number of instruments
51
51
52
441
35196.80***
3216.39***
10012.35***
2740.95***
Wald
2
[7],
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-1.3473{0.178}
-0.1804{0.857}
-1.2995{0.194}
-0.8380{0.402}
Sargan test of over-identifying restrictions:
[2 ],
Source: Author‟s estimation
[43], 17.7853
[43], 27.6982
[43], 27.1016
[433], 29.8626
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
2-step robust z-statistics are in ( ), z-probabilities are in { }
During the turbulent times of the 1990s (see Table 3.1), however, migrant remittance inflows
impacted negatively (-0.0350) on bank credit allocation in SSA probably because during this
era, migrant remittances were instantaneously spent on consumables rather than saved to
enable banks create more money. In fact, it was shown in Adenutsi et al. (2012) that official
remittances per capita and domestic savings are highly and positively correlated in SSA, a
coefficient of 0.867 over the period 1980-2009. Even though correlation does not necessarily
imply causation, it is evident from this study that when gross domestic savings as a percentage
of GDP were relatively high at 22.23 and 23.50 in the 1980s and 2000s respectively (see Table
3.1), remittances impacted more and positively on private sector credit allocation (Table 7.5.1).
However, when gross domestic savings ratio declined to 14.48 per cent in the 1990s (Table
3.1), the impact of remittances on credit allocation turned negative and with lower coefficient.
328
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.1.1 provides statistical justification for the decade-by-decade changing impact of
international migrant remittances on bank credit to the private sector in SSA over the past three
decades – 1980-89, 1990-99 and 2000-09. Indeed, in columns A-B, B-C and A-C of Table
7.5.1.1, the statistical significance of the computed „differential‟ z-statistics affirms the rejection
of the null hypothesis at one per cent level of statistical significance that the corresponding
estimated decade-based coefficients are not statistically different from one another. With the
exception of real lending rate, this conclusion actually holds for all other determinants of bank
credit to the private sector as reported in Table 7.5.1. The computed „differential‟ z-statistics
associated with real lending rate reported in columns A-B and A-C of Table 7.5.1.1 shows that
the estimated coefficient of real lending rate for the 1980-89 decade is not statistically different
from the corresponding estimates for the 1990-99 decade and the 2000-09 decade
respectively.
From the computed z-statistics reported in columns A-D, B-D, B-E and C-E of Table 7.5.1.1, it
can be concluded that, generally, the variations in the estimated decade-based parameters of
migrant remittances are statistically stable over time. Thus, there is instability in the decadebased parameter estimates reported in Tale 7.5.1. Apart from a few violations like initial private
sector credit and trade openness (as reported in column B-D), real GDP per capita PPP (with
reference to column B-E) and government expenditure and real lending rate (as reported in
column C-E), there is a substantially significant statistical evidence for coefficient instability
over time for each of the explanatory variables. Therefore, it can be concluded that, on the
average, the estimated decade-based coefficients of migrant remittance inflows on bank credit
to the private sector in SSA are evolving and the evolution is instable across the three decades,
1980-89 1990-99 and 2000-09.
329
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.1.1: Results of Parameter Evolution and Instability Tests for Impact of Migrant Remittances on Private Sector Credit in SSA
Decade-Based Rolling
Estimated Results
Estimated Decade-Based Results
Initial private sector credit (lnPSC_1)
Migrant remittances (lnREMPC)
Real GDP per capita (InY_PPP)
Rate of inflation (INF)
Government expenditure (lnGXP)
Real lending rate (RLR)
Trade openness (lnOPN)
Constant term
A
B
C
D
E
1980-1989
1990-1999
2000-09
1985-1994
1995-2004
Non-Overlapping Decade-Based
Coefficient Stability Test Results
Overlapping Decade-Based Coefficient
Stability Test Results
A-B
B-C
A-C
A-D
0.5577
0.7830
0.9892
0.7256
0.8917
-0.2254
-0.2061
-0.4315
-0.1680
B-D
0.0574
B-E
-0.1086
C-E
0.0975
[0.0213]
[0.0574]
[0.0207]
[0.0176]
[0.0209]
[0.0360]
[0.0366]
[0.0006]
[0.0037]
[0.0398]
[0.0364]
[0.0002]
{26.13}***
{13.65}***
{47.66}***
{41.22}***
{42.58}***
{-6.25}***
{-5.63}*** {-731.37}***
{-44.91}***
{1.44}
0.0429
-0.0350
0.0684
0.0000
0.0115
0.0779
-0.1034
-0.0255
0.0429
-0.0350
-0.0465
0.0569
[0.0060]
[0.0054]
[0.0079]
[0.0040]
[0.0038]
[0.0006]
[0.0025]
[0.0019]
[0.0020]
[0.0014]
[0.0016]
[0.0041]
{7.14}***
{-6.50}***
{8.68}***
{0.01}
{3.04}***
{123.63}***
{-41.36}***
{-13.64}***
{21.34}***
{-25.36}***
{-29.24}***
{13.91}***
0.2062
0.2643
-0.0161
0.4641
0.2466
-0.0581
0.2804
0.2223
-0.2579
-0.1998
0.0177
-0.2627
[0.0311]
[0.0664]
[0.0160]
[0.0157]
[0.0375]
[0.0353]
[0.0503]
[0.0149]
[0.0154]
[0.0507]
[0.0289]
[0.0214]
{6.64}***
{3.98}***
{-1.00}
{29.54}***
{6.58}***
{-1.64}*
{5.57}***
{14.87}***
{-16.80}***
{-3.94}***
{0.61}
{-12.29}***
-0.0015
-0.0020
0.0035
-0.0027
-0.0003
0.0006
-0.0055
-0.0049
0.0012
0.0007
-0.0017
0.0038
[0.0001]
[0.0003]
[0.0008]
[0.0003]
[0.0002]
[0.0002]
[0.0005]
[0.0007]
[0.0002]
[0.0000]
[0.0001]
[0.0006]
{-10.69}***
{-6.91}***
{4.17}***
{-8.58}***
{-1.49}
{3.73}***
{-10.15}***
{-7.10}***
{7.29}***
{34.00}***
{-19.11}***
{5.97}***
-0.0232
0.2611
0.0407
0.1832
0.0213
-0.2843
0.2204
-0.0639
-0.2064
0.0779
0.2397
0.0193
[0.0190]
[0.0448]
[0.0053]
[0.0367]
[0.0260]
[0.0257]
[0.0395]
[0.0138]
[0.0177]
[0.0081]
[0.0188]
[0.0208]
{-1.22}
{5.83}***
{7.74}***
{4.99}***
{0.82}
{-11.04}***
{5.58}***
{-4.63}***
{-11.68}***
{9.65}***
{12.78}***
{0.93}
-0.0004
0.0045
0.0037
0.0054
0.0039
-0.0048
0.0008
-0.0041
-0.0058
-0.0009
0.0005
-0.0002
[0.0002]
{-2.98}** {513.26}***
[0.0036]
[0.0003]
[0.0006]
[0.0004]
[0.0004]
[0.0033]
[0.0002]
[0.0030]
[0.0032]
[0.0001]
[0.0000]
{-0.10}
{13.26}***
{6.59}***
{12.67}***
{10.78}***
{-1.48}
{3.50}***
{-1.34}
{-1.82}**
{-10.22}***
{18.00}***
{-1.18}
0.0742
-0.1762
-0.1105
-0.1608
-0.0102
0.2504
-0.0657
0.1848
0.2350
-0.0154
-0.1660
-0.1004
[0.0203]
[0.0001]
[0.0387]
[0.0410]
[0.0206]
[0.0177]
[0.0207]
[0.0023]
[0.0204]
[0.0181]
[0.0210]
[0.0233]
{1.92}**
{-4.30}***
{-5.37}***
{-9.11}***
{-0.49}
{108.88}***
{-3.22}***
{10.21}***
{11.18}***
{-0.66}
{-8.19}*** {-772.08}***
-0.5467
-1.3059
0.3440
-2.4567
-1.6032
0.7592
-1.6499
-0.8907
1.9100
1.1508
0.2973
1.9472
[0.1603]
[0.3699]
[0.1476]
[0.1810]
[0.2969]
[0.2096]
[0.2224]
[0.0127]
[0.0207]
[0.1889]
[0.0730]
[0.1493]
{-3.41}***
{-3.53}***
{2.31}**
{-13.57}***
{-5.40}***
{3.62}***
{-7.42}***
{-69.91}***
{92.18}***
{6.09}***
{4.07}***
{13.04}***
275
298
312
289
308
286.5
305
293.5
282
293.5
303
310
Number of groups
33
35
35
34
35
34
35
34
34
35
35
35
Number of instruments
51
51
51
51
51
51
51
51
51
51
51
51
19206.8***
Number of observations
Wald statistic
A-B 2ⁿᵈ-order autocorrelation test
35196.80***
3216.39***
10012.35*** 25567.30*** 12671.59***
-1.347(0.18)
-0180(0.86)
-1.300(0.19) -1.374(0.17) 0.574(0.57)
-
-
-
-
-
-
-
Sargan over-identifying restrictions
17.785(0.99) 27.700(0.97) 27.102(0.97) 20.915(0.99) 28.602(0.95)
-
-
-
-
-
-
-
Source: Author‟s estimation
6614.37*** 22604.58*** 30382.05*** 14391.85***
*/**/*** denotes significant at 10/5/1 per cent statistical levels respectively.
Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
330
7943.99*** 11341.97***
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.1.2 reveals that, under favourable macroeconomic conditions, SSA countries with
relatively higher levels of financial development measured in terms of private sector credit
stand a better chance to develop their financial systems through higher remittance inflows.
During periods of harsh economic conditions, the negative impact of migrant remittances on
private sector credit allocation is higher (-0.0376) in countries with fairly developed financial
markets than in those with relatively underdeveloped financial markets, given that, for the entire
group, the impact was 0.0350. The finding that remittances directly contribute to financial
development as far as private sector credit allocation is concerned, confirms the results
obtained in previous studies notably those of Aggarwal et al. (2006), Shahbaz et al. (2007),
Giuliano and Ruiz-Arranz (2009), Ambrosius (2011), and Gani and Sharma (2013).
Table 7.5.1.2: Comparative Analysis of Remittance Effects on Private Sector Credit in SSA
Type of Duumy Effect
Independent MDV
1980-1989
1990-1999
0.2718 (17.55)***
0.1522 (9.73)***
MDV-Remittance Interactive 0.0455 (3.31)***
2000-2009
1980-2009
0.1955 (11.32)*** 0.2065 (12.33)***
-0.0376 (-2.81)***
0.0753 (7.46)***
0.0350 (6.86)***
Number of observations
275
298
312
952
Number of groups
33
35
35
35
Instruments
Wald ( χ²₈ )
52
52
52
442
34185.29***
2321.43***
4212.82***
2881.88***
-1.3538{0.1758}
[43], 22.1234
-0.3827{0.7019}
[43], 27.6027
-1.0641{0.2873}
[43], 27.6188
-0.7925{0.4280}
[443], 29.5662
Arellano-Bond Test
Sargan Test (χ² ₍₀₎ )
Source: Author‟s estimation
Note: *** denotes statistical significance at 1 per cent respectively
2-step robust z-statistics are in ( ); z-probabilities are in { }
As a final point, a mere academic exercise was undertaken to show that empirically, in
comparison with robust static panel-data modelling, the estimated robust dynamic panel-data
modelling of the impact of migrant remittance inflows on financial development in SSA over the
period 1980-2009 actually produced more convincing results. The static panel-data version of
the estimated impact of migrant remittance inflows on bank credit to the private sector in SSA
over the period, 1980-2009, is reported in Table A7.4 in the Appendix. Static panel-data FE and
RE models were estimated and the Hausman test was performed to select the estimated FE
model as the better of the two in the empirical context of this estimation. When the BreuschPagan test for heteroskedasticity was carried out, it was revealed that, at 10 per cent level of
statistical significance, the standard errors of the estimated conventional models are not
homoscedastic, hence the need to re-estimate and rely upon heteroskedasticity-corrected
robust estimations. By implication, the most reliable and efficient estimated result among the
class of empirical static panel-data estimations in this very context is the estimated robust FE
331
Stellenbosch University http://scholar.sun.ac.za
model reported in Table A7.4. The reported R 2 of 0.2365 implies that the empirical robust FE
model estimating the impact of migrant remittance inflows on private sector credit in SSA has
an overall explanatory power only 23.65 per cent. Thus, the overall performance of the
empirical robust FE static panel-data model perform very poor since the percentage of the
residuals sum of squares to the total sum of squares is as much as 76.35. The reported z statistics of the robust FE estimation show that, in SSA, migrant remittances, rate of inflation,
and government expenditure do not explain bank credit to the private sector whilst trade
openness and real GDP per capita PPP do positively impact on private sector credit. To a
reasonable extent, the results of the robust FE static panel-data estimation (as reported in
Table A7.4) validate the results of the two-step sys-GMM dynamic panel-data estimation
(reported in Table 7.5.1) as far as the impact of remittance inflows on private sector credit is
concerned for the overall period, 1980-2009 as in both cases, though with contradictory signs,
migrant remittances do not statistically influence bank credit allocation to the private sector in
SSA.
7.4.4.2 The Impact of Remittances on M 2 / GDP in SSA
Table 7.5.2 presents the result of a consistent positive trend of the impact of remittances on
M 2 / GDP as a complementary indicator of financial development over the past three decades.
In the early years of the adoption of financial liberalisation programmes (i.e. in the 1980s), the
impact of a 100 percentage rise in migrant remittances per capita on M 2 / GDP was
approximately negative 5.16 per cent. The impact of remittances on M 2 / GDP turned positive
(3.41 per cent) in the 1990s and this positive impact became even more robust (4.08 per cent)
in the 2000s. Therefore, it can be concluded that the contribution of migrant remittances to
financial development when measured as M 2 / GDP , was increasingly positive and robust in
SSA between 1980 and 2009.
332
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.2: Impact of Remittances on Broad Money-GDP Ratio in SSA, 1980-2009
Group variable: Country Code
Time variable: Year
Two-Step Estimation by Blundell-Bond System Dynamic Panel-Data Procedure
1980-1989
1990-1999
2000-2009
1989-2009
Lagged dependent variable (lnM2/GDP _1)
0.6088
(37.65)***
0.5812
(25.61)***
0.8990
(18.45)***
0.6884
(12.73)***
Migrant remittances (lnREMPC)
-0.0516
(-4.94)***
0.0341
(8.06)***
0.0408
(5.39)***
0.0058
(2.34)**
Human capital accumulation (lnHCA)
-0.1847
(-4.62)***
0.0814
(4.32)***
0.0166
(0.55)
0.1316
(3.92)***
Real GDP per capita (lnY_PPP)
0.4234
(7.05)***
0.0403
(1.65)*
-0.2401
(14.19)***
0.0446
(0.83)
Official development assistance (lnODA)
-0.0228
(-1.00)
0.0135
(1.55)
0.0121
(2.15)**
0.0175
(2.84)***
Rate of inflation (INF)
0.0056
(6.08)***
-0.0007
(-2.72)***
0.0025
(7.48)***
0.0015
(5.23)***
Government expenditure (lnGXP)
-0.1558
(-3.04)***
0.2322
(11.33)***
0.0652
(16.52)***
0.1016
(1.93)*
Real exchange rate (lnRXR)
0.0621
(3.54)***
-0.0109
(-0.87)
-0.0043
(-0.33)
0.0254
(2.15)**
Real lending rate (RLR)
0.0135
(9.05)***
0.0028
(10.01)***
0.0068
(17.72)***
0.0064
(25.79)***
Constant term
-1.1817
(-2.40)**
0.0500
(0.19)
1.4746
(4.15)***
-0.2946
(-1.03)
Number of observations
266
298
312
942
Number of groups
32
35
35
35
Number of instruments
53
53
53
443
328835.92***
7618.21***
2337.83***
3042.13***
Wald
2
[9],
Arellano-Bond test for zero autocorrelation in first-difference errors (order 2):
-0.9692{0.332}
-1.5505{0.121}
-1.8405{0.066}
*
-1.0435{0.296}
Sargan test of over-identifying restrictions:
[2 ],
Source: Author‟s estimation
[43], 18.4130
[43], 29.0973
[43], 28.5025
[433], 27.0762
*/**/*** denotes statistical significance at 10/5/1 per cent respectively
2-step robust z-statistics in ( ), z-probabilities in { }
In a similar fashion, remittance-receiving SSA countries within the sub-region with relatively
higher levels of liquidity received an overall higher impact of 0.0207 remittances over the entire
study period (Table 7.5.2.1), compared to the general impact of 0.0058 for the full sample
(Table 7.5.2). This also holds for the most recent decade, 2000-2009. Nevertheless, in the
early years of financial liberalisation, the impact on remittances on SSA countries with relatively
higher levels of M 2 / GDP was lower (-2.98 per cent and 1.90 per cent for the 1980s and the
1990s respectively) than the average impact of -5.16 per cent and 3.41 per cent for the entire
sampled 36 SSA countries.
333
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.2.1: Comparative Analysis of Remittance Effects on Broad Money-GDP Ratio in SSA
Type of Dummy Effect
Independent Dummy
1980-1989
1990-1999
2000-2009
1980-2009
0.2258 (3.50)***
0.1349 (9.03)***
0.1710 (9.71)***
0.1950 (6.54)***
0.0190 (4.90)***
0.0513 (8.27)***
0.0207 (4.40)***
MDV-Remittance Interactive -0.0298 (-4.50)***
Number of observations
266
298
312
942
Number of groups
32
35
35
35
Instruments
54
54
54
444
9001.79***
11688.22***
11632.52***
-1.0068{0.3140}
-1.4279{0.1533}
-1.6676{0.0954}*
-1.0420{0.2974}
[43], 18.5894
[43], 28.9679
[43], 26.7313
[443], 22.2294
Wald ( χ²₁₀ )
Arellano-Bond Test
Sargan Test (χ² ₍₀₎ )
Source: Author‟s estimation
36401.62
***
Note: *(***) denotes statistical significance at 10(1) per cent respectively
2-step robust z-statistics in ( ); z-probabilities in { }
The finding that migrant remittances generally impact positively on M 2 / GDP is in consonance
with the results obtained in earlier related studies reviewed in this study (see Table A7.1).
In Table 7.5.2.2, the results of the statistical differences and stability of the estimated decadebased coefficients of the impact of migrant remittances on broad money as a ratio of GDP
(M2/GDP) in SSA are reported. Columns A-B, B-C and A-C report the „differential‟ z-statistics of
the statistical differences between the estimated coefficients of 1980s and 1990s, 1990s and
2000s, and 1980s and 2000s respectively. Under the working hypothesis that the estimated
decade-based coefficients truly differ from each other statistically when the computed
„differential‟ z -statistics centre further away from zero, the results suggest that the estimated
decade-based coefficients are actually different from each other at five per cent level of
statistical significance. Thus, international migrant remittance inflows have a decade-based
changing impact on M2/GDP in SSA between 1980 and 2009. In fact, for the period 1980-2009,
the changing impact of the other determinants of M2/GDP as reported in Table 7.5.2 generally
holds.
334
Stellenbosch University http://scholar.sun.ac.za
Table 7.5.2.2: Results of Parameter Evolution and Instability Tests for Impact of Migrant Remittances on Broad Money Supply in SSA
Decade-Based Rolling
Estimated Results
Estimated Decade-Based Results
Lagged dependent variable (lnM₂/GDP)
Migrant remittances (lnREMPC)
Human capital accumulation (lnHCA)
Real GDP per capita (lnY_PPP)
Official development assistance (lnODA)
Rate of inflation (INF)
Government expenditure (lnGXP)
Real exchange rate (lnRXR)
Non-Overlapping Decade-Based
Coefficient Stability Test Results
Overlapping Decade-Based Coefficient Stability
Test Results
A
B
C
D
E
1980-1989
1990-1999
2000-2009
1985-1994
1995-2004
0.6088
0.5812
0.8990
0.1661
0.9227
0.0276
-0.3178
-0.2902
0.4428
0.4151
-0.3414
-0.0236
[0.0162]
[0.0227]
[0.0487]
[0.0061]
[0.0318]
[0.0065]
[0.0260]
[0.0326]
[0.0101]
[0.0166]
[0.0091]
[0.0169]
{37.65}***
{25.61}***
{18.45}***
{27.30}***
{29.02}***
{4.24}***
{-12.2}***
{-8.91}***
{43.88}***
{24.99}***
{-37.52}***
{-1.40}
-0.0516
0.0341
0.0408
-0.0239
0.0020
-0.0857
-0.0067
-0.0924
-0.0277
0.0581
0.0321
0.0388
A-B
B-C
A-C
A-D
B-D
B-E
C-E
[0.0105]
[0.0042]
[0.0076]
[0.0041]
[0.0060]
[0.0062]
[0.0033]
[0.0029]
[0.0063]
[0.0001]
[0.0017]
[0.0016]
{-4.94}***
{8.06}***
{5.39}****
{-5.83}***
{0.34}
{-13.809}***
{-2.00)**
{-32.14}***
{-4.37}***
{460.95}***
{18.52}***
{24.23}***
-0.1847
0.0814
0.0166
-0.2430
0.1778
-0.2661
0.0648
-0.2013
0.0583
0.3244
-0.0964
-0.1612
[0.0400]
[0.0188]
[0.0302]
[0.0195]
[0.0258]
[0.0212]
[0.0113]
[0.0098]
[0.0205]
[0.0007]
[0.0069]
[0.0044]
{-4.62}***
{4.32}***
{0.55}
{-12.45}***
{6.90}***
{-12.58}***
{5.72}***
{-20.50}***
{2.85}**
{470.12}***
{-13.89}***
{-36.72}***
0.4234
0.0403
-0.2401
-0.1948
-0.1756
0.3832
0.2804
0.6636
0.6182
0.2350
0.2159
-0.0645
[0.0601]
[0.0244]
[0.0169]
[0.0148]
[0.0246]
[0.0356]
[0.0075]
[0.0431]
[0.0453]
[0.0096]
[0.0002]
[0.0077]
{7.05}***
{1.65}*
{-14.19}***
{-13.15}***
{-7.14}***
{10.75}***
{37.43}***
{15.38}***
{13.66}***
{24.48}*** {1199.28}***
{-8.41}***
-0.0228
0.0135
0.0121
-0.0151
0.0082
-0.0363
0.0014
-0.0348
-0.0077
0.0285
0.0053
0.0039
[0.0228]
[0.0087]
[0.0056]
[0.0120]
[0.0038]
[0.0141]
[0.0031]
[0.0172]
[0.0107]
[0.0034]
[0.0048]
[0.0017]
{2.20}**
{-1.00}
{1.55}
{2.15}**
{-1.25}
{2.13}**
{-2.58}**
{0.46}
{-2.03}**
{-0.72}
{8.52}***
{1.09}
0.0056
-0.0007
0.0025
0.0041
0.0005
0.0063
-0.0032
0.0031
0.0015
-0.0048
-0.0012
0.0020
[0.0009]
[0.0003]
[0.0003]
[0.0002]
[0.0004]
[0.0007]
[0.0001]
[0.0006]
[0.0008]
[0.0001]
[0.0001]
[0.0000]
{6.08}***
{-2.72}**
{7.48}***
{25.40}***
{1.33}
{9.37}***
{-40.13}***
{5.20}***
{1.91}**
{-53.67}***
{-10.67}***
{67.87}***
-0.1558
0.2322
0.0652
0.0909
0.0244
-0.3880
0.1671
-0.2210
-0.2467
0.1413
0.2079
0.0408
[0.0513]
[0.0205]
[0.0039]
[0.0112]
[0.0200]
[0.0307]
[0.0166]
[0.0473]
[0.0401]
[0.0093]
[0.0005]
[0.0160]
{-3.04}***
{11.33}***
{16.52}***
{8.15}***
{1.22}
{-12.62}***
{10.09}***
{-4.67}***
{-6.15}***
{15.13*** {392.20}***
{2.55}**
0.0621
-0.0109
-0.0043
-0.1896
0.0059
0.0730
-0.0066
0.0665
0.2518
0.1787
-0.0169
-0.0103
[0.0175]
[0.0125]
[0.0132]
[0.0124]
[0.0099]
[0.0050]
[0.0006]
[0.0044]
[0.0051]
[0.0001]
[0.0026]
[0.0032]
{49.27}*** {1624.64}***
{-6.36}***
{-3.16}***
{3.54}***
{-0.87}
{-0.33}
{-15.24}***
{0.60}
{14.61}***
{-10.97}***
{15.10}***
0.0135
0.0028
0.0068
0.0071
0.0040
0.0107
-0.0039
0.0068
0.0064
-0.0043
-0.0011
0.0028
[0.0015]
[0.0003]
[0.0004]
[0.0002]
[0.0003]
[0.0012]
[0.0001]
[0.0011]
[0.0013]
[0.0001]
[0.0000]
[0.0001]
{9.05}***
{10.01}***
{17.72}***
{35.18}***
{14.38}***
{8.77}***
{-39.30}***
{6.04}***
{4.93}***
{-53.63}***{-1140.00}***
{27.90}***
-1.1817
0.0500
1.4746
5.4343
0.8237
-1.2318
-1.4246
-2.6564
-6.6160
-5.3842
-0.7736
0.6510
[0.4924]
[0.2634]
[0.3553]
[0.1521]
[0.2416]
[0.2290]
[0.0920]
[0.1370]
[0.3402]
[0.1112]
[0.0218]
[0.1138]
{-2.40}**
{0.19}
{4.15}***
{35.72}***
{3.41}***
{-5.38}***
{-15.49}***
{-19.38}***
{-19.45}***
{-48.41}***
{-35.45}***
{5.72}***
266
298
312
284
308
282
305
289
275
291
303
310
Number of groups
32
35
35
35
35
34
35
34
34
35
35
35
Number of instruments
53
53
53
53
53
53
53
53
53
53
53
53
Wald statistic
A-B 2ⁿᵈ-order autocorrelation test
328835.92***
7618.21***
2337.83***
6078.18***
6848.20*** 10049.56***
7409.40***
-0.969(0.33) -1.551(0.12)
-1.841(0.07) -1.146(0.252) -2.009(0.045)
-
-
-
-
-
-
-
Sargan over-identifying restrictions
18.413(0.99) 29.097(0.95)
28.503(0.956) 23.018(0.995) 26.477(0.978)
-
-
-
-
-
-
-
Real lending rate (RLR)
Constant term
Number of observations
Source: Author‟s estimation
12480.96*** 168227.07*** 4978.02*** 165586.88*** 167457.05***
*/**/*** denotes significant at 10/5/1 per cent statistical levels respectively.
Standard errors in [ ], z-statistics in { }, 2 probabilities in ( )
335
Stellenbosch University http://scholar.sun.ac.za
As reported in columns A-D, B-D, B-E and C-E of Table 7.5.2.2, the differences in the
estimated decade-based coefficients of international migrant remittance inflows are statistically
consistent showing coefficient instability across the three decades within 99 per cent
confidence interval. Indeed, the evolution in the estimated decade-based coefficients of all
other explanatory variables reported in Table 7.5.2.2 is consistently and statistically unstable
across the three decades between 1980 and 2009. The only exception to this general
observation is the estimates of official development assistance (ODA) as shown in columns AD and B-E in Table 7.5.2.2. In conclusion, the estimated decade-based coefficients reported in
Table 7.5.2.2 are actually different and this evolution is statistically consistent, implying
parameter instability from decade to decade over the 1980-2009 period.
Finally, to satisfy the intellectual curiosity of persons who may be interested in seeing how
migrant remittance inflows affected broad money-GDP ratio in SSA between 1980 and 2009
within the confines of static panel-data analytical framework, FE and RE models were
estimated. The static panel-data empirical results of the impact of international migrant
remittance inflows on broad money as a ratio of GDP in SSA over the period, 1980-2009 are
presented in Table A7.5. Out of the four static panel-data estimations conducted comprising
conventional FE, conventional RE, robust FE and robust RE estimations, the Hausman
specification test statistic and the Breusch-Pagan heteroskedasticity test statistic jointly suggest
that the most efficient and reliable result in the context of static estimations is reported under
robust GLS RE empirical model presented in Table A7.5 in the Appendix. Although in the
presence of heteroskedasticity the conventional RE results indicate that migrant remittance
inflows, real GDP per capita PPP, human capital accumulation, and real lending rate positively
stimulate variations in broad money-GDP ratio, government expenditure and real exchange
rate negatively affect broad money-GDP ratio with reference to SSA between 1980 and 2009.
However, when the robust RE estimation to correct heteroskedasticity was undertaken, besides
human capital accumulation, none of the explanatory variables statistically impact on broad
money-GDP ratio, thereby sharply contradicting the results of the dynamic panel-data
estimation by sys-GMM as reported under Table 7.5.2.
336
Stellenbosch University http://scholar.sun.ac.za
7.5 CONCLUSIONS AND POLICY RECOMMENDATIONS
This chapter explored the impact of migrant remittances on various aspects of economic
development, viz. poverty, income inequality, unemployment, labour participation, labour
productivity, human welfare, educational attainment, life expectancy, and financial development
in SSA. In general, an affirmative common answer to the research questions of this chapter is
hereby given, that, migrant remittance inflows have a positive developmental impact on SSA,
with no evidence for moral hazard effects on labour market outcomes. Based on the empirical
findings, this study concludes that, in remittance-receiving SSA countries:
i.
Remittances have huge potential for promoting economic development by way of
helping to reduce poverty, but not necessarily income inequality. In fact, there is some
evidence for an income inequality-equalising effect of remittances, but this is not
statistically significant at the conventional levels, probably because the amount
remittances received per migrant is too low in the sub-region. In countries with a
relatively higher incidence of poverty and income inequality, remittances actually
exacerbate poverty and income inequality.
ii.
Remittances contribute directly to reducing the rate of unemployment. However, in
countries with mild unemployment problems, increased inflows of remittances alone
may not be sufficient to reduce or solve the perennial frictions in their labour markets. It
is in countries with relatively „unmanageable‟ unemployment rates that migrant
remittances directly contribute more substantially to reducing the unemployment rates.
iii.
Although, in general, remittances have no long-run impact on labour force participation,
there are prospects that if more remittances are received, they can actually contribute
positively to higher labour force participation in countries with relatively higher rates of
labour market participation. In countries where labour productivity rates are relatively
high, remittances impact positively on productivity. In either case, the findings of this
study indicate that migrant remittance inflows to SSA do not lead to moral hazard
effects in the labour market, probably because the amount of remittances received by
the sub-region is too low for recipients to solely rely on as a means of livelihood.
iv.
Over the past three decades, remittances contributed significantly to promoting human
welfare, educational attainment and life expectancy. But, whereas remittances
promoted higher general human welfare in countries with relatively lower HDI and
337
Stellenbosch University http://scholar.sun.ac.za
lower rates of educational attainment, it is in countries with relatively higher years of life
expectancy that remittances contributed more to increasing life expectancy. This may
be due to the fact that in very poor countries, availability of quality healthcare facilities
and access to modern medical services are more challenging. This implies that in the
absence of quality healthcare infrastructure and motivated professionals, migrant
remittances may not contribute significantly to increasing life expectancy in remittancerecipient countries.
v.
Since the adoption of financial liberalisation programmes in the 1980s, migrant
remittances have been stimulating financial development through higher private sector
credit and broad money to GDP ratio. However, without a sound macroeconomic policy
environment, the impact of remittances on financial development can be negligible and
even negative under extreme adverse economic conditions. In the contemporary world,
the higher the level of financial development of a country, the more robust the impact of
migrant remittances on financial market development in response to macroeconomic
conditions.
vi.
The contribution of remittances to financial development has been increasing steadily
over the years under sound macroeconomic policy environment. The contribution of
remittances to financial development in the recent past decade (2000-2009) has been
greater than at the onset of adopting financial liberalisation programme in the 1980s.
This trend points to the fact that there is a higher prospect of migrant remittances
contributing to the development of financial markets in SSA provided prudent policy
measures are put in place to stabilise the macroeconomic environment.
The main implications of the conclusions158 enumerated above are that migrant remittances
promoted economic development in SSA over the past three decades, but the optimal
contribution of remittances to the development of SSA has not been realised, mainly because
the sub-region failed to: (i) mobilise adequate remittances from its migrants; and (ii) create the
ideal macroeconomic policy environment for remittances to contribute effectively towards
economic development. This is because even though remittances have a potential to engender
158
It is important to note that with regard to the conclusions based on the distributive effects of the comparative
analyses in particular, given that higher unofficial remittances are more likely to be received in poorer countries with
weaker institutions and financial infrastructure, the developmental-impact of remittances in these countries could be
underestimated if only official remittances are used, as happened in the case of this study because data on unofficial
remittances are not available over the study period and across the sampled countries.
338
Stellenbosch University http://scholar.sun.ac.za
economic development in SSA, the specific socioeconomic and geopolitical conditions of the
migrant-home country, by and large, determine the extent to which the benefits of remittances
can be exploited.
Consequently, there is the need to attract optimal official remittances in order to realise the full
developmental-impact of remittances in SSA. In addition to the specific policy measures that
can be put in place to attract higher inflows of official remittances noted in the preceding
chapters, this study recommends the need for governments to stabilise the macroeconomic
policy environment by ensuring lower and stable prices, creating an ideal investment climate,
and being committed to pursuing pro-growth policies which have equitable distributive effects.
Besides pursuing sound macroeconomic policies, this study prescribes the following specific
policies towards enhancing the developmental-impact of migrant remittances in SSA:
i.
Pursue complementary development strategies towards eliminating market distortions
in favour of the rich, so as to reduce poverty, income inequality and higher
unemployment rates. Poverty alleviation policies such as capacity building, vocational
training, access to venture capital and microcredit, and other SME incentive packages
through which the economically-disadvantaged and the vulnerable groups such as
deprived rural dwellers and marginalised women stand a better chance to gain and
improve upon their welfare, can be useful in enhancing the poverty-alleviating and
inequality-reducing effects of remittances.
ii.
Adopt an integrated economic development programme in which fiscal and monetary
policies are well-coordinated in a manner that will ensure that progressive price and
income policies are designed and implemented towards bridging the gap between highincome and low-income earners at all levels of economic development.
iii.
Enhance the institutional capacity of banks and other financial institutions including
social security and pension funds, credit unions and microfinance institutions so that
these institutions can develop innovative products for the Diaspora population and at
the same time ensure optimal financial inclusiveness. In this way, more official
remittances can be received and managed more effectively for sustainable
socioeconomic development.
339
Stellenbosch University http://scholar.sun.ac.za
iv.
Encourage the formation of vibrant and progressive migrant hometown associations in
all major migrant host countries as a means to mobilise more remittances to finance
development projects at the community level. Community-based projects financed by
remittances are often of public nature and the benefits accrue to all users irrespective of
income status and, whether or not, a family has an international migrant.
v.
Enhance the institutional capacity of local governments to enable them to design and
implement cutting-edge attractive development programmes according to their unique
culture, history, heritage, and to develop tourist attractions. Through this, effective social
and business networks can be formed to facilitate the mobilisation of resources to
finance critical development projects. Furthermore, through these networks a platform
can be created to harness the intellectual capital of migrants for the benefit of their
native countries.
340
Stellenbosch University http://scholar.sun.ac.za
APPENDIX 7
Table A7.1: Summary of Empirical Studies on the Impact of Remittances on Economic Development
Author(s), Year
Case Study
Study Period
Model & Estimation
Method
OLS,
simultaneous
equations
with
distributed lag feature
estimated by 2SLS
Glytsos (2002)
Egypt,
Greece,
Jordan,
Morocco,
Portugal
1969-1998
Adams and Page
(2005)
71
Developing
countries
New
dataset
collected
by
authors (19801999)
depending upon
country-specific
availability
OLS and 1 -Stage IV
estimation
López-Córdova
(2005)
Mexican
municipalities
(a
survey
data of more
than
2400
households)
2000
crosssectional data
2SLS estimation with
remittances
as
an
instrumental variable
Acosta et al. (2006)
10
LAC
countries
Cross-sectional
data based on
most
recent
National
Cross-sectional
country-specific
data
analysis by OLS and
Instrumental
variable
st
Variables Included
Key Finding(s)
Dependent: Private consumption expenditure, gross
domestic investment (private and public) including
changes in stocks, imports (M), disposable income
comprising GDP and volume of remittances.
Explanatory:
GDP
growth
rate,
government
spending/GDP,
cumulative
gross
domestic
investment/GDP, goods exports/GDP, real migrant
remittances proxied by (WR+CE+MT)/CPI
The impact of remittances on all
development
outcome
indicators
(dependent variables) except investment
in Egypt is positive for both short- and
long run and so are the distributed time
effects with only one negative interim
multiplier
effect
in
Morocco
for
consumption and income but with
positive overall effects in each case.
Remittances,
just
as
migration,
significantly reduce the incidence, depth
and severity of poverty in developing
countries. A 10 percentage rise in
international remittances leads to a 3.5
percentage decrease in the share of the
population living in poverty.
Dependent: (Log of) Poverty (less than US$1 per
person; gap; squared gap), income inequality (Gini
index).
Explanatory: (Log of) Remittances per capita, per capita
GDP PPP, regional dummies, survey mean income o
per capita GDP, distance from remittance-sending
area, government stability, percentage of population
over 25 years with secondary education.
Dependent: Various developmental outcome variables
including poverty, educational attainment, infant
mortality, illiteracy rate and healthcare indicators.
Explanatory: Remittances received by households.
Control variables (squared remittances, distance
between US-Mexico border from municipalities,
historical migration rates, municipal per capita income,
percentage of population in rural communities, fraction
of indigenous people, Gini coefficient, share of
employment in agriculture and in government sector,
unemployment rate, homicide rate at municipal level,
and governance.
Dependent: (Log of) Poverty (headcount, gap, squared
gap).
Explanatory: (Log of) Remittances (WR+CE+MT) per
capita or Remittances/GDP, GDP per capita or
341
An increase in the fraction of households
receiving international remittances is
positively
correlated
with
higher
schooling, healthcare indicators and
reduced
poverty.
International
remittances lead to improved welfare,
reduced infant mortality and illiteracy
among children aged 6-14 while
increasing school attendance among the
latter group. Also, poverty levels and
marginalisation index decline.
Remittances do not have a significant
inequality-alleviating effect, but they do
reduce poverty headcounts significantly.
A percentage rise in remittance ratio to
Stellenbosch University http://scholar.sun.ac.za
Aggarwal
(2006)
Household
Survey
Data
(NHSD)
estimation
al.
99
developing
countries
1975-2003
Panel FE, RE and
system dynamic GMM
estimations
al.
19
remittancedependent
countries
(remittances
of at least 1
per cent of
GDP)
1976-2003
(unbalanced
panel)
OLS and FE modelling
Faini (2006a)
37 countries
with migrants
in Europe as
reported
in
the European
Community
Households
Survey 19942001
1990-2000
(Pooled data)
De Leon-Manlagnit
(2006)
40
developing
countries
1975-2003
Aggregate
data
modelling and pooled
data modelling. Linear,
Log-linear
functional
form estimations. Log
of distance between
home-country and hostcountry used as an
instrument in a reestimated IV model
Single equation by
OLS, 2SLS based on 5year moving average
data) with investment
Drinkwater
(2006)
et
et
household
mean
income,
Gini
index,
remittances*regional dummy variable.
Instrumental variables: Distance between home and
host countries, percentage of population over age 25
that have completed secondary education, government
stability, growth of sender countries weighted by
distance, growth of sender countries weighted by stock
of migrants of receiving countries
Dependent: Financial development (either bank credit
to private sector/GDP or bank deposit/GDP).
Explanatory: Dynamic Panel Model (Log of)
Remittances(WR+CE+MT)/GDP, lag of dependent
variable, control variables (real GDP, real GDP per
capita, inflation, current and capital account openness,
dummy for dual exchange rate regimes).
nd
2 set of control variables are exports/GDP, AID/GDP,
FDI/GDP, portfolio equity/GDP.
Dependent:
Unemployment
rate;
investment
(GCF)/GDP.
Explanatory:
Remittances
(WR+CE+MT)/GDP,
M2/GDP, openness (X+M)/GDP, fiscal policy (budget
deficit)/GDP, uncertainty (5-year moving average of
CPI), economic activity (real GDP growth rate), real
interest rate, AID/GDP
Dependent: Remittances (WR+CE+MT)/population,
Explanatory: Total migration/population, home-country
income per capita, ratio of skilled migrants/population,
per capita income*migration of skilled and unskilled
workers.
In pooled-data regression, time dummy, and time
dummy*migration stock included.
Dependent: Private household consumption/GDP;
private household investment (GFC)/GDP
Explanatory (consumption regression): Remittances
342
GDP reduces the fraction of population
living in poverty by about 0.4 per cent.
Remittances have positive impact on
financial development and this finding is
robust for all estimations even after
accounting for all possible forms of
endogeneity.
Remittances help in solving credit
constraint problems in recipient countries.
Remittances have a small negative effect
on unemployment, but a significant
positive effect on investment. The
positive investment effect outweighs the
negative search income effect and so
remittances can reduce unemployment
rate in the long run.
Brain drain is associated with a smaller
flow of remittances. Migration stock leads
to higher inflow of remittances. Skilled
migrants remit less, hence the cost of
migration outweighs the benefit from
remittances
in
migrant-exporting
countries.
Remittances have significant positive
impact on private household consumption
with or without interaction effects.
Transitory remittance incomes not
Stellenbosch University http://scholar.sun.ac.za
as IV
Acosta
(2008a)
al.
10
LAC
countries
Household
survey
data
2000-2004
Dynamic
panel-data
modelling by GMM
Jongwanich (2007)
17
AsiaPacific
developing
countries
1993-2003
Panel IV-FE modelling
of poverty and human
capital
development
functions.
Dynamic
panel-data modelling of
investment function by
system GMM
Shahbaz
(2007)
Pakistan
1971-2001
ARDL and Johansen
co-integration approach
64
developing
countries
including 10
SSA
2001-2005
(unbalanced
data
mostly
starting
from
2003)
Single equation by OLS
et
et
al.
Toxopeus
and
Lensink (2007)
(WR+CE+MT)/GDP, Control variables – domestic
permanent
income,
bank
credit/GDP,
quasi
money/GDP, economic openness (X+M)/GDP, current
fiscal balance/GDP, remittances*GDP, transitory
remittances*financial development (FDV) indicator,
transitory
remittances*openness,
transitory
remittances*FDV*openness.
Explanatory (investment regression): Remittances
(WR+CE+MT)/GDP, Control variables – lending rate,
current
fiscal
balance/GDP,
transitory
remittances*openness,
permanent
remittances*openness,
openness*financial
development indicator
Dependent: Log difference in inequality (Gini index)
Explanatory: (Log of) Gini indext_1, remittances
(WR+CE+MT) but with some exceptions at time t_1,
average years of secondary education of the female
population t_1, average years of secondary education of
the male population t_1, price of capital t_1, remittances
t_1*LAC regional dummy or remittances in LAC
Dependent: Poverty (headcount ratio of US$1 per day
Explanatory: Economic growth, inequality, remittances
(WR+CE+MT)/GDP and control variables (human
capital, inflation, trade openness).
Human
capital
model:
initial
income
and
remittances/GDP.
Investment model: Lag of investment (GFCF)/GDP,
growth, openness, inflation, remittances, and real
interest rate.
Dependent: Log of financial development proxied by
private sector credit/GDP
Explanatory: (Log of) Remittances(WR+CE+MT)/GDP,
real GNP per capita, CPI, exports
Dependent: Share of households with bank accounts as
a proxy for financial inclusion
Explanatory: Log of remittances per capita; governance
index; population density; log of GDP; GDP per capita;
communication infrastructure; concentration ratio; credit
343
significant in both models and so are all
its interactive effects.
Remittances reduce income inequality.
Remittances have significant impact on
poverty reduction through increasing
income, smoothing consumption and
easing capital constraints of the poor.
Remittances
promote
financial
development in the long run, and
although remittances carry a positive
sign, statistically, they are not a short-run
determinant of financial development
Remittances have a development impact
through their positive effects on financial
inclusion.
Stellenbosch University http://scholar.sun.ac.za
countries
Acosta
(2008b)
information index; share of assets in governmentowned banks; restrictions on bank activities;
requirements for entry into banking
Dependent: Log of changes in Gini index (gt-gt-1)
Explanatory: (Log of) Initial inequality (gt-1), one lag
remittances/GDP
(remt-1),
(remt-1)*LAC
regional
dummy. Control variables: average years of secondary
education of female and male population, per capita
income, distance between host country and home
country, market distortions proxied by price of
investment goods.
Dependent: Outcome variable (household income,
consumption per capita, income inequality (Gini index,
Theil‟s L index, Theil‟s T index), poverty (headcount
index, gap and squared gap)
Explanatory: Secondary school enrolment, number of
dependants, household size, squared household size,
type of employment of household head, locality,
domestic remittances, type of residential apartment,
international remittances
Dependent: Development outcome variables (school
enrolment of children aged 6-15; child malnutrition;
respiratory diseases and diarrhoea infection rate among
children under 5 years; access to health services; log of
per capita consumption; log of consumption of food; log
of educational expenditure; log of health expenditure)
Explanatory: Individual demographic features such as
age, sex, educational attainment, ethnicity of household
head, and monthly remittances received by household
101
developing
countries
1970-2003
System
GMM
estimation of a dynamic
panel model
Nguyen (2008)
Vietnam
National Survey
Data
2002,
2004 on 4008
households
Panel FE estimation
Ponce (2008)
937
Households
in
Ecuador
(National
Living
Standard
Measurement
Survey)
2006
Several estimations of
single
equation
modelling
OLS
ad
2SLS using remittances
as instrumental variable
Acosta, Baerg and
Mandelman (2009)
109
developing
and transition
countries
1990-2003
(unbalanced
data)
Dynamic
panel-data
estimation
by
firstdifference GMM
Dependent: Real exchange rate index
Explanatory: Remittances(WR+CE+MT)/GDP; bank
credit/GDP; bank deposit/GDP; remittances*bank
credit/GDP; remittances*bank deposit/GDP; Control
variables: ToT, excess money growth, trade openness,
GDP per capita, GDP growth
Ajayi et al. (2009)
38
SSA
countries
2007
Simple linear equation
by OLS
Dependent: Life expectancy at birth
Explanatory:
Remittances/GDP,
et
al.
344
migration
rate,
Generally, remittances contribute to
improved higher income and inequality.
In LAC. Remittances either reduce or
have no impact on inequality as a one
percentage increase in remittances lead
to a decrease in the number of persons
living in poverty by 0.4 per cent.
Remittances increase household income
and consumption remarkably, and
although they help decrease poverty
marginally, there is also evidence for
marginal increase in income inequality.
Remittances have positive effect on
consumption, and access to education
and healthcare. Remittances lead to
higher probability of children attending
private schools. Remittances increase
consumption by about 9 per cent. If
remittances
increase
by
US$10,
educational spending increase by 18 per
cent; and 25 per cent in case of
healthcare expenditure
Remittances exert pressure on real
exchange rate, with weaker impact on
countries with deeper and more
sophisticated financial markets which
seem to retain trade competitiveness.
Well-developed financial market can
more effectively channel remittances into
productive investment.
Remittances have significant positive
impact on life expectancy.
Stellenbosch University http://scholar.sun.ac.za
inequality as measured by Gini coefficient
Dependent: Gini coefficient
Explanatory: Level of development proxied by log of
GDP per capita, dependency ratio, migration costs
proxied by cost of passport, brain-drain level, log of
remittances (WR+CE+MT) per capita. Control variable:
FDI/GDP, M2/GDP, inflation, log of government
consumption/GDP, trade openness; state of institutions.
REMPC*logGDP;
REMPC*initial
brain
drain;
REMPC*passport costs; REMPC*distance
Dependent: Bank deposit/GDP, bank loans/GDP, bank
credit/GDP, M2/GDP
Explanatory: Remittances (WR+CE+MT)/GDP with
investment as a control variable.
Ebeke and Le Goff
(2009)
80
developing
countries
1970-2000
Single equation OLS,
IV
estimation
and
system GMM in which
the dynamic properties
of
the
dependent
variable was excluded
Giuliano and RuizArranz (2009)
73
developing
countries
1975-2002
System
GMM
estimation of a dynamic
panel-data model
Gupta, Pattillo and
Wagh (2009)
76
developing
countries for
poverty
estimations.
Survey
data
beginning 1980
(unbalanced
data).
1975-2004
5-year averages
Model I:
Pooled OLS 3SLS for
poverty models
Dependent: (Log of): Poverty (headcount, gap and
squared gap)
Explanatory: (Log of): remittances (WR+CE+MT)/GDP,
real GDP for country size, real GDP per capita for
institutional development, CPI-based inflation, gini
coefficient, distance, trade openness (X+M)/GDP, dual
exchange rate market dummy for capital account
openness, school (average school years among over
25 years of population).
Model II:
Panel RE, Panel FE,
Plus FE-IV for financial
development model.
Dependent: (Log of): Bank deposits/GDP; M2/GDP
Explanatory: (Log of): Remittances/GDP instrumental
variable. (Exogenous variables): real GDP, real GDP
per capita, inflation, trade openness, (FDI+ODA)/GDP
for capital account openness 1, dual exchange rate
dummy for capital account openness 2.
IV-2
Stage
Probit
modelling for the crosssectional data analysis.
Dependent: Poverty (headcount, gap, squared gap)
Explanatory: Domestic remittances, international
remittances, age, squared age, gender, ethnicity,
education, household size, rural location, number of
adult workers, gender*abroad (i.e. interacting gender
44
SSA
countries for
financial
development
estimation
Gyimah-Brempong
and Asiedu (2009)
Ghana
National survey
data
from
Ghana
Living
Standards
Survey (GLSS5)
Dynamic pseudo panel-
345
The more the mean income of the
recipient country is high, the more
remittances reduce income inequality.
International
remittances
are
less
inequality-mitigating as the costs of
migration rise. The more the brain drain
is important, the more remittances
promote income inequality.
For the mean and median categories of
countries, remittances have a robust
positive effect on financial development
with this impact increasing in the
presence of investment. Above the
median level of financial depth, however,
remittances have dampening effect.
Remittances have a direct poverty
mitigating effect and enhance financial
development in the long run. Remittances
negatively impact on poverty headcount
(-0.13) and poverty gap (-0.13).
Remittances reduce income inequality.
In all estimations, remittances exerted
positive and robust impact on financial
development
Remittances increase the probability of a
family to escape poverty or being
chronically
poor.
International
remittances
have
higher
poverty
mitigating
effects
than
domestic
Stellenbosch University http://scholar.sun.ac.za
in
2005/2006.
Psuedo-panel
GLSS3-5
data
GMM
estimation
by
Kalim and Shahbaz
(2009)
Pakistan
1973-2006
Fully-Modified
OLS
estimation of log-linear
single
equation
specification
Portes (2009)
46 countries
1970-2000
Seemingly
Unrelated
Regression (SUR) by
parsimonious approach
alongside pooled OLS
and SUR FE
Adenutsi (2010a)
15
SSA
countries
1987-2007
Fixed Effects
data modelling
panel-
Adenutsi (2010b)
18
SSA
countries
1987-2007
Fixed Effects
data modelling
panel-
Adenutsi
and
Ahortor (2010)
31
developing
countries
from
LAC
(16) and SSA
(15)
1986-2006
Dynamic
modelling
GMM
panel-data
by system
with gender of international migrant)
Dependent: Poverty headcount ratio
Explanatory: (Log of) Remittances/GDP, lag of
dependent variables as proxy for economic shocks,
trade openness, lag one of GDP growth, inflation,
urbanisation (share of urban population), tax
revenue/GDP, FDI/GDP.
In another estimation, these variables were used in
addition to the square of remittances
Dependent: Average income of a decile
Explanatory: Log of real remittances (WR+CE),
controlling for average income level, human capital
accumulation proxied by secondary school enrolment,
inflation, trade openness (X+M)/GDP, time dummies
and other country-specific characteristics
Dependent: Marginal deviations in human Development
Index (HDI)
Explanatory: (Log of): Investment (GFCF/GDP), human
capital (secondary school enrolment rate), remittances
(WR+CE+MT), trade openness, inflation (logCPI),
government expenditure/GDP, time dummy (a
dichotomous variable)
Dependent: Marginal deviations in Human development
index (HDI)
Explanatory: (Log of): Investment (GFCF/GDP), human
capital (secondary school enrolment rate), remittances
(WR+CE+MT), trade openness, inflation (logCPI),
government expenditure/GDP, time dummy (a
dichotomous variable)
Dependent: Marginal deviations in Human development
index (HDI)
Explanatory: (Log of): HDIt-1, investment (GFCF/GDP),
terms of trade (ToT), remittances (WR+CE+MT),
human capital (secondary school enrolment rate), trade
openness,
inflation
(logCPI),
government
expenditure/GDP, time dummy (a dichotomous
346
remittances. Remittances also increase
the number of children in remittancerecipient households attending school,
hence raises human capital formation.
Remittances have poverty-mitigating
effects both in the short- and long run.
The impact of remittances is positive and
decreasing in income for the bottom 70
per cent of the population and negative
and increasing in income for the top-20
countries of the population. Remittances,
thus, have a huge potential for reducing
income inequality across countries.
Remittances impact positively and
robustly on human development.
Remittances have a positive and a
significant impact on socioeconomic
development.
Remittances
impact
positively
on
socioeconomic development of SSA but
impact negatively on LAC countries
Stellenbosch University http://scholar.sun.ac.za
variable)
Gheeraert
(2010)
et
al.
100
developing
countries
Cross country;
and panel data
(1975-2004)
OLS for cross-country
analysis.
FE and Least Squares
Dummy Variable for
panel data analysis
Gubert et al. (2010)
Mali
households
2006
Single equation by OLS
with
counterfactual
approach
(with
or
without migration and
remittances)
Orrenius
(2010)
Mexican
states
2003-2007
Pooled OLS, and Panel
2SLS, FE modelling.
Quarterly and annual
data were used
Mexican
households
2002 and 2005
Mexico Family
Life Survey
Treatment-effect
Modelling. Analysis at
overall, rural and urban
subsets
et
al.
Ambrosius (2011)
Investment Model:
Dependent: Total investment/GDP
Explanatory:
Remittances(WR+CE+MT)/GDP
i.e.
REMGDP, cost of bank depositing (CDEP), marginal
cost of external finance(CEXF), REMGDP*CDEP,
REMGDP*CEXF, REMGDP*CDEP*CEXF, and set of
control variables viz. business cycle proxied by GDP
growth trend, level of economic development proxied by
GDP PPP, REMGDP*GDP PPP
Bank Deposit Model:
Dependent: Difference in total bank deposits
Explanatory: REMGDP, CDEP, REMGDP*CDEP.
Control variables: business cycle proxied by GDP
growth trend, level of economic development proxied by
GDP PPP, REMGDP*GDP PPP, money creation by
Central Banks proxied by differences in reserve
money/GDP
Dependent: Poverty (headcount), Gini index
Explanatory: Household non-remittance income,
remittance income, household human capital,
household physical assets, regional dummies;
household head characteristics – age, marital status,
sex, occupation
Dependent: Employment, wage, unemployment, school
enrolment (primary, secondary, technical and
university)
Explanatory: Remittances also used as instrumental
variable (defined as all forms of private transfers), FDI,
net migration stock, formal and informal labour force.
Dependent: Change in access to financial services
Explanatory: Change in status of remittances received;
vector of pre-determined control variables.
347
Remittances have significant positive
impact on investment and bank deposits,
implying remittances can have positive
effect on financial development in
remittance-recipient countries.
Remittances reduce poverty by 5 to11
per cent; and income inequality by
approximately 5 per cent, with greater
impact on bottom quintiles.
Remittances lead to improved labour
market conditions, higher employment,
and
lower
unemployment
rates.
Remittances may also reduce the fraction
of workers earning minimum wages or
less. Remittances, generally, have no
effects on high income earners (those
who earned 5 times or more above the
minimum wage)
Remittances have significant positive
impact on ownership of savings accounts
and the availability of borrowing options
for rural dwellers, but not urban dwellers.
This effect is relevant only in the case of
Stellenbosch University http://scholar.sun.ac.za
Serino
(2011)
and
Kim
66
developing
countries
1981-2005
OLS for full sample and
quantile regression
Singh and
(2011)
Hari
India
1971-2008
Bivariate
descriptive
statistics
including
trend analysis
106
developing
countries
1980-2011:
(1980-2007) for
pre-crisis era
(2008-2011) for
crisis era
Single equation OLS
Orzell (2012)
Dependent: Poverty measures (headcount, gap, gap
squared based on US$1/day threshold)
Explanatory: Logarithm of: Gini index, real GDP per
capita, remittances (total amount of remittances that
flow through banks as a percentage of GDP), plus
control variables – FDI, ODA, regional dummies and
various year dummies
Dependent: GDP at current market prices, gross
domestic savings, gross domestic capital formation,
exports, imports, foreign exchange reserve, private final
consumption expenditure, FDI, balance of trade deficit,
exchange rate, poverty
Explanatory: Total remittances that can be
approximated by (WR+CE+MT+other current transfers).
Dependent: Natural logarithm of Poverty headcount on
$1.25 per day PPP
Explanatory: Natural logarithm of: Remittances
(WR+CE), real GDP per capita, GINI, and REM*crisis
dummy. Also included are dummies for crisis, and subregional grouping of sampled countries
microfinance institutions but not for
traditional banks.
Remittances have poverty-alleviating
effects and this effect is more
pronounced with worst-off group
Remittances impacted positively on
household sector as well as the general
economy, foreign exchange reserve,
reduce poverty. Remittances appear to
have been used to finance more for
investment than consumption goods
Despite the marginal decline in
remittances in 2009, there was
resurgence in 2010. Remittances have a
decreasing effect on poverty rates during
the period 1980-2011 and the effect was
not statistically changed during the crisis.
Thus, the effect of remittances on poverty
during 1980-2007 is not statistically
different from during the 2008-2011 crisis
Remittance inflows in the low and upper
middle income countries are positively
and significantly related to domestic
credit provided by the banking sector
under all three forms of estimations.
1999-2008 for FE, RE and standard- Dependent: Credit proxied by domestic credit/GDP ratio
low income;
corrected FE & RE Explanatory: Remittances (WR+CE)/GDP, GDP per
capita growth, real interest rate, annual percentage
1996-2008 for models
lower
middle
changes in GDP deflation as proxy for inflation,
income;
technology measured as internet users per 100 people,
real per capita GDP
1995-2008 for
upper
middle
income
Source: Author‟s compilation. Note: WR, CE, MT denote workers‟ remittances, compensation of employees and migrants‟ transfers respectively as defined in Chapter Two.
Gani and Sharma
(2013)
57 countries
(9
low
income, 24
lower middle,
24
upper
middle
countries)
348
Stellenbosch University http://scholar.sun.ac.za
Table A7.2: Set of Control Variables in the Empirical Models
The Empirical Economic Development Models Involving……………………
Control Variables
Poverty Inequality lnPSC lnM₂/GDP lnHDI lnLIF lnEDU lnUNE lnLFP lnPRO
Human capital accumulation (lnHCA)
Yes
Yes
Yes
Yes
No
Yes
No
Yes
Yes
Yes
Real GDP per capita PPP (lnY_PPP)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Physical capital accumulation (lnINV)*
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Government expenditure (lnGXP)*
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Real lending rate (RLR)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Rate of Inflation (INF)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Trade openness (lnOPN)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Foreign direct investment, net (FDI)*
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Overseas development assistance (lnODA)*
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Real exchange rate (lnRXR)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Business cycle (BZC)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Institutional quality (INS)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
Yes
Yes
Yes
Adult literacy rate (lnLIT)
Source: Author
Note: *Variable expressed as a percentage of nominal GDP. LFP and PRO denote
labour force participation and labour productivity respectively.
Table A7.3: Median of Endogenous Variables and Specification of Median Dummy Variables
1980-1989
1990-1999
2000-2009
1980-2009 MDV i,t =1;
if Ě i,t >MDV i,t
Poverty and Income Inequality Models:
Poverty headcount (lnPovH)
3.94051
YES
Poverty gap (lnPovG)
3.08024
YES
Poverty severity (lnPovS)
6.16048
YES
Income inequality (lnGini)
3.86231
YES
Labour Market Outcome Models:
Unemployment rate (lnUNE)
2.30347
NO
Labour force participation (lnLFP)
4.28693
YES
Labour productivity (lnPRO)
6.84166
YES
Socioeconomic development (lnHDI)
-1.01943
YES
Educational attainment (lnEDU)
3.22312
YES
Life expectancy (lnLIF)
4.00068
YES
Human Welfare and Development Models:
Financial Development Models:
Bank credit to private sector (lnPSC)
2.77029
2.57077
2.63619
2.69880
YES
Broad money as ratio of GDP (lnM₂/GDP)
3.09268
3.03179
3.22617
3.10697
YES
Source: Author
349
Stellenbosch University http://scholar.sun.ac.za
Table A7.4:
Static Panel-Data Modelling of Remittances on Private Sector Bank Credit in SSA, 1980-2009
Fixed
Effects (FE)
Random GLS
Effects (RE)
Robust FE
Migrant remittances (lnREMPC)
-0.0105
(-0.83)
-0.0060
(-0.47)
-0.0105
(-0.39)
-0.0060
(-0.24)
Real GDP per capita (lnY_PPP)
1.0381
(14.23)***
0.7748
(12.57)***
1.0381
(5.11)***
0.7748
(5.19)***
Rate of inflation (INF)
-0.0012
(-1.16)
-0.0013
(-1.32)
-0.0012
(-0.85)
-0.0013
(-1.01)
Government expenditure (lnGXP)
0.1649
(3.50)***
0.1559
(3.28)***
0.1649
(1.24)
0.1559
(1.12)
Real lending rate (RLR)
-0.0005
(-0.33)
0.0003
(0.18)
-0.0005
(-0.22)
0.0003
(-0.12)
Trade openness (lnOPN)
0.3743
(6.37)***
0.3543
(6.01)***
0.3743
(2.67)**
0.3543
(2.52)**
Constant term
-6.9949
(-12.74)***
-4.9395
(-10.44)***
-6.9949
(-5.49)***
-4.9395
(-5.11)***
Number of observations
981
981
981
981
Number of groups
35
35
35
35
0.2365
0.2410
0.2365
0.2410
Overall R
2
F-statistics
Hausman_FE
Breusch-Pagan (B-P) statistics
Source: Author‟s estimation
***
***
++
Robust Random
GLS (RE)
***
55.87{0.000}
249.93{0.000}
13.69{0.000}***
76.24{0.000}
***
n/a
n/a
n/a
25.18{0.000}
***
n/a
n/a
n/a
4215.11{0.000}
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
robust z-statistics in ( ), probabilities in { }, n/a denotes not available or required
++
most efficient and reliable results based on Hausman test and B-P statistics
350
Stellenbosch University http://scholar.sun.ac.za
Table A7.5: Static Panel-Data Modelling of Remittances on M2/GDP in SSA, 1980-2009
Fixed
Effects (FE)
Random GLS
Effects (RE)
Robust FE
Robust Random
++
GLS (RE)
Migrant remittances (lnREMPC)
0.0192
(1.33)
0.0308
(2.34)**
0.0192
(0.75)
0.0308
(1.20)
Human capital accumulation (lnHCA)
0.2735
(5.42)***
0.2836
(6.20)***
0.2735
(2.82)***
0.2836
(3.09)***
Real GDP per capita (lnY_PPP)
0.2784
(3.08)***
0.1278
(2.13)**
0.2784
(1.56)
0.1278
(1.21)
Official development assistance (lnODA)
0.0025
(0.10)
-0.0011
(-0.04)
0.0025
(0.06)
-0.0011
(-0.03)
Rate of inflation (INF)
0.0018
(1.62)*
0.0016
(1.49)
0.0018
(0.77)
0.0016
(0.72)
Government expenditure (lnGXP)
-0.1990
(-3.82)***
-0.1770
(-3.49)***
-0.1990
(-0.90)
-0.1770
(-0.82)
Real exchange rate (lnRXR)
-0.0188
(-0.43)
-0.0468
(-2.09)**
-0.0188
(-0.24)
-0.0468
(1.49)
Real lending rate (RLR)
0.0116
(7.55)***
0.0118
(7.71)***
0.0116
(1.38)
0.0118
(1.36)
Constant term
0.6903
(1.08)
1.8406
(4.18)***
0.6903
(0.55)
1.8406
(2.33)**
Number of observations
970
970
970
970
Number of groups
35
35
35
35
2
Overall R
F-statistics
Hausman_FE
Breusch-Pagan (B-P) statistics
Source: Author‟s estimation
0.2507
0.3046
0.2507
0.3046
***
***
***
***
27.58{0.000}
244.08{0.000}
6.94{0.000}
90.95{0.000}
n/a
n/a
n/a
12.75{0.121}
***
n/a
n/a
n/a
1312.18{0.000}
*/**/*** denotes statistical significance at 10%, 5%, 1% respectively
robust z-statistics in ( ), probabilities in { }, n/a denotes not available or required
++
most efficient and reliable results based on Hausman test and B-P statistics
351
Stellenbosch University http://scholar.sun.ac.za
Table A7.6: Data Description, Measurement and Sources
VARIABLE
NOTATION
DESCRIPTION, MEASUREMENT AND MAIN SOURCES
Dependent Variables: Poverty & Income Inequality
Poverty headcount
lnPovH
Share of the population living on less than US$1.25 per day at 2005 international
prices. Source: WDI.
Poverty gap
lnPovG
The mean shortfall from the poverty line 159 (counting the non-poor as having zero
shortfall) expressed as a percentage of the poverty line. Source: WDI.
Poverty severity
lnPovS
Squared value of the PovG. Source: Author‟s computation based on WDI.
Income Inequality
lnGini
Gini index measuring the extent to which the distribution of income (or
consumption expenditure) among individuals or households within an economy
deviates from a perfectly equal distribution. Source: WDI.
Dependent Variables: Labour Market Outcome
Unemployment
lnUNE
Share of labour force that is without work but available for, and seeking, work.
Source: WDI, WEO and African Development Indicators (ADI).
Labour participation lnLPR
Proportion of the population aged 15 and above that is economically active: all
rate
people who supply labour for the production of goods and services during a
particular year. Source: WDI and ADI.
Labour productivity
lnPRO
Total output proxied by GDP weighted by labour force. Source: Author based on
WDI, WEO and IMF country-specific desk information.
Dependent Variables: Human Welfare and Development
Human welfare
lnHDI
A weighted composite statistical index, ranging between zero (worst scenario)
and one (best scenario), involving three key human development indicators: life
expectancy at birth; knowledge and education; and living standard. Source: WDI
and Human Development Reports by the UN.
Educational
lnEDU
Net enrolment ratio of children of official school age based on International
attainment
Standard Classification of Education 1997, who enrolled in post-primary school
relative to the population of the corresponding official school age. Source: WDI.
Life expectancy
lnLIF
The number of years a new-born child will live if prevailing patterns of mortality at
the time of birth were to remain unchanged. Source: WDI.
Dependent Variables: Financial Market Development
Domestic credit to
Total financial resources in the form of loans, purchases of non-equity securities,
private sector
lnPSC
trade credits and other accounts receivable as a ratio of GDP extended to the
private sector that establish a claim for repayment. Source: WDI and author
based on IFS and WEO.
Broad money to lnM2 /GDP
Sum of currency outside banks, demand deposit other than those of the central
GDP ratio
government, and time, savings and foreign currency deposits of resident sectors
other than the central bank as ratio of GDP. Source: WDI and author based on
IFS and WEO.
‘Uncontrolled’ Explanatory Variables
Initial
level
of
The immediate past values of the dependent variable. Source: Author‟s
economic
computation from the specific dependent variable.
ln E_1
development
Migrant remittances
per capita
The sum of workers‟ remittances and compensation of employees as ratio of
population. Source: WDI, BoPS, MRF-2011 CD-ROMs and e-databases and
estimates based on country-specific information obtained from country-desk
officials of the IMF and the World Bank.
Newly Introduced Controlled Variables*
+/Real GDP per capita
GDP per capita based on purchasing power parity (PPP) at constant 2005
lnY_PPP
PPP
international prices in US dollars. Source: WDI and WEO.
+/Business Cycle
Annual growth in real GDP. Source: Author based on WDI, IFS and WEO.
BZC
+/Real lending rate
Average annual rate charged by banks on loans to prime customers minus the
RLR
annual rate of inflation. Source: Author based on WDI, IFS and WEO.
Official development lnODA+/Disbursement flows (net of repayments) from official donors to a country as a
assistance
percentage of nominal GDP. Source: WDI
+/Adult literacy rate
Percentage of people aged 15 and above who can, with understanding, read and
lnLIT
write a short, simple statement on their everyday lives. Source: Author based on
WDI and ADI.
lnREMPC
Note: The a priori sign is indicated by +/- by the notation column of each variable. *Other explanatory variables are
as defined and measured in preceding chapters. In each case, the April 2011 edition was primarily used.
159
The World Bank defines poverty line as the annual cost of obtaining the standardised minimum daily caloric 2172
requirement of 2172 calories per person plus basic non-food essential items such as food and education.
352
Stellenbosch University http://scholar.sun.ac.za
CHAPTER EIGHT
SUMMARY, CONCLUSIONS, POLICY IMPLICATIONS AND RECOMMENDATIONS
8.0 INTRODUCTION
This chapter draws the curtain on the study with an overall summary, general conclusions,
policy implications and recommendations, contributions to knowledge, and suggested areas for
future research. By way of organisation, Section 8.1 summarises the entire study with
emphasis on the key findings. Section 8.2 concludes by drawing attention to how the study
responded to the research questions posed, evaluated the hypotheses, and how the research
objectives were achieved. The policy implications and recommendations of the key findings of
the study follow in Section 8.3. The contributions of this study to knowledge are outlined in
Section 8.4, whilst some areas for future research are suggested in Section 8.5.
8.1 SUMMARY
This study was embarked upon to investigate the role macroeconomic policy can play in
attracting optimal migrant remittances through official channels, and in enhancing the economic
growth and developmental potentials of international migrant remittances in sub-Saharan Africa
(SSA). In pursuance of these objectives, this study was organised into eight chapters of which
this final one serves as the concluding one.
Chapter One set the stage with a general introduction to the study by presenting the
background information, specifying the research problem, the research questions, the
underlying motivation, the objectives and the hypotheses. The scope as well as the structure of
the study was also discussed in Chapter One. It is observed that even though SSA is a leading
„exporter of labour‟, the sub-region has been the least recipient of officially reported remittances
in terms of actual volume received, per migrant and per capita. At the same time, however,
some studies including that of Freund and Spatafora (2005), identify SSA as the leading
recipient of informal remittances due to the high cost of remitting through the formal financial
system, which is associated with the underdevelopment of the domestic financial markets.
Therefore, it is imperative to explore the role macroeconomics can play in policy formulation
towards increasing the flow of migrant remittances to SSA through official channels, especially
because the continuous inflow of informal remittances can destabilise not only the
353
Stellenbosch University http://scholar.sun.ac.za
socioeconomic and geopolitical fabrics of the sub-region, but also threaten global security. It is
further noted that SSA is the only sub-region in the world today that still depends more on
foreign aid than other non-trade external finances despite the desperate attempts in search of
the required finance to close the wide resource gap towards meeting the Millennium
Development Goals (MDGs). Finally, it is demonstrated in Chapter One why the scope of this
study is restricted to 36 SSA countries over the period, 1980 to 2009.
The study proceeds with a discussion on the definition and measurement of the key concepts
in Chapter Two. These concepts are international remittances and financial liberalisation. It is
noted that, unlike financial development which is essentially a de facto concept, financial
liberalisation is de jure synonymous with the timing of the gradual implementation of policy
reforms and developments within the financial system. The reasons justifying why the financial
liberalisation index developed by Abiad et al. (2010) was used in this study despite a host of
other alternative measures are discussed. It is evident that there can either be a narrow
definition and measurement of international remittances or a broad definition and
measurement. It is the narrow definition which is concerned with regular transfers of funds by
international migrants that is considered the more appropriate measure in the context of this
study. In effect, the definition and measurement of international remittances is narrowed to the
sum of workers‟ remittances and compensation of employees. This implies that, contrary to the
frequently used definition and measurement of remittances, migrant transfers are excluded
from the definition of international migrant remittances in this study. Rationalised reasons for
this exclusion are duly assigned. Finally, the conventional problems of remittance data
deficiency and low quality are also discussed with the acknowledgement that, as has become a
fact in macro-level cross-country time-series analyses, the officially reported remittance data
used in this study are underestimated because remittances received through informal channels
are not captured and incorporated due to a myriad of complexities.
Having resolved the pertinent definitional and measurement issues surrounding the main
concepts in this study, in Chapter Three, efforts are directed at providing a broad overview of
macroeconomic performance and policy environment of SSA since independence in the 1960s.
This is to provide an insight into understanding the observed trends in the macroeconomic
performance as well as the composition and pattern of remittances and other external capital
flows to SSA over the years. It is observed that, generally, very little is achieved by the subregion in terms of real per capita income growth, investment and resource mobilisation, so that
354
Stellenbosch University http://scholar.sun.ac.za
the unfavourable structural features of SSA which existed during the pre-reforms era are all still
present today. Admittedly, though, there seems to be some glimmer of hope regarding
economic recovery and prosperity in very recent years in terms of international trade, savings,
and financial development. It is noted that what would have been the initial gains from the
pursuit of financial liberalisation in SSA are eroded by unfavourable macroeconomic conditions
and widespread political unrest, civil strife, and bad governance in the 1990s. Following this,
SSA recorded its worst macroeconomic performance during the 1990s so that most of the
performance indicators of the 1980s are closely comparable with those of the 2000s. Unlike
other regions of the developing world, SSA attracts the least migrant remittances but the most
foreign aid. This is notwithstanding the fact that the amount of remittances received by the subregion has been increasing steadily over the years and has increasingly been becoming more
and more robust since the implementation of financial liberalisation in the 1980s. It is also
observed that of the various forms of external capital, migrant remittances have been the most
resilient to adverse economic shocks and the least volatile in the developing world where SSA
enjoys the highest stability. A significant positive correlation exists between migrant remittance
inflows and bank-based financial development indicators in SSA; and this is more robust for the
highest remittance-recipient countries in comparison with the least remittance-recipient
countries.
The study then progressed to Chapter Four with the determination of the macroeconomic
factors that explain the changing trends in migrant remittances received in SSA. This task is
executed at both the aggregated and the disaggregated levels. At the aggregated level, the
macroeconomic factors that affect migrant remittances are determined whereas at the
disaggregated level, the macroeconomic determinants of the components of migrant
remittances (workers‟ remittances and compensation of employees) are separately estimated.
In order to provide a further understanding of the changing cyclical behaviour of remittances
received in SSA, in each case, decade-based estimations are undertaken alongside the overall
study period analysis. It is evident from the literature reviewed that the motives behind the flow
of migrant remittances can be broadly categorised into either altruism or self-interest.
Conceptually, however, it is noted that altruism and self-interest may not necessarily be
mutually exclusive as, in many cases, a remitting migrant is often motivated by both motives.
The system Generalised Method of Moment (sys-GMM) estimation procedure for dynamic
panel-data models is adopted. It is found out that apart from asynchronous effects both hostcountry and home-country macroeconomic factors play crucial roles in determining the amount
355
Stellenbosch University http://scholar.sun.ac.za
of officially reported remittances received in SSA between 1980 and 2009. Although, in
response to the pursuit of a financial liberalisation programme, there seems to be an increasing
impact of these macroeconomic factors on the changing trends in migrant remittance inflows,
this is largely dependent upon the general macroeconomic performance of the sub-region.
Macroeconomic variables, to a reasonable extent, have impacted differently on workers‟
remittances and migrant remittances received in SSA over the past three decades, although
the results of these two are more consistent than in comparison with compensation of
employees. Whereas workers‟ remittances and migrant remittances seem to be driven primarily
by the self-interest economic motive, compensation of employees seems relatively more
altruistic in nature. However, it does seem as though altruism is fading gradually.
In Chapter Five, the direct causal effects and impacts of financial liberalisation on international
migrant remittance inflows in SSA are investigated using a set of bivariate empirical models,
notably, panel Granger-causality and panel GLS Random Effects (RE). For this analysis, the
cross-sectional sampled size dropped from 36 to 13 SSA countries for which relevant data was
available. The results suggest that financial liberalisation has contributed positively to
international remittance inflows through official channels in SSA, but banks have not been
active participants in the international remittance market, so that most of the officially reported
migrant remittances received in SSA are received outside the formal financial system.
Generally, the positive impact of financial liberalisation on international remittance inflows has
been increasing over time, but it is relatively more robust for SSA countries with frontier and
emerging financial markets than for their counterparts with underdeveloped financial markets.
Policy reforms on stock market developments impact more on official remittance inflows than
those directly related to the banking sector. Of the various financial policy reforms on the
banking sector, bank supervision and prudential regulation, deregulation of credit allocation and
reduction in reserve requirements, and elimination of entry barriers for competition in the
banking industry exert the most significant positive impact on remittance inflows. Furthermore,
it is found out that financial liberalisation Granger-causes international remittance inflows in
SSA with a low statistical evidence of reverse causality. Policy reforms on stock market
developments are found to have had the most significant uni-directional causal effect, which
runs from stock market development to international remittance inflows. It is noted that the low
participation of SSA banks in the remittance market, may be one of the reasons why the cost of
international money transfers is relatively higher on the various remittance-corridors linking
SSA countries as compared to the remittance corridors linking migrant-home countries in other
356
Stellenbosch University http://scholar.sun.ac.za
parts of the developing world.
The study then proceeds to Chapter Six where a dynamic panel-data model involving 36 SSA
countries is estimated following the sys-GMM estimation procedure, to examine the impact of
migrant remittances on economic growth over the period, 1980-2009. Based on the assumption
that the effects of international remittances on economic growth could vary over time in
response to changing macroeconomic policy environment in remittance-recipient countries, a
decade-based analysis is undertaken alongside the overall study period analysis. Also explored
are the possible economic growth size-effects of migrant remittances in SSA. It is discovered
that, between 1980 and 2009, migrant remittances impact positively on economic growth in the
sampled 36 SSA countries. Again, for these sampled 36 countries, the impact of remittances
on economic growth varies over time in response to changing macroeconomic environment
such that, during the decades (the 1980s and the 1990s) that remittances have significant
impact on growth, the positive impact is more significant during „good times‟ (the 1980s) than
during „bad times‟ (the 1990s). It is, thus, found out that, broadly, since the implementation of
financial liberalisation in SSA, the economic significance of international remittances to growth
in the sub-region has been declining until it becomes zero in the 2000s. Another important
finding in Chapter Six is that the positive impact of migrant remittances on economic growth
has varying effects in response to the rate of economic growth in recipient countries such as
SSA countries with relatively higher growth rates that have benefitted more directly from
remittances. For SSA countries with relatively higher growth rates, the positive effects of
remittances on economic growth have been increasing consistently both statistically and
economically over the past three decades, even though remittances have had a zero-effect on
growth in this category of countries in the 1980s. Moreover, the impact of migrant remittances
on economic growth is more asynchronous with one year lag effect than contemporaneously.
Finally, the implications of migrant remittances on various aspects of economic development in
SSA are examined in Chapter Seven. In doing this, a set of dynamic panel-data models is
estimated following the sys-GMM technique. Aspects of economic development covered are
poverty, income inequality, unemployment, labour participation, labour productivity, human
welfare, educational attainment, life expectancy, and financial development. For each
estimated model, except for the model involving financial development, the time frequency of
the data used changed to a 5-year non-overlapping panel data due to the unavailability of
annual data for most of the development outcome indicators aforementioned. Hence, the
357
Stellenbosch University http://scholar.sun.ac.za
enquiry into the time-varying effects of international remittances on economic development
since the implementation of financial liberalisation programme in SSA is limited to models in
which the effects of remittances on financial development are explored. On the average,
migrant remittance inflows have poverty-alleviating effects with the potential of income
equalisation in SSA, but in migrant-home SSA countries with relatively higher levels of poverty
and inequality, remittances intensify poverty and income inequality. It is also found out that
remittances significantly reduce unemployment rates in SSA with no moral hazard effects in
countries with relatively lower rates of unemployment. It is further realised that remittances
have no direct effects on labour participation rates in SSA, but countries with higher rates of
labour participation have higher prospects of benefitting from remittances in this regard.
Similarly, the overall impact of international migrant remittances on labour productivity is
statistically insignificant in SSA, but in SSA countries with relatively higher rates of labour
productivity, remittances contribute significantly to labour productivity.
Also, in Chapter Seven, it is noted that, overall, migrant remittances have contributed positively
to enhancing human welfare, educational attainment and life expectancy in migrant-home SSA
countries over the past three decades. However, the positive effects of migrant remittances on
human welfare and educational attainment are more beneficial to SSA countries with relatively
lower levels of human welfare and educational attainment. In fact, migrant remittances impact
negatively on migrant remittance-receiving SSA countries with relatively higher rates of
educational attainment. Concerning the size effects of international remittances on life
expectancy in SSA, the positive impact is more robust in countries with relatively higher years
of life expectancy at birth than those with relatively lower years of life expectancy. Furthermore,
it is found out that between 1980 and 2009, migrant remittances impacted positively on
financial market development in SSA and this impact has been increasing over time when all
other factors are held constant. In particular, migrant remittances promote bank credit to the
private sector as an indicator of financial development in the 1980s and the 2000s when the
macroeconomic environment in SSA is relatively stable with higher real GDP per capita, whilst
during the recession period of the 1990s, migrant remittances contribute negatively to private
sector access to bank credit. Over the period 1980-2009, the impact of remittances on private
sector credit has been more pronounced in migrant-home SSA countries with higher private
sector access to bank credit than in other SSA countries with lower private sector access to
bank credit. With regard to the impact of international remittances on broad money-GDP ratio
(M 2 / GDP) as an indicator of financial market development in SSA, it is negative in the 1980s
358
Stellenbosch University http://scholar.sun.ac.za
and positive for the subsequent decades with the economic significance of the 1990s being the
least, probably due to the harsh economic conditions during that decade. The size of the
economic impact of migrant remittances on M 2 / GDP is more significant in countries with
relatively lower M 2 / GDP in the 1980s and the 1990s, but this has since switched in favour of
countries with relatively higher M 2 / GDP in the 2000s.
The panel Granger-causality empirical results reported in Chapter Five are based on
Econometric Views version 7.0 whilst all other empirical results are based on STATA version
11.0. Tables and Figures are mainly based on Microsoft Office Excel 2007 edition.
8.2 CONCLUSIONS
From the foregoing, it can be concluded that the objectives of this study have been achieved in
view of the fact that:
i.
the study presents the facts on the extent to which macroeconomic environment has
transformed following the adoption of economic reforms in SSA; and the trend in
migrant remittance flows to SSA since financial liberalisation in the 1980s;
ii. the macroeconomic factors that influence international remittances received in SSA
under liberalised financial regime are identified;
iii. the causal effect as well as the impact of financial liberalisation on international
remittance inflows in SSA are examined;
iv. the impact of international remittance inflows on long-run economic growth in SSA is
analysed; and
v. the developmental-impact of international remittances in SSA is determined.
By achieving the underlying research objectives, the study, in effect, has responded to the
pertinent research questions as follows:
i.
The macroeconomy of SSA has not demonstrated any significant improvement towards
development since the adoption of economic reforms in the 1980s except in terms of
financial market development and international trade. There has been a strong positive
trend in the international migrant remittance inflows in SSA since the implementation of
financial liberalisation programmes despite the fact that the sub-region remains the
least remittance-recipient in the world.
359
Stellenbosch University http://scholar.sun.ac.za
ii. The macroeconomic determinants of international remittance inflows in SSA are hostcountry income and law enforcement on banning the use of unofficial channels to remit.
Other factors are home-country variables, notably real bilateral exchange rate, bank
credit to private sector, home-country income, rate of inflation, institutional quality, and
real deposit interest rate. The macroeconomic determinants of workers‟ remittances are
host-country income, law enforcement on the use of official money transfer channels,
home-country income, real bilateral exchange rate, institutional quality, and bank credit
to private sector. Broadly, host-country income, home-country income, real bilateral
exchange rate, law enforcement on use of official money transfer channels, bank credit
to private sector, institutional quality, inflation rate, and real deposit interest rate are
determinants of compensation of employees.
iii. The impact and causal effect of financial liberalisation on international migrant
remittance inflows in SSA are significantly robust. Overall, each specific policy
implemented under financial liberalisation programme impacts positively on migrant
remittance inflows. In a descending order of economic significance these policies are
stock market development, prudential regulation and supervision of banks, elimination
of entry barriers to the banking industry, deregulation of international capital flows,
interest rate deregulation, and privatisation of banks. Also, financial liberalisation
Granger-causes migrant remittance inflows through official channels. All other things
remaining equal, SSA countries with frontier and emerging markets receive relatively
higher migrant remittances through official channels of which the formal financial
system is a part rather than other countries within the sub-region with relatively less
developed financial markets. This is notwithstanding the fact that SSA countries with
underdeveloped financial markets are the recipients of the most migrant remittances
through official channels in which money transfer operators (MTOs) are dominant.
iv. International migrant remittance inflows stimulate economic growth in SSA, and have a
varying impact in response to macroeconomic policy environment rather than merely in
reaction to improved financial market liberalisation. However, international remittances
are not a panacea for long-run economic growth in every SSA country. International
migrant remittances are more advantageous in enhancing growth in SSA countries with
relatively higher rates of economic growth than in other SSA countries with relatively
360
Stellenbosch University http://scholar.sun.ac.za
lower rates of economic growth.
v. To a very large extent, international migrant remittance inflows serve as a catalyst for
economic development as far as poverty alleviation, unemployment, human welfare,
educational attainment, life expectancy, and financial market development are
concerned. Generally, migrant remittances contribute more to the economic
development in „labour-exporting‟ SSA countries with relatively higher levels of
economic development than what other countries within the sub-region with relatively
lower levels of economic development save with reference to human welfare and
educational attainment.
Consequently, exclusive of hypotheses H2, H7, H9 and H10, at the conventional levels of
statistical significance, this study fails to accept all the hypotheses specified under Section 1.6
but, instead, concludes that, overall, with reference to SSA:
i.
macroeconomic factors are determinants of international remittance inflows;
ii.
macroeconomic determinants do not have exact influence on attracting remittances
from permanent migrants (workers‟ remittances) and remittances from temporary
migrants (compensation of employees);
iii.
financial liberalisation Granger-causes international remittance inflows with an evidence
of weak reversal;
iv.
financial liberalisation impacts positively on international remittance inflows;
v.
international remittance inflows stimulate long-run economic growth;
vi.
international remittance inflows impact negatively on poverty (i.e. poverty headcount,
poverty gap and poverty severity);
vii.
international remittance inflows do not influence income inequality;
viii.
international remittance inflows impact negatively on unemployment;
ix.
international remittance inflows do not affect labour participation;
x.
international remittance inflows do not influence labour productivity;
xi.
international remittance inflows promote human welfare development;
xii.
international remittance inflows impact positively on educational attainment;
xiii.
international remittance inflows impact positively on life expectancy at birth; and
xiv.
international remittance inflows promote financial development with reference to broad
money to GDP ratio, and to some extent, access to private sector credit.
361
Stellenbosch University http://scholar.sun.ac.za
8.3 POLICY IMPLICATIONS AND RECOMMENDATIONS
A number of policy issues have been put forward by this study; the overriding ones are outlined
below.
Firstly, as far as an appropriate inquiry into the stability in the flow and the impact of
international migrant remittances in remittance-receiving SSA countries is concerned, it is
implied that the best possible measure of migrant remittances is the sum of workers‟
remittances and compensation of employees, relative to population size. Secondly, it is implied
that the pursuit of financial liberalisation is necessary but not a sufficient condition for receiving
optimal migrant remittances through official channels as the macroeconomic performance and
policy environment affect the potential contribution of financial liberalisation in this context.
Thirdly, an important implication of the significant changing impact of macroeconomic factors
on migrant remittance inflows in SSA is that one of the reasons why SSA is the least recipient
of migrant remittances is the absence of appropriate and effective macroeconomic policies on
the mobilisation of remittances from their citizens living abroad. Fourthly, it is implied that SSA
countries with frontier and emerging financial markets receive more remittances through the
formal financial channels but that the majority of the remittances received by the sub-region
through the official channels are outside the financial system with low bank participation. It is
further implied that, compared with the financial liberalisation policies on the banking system,
policies on stock market development have been the most successful in the mobilisation of
migrant remittances through the formal financial system. Another implication from this study is
that in order to maximise the growth-enhancing impact of migrant remittances in the long run,
remittance-receiving SSA countries must ensure that other pro-growth policies are vigorously
pursued. Finally, it is implied that the implementation of a comprehensive economic
development strategy is inevitable to maximise the developmental-impact of migrant
remittances. In other words, it is imprudent to rely on migrant remittances as the main source of
long-run growth in remittance-receiving SSA countries.
Therefore, the most important recommendations of this study to inform policy initiatives are:
i.
The need for policy makers to ensure stable and credible macroeconomic policy
environment through reduction in the rate of inflation, improvement in economic
performance which reflects in higher real per capita income, and a stronger national
currency in the international financial market so as to encourage private sector savings
362
Stellenbosch University http://scholar.sun.ac.za
and investment. Furthermore, it is imperative to ensure that policy makers in „labourexporting‟ SSA countries devise ways to strengthen institutions through improved
democratic governance and freedom from civil and political strife as in more recent
years, quality institutions impact positively on the inflows of workers‟ remittances and
compensation of employees. To achieve this, measures must be put in place to reduce
corruption (or the perception thereof), improve national security and peace, and create
a conducive investment environment through the enactment of laws that protect the
interest of investors and entrepreneurs, whether resident at home (potential recipients
of remittances) or abroad (potential remitting migrants).
ii. Policies under financial liberalisation programme such as stock market development,
prudential regulation and supervision of banks, deregulation of credit allocation,
relaxation of entry barriers to the banking industry, privatisation of banks, deregulation
of interest rates and external account liberalisation must be rigorously pursued. When
the pursuit of financial liberalisation leads to higher competition in the financial market,
financial institutions will become more efficient resulting in reduced money transfer fees,
introduction of innovative and diversified financial products and services, expansion and
wider coverage with more outlets at home and abroad. This is essential because when
SSA migrants find the patronage of informal money transfer channels cheaper, safer,
more reliable, convenient and accessible; it will be difficult, if not impossible, for the subregion to mobilise optimal remittances through the formal money transfer channels
which include the formal financial system.
iii. Efforts must be directed at further deepening the pursuit of financial liberalisation
programme in a bid to foster competition among banks towards mobilisation of optimal
remittances in order to maximise the potential benefits of remittances in SSA. For
example, banks can introduce differentiated services and develop remittance products
such as online and automation in payment systems through technological innovation,
measures which are needed to reduce the cost of handling small cross-border money
transfers. Also, banks can open overseas branches and offer more offshore services to
residents at home to facilitate payments and receipts of remittances internationally.
And, when banks offer customers relatively lower cost on remittance services and
provide relevant information on investment opportunities at home, offering mortgages
and housing loans, and assisting migrants in planning for their retirement and in
363
Stellenbosch University http://scholar.sun.ac.za
insuring their valuable assets, migrants may be encouraged to remit through the
banking system. Besides, when more remittances are received through the banking
system, this promotes higher „bancarisation‟, hence a higher propensity to save and put
remittances received into productive use rather than a conspicuous consumption
because banks, unlike MTOs, offer additional services such as financial intermediation
necessary for economic growth and sustainable development.
iv. SSA banks must establish bilateral and multilateral partnerships and networks with one
another not excluding rural and community banks, and with post offices and foreign
banks in order to build efficient and reliable national and international payment systems
among collaborating banks and institutions. This will also make SSA banks more visible
and conveniently accessible in the remittance market at home and overseas. The
strategic partnerships, networks, and negotiated alliances and franchise should be
effective in enabling local banks to overcome the challenges of high operational costs
and the geographic fragmentation of the remittance markets. Banks should design
special incentive packages, including zero tax on remittances received, and special
remittance agreements with major migrant-host countries. There should be the
regulation of informal intermediaries in the money transfer market, and banks should
issue special foreign currency denominated bonds targeted at the Diaspora
communities, establish „remittance banks‟ at home with overseas branches or outlets,
and opportunities for social security contributions from abroad, to attract migrants to
remit funds home using official channels.
v. Policy makers in SSA must pursue complementary pro-growth and development
strategies towards eliminating market distortions in favour of the rich so as to reduce
poverty, income inequality and high unemployment rates. Poverty alleviation policies
such as capacity building, vocational training, access to venture capital and microcredit,
and other SME incentive packages through which the economically disadvantaged and
the vulnerable groups such as rural dwellers and women stand a better chance to gain
and improve upon their welfare can be useful in enhancing the poverty-alleviating and
inequality-reducing effects of remittances. Policy makers should adopt an integrated
economic development programme in which fiscal and monetary policies are well coordinated in a manner that will ensure that progressive price and income policies are
designed and implemented towards bridging the gap between high-income and
364
Stellenbosch University http://scholar.sun.ac.za
economically vulnerable groups of society at all levels of economic development.
8.4 CONTRIBUTIONS TO KNOWLEDGE
This study has contributed to existing body of knowledge in a number of ways, notable of these
contributions are:
i.
First and foremost, this study is currently most comprehensive on SSA taking into
account concurrently both the time span of 30 years (1980-2009) and cross-sectional
dimension of 36 countries. All known related available studies on SSA either fall short of
the time span used in this study or the number of sampled countries covered.
Therefore, as far as SSA as a sub-region is concerned, this is currently the most
representative and comprehensive macro-level study on international migrant
remittances.
ii. Secondly, this study is the most detailed on the macroeconomic determinants of
migrant remittances and the implications of remittance inflows for economic growth and
development on SSA, and arguably on remittance-receiving developing economies,
considering the carefully detailed and systematic empirical analyses. For instance,
unlike all other known related studies on international migrant remittances, this study
has shown that the impact of macroeconomic factors that determine migrant remittance
inflows can vary over time depending upon the macroeconomic fundamentals of the
recipient countries. It has also shown that macroeconomic variables can have a varying
impact on the two components of migrant remittances - workers‟ remittances and
compensation of employees. This was achieved in this study by undertaking a
systematic decade-by-decade analysis as well as by estimating the macroeconomic
determinants of migrant remittances at the aggregated and the disaggregated levels.
Similarly, a detailed analysis of the contributions of migrant remittances to economic
growth and financial development as an aspect of economic development has been
done to determine the time-varying remittance impact under different macroeconomic
conditions. Another important novelty with regard to a detailed analysis achieved by this
study is the fact that, on examining the economic growth and developmental-impact of
migrant remittances in SSA, size-effects have been taken into account.
365
Stellenbosch University http://scholar.sun.ac.za
iii. Thirdly, the use of migrant remittances per capita as the best alternative measure of
remittances per migrant is a notable contribution of this study to the existing body of
knowledge. The use of remittances per migrant, hence remittances per capita, in place
of the commonly used remittances as a percentage of GDP is important in reducing
obvious endogeneity bias as lower income countries are more likely to suffer higher
emigration of active labour and hence attract higher remittances. This premise is based
on the predictions of the altruistic theory. Also, the pure self-interest economic theory of
remittances predict pro-cyclicality in the flow of remittances by hypothesising that
migrant-home countries with higher growth rates are more likely to attract remittances
from their migrants for essentially investment motives. Going by either theory, there is
an apparently high likelihood that the use of remittances as a percentage of GDP rather
than remittances per capita can yield unreliable results. A related contribution of this
study is the identification and subsequent use of the most appropriate measurement of
migrant remittance inflows to include only the relevant credit entries in the BoP current
account (workers‟ remittances and compensation of employees), thereby excluding
migrant transfers which are a BoP capital account credit transaction. This is contrary to
common practice in most previous studies in which migrant remittances were defined as
the sum of the three aforementioned items even though migrant transfers have a set of
completely different features to workers‟ remittances and compensation of employees.
iv. Fourthly, with reference to SSA, this study has contributed to widening the knowledge
horizon by unrestricting the possible direct impact of remittances on economic growth in
migrant-home countries to contemporaneous effects, as has been the common practice
in virtually all known related studies. The finding that the direct growth-impact of migrant
remittances is asynchronous rather than contemporaneous is considered a valuable
contribution to knowledge in the remittances-growth literature because most scholars
tend to model the impact of remittances on economic growth as if to suggest that the
growth-impact of remittances should always be instantaneous. Both theoretically and as
shown empirically in this study, restricting the possible impact of remittances on growth
to only contemporaneous effects even when remittances are spent on investment
goods cannot only be an underestimation but also technically erroneous.
v. The fifth ground-breaking contribution of this study to the literature on remittances is the
empirical analysis of the direct impact and causal effects of financial liberalisation on
366
Stellenbosch University http://scholar.sun.ac.za
international remittance inflows through official channels in migrant-home countries,
using SSA countries as a case study. This is the first known attempt at an empirical
exercise in this endeavour and, more importantly, the first time that a comprehensive de
jure measure of financial liberalisation is used in this context. Another crucial
contribution of this study is that the implications of financial liberalisation for migrant
remittance inflows in SSA are further explored at the financial liberalisation policyspecific level. Furthermore, the exclusive impact and the causal effect of financial
liberalisation on international migrant remittance inflows in SSA countries with frontier
and emerging markets and other SSA countries with underdeveloped financial markets
have been explored in a decade-by-decade analysis.
8.5 LIMITATIONS AND DIRECTIONS OF FUTURE RESEARCH
The quality of the data used can be considered as a major limitation of this study. This is
because the study relied on secondary sources especially the World Bank, the International
Monetary Fund (IMF), and Abdul Abiad and Thierry Tressel of the IMF who, together with
Enrica Detriagiache published “A New Database of Financial Reforms” in IMF Staff Papers,
57(2): 281-302 in the year 2010. Under some circumstances, additional information which was
largely on the implementation of financial sector adjustment programmes was sourced from the
various Central Bank reports and recent series of IMF‟s World Economic Financial Surveys.
For some few countries, notably Guinea (1980-1985), Guinea-Bissau and Seychelles (19801987), Malawi and Mauritius (1980-1993), Namibia (1980-1989), Tanzania and Uganda (19801994), and São Tomé and Príncipe (1991-1995) missing published data on migrant remittances
was filled with estimates based on country-specific information obtained from country-desk
officials at the Headquarters of the IMF and the World Bank in Washington, D.C., USA. The
extrapolations were based on a three-year data average of per capita remittances aligned with
the receipt of the country involved relative to the total receipt of SSA as a sub-region. It is,
however, considered that since a substantial amount of the core data for analytical purposes
was obtained from the same credible sources (the World Bank and the IMF), the empirical
results obtained from this study are not negatively affected by poor quality data. Besides, in the
most recent and comprehensive related study on the SSA, Singh et al. (2010) used a similar
data compilation technique. Accordingly, as far as the quality of data is concerned, the
empirical results of this study can be considered as reliable.
367
Stellenbosch University http://scholar.sun.ac.za
And, notwithstanding the aforementioned contributions of this study, it is suggested that future
related research should be directed at addressing the following pertinent issues which could
not be attended to in this present study.
i.
A survey study aimed at estimating the amount of migrant remittances received in SSA
through the informal channels, whether or not there are country and regional differences
in the amount of informal remittances received relative to the officially reported
remittances and, if so, to determine the reasons behind this disparity to inform an
effective policy design to mitigate them.
ii. A counterfactual analysis should be undertaken to examine the effects of international
migration on the host-country and on the home-country in order to determine whether
„labour-exporting‟ developing countries are actually the net beneficiaries despite losing
their active labour to the industrialised world. In doing so, the impact of remittances and
other non-financial gains from international migration will have to be analysed under
various hypotheses such as “a with no emigration no remittances plus other gains”
scenario against “a with emigration remittances earnings plus other gains” scenario,
and “a with emigration remittance earnings but no other gains” scenario. It should also
be interesting to explore a counterfactual condition of a world of no South-North
migration, hence zero remittances, but with optimal foreign aid from the would-have
been North migrant-host countries. This is considered worthy of investigation because
the facts presented in this study reveal that regions that receive higher remittances are
recipients of lower foreign aid and vice versa. Therefore, an important question that
must be addressed in the future is: To what extent would the above mentioned
scenarios have impacted on global economic growth and critical development indicators
such as poverty and income inequality?
iii. So far, empirical studies on international migrant remittances have been confined to
remittance inflows, but as revealed in this study, concentrating on the implications of
migrant remittance inflows as a direct financial gain from losing active labour to the
outside world can be misleading when remittance outflows are not accounted for. SSA,
for example, has not only been a destination of migrant remittances but also a major
source of migrant remittances to the outside world. Therefore, it should be interesting to
explore factors that determine net remittance inflows and the implications of net migrant
368
Stellenbosch University http://scholar.sun.ac.za
remittances for economic growth and development in SSA.
iv. One of the striking findings from this study is the fact that migrant remittances impacted
positively on financial liberalisation. Therefore, it is hereby suggested that future related
studies should aim at identifying the appropriate sequencing of the specific reform
policies under financial liberalisation programme so as to enhance an optimal flow of
migrant remittances to SSA. Also, it should be interesting to determine how international
remittance inflows will react to shocks emanating from each specific policy reform
component of a financial liberalisation programme in developing countries. Furthermore,
the question as to whether migrant remittance inflows could have a threshold effect on
the financial market development could be explored in the future.
v. Finally, because it is found in this study that financial liberalisation Granger-causes
international migrant remittance inflows and remittance inflows promote financial market
development in SSA, the implications of remittance inflows for financial inclusion and
international financial integration should be explored in the future.
369
Stellenbosch University http://scholar.sun.ac.za
REFERENCES
Abdel-Rahman, Abdel-Mahmoud M. 2006. “Determinants of Foreign Worker Remittances in the
Kingdom of Saudi Arabia.” Journal of King Saud University, 18(2): 93-121.
Abe, Shigeyuki, Maxwell J. Fry, Byoung K. Min, Pairoj Vongvipanond and Teh-Pei Yu. 1977.
“Financial Liberalisation and Domestic Saving in Economic Development: An Empirical
Test for Six Countries.” Pakistan Development Review, 16(3): 298-308.
Abiad, Abdul, Enrica Detragiache and Thierry Tressel. 2008. “A New Database of Financial
Reforms.” IMF Working Paper 08/266. Washington, D.C.: International Monetary Fund.
Abiad, Abdul, Enrica Detragiache and Thierry Tressel. 2010. “A New Database of Financial
Reforms.” International Monetary Fund Staff Papers, 57(2): 281-302.
Abiad, Abdul and Ashoka Mody. 2005. “Financial Reform: What Shakes It? What Shapes It?”
American Economic Journal, 95(1): 66-88.
Abu-Bader, Suleiman and Aamer S. Abu-Qarn. 2008. “Financial Development and Economic
Growth: Empirical Evidence from MENA Countries.” Review of Development
Economics, 12(4): 803-817.
Acemoglu, Daron, Simon Johnson and James A. Robinson. 2001. “The Colonial Origins of
Comparative Development: An Empirical Investigation.” American Economic Review,
91(5): 1369-1401.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2002. “Reversal of Fortune:
Geography and Institution in the Making of the Modern World Income Distribution.”
Quarterly Journal of Economics, 117(4): 1231-1294.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2005. “Institutions as the
Fundamental Cause of Long-Run Growth.” In Handbook of Economic Growth - 1, 385472. Philippe Aghion and Steven Durlauf, (eds.). Amsterdam: Elsevier.
Acosta, Pablo A., Nicole R. Baerg and Federico S. Mandelman. 2009. “Financial Development,
Remittances, and Real Exchange Rate Appreciation.” Economic Review, 1(2009).
Atlanta: Federal Reserve Bank of Atlanta.
Acosta, Pablo A., César Calderón, Pablo Fajnzylber and J. Humberto López. 2006.
“Remittances and Development in Latin America.” The World Economy, 29(7): 957-987.
Acosta, Pablo A., César Calderón, Pablo Fajnzylber and J. Humberto López. 2008a. “Do
Remittances Lower Poverty Levels in Latin America?” In Remittances and
Development: Lessons from Latin America, 87-132. Pablo Fajnzylber and J. Humberto
López, (eds.). Washington, D.C.: World Bank.
370
Stellenbosch University http://scholar.sun.ac.za
Acosta, Pablo A., César Calderón, Pablo Fajnzylber and J. Humberto López. 2008b. “What is
the Impact of International Remittances on Poverty and Inequality in Latin America?”
World Development, 36(1): 88-114.
Adams, Richard H. jr. 1989. “Worker Remittances and Inequality in Rural Egypt.” Economic
Development and Cultural Change, 38(1): 45-71.
Adams Richard H. jr. 1991. “The Economic Uses and Impact of International Remittances in
Rural Egypt.” Economic Development and Cultural Change, 39(4): 695-722.
Adams, Richard H. jr. 1998. “Remittances, Investment, and Rural Asset Accumulation in
Pakistan.” Economic Development and Cultural Change, 47(1): 155-173.
Adams, Richard H. jr. 2006. “Remittances and Poverty in Ghana.” World Bank Policy Research
Working Paper 3838. Washington, D.C.: World Bank.
Adams, Richard H. jr. 2009. “The Determinants of International Remittances in Developing
Countries.” World Development, 37(1): 93-103.
Adams, Richard H. Jr. and John Page. 2005. “Do International Migration and Remittances
Reduce Poverty in Developing Countries?” World Development, 33(10): 1645-1669.
Adelman, Irma and J. Edward Taylor. 1990. “Is Structural Adjustment with a Human Face
Possible? The Case of Mexico.” Journal of Development Studies, 26(3): 387-407.
Adenutsi, Deodat E. 2008. “The Policy Dilemma of Openness and Seigniorage-Maximizing
Inflation in Dollarized Developing Countries: The Ghanaian Experience.” West African
Journal of Monetary and Economic Integration, 8(1): 91-129.
Adenutsi, Deodat E. 2010a. “Long-Run Macroeconomic Impact of International Migrant
Remittances on Human Development in Low-Income Countries: A Panel Analysis of
Sub-Saharan Africa.” Journal of International Economic Studies, 24(March): 113-132.
Adenutsi, Deodat E. 2010b. “Do International Remittances Promote Human Development in
Poor Countries? Empirical Evidence from Sub-Saharan Africa.” International Journal of
Applied Economics and Finance, 4(1): 31-45.
Adenutsi, Deodat E. 2011. “Financial Development, International Migrant Remittances and
Endogenous Growth in Ghana.” Studies in Economics and Finance, 28(1): 68-89.
Adenutsi, Deodat E. and Christian R.K. Ahortor. 2008. “Remittances, Exchange Rate and
Monetary Policy in Ghana.” West African Journal of Monetary and Economic
Integration, 8(2): 1-42.
Adenutsi, Deodat E, Meshach J. Aziakpono and Matthew K. Ocran. 2012. “Macroeconomic
Environment and Remittances in Post-Independent Sub-Saharan Africa: Magnitudes,
Trends and Stylised Facts.” Studies in Economics and Econometrics, 36(2): 1-22.
371
Stellenbosch University http://scholar.sun.ac.za
Adenutsi, Deodat E. and Charles A. Yartey. 2007. “Financial Sector Development and the
Macrodynamics of „de facto‟ Dollarisation in Developing Countries: The Case of Ghana.”
West African Journal of Monetary and Economic Integration, 7(2): 49-90.
AfDB (African Development Bank). 1994. “Financial Structures, Reforms and Economic
Development in Africa.” African Development Report, 1994: 192-256.
AfDB (African Development Bank). 2002. Achieving the Millennium Development Goals in
Africa: Progress, Prospects and Policy Implications. Abidjan: African Development
Bank.
AFRODAD (African Forum and Network on Debt and Development). 2007. The Impact of
Economic Reform Programmes on Social Services: The Case of Malawi. March 2007.
Harare: African Forum and Network on Debt and Development.
Aggarwal, Reena, Asli Demirgüç-Kunt and Maria S.M. Peria. 2006. “Do Workers‟ Remittances
Promote Financial Development?” World Bank Policy Research Paper 3957.
Washington, D.C.: World Bank.
Aghion, Philippe and Peter Howitt. 1992. “A Model of Growth through Creative Destruction.”
Econometrica, 60(2): 323-351.
Aghion, Philippe and Peter Howitt. 1997. Endogenous Growth Theory. Cambridge, MA:
Massachusetts Institute of Technology (MIT) Press.
Aghion, Philippe and Peter Howitt. 1998. “Capital Accumulation and Innovation as
Complementary Factors in Long-Run Growth.” Journal of Economic Growth, 3(2): 111130.
Ahlburg, Dennis A. 1991. Remittances and their Impact: A Study of Tonga and Western
Samoa. Canberra: National Centre for Development Studies.
Ahmed, Ismail I. 2000. “Remittances and their Economic Impact in Post-War Somaliland.”
Disasters, 24(4): 380-389.
Ahmed, Junaid, Khalid Zaman and Iqtidar Shah. 2011. “An Empirical Analysis of RemittancesGrowth Nexus in Pakistan using Bounds Testing Approach.” Journal of Economics and
International Finance, 3(3): 176-186.
Ahmed, Mohammad. 1988. “Financial Repression in the LDCs: A Survey of Issues and
Reappraisal.” Pakistan Journal of Applied Economics, 7(1): 19-41.
Ahortor, Christian R.K. and Deodat E. Adenutsi. 2009. “The Impact of Remittances on
Economic Growth in Small-Open Developing Economies.” Journal of Applied Sciences,
9(18): 3275-3286.
372
Stellenbosch University http://scholar.sun.ac.za
Akinlo, A. Enisan. 2004. “Foreign Direct Investment and Growth in Nigeria: An Empirical
Investigation.” Journal of Policy Modelling, 26(5): 627-639.
Akkoyunlu, Sule. 2010. “Are Turkish Migrants Altruistic? Evidence from Macro Data.” KOF
Working Paper 246. Zurich: KOF Swiss Economic Institute.
Akkoyunlu, Sule and Konstantin A. Kholodilin. 2006. “What Affects the Remittances of Turkish
Workers: Turkish or German Output?” GIER Discussion Paper 622. Berlin: German
Institute for Economic Research.
Alberola, Enrique and Rodrigo C. Salvado. 2006. “Banks, Remittances and Financial
Deepening in Receiving Countries - A Model.” Banco de España Working Paper 0621.
Madrid: Banco de España.
Alesina, Alberto, Enrico Spolaore and Romain Wacziarg. 2000. “Economic Integration and
Political Disintegration.” American Economic Review, 90(5): 1276-1296.
Alesina, Alberto, Enrico Spolaore and Romain Wacziarg. 2005. “Trade, Growth and the Size of
Countries.” In Handbook of Economic Growth, Vol. 1B, 1499-1542. Phillipe Aghion and
Steven N. Durlauf, (eds.). Amsterdam: North-Holland.
Alexious, Constantinos. 2007. “Unravelling the „Mystery‟ between Public Expenditure and
Growth: Empirical Evidence from Greece.” International Journal of Economics, 1(1): 2131.
Allen, Donald S. and Leonce Ndikumana. 2000. “Financial Intermediation and Economic
Growth in Southern Africa.” Journal of African Economies, 9(2): 132-160.
Almeida, Carlos C. 1973. “Emigration, espace et sous-Développement.” International Migration,
11(3): 112-117.
Altonji, Joseph G. and Lewis M. Segal. 1996. “Small-Sample Bias in GMM estimation of
Covariance Structures.” Journal of Business and Economic Statistics, 14(3): 353-366.
Ambrosius, Christian. 2011. “Are Remittances a „Catalyst‟ for Financial Access? Evidence from
Mexico.” Mimeo. Berlin: Freie Universität Berlin.
Amuedo-Dorantes, Catalina and Susan Pozo. 2004. “Workers‟ Remittances and the Real
Exchange Rate: A Paradox of Gifts.” World Development, 32(4): 1407-1417.
Amuedo-Dorantes, Catalina and Susan Pozo. 2006. “Remittances as Insurance: Evidence from
Mexican Immigrants.” Journal of Population Economics, 19(2): 227-254.
Amuedo-Dorantes, Catalina, Susan Pozo and Carlos Vargas-Silva. 2007. “Remittances and the
Macroeconomy: The Case of Small Island Developing States.” UNU-WIDER Research
Paper 2007/22. Helsinki: United Nations University, World Institute for Development
Economics Research.
373
Stellenbosch University http://scholar.sun.ac.za
Anderson, Torben G. and Bent E. Sørensen. 1996. “GMM Estimation of a Stochastic Volatility
Model: A Monte Carlo Study.” Journal of Business and Economic Statistics, 14(3): 328352.
Andrews, Donald W.K. 1993. “Tests for Parameter Instability and Structural Change with
Unknown Change Point.” Econometrica, 61(4): 821-856.
Andrews, Donald W.K. 2003. “End-of-Sample Instability Tests.” Econometrica, 71(6): 16611694.
Andrews, Donald W.K. and Werner Ploberger. 1994. “Optimal Tests when a Nuisance
Parameter is Present only under the Alternative.” Econometrica, 62(6): 1383-1414.
Ang, James B. 2010. “Savings Mobilisation, Financial Development and Liberalisation: The
Case of Malaysia.” Munich Personal RePEc Archive Paper 21718. [Online] Available:
http://mpra.ub.uni-muenchen.de/21718/. Posted: March 28, 2010. Accessed: July 18,
2010.
Ang, James B. and Warwick J. McKibbin. 2007. “Financial Liberalisation, Financial Sector
Development and Growth: Evidence from Malaysia.” Journal of Development
Economics, 84(1): 215-233.
Ang, James B. and Kunal Sen. 2011. “Private Saving in India and Malaysia Compared: The
Roles of Financial Liberalisation and Expected Pension Benefits.” Empirical Economics,
41(2): 247-267.
Appleyard, Reginald T. 1989. “Migration and Development: Myths and Reality.” International
Migration Review, 23(3): 486-499.
Arcand, Jean-Louis, Enrico Berkes and Ugo Panizza. 2012. “Too Much Finance.” IMF Working
Paper 12/161. Washington, D.C.: International Monetary Fund.
Arellano, Manuel. 2003. Panel Data Econometrics. Oxford: Oxford University Press.
Arellano, Manuel and Stephen Bond. 1991. “Some Tests of Specification for Panel Data: Monte
Carlo Evidence and an Application to Employment Equations.” Review of Economic
Studies, 58(2): 277-297.
Arellano, Manuel and Olympia Bover. 1995. “Another Look at the Instrumental Variable
Estimation of Error-Components Models.” Journal of Econometrics, 68(1): 29-51.
Arestis, Philip. 2005. “Financial Liberalisation and the Relationship between Finance and
Growth.” CEPP Working Paper 05/05. Cambridge: Centre for Economic and Public
Policy, University of Cambridge.
374
Stellenbosch University http://scholar.sun.ac.za
Arestis, Philip and Panicos O. Demetriades. 1992. “Investment and Financial Repression:
Theory and Evidence from 63 LDCs.” Department of Economics Working Paper 92/16.
Keele: Keele University.
Arestis, Philip and Panicos O. Demetriades. 1993. “Financial Liberalisation and Economic
Development: A Critical Exposition.” In Money and Banking: Issues for the Twenty First
Century. Essays in Honour of Stephen F. Frowen, 287-303. Philip Arestis, (ed.).
Basingstoke: Macmillan.
Arestis, Philip and Panicos O. Demetriades. 1997. “Financial Development and Economic
Growth: Assessing the Evidence.” Economic Journal, 107(442): 783-799.
Artadi, Elsa V. and Xavier X. Sala-i-Martin. 2003. “The Economics Tragedy of the XXth
Century: Growth in Africa.” Department of Economics Working Paper 684. Universitat
Pompeu Fabra.
Aschauer, David A. 1990. “Is Government Spending Stimulative?” Contemporary Policy Issues,
8(4): 30-45.
Asteriou, Dimitrios. 2006. Applied Econometrics: A Modern Approach using Eviews and
Microfit. Hampshire and New York: Palgrave Macmillan.
Ayadi, Rym, Emrah Arbak, Samy B. Naceur and Willen Pieter de Groen. 2013. “Financial
Development, Bank Efficiency and Economic Growth across the Mediterranean.”
MEDPRO Technical Report No. 30/March 2013. Mediterranean Prospects and the
Socio-Economic Sciences and Humanities Programme of the European Commission.
Aydaş, Osman T., Bilin Neyapti and Kivilcim Metin-Özcan. 2004. “Determinants of Workers‟
Remittances: The Case of Turkey.” Department of Economics Working Paper 0405.
Ankara: Bilkent University.
Azam, Jean-Paul and Flore Gubert. 2005. “Those in Kayes: The Impact of Remittances on their
Recipients in Africa.” Revue Économique, 56(6): 1331-1358.
Bach, Stephen. 2006. “International Mobility of Health Professionals: Brain Drain or Brain
Exchange?” UNU-WIDER Research Paper 2006/82. Helsinki: United Nations
University, World Institute for Development Economics Research.
Bagehot, Walter. 1873. Lombard Street, 1962 Edition. Homewood, IL: Richard Irwin.
Bai, Jushan and Pierre Perron. 1998. “Estimating and Testing Linear Models with Multiple
Structural Changes.” Econometrica, 66(1): 47-78.
Baldé, Yéro. 2009. “Migrants‟ Remittances and Economic Growth in Sub-Saharan Africa.”
LAPE Monograph. Limoges: Laboratoire d‟Analyse et de Prospective Economique
(LAPE), Université de Limoges.
375
Stellenbosch University http://scholar.sun.ac.za
Ballard, Roger. 2004. Migration and Development: How to Make Migration Work for Poverty
Reduction. London: House of Commons International Development Committee.
Baltagi, Badi H. 2008. Econometric Analysis of Panel Data, 4th Edition. Chichester: John Wiley
and Sons.
Bandiera, Oriana, Gerard Caprio, jr., Patrick Honohan and Fabio Schiantarelli. 2000. “Does
Financial Reform Raise or Reduce Savings?” Review of Economics and Statistics,
82(2): 239-263.
Barajas, Adolfo, Ralph Chami, Connel Fullenkamp, Michael Gapen and Peter Montiel 2009.
“Do Workers‟ Remittances Promote Economic Growth?” IMF Working Paper 09/153.
Washington, D.C.: International Monetary Fund.
Barajas, Adolfo, Ralph Chami, Connel Fullenkamp and Anjali Garg. 2010. “The Global
Financial Crisis and Workers‟ Remittances to Africa: What is the Damage?” IMF
Working Paper 10/24. Washington, D.C.: International Monetary Fund.
Barham, Bradford and Stephen Boucher. 1998. “Migration, Remittances, and Inequality:
Estimating the Net Effects of Migration on Income Distribution in Bluefields, Nicaragua.”
Journal of Development Economics, 55(2): 307-331.
Barro, Robert J. 1990. “Government Spending in a Simple Model of Endogenous Growth.”
Journal of Political Economy, 98(special issue): 103-125.
Barro, Robert J. 1991. “Economic Growth in Cross Section of Countries.” Quarterly Journal of
Economics, 56(2): 407-443.
Barro, Robert J. 1998. “Determinants of Economic Growth: A Cross-Country Empirical Study.”
Cambridge, MA: MIT Press.
Barro, Robert J. 2003. “Determinants of Economic Growth in a Panel of Countries.” Annals of
Economics and Finance, 4(2): 231-274.
Barro, Robert J. and Jong-Wha Lee. 1996. “International Comparison of Educational
Attainment.” Journal of Monetary Economics, 32(3): 363-394.
Barro, Robert J. and Xavier X. Sala-í-Martin. 1995. Economic Growth. New York: McGraw-Hill.
Barro, Robert J. and Xavier X. Sala-í-Martin. 1997. “Technological Diffusion, Convergence and
Growth.” Journal of Economic Growth, 2(1): 1-26.
Barro, Robert J. and Xavier X. Sala-í-Martin. 2003. Economic Growth, 2nd Edition. Cambridge,
MA: MIT Press.
Bauer, Peter T. 1976. Dissent on Development. Cambridge, MA: Harvard University Press.
Baum, Christopher F., Mark E. Schaffer and Steven Stillman. 2003. “Instrumental Variables
and GMM: Estimation and Testing.” The Stata Journal, 3(1):1-31.
376
Stellenbosch University http://scholar.sun.ac.za
Baumol, William J. 1986. “Productivity Growth, Convergence and Welfare: What the Long-Run
Data Show.” American Economic Review, 76(5): 1072-1085.
Bayoumi, Tamim A. 1993. “Financial Deregulation and Household Saving.” Economic Journal,
103(421): 1432-1443.
Bayoumi, Tamim A. 1997. Financial Integration and Real Activity. Manchester: Manchester
University Press.
Bayoumi, Tamin A. and Roland MacDonald. 1995. “Consumption, Income, and International
Capital Market Integration.” International Monetary Fund Staff Papers, 42(3): 552-576.
Beck, Thorsten, Asli Demirgüҫ-Kunt and Ross E. Levine. 2000. “A New Database on Financial
Development and Structure.” World Bank Economic Review, 14(3): 597-605.
Becker, Gary S. 1974. “A Theory of Social Interactions.” Journal of Political Economy, 82(6):
1063-1093.
Behr, Andreas. 2003. “A Comparison of Dynamic Panel Data Estimators: Monte Carlo
Evidence and an Application to the Investment Function.” Economic Research Centre
Discussion Paper 05/03. Frankfurt am Main: Deutsche Bundesbank.
Beijer, G. 1970. “International and National Migratory Movements.” International Migration,
8(3): 93-109.
Beim, David O. and Charles W. Calomiris. 2001. Emerging Financial Markets. New York:
McGraw-Hill.
Beine, Michel, Elisabetta Lodigiani and Robert Vermeulen. 2011. “Remittances and Financial
Openness.” DNB Working Paper 317. Amsterdam: De Nederlandsche Bank.
Bekaert, Geert, Campbell R. Harvey and Christian Lundblad. 2005. “Does Financial
Liberalisation Spur Growth?” Journal of Financial Economics, 77(1): 3-55.
Bencivenga, Valerie R. and Bruce D. Smith. 1991. “Financial Intermediation and Endogenous
Growth.” Review of Economic Studies, 58(2): 195-209.
Bengelsdijk, Sjoerd, Smeets Roger and Remco Zwinkels. 2008. “The Impact of Horizontal and
Vertical FDI on Host Country Economic Growth.” International Business Review,
17(2008): 452-472.
Benhabib, Jess and Mark M. Spiegel. 1994. “The Role of Human Capital in Economic
Development: Evidence from Aggregate Cross-Country and Regional US Data.” Journal
of Monetary Economics, 34(1): 143-173.
Bernheim, B. Douglas, Andrei Shleifer and Lawrence Summers. 1985. “The Strategic Bequest
Motive.” Journal of Political Economy, 93(6): 1045-1076.
377
Stellenbosch University http://scholar.sun.ac.za
Bhagwati, Jagdish N. 1973. “The Theory of Immersing Growth: Further Applications.” In
International Trade and Money, 45-54. Michael B. Connolly and Alexander K. Swoboda,
(eds.). Toronto: University of Toronto.
Bhagwati, Jagdish N. 1976. “The Brain Drain.” International Social Science Journal, 28(4): 691729.
Bhatia, Rattan J.1960. “Inflation Deflation, and Economic Development.” IMF Staff Papers,
8(1): 101-114.
Bick, Alexander, Stephanie Kremer and Dieter Nautz. 2009. “Inflation and Growth: New
Evidence from a Dynamic Panel Threshold Analysis.” SFB 649 Discussion Paper 036:
33-47.
de la Bière, Benedicte, Elisabeth Sadoulet, Alain de Janvry and Sylvie Lambert. 2002. “The
Roles of Destination, Gender and Household Composition in Explaining Remittances:
An Analysis for the Dominican Sierra.” Journal of Development Economics, 68(2): 309328.
Bils, Mark and Peter J. Klenow. 2000. “Does Schooling Cause Growth?” American Economic
Review, 90(5): 1160-1183.
Blackburn, Keith and Victor T.Y. Hung. 1996. “A Theory of Growth, Financial Development and
Trade.” Economica, 65: 107-124.
Blom, Svein and Kristin Henriksen, (eds.). 2008. Levekár blant Innvandrere i Norge 2005/2006.
Oslo: Statistics Norway.
Blomström, Magnus, Robert E. Lipsey and Mario Zejan. 1996. “Is Fixed Investment the Key to
Economic Growth?” Quarterly Journal of Economics, 111(1): 269-276.
Blundell, Richard W. and Stephen Bond. 1998. “Initial Conditions and Moment Restrictions in
Dynamic Panel Data Models.” Journal of Econometrics, 87(1): 115-143.
Blundell, Richard and Stephen Bond. 2000. “GMM Estimation with Persistent Panel Data: An
Application to Production Functions.” Econometric Reviews, 19(3): 321-340.
Bobba, Matteo and Andrew Powell. 2007. “Aid and Growth: Politics Matters.” IADB Working
Paper 601. Washington, D.C.: Inter-American Development Bank.
Bolbol, Ali A, Ayten Fatheldin and Mohammed M. Qmran. 2005. “Financial Development,
Structure, and Economic Growth: The Case of Egypt, 1974-2002.” Research in
International Business and Finance, 19(1): 171-194.
Bollard, Albert, David McKenzie and Melanie Morten. 2010. “The Remitting Patterns of African
Migrants in the OECD.” Policy Research Working Paper 5260. Washington, D.C.: World
Bank.
378
Stellenbosch University http://scholar.sun.ac.za
Bollard, Albert, David McKenzie, Melanie Morten and Hillel Rapoport. 2009. “Remittances and
the Brain Drain Revisited: The Microdata show that More Educated Migrants Remit
More.” Mimeo. Washington, D.C.: World Bank.
Bond, Eric W. and Ping Wang. 1996. “A General Two-Sector Model of Endogenous Growth
with Human and Physical Capital: Balanced Growth and Transitional Dynamics.”
Journal of Economic Theory, 68(1): 149-173.
Bond, Stephen. 2002. “Dynamic Panel Data Models: A Guide to Micro Data Methods and
Practice.” Cemmap Working Paper CWP 09/02. London: Centre for Microdata Methods
and Practice, The Institute for Fiscal Studies.
Bond, Stephen R., Anke Hoeffler and Jonathan Temple. 2001. “GMM Estimation of Empirical
Growth Models.” CEPR Discussion Paper 3048. London: Centre for Economic Policy
Research.
Boone, Peter. 1994. “The Impact of Foreign Aid on Savings and Growth.” CEP Working Paper
677. London: Centre for Economic Performance, London School of Economics.
Boone, Peter. 1996. “Politics and the Effectiveness of Foreign Aid.” European Economic
Review, 40(2): 289-329.
Borensztein, Eduardo J., José de Gregorio and Jong-Wha Lee. 1998. “How Does Foreign
Direct Investment Affect Growth?” Journal of International Economics, 45(1): 115-135.
Borjas, George J. 1989. “Immigrant and Emigrant Earnings: A Longitudinal Study.” Economic
Inquiry, 27(1): 21-37.
Borovnik, Maria. 2003. “Seafarers in Kiribati – Consequences of International Labour
Circulation.” Unpublished PhD Thesis. Christchurch: University of Canterbury.
Bos, Hendricus C., Martin Sanders and Carlo Secchi. 1974. Private Foreign Investment in
Developing Countries: A Quantitative Study on the Macroeconomic Effects. Dardrecht:
Reidel.
Bouhga-Hagbe, Jacques. 2004. “A Theory of Workers‟ Remittances with an Application to
Morocco.” IMF Working Paper 04/194. Washington, D.C.: International Monetary Fund.
Bouhga-Hagbe, Jacques. 2006. “Altruism and Workers‟ Remittances: Evidence from Selected
Countries in the Middle East and Central Asia.” IMF Working Paper 06/130.
Washington, D.C.: International Monetary Fund.
Bowsher, Clive G. 2002. “On Testing Overidentifying Restrictions in Dynamic Panel Data
Models.” Economics Letters, 77(2): 211-220.
Boyd, David J. 1990. “New Wealth and Old Power: Circulation, Remittances and the Control of
Inequality in an Eastern Highlands Community, Papua New Guinea.” In Migration and
379
Stellenbosch University http://scholar.sun.ac.za
Development in the South Pacific, 97-106. John Connell, (ed.). Canberra: National
Centre for Development Studies, Australian National University.
Bracking, Sarah. 2003. “Sending Money Home: Are Remittances Always Beneficial to those
Who Stay Behind?” Journal of International Development, 15(5): 633-644.
Breitung, Jörg. 2000. “The Local Power of Some Unit Root Tests for Panel Data.” Advances in
Economics, 15(2): 161-177.
Breusch, Trevor S. and Rodney A. Pagan. 1980. “The Lagrange Multiplier Test and its
Applications to Model Specification in Econometrics.” Review of Economic Studies,
47(1): 239-253.
Brossard, Mathieu and Luc-Charles Gacougnolle. 2001. Financing Primary Education for All:
Yesterday, Today and Tomorrow. Paris: United Nations Educational, Scientific and
Cultural Organisation (UNESCO).
Brown, Richard P.C. 1994. “Migrants‟ Remittances, Savings, and Investment in the South
Pacific.” International Labour Review, 133(3): 347-367.
Brown, Richard P.C. 1995. Consumption and Investments from Migrants‟ Remittances in the
South Pacific. Geneva: International Labour Organisation.
Brown, Richard P.C. 1997. “Estimating Remittance Functions for Pacific Migrants.” World
Development, 25(4): 613-626.
Brown, Richard P.C. and John Connell. 1993. “The Global Flea Market: Migration, Remittances
and the Informal Economy in Tonga.” Development and Change, 24(4): 611-647.
Brown, Richard P.C. and Bernard Poirine. 2005. “A Model of Migrants‟ Remittances with
Human Capital Investment and Intrafamilial Transfers.” International Migration Review,
39(2): 407-438.
Brown, R.L., James Durbin, and John M. Evans. 1975. “Techniques for Testing the Constancy
of Regression Relationships over Time.” Journal of the Royal Statistical Society,
Series B. 37(2): 149-172.
Bruns, Barbara, Alain Mingat and Ramahatra Rakotomalala. 2003. Achieving Universal Primary
Education by 2015: A Chance for Every Child. Washington, D.C.: World Bank.
Buch, Claudia M., Anja Kuckulenz and Marie-Helene Le Manchec. 2002. “Worker Remittances
and Capital Flows.” Kiel Working Paper 1130. Kiel: Kiel Institute for World Economics.
Buffie, Edward F. 1984. “Financial Repression, the New Structuralists, and Stabilisation Policy
in Semi-Industrialized Economies.” Journal of Development Economics, 14(3): 305-322.
Bugamelli, Matteo and Francesco Paternò. 2008. “Output Growth Volatility and Remittances.”
Temi di Discussioné 673. Rome: Banca D‟Italia.
380
Stellenbosch University http://scholar.sun.ac.za
Burnside, Craig and David Dollar. 2000. “Aid, Policies, and Growth.” American Economic
Review, 90(4): 849-868.
Cai, Qian. 2003. “Migrant Remittances and Family Ties: A Case Study of China.” International
Journal of Population and Geography, 9(6): 471-483.
Calderón, Céasr and Lin Liu. 2003. “The Direction of Causality between Financial Development
and Economic Growth.” Journal of Development Economics, 72(1): 321-334.
Cameron, Rondo. 1967. Banking in the Early Stages of Industrialisation. New York: Oxford
University Press.
Caprio, Gerard, Patrick Honohan and Joseph E. Stiglitz, (eds.). 2001. Financial Liberalisation:
How Far, How Fast? New York: Cambridge University Press.
Carceles, Gabriel, Birger Fredriksen and Patrick Watt. 2001. “Can Sub-Saharan Africa Reach
the International Targets for Human Development?” Africa Region Human
Development Working Paper Series 2001. Washington, D.C.: World Bank.
Carkovic, Maria and Ross Levine. 2005. “Does Foreign Direct Investment Accelerate Economic
Growth?” In Does Foreign Direct Investment Promote Development, 195-220. Theodore
H. Moran, Edward M. Graham and Magnus Blomström, (eds.). Washington, D.C.:
Institute for International Economics.
Carling, Jorgen. 2004. “Policy Options for Increasing the Benefits of Remittances.” CMPS
Working Paper 8. Oxford: Centre on Migration, Policy and Society, University of Oxford.
Cartagena, Edgar R. 2004. “Las Remesas Familiares Salvadoreňas y nas relación con los
Ciclos y Perturbaciones Económicas.” Documentos Ocasionales 2004-01. San
Salvador: Banco Central de Reserva de El Salvador.
Caselli, Francesco, Gerardo Esquivel and Fernando Lefort. 1996. “Reopening the
Convergence Debate: A New Look at Cross-Country Growth Empirics.” Journal of
Economic Growth, 1(3): 363-389.
Catrinescu, Natalia, Miguel León-Ledesma, Matloob Piracha and Bryce Quillin. 2009.
“Remittances, Institutions and Economic Growth.” World Development, 37(1): 81-92.
CfA (Commission for Africa). 2005. Our Common Interest: Report of the Commission for Africa.
London: Department for International Development, Commission for Africa.
Chami, Ralph, Adolfo Barajas, Thomas F. Cosimano, Connel Fullenkamp, Michael Gapen and
Peter Montiel. 2008. “Macroeconomic Consequences of Remittances.” IMF Occasional
Paper 259. Washington, D.C.: International Monetary Fund.
381
Stellenbosch University http://scholar.sun.ac.za
Chami, Ralph, Connell Fullenkamp and S. Jahjah. 2005. “Are Immigrant Remittance Flows a
Source of Capital for Development?” International Monetary Fund Staff Paper, 52(1):
55-81.
Chami, Ralph, Dalia Hakura and Peter Montiel. 2009. “Remittances: An Automatic Output
Stabiliser?” IMF Working Paper 09/91. Washington, D.C.: International Monetary Fund.
Chan, Felix, Tommaso Mancini-Griffoli and Laurent L. Pauwels. 2012. “Testing Structural
Stability in Heterogeneous Panels with an Application to the Euro‟s Trade Effect.”
Journal of Time Series Econometrics, 4(2): 1-35.
Chang, Doug and Woo S. Jung. 1984. “Unorganised Money Markets in LDCs: The McKinnonShaw Hypothesis versus the van Wijnbegen Hypothesis.” Department of Economics
Working Paper 84-W21. Nashville: University of Vanderbilt.
Chang, Roberto, Linda Kaltani and Norman Loayza. 2009. “Openness is Good for Growth: The
Role of Policy Complementaries.” Journal of Development Economics, 90(1): 33-49.
Chenery, Hollis B. and Alan M. Strout. 1966. “Foreign Assistance and Economic Development.”
American Economic Review, 56(4): 679-732.
Chilivumbu, A. 1985. Migration and Uneven Rural Development in Africa: The Case of Zambia.
Lanham: University Press of America.
Chipeta, Chinyamata and Willie Kachaka. 2005. “Role of Migrants‟ Remittances in Unstable
Low-Income Economy: A Case Study of Malawi.” Chancellor College Working Paper
2005/05. Zomba: University of Malawi.
Cho, Yoon Je. 1986. The Effects of Financial Liberalisation on the Development of the
Financial Market and the Allocation of Credit to Corporate Sectors: The Korean Case.
Washington, D.C.: World Bank.
Cho, Yoon Je. 1988. “The Effect of Financial Liberalisation on the Efficiency of Credit
Allocation: Some Evidence from Korea.” Journal of Development Economics, 29(1):
101-110.
Chow, Gregory C. 1960. “Tests of Equality between Sets of Coefficients in Two Linear
Regressions.” Econometrica, 28(3): 591-605.
Chow, Gregory C. 1984. “Random and Changing Coefficient Models.” In Handbook of
Econometrics, Vol. 2, 1213-1245. Zvi Griliches and Michael D. Intriligator, (eds.).
Amsterdam: North Holland.
Christopoulos, Dimitris K. and Efthymios G. Tsionas. 2004. “Financial Development and
Economic Growth: Evidence from Panel Unit Root and Co-integration Test.” Journal of
Development Economics, 73(1): 55-74.
382
Stellenbosch University http://scholar.sun.ac.za
Clark, P. 2004. “The Economic Impact of Contracted Labour upon the Livelihoods of Small
Pacific Island States: An Examination of the Expenditure Patterns of I-Kiribati and
Tuvaluan Seafarers and the Dependents.” Unpublished Masters of Social Planning and
Development Thesis. Queensland: University of Queensland.
Clarke, George R.G. and Scott J. Wallsten. 2003. “Do Remittances Act Like Insurance?
Evidence from a Natural Disaster in Jamaica.” Unpublished Working Paper,
Development Research Group: January 2003. Washington, D.C.: World Bank.
Clarke, George R.G and Scott Wallsten. 2004. “Do Remittances Protect Households in
Developing Countries against Shocks? Evidence from a Natural Disaster in Jamaica.”
Unpublished Paper. Washington, D.C.: World Bank.
Clemens, Michael A., Steven Radelet and Rikil R. Bhavnani. 2004. “Counting Chickens when
they Hatch: The Short-Term Effect of Aid on Growth.” CED Working Paper 44.
Washington, D.C.: Centre for Global Development.
Cobb, Charles W. and Paul H. Douglas. 1928. “A Theory of Production.” American Economic
Review, XVIII(Supplement): 139-165.
Coe, David T. and Elhanan Helpman. 1995. “International R&D Spillovers.” European
Economic Reviewers, 39(5): 859-887.
Collier, Paul and David Dollar. 2001. “Can the World cut Poverty in Half? How Policy Reform
and Effective Aid can meet International Development Goals.” World Development,
29(11): 1787-1802.
Collier Paul and David Dollar. 2002. “Aid Allocation and Poverty Reduction.” European
Economic Review, 46(8): 1475-1500.
Collier Paul and Anke E. Hoeffler. 2004. “Aid, Policy and Growth in Post-Conflict Societies.”
European Economic Review, 48(5): 1125-1145.
Connell, John. 1980. “Remittances and Rural Development: Migration, Dependency and
Inequality in the South Pacific.” National Centre for Development Studies Occasional
Paper 22. Canberra: Australia National University.
Cooray, Arusha V. 2012. “The Impact of Migrant Remittances on Economic Growth: Evidence
from South Asia.” Review of International Economies, 20(5): 985-998.
Courakis, Anthony S. 1984. “Constraints on Bank Choices and Financial Repression in Less
Developed Countries.” Oxford Bulletin of Economics and Statistics, 46(4): 341-370.
Courakis, Anthony S. 1986. “In What Sense do Compulsory Ratios reduce the Volume of
Deposits?” In The Operation and Regulation of Financial Markets, 150-186. Charles
E.A. Goodhart, David Curie and David T. Llewellyn, (eds.). London: Macmillan.
383
Stellenbosch University http://scholar.sun.ac.za
Cox, Donald. 1987. “Motives for Private Income Transfers.” Journal of Political Economy, 95(3):
508-546.
Cox, Donald and Emmanuel Jimenez. 1992. “Social Security and Private Transfers in a
Developing Country: The Case of Peru.” World Bank Economic Review, 6(1): 155-169.
Cox, Donald and Oded Stark. 1994. “Intergenerational Transfers and the Demonstration
Effect.” Boston College Working Papers in Economics 329. Boston: Boston College.
Cox-Edwards, Alejandro and Manuelita Ureta. 2003. “International Migration, Remittances and
Schooling: Evidence from El Salvador.” Journal of Development Economics, 72(2): 429461.
Crãciun, Cristina. 2006. “Migration and Remittances in the Republic of Moldova: Empirical
Evidence at Micro Level.” Unpublished Master of Arts Thesis. Kiev: Economics,
National University „Kyiv-Mohyla Academic‟.
Cukierman, Alex, Pantelis Kalaitzidakis, Lawrence H. Summers and Steven B. Webb. 1993.
“Central Bank Independence, Growth, Investment, and Real Rates.” CarnegieRochester Conference Series on Public Policy, 39(Autumn): 95-140.
van Dalen, Hendrik P., George Groenewold and Tineke Fokkema. 2005. “The Effect of
Remittances on Emigration Intentions in Egypt, Morocco and Turkey.” Population
Studies - A Journal of Demography, 59(3): 375-392.
Dalgaard, Carl-Johan, Henrik Hansen and Finn Tarp. 2004. “On the Empirics of Foreign Aid
and Growth.” The Economic Journal, 114(496): 191-216.
Darrat, Ali F., K.P. LaBarge and R.A. LaBarge. 1989. “Is Financial Deepening a Reliable
Prescription for Economic Growth.” American Economist, 33(2): 25-33.
Darrat, Ali F., Samer Kherfi and Mohamed Soliman. 2005. “FDI and Economic Growth in CEE
and MENA Countries: A Tale of Two Regions.” Economic Research Forum - 12th
Annual Conference, 19-21 December. Cairo: Cairo University.
Datta, Kavita, Cathy Mcllwaine, Jane Wills, Yara Evans, Joana Herbert and Jon May. 2006.
Challenging Remittances as the New Development Mantra Perspectives from Low-Paid
Migrant Workers in London. Queen Mary: University of London.
Demetriades, Panicos O. and Khaled Hussein. 1996. “Does Financial Development Cause
Growth? Time Series Evidence from 16 Countries.” Journal of Development Economics,
51(2): 387-411.
Dennis, Jennifer. 2003. Pacific Island Seafarers: A Study of the Economic and Social
Implications of Seafaring on Dependants and Communities. Research Report for the
384
Stellenbosch University http://scholar.sun.ac.za
Pacific Seafarers Training Project, Regional Maritime Programme. Suva: Secretariat of
the Pacific Community.
Devarajan, Shantayanan, Margaret J. Miller and Eric V. Swanson. 2002. “Goals for
Development: History, Prospects and Costs.” World Bank Policy Research Working
Paper 2819. Washington, D.C.: World Bank.
Diatta, Marie A. and Ndiaga Mbow. 1999. “Releasing the Development Potential of Return
Migration: The Case of Senegal.” International Migration, 37(1): 243-266.
Dollar, David. 1992. “Outward-Oriented Developing Economies Really do Grow More Rapidly:
Evidence from 95 LDCs, 1976-1985.” Economic Development and Cultural Change,
40(3): 523-544.
Dollar, David and Aart Kraay. 2003. “Institutions, Trade, and Growth.” Journal of Monetary
Economics, 50(1): 133-162.
Dollar, David and Aart Kraay. 2004. “Trade, Growth and Poverty.” The Economic Journal,
114(February): 22-49.
Dollar, David and Jakob Svensson. 2000. “What Explains the Success or Failure of Structural
Adjustment Programmes?” The Economic Journal, 110(466): 894-917.
Domar, Evsey D. 1947. “Expansion and Employment.” American Economic Review, 37(1,
March): 34-55.
Dorrance, Graeme S. 1963. “The Effect of Inflation on Economic Development.” IMF Staff
Papers, 10(1): 1-47.
Drazen, Allan. 1981. “Inflation and Capital Accumulation under a Finite Horizon.” Journal of
Monetary Economics, 8(2): 245-262.
Drinkwater, Stephen, Paul Levine and Emanuela Lotti. 2006. “Labour Market and Investment
Effects of Remittances.” Mimeo. Surrey: University of Surrey.
Duc, Cindy and Emmanuelle Lavallée. 2004. “Les Bases de Données sur la Gouvernance.”
Cahiers de recherches de l‟EURISCO, 2004-11, Université Paris-Dauphine.
Durand, Jorge, William Kandel, Emilio A. Parrado and Douglas S. Massey. 1996a.
“International Migration and Development in Mexican Communities.” Demography,
33(2): 249-264.
Durand, Jorge, William Kandel, Emilio A. Parrado and Douglas S. Massey. 1996b.
“Migradollars and Development: A Reconsideration of the Mexican Case.” International
Migration Review, 30(2): 423-444.
385
Stellenbosch University http://scholar.sun.ac.za
Durham, Benson J. 2004. “Absorptive Capacity and the Effects of Foreign Direct Investment
and Equity Foreign Investment on Economic Growth.” European Economic Review,
48(2): 285-306.
Duryea, Suzanne, Ernesto López-Córdova, and Alexandra Olmedo. 2005. “Migrant
Remittances and Infant Mortality: Evidence from Mexico.” Mimeo. Washington, D.C.:
Inter-American Development Bank.
Easterly, William. 2001. The Elusive Quest for Economic Growth: Economists‟ Adventures and
Misadventures in the Tropics. Cambridge, MA: MIT Press.
Easterly, William. 2003. “Can Foreign Aid Buy Growth?” Journal of Economic Perspective,
17(3): 23-48.
Easterly, William. 2005. “What did Structural Adjustment Adjust? The Association of Politics
and Growth with Repeated IMF and World Bank Adjustment Loans.” Journal of
Development Economics, 76(1): 1-22.
Easterly, William. 2007a. “Are Aid Agencies Improving?” Economic Policy, 22(52): 633-678.
Easterly, William. 2007b. “Was Development Association a Mistake?” American Economic
Review Papers and Proceedings, 97(1): 328-332.
Easterly, William and Ross Levine. 2001. “It‟s Not for Factor Accumulation: Stylised Facts and
Growth Models.” World Bank Economic Review, 15(2): 177-219.
Easterly, William, Ross Levine and David Roodman. 2004. “Aid, Politics, and Growth:
Comment.” American Economic Review Papers and Proceedings, 94(3): 774-780.
Ebeke, Christian and Maëlan Le Goff. 2009. “Why Migrants‟ Remittances Reduce Income
Inequality in Some Countries and Not in Others.” Paper Presented at International
Conference on Inequalities and Development in the Mediterranean Countries, May 2123, 2009. Istanbul: Galatasaray University.
Edwards, Sebastián. 1989. “On the Sequencing of Structural Reforms.” OECD Department of
Economics and Statistics Working Paper 70. Paris: Organisation for Economic Cooperation and Development (OECD) Publishing.
Edwards, Sebastián. 1992. “Trade Orientation, Distortions and Growth in Developing
Countries,” Journal of Development Economics, 39(1): 31-57.
Edwards, Sebastián. 1998. “Openness, Productivity and Growth: What Do We Really Know?”
Economic Journal, 108(447): 383-398.
Elbadawi, Ibrahim A. and Roberto de Rezend Rocha. 1992. “Determinants of Expatriate
Workers‟ Remittances in North Africa and Europe.” World Bank Policy Research
Working Paper Series 1038. Washington, D.C.: World Bank.
386
Stellenbosch University http://scholar.sun.ac.za
Elkhider, Abdelkader, Abdelhamid El Bouhadi and El Mustapha Kchirid. 2008. “The
Determinants of Remittances by Moroccans Resident Abroad: Empirical Study from
1970-2006.” Proceedings for the African Economic Conference 2008: 133-160.
El-Sakka, Mohammed I.T. and Robert McNabb. 1999. “The Macroeconomic Determinants of
Emigrant Remittances.” World Development, 27(8): 1493-1502.
Engen, Eric M. and Jonathan Skinner. 1992. “Fiscal Policy and Economic Growth.” NBER
Working Paper 4223. Cambridge, MA: National Bureau of Economic Research.
Engle, Robert F. and Clive W.J. Granger. 1987. “Co-integration and Error Correction
Representation, Estimation, and Testing.” Econometrica, 55(2): 251-276.
Entorf, Horst. 1997. “Random Walks with Drifts: Nonsense Regression and Spurious FixedEffect Estimation.” Journal of Econometrics, 80(2): 287-296.
Entzinger, Han. 1985. “Return Migration in Western Europe: Current Policy Trends and their
Implications in Particular for the 2nd Generation.” International Migration, 23(2): 263-290.
Epstein, Gerald A. 1999. UNCTAD‟s World Investment Report, 1999: A Critique of Neo-Liberal
Globalisation? Amherst: University of Massachusetts.
Esquivel, Gerardo and Alejandra Huerta-Pineda. 2007. “Remittances and Poverty in Mexico: A
Propensity Score Matching Approach.” Integration and Trade Journal, 27(2): 45-57.
Faeamani, Sione U. 1995. “The Impact of Remittances on Rural Development in Tongan
Villages.” Asian and Pacific Migration Journal, 4(1): 139-156.
Fafchamps, Marcel and Matthias Schündeln. 2013. “Local Financial Development and Firm
Performance: Evidence from Morocco.” Journal of Development Economics, 103(2013):
15-28
Faini, Riccardo. 2002. “Development, Trade, and Migration.” Proceedings from the Annual
World Bank Conference on Development Economics for Europe, 1-2: 85-116.
Faini, Riccardo. 2003. “Is the Brain Drain an Unmitigated Blessing?” UNU-WIDER Discussion
Paper 2003/64. Helsinki: United Nations University, World Institute for Development
Economics Research.
Faini, Riccardo. 2006a. “Migration and Remittances: The Impact on the Countries of Origin.”
Unpublished Paper. Rome: Università di Roma Tor Vergata. [Online] Available:
http://www.eudnet.net/download/Faini.pdf. Posted: (unspecified). Accessed: July 12,
2010.
Faini, Riccardo. 2006b. “Remittances and the Brain Drain: Do More Skilled Migrants Remit
More?” The World Bank Economic Review, 21(2): 177-191.
387
Stellenbosch University http://scholar.sun.ac.za
Fajnzylber, Pablo and J. Humberto López. 2007. “Close to Home: The Development Impact of
Remittances in Latin America.” Mimeo. Washington, D.C.: World Bank.
Faria, Joao R. and Francisco G. Carneiro. 2001. “Does High Inflation Affect Growth in the Long
and Short-Run?” Journal of Applied Economics, IV(1): 89-105.
Fase, Martin M.G. and R.C.N. Abma. 2003. “Financial Environment and Economic Growth in
Selected Asian Countries.” Journal of Asian Economics, 14(2013): 11-21.
Favara, Giovanni. 2003. “An Empirical Reassessment of the Relationship between Finance and
Growth.” IMF Working Paper 03/123. Washington, D.C.: International Monetary Fund.
Fay, Marianne, Danny Leipziger, Quentin Wodon and Tito Yepes. 2005. “Achieving ChildRelated Millennium Development Goals: The Goals of Infrastructure.” World
Development, 33(8): 1267-1284.
Fayissa, Bichaka and Christian Nsiah. 2008. “The Impact of Remittances on Economic Growth
and Development in Africa.” Department of Economics and Finance Working Paper
Series February 2008. Murfreesboro: Middle Tennessee State University.
Fayissa, Bichaka and Christian Nsiah. 2010. “Can Remittances spur Economic Growth and
Development? Evidence from Latin American Countries.” Department of Economics
and Finance Working Paper Series March 2010. Murfreesboro: Middle Tennessee State
University.
Fayissa, Bichaka and Christian Nsiah. 2011. “Remittances and Economic Growth in Africa,
Asia and Latin America-Caribbean Countries: A Panel Unit Root and Panel Cointegration Analysis.” Department of Economics and Finance Working Paper Series
June 2011. Murfreesboro: Middle Tennessee State University.
Fedderke, Johannes W. 2005. “Technology, Human Capital and Growth.” UCT School of
Economics Working Paper 27. Cape Town: University of Cape Town.
Feenstra, Roberto C. 1996. “Trade and Uneven Growth.” Journal of Development Economics,
49(1): 229-256.
Ferreira, Cândida. 2008. “Public Debt and Economic Growth: A Granger Causality Panel Data
Approach.” ISEG-UTL Working Paper WP24/2009/DE/UECE. Lisbon: Technical
University of Lisbon.
Findley, Sally E. and Salif Sow. 1998. “From Season to Season: Agriculture, Poverty, and
Migration in the Senegal River Valley, Mali.” In Emigration Dynamics in Developing
Countries: Sub-Saharan Africa, 69-144. Reginald Appleyard, (ed.). London: Ahsgate
Publishing Limited.
388
Stellenbosch University http://scholar.sun.ac.za
Fischer, Stanley. 1979a. “Capital Accumulation on the Transition Path in a Monetary Optimising
Model.” Econometrica, 47(6): 1433-1439.
Fischer, Stanley. 1979b. “Anticipations and the Nonneutrality of Money.” Journal of Political
Economy, 87(2): 225-252.
Fischer, Stanley. 1981. “Towards an Understanding of the Costs of Inflation: II.” CarnegieRochester Conference Series on Public Policy, 15(Autumn): 5-41.
Fischer, Stanley. 1993. “The Role of Macroeconomic Factors in Growth.” Journal of Monetary
Economics, 32(3): 485-512.
Fisman, Raymond and Inessa Love. 2003. “Trade Credit, Financial Intermediary Development,
and Industry Growth.” Journal of Finance, 58(1): 353-374.
Fisman, Raymond and Inessa Love. 2004. “Financial Development and Growth in the Shortand Long-Run.” World Bank Policy Research Working Paper 3319. Washington, D.C.:
World Bank.
Fölster, Stefan and Magnus Henrekson. 1999. “Growth and the Public Sector: A Critique of the
Critics.” European Journal of Political Economy, 15(2): 337-358.
Fölster, Stefan and Magnus Henrekson. 2001. “Growth Effects of Government Expenditure and
Taxation in Rich Countries.” European Economic Review, 45(8): 1501-1520.
Fomby, Thomas, R. Carter Hill and Stanley Johnson. 1984. Advanced Econometric Methods.
New York: Springler-Verlag.
Foster, Neil. 2008. “The Impact of Trade Liberalisation on Economic Growth: Evidence from a
Quantile Regression Analysis.” Kyklos, 61(4): 543-567.
Frankel, Jeffrey A. and David H. Romer. 1999. “Does Trade cause Growth?” America
Economic Review, 89(3): 379-399.
Frankel, Jeffrey A. and Andrew K. Rose. 2002. “An Estimate of the Effect of Common
Currencies on Trade and Income.” The Quarterly Journal of Economics, 117(2): 437466.
Fraser Institute. 2008. Economic Freedom of the World: 2008 Annual Report. Vancouver, BC:
The Fraser Institute.
Freund, Caroline and Nikola Spatafora. 2005. “Remittances: Transaction Costs, Determinants,
and Informal Flows.” World Bank Policy Research Working Paper 3704. Washington,
D.C.: World Bank.
Fry, Maxwell J. 1989. “Foreign Debt Instability: An Analysis of National Saving and Domestic
Investment Responses to Foreign Debt Accumulation in 28 Developing Countries.”
Journal of International Money and Finance, 8(3): 315-344.
389
Stellenbosch University http://scholar.sun.ac.za
Fry, Maxwell J. 1993. “Foreign Debt Accumulation: Financial and Fiscal Effects and Monetary
Policy Reactions in Developing Countries.” Journal of International Money and Finance,
12(4): 347-367.
Fry, Maxwell J. 1995. Money, Interest, and Banking in Economic Development, 2nd Edition.
Baltimore: The Johns Hopkins University Press.
de la Fuente, Angel and Rafael Domenech. 2002. “Human Capital in Growth Regressions: How
much Difference does Data Quality Make? An Update and Further Results.” CEPR
Discussion Paper 3587. London: Centre for Economic Policy Research.
Funke, Norbet and Saleh M. Nsouli. 2003. “The New Partnership for Africa‟s Development
(NEPAD): Opportunities and Challenges.” IMF Working Paper 03/69. Washington, D.C.:
International Monetary Fund.
Funkhouser, Edward. 1992. “Migration from Nicaragua: Some Recent Evidence.” World
Development, 20(8): 1209-1218.
Funkhouser, Edward. 1995. “Remittances from International Migration: A Comparison of El
Salvador and Nicaragua.” The Review of Economics and Statistics, 77(1): 137-146.
Gani, Azmat and Basu Sharma. 2013. “Remittances and Credit Provided by the Banking Sector
in Developing Countries.” International Review of Business Research Papers, 9(3): 8598.
Garcia-Fuentes, Pablo A. and P. Lynn Kennedy. 2009. “Remittances and Economic Growth in
Latin America and the Caribbean: The Impact of Human Capital Development.” A Paper
Presented at the Southern Agricultural Economics Association Annual Meeting, January
31-February 3, 2009. Atlanta, Georgia.
Gaulier, Guillaume, Christophe Hurlin and Philippe Jean-Pierre. 1999. “Testing Convergence: A
Panel Data Approach.” Annales d‟Economie et le Statistique, ENSAE, 55-56: 411-427.
Gelbard, Enrique and Sérgio P. Leite. 1999. “Measuring Financial Development in SubSaharan Africa.” IMF Working Paper 99/105. Washington, D.C.: International Monetary
Fund.
Gemech, Firdu and John Struthers. 2003. “The McKinnon-Shaw Hypothesis -Thirty Years On:
A Review of Recent Developments in Financial Liberalisation Theory.” A Paper
Presented at Development Studies Association Annual Conference on Globalisation
and Development. Glasgow: September 2003.
Georges, Eugenia. 1990. The Making of a Transnational Community: Migration, Development,
and Cultural Change in the Dominican Republic. New York: Columbia University Press.
390
Stellenbosch University http://scholar.sun.ac.za
Gerschenkron, Alexander. 1962. Economic Backwardness in Historical Perspective: A Book of
Essays. Cambridge: Harvard University Press.
Gesell, Silvio. 1911. Die natürliche Wirstschaftsordnung durch Freiland und Freigeld. Berlin:
Freiland-Freigeldverlag. [The Natural Economic Order. Translated from the 6th German
Edition by Philip Pye. Berlin: Neo-Verlag, 1929].
Gheeraert, Laurent, Ritha S. Mata and Daniel Traça. 2010. “Remittances and Domestic
Investment in Developing Countries: An Analysis of the Role of Financial Sector
Development.” CEB Working Paper 10/013. Brussels: Centre Emile Bernheim,
Université Libre de Bruxelles.
Ghosh, Atish R. and Steven Phillips. 1998. “Inflation, Disinflation, and Growth.” IMF Working
Paper 98/88. Washington, D.C.: International Monetary Fund.
Ghosh, Bimal. 1997. Gains from Global Linkages: Trade in Services and Movement of
Persons. Hound Mills: Macmillan Press.
Ghosh, Bimal. 2006. Migrants‟ Remittances and Development: Myths, Rhetorics and Realities.
The Hague: International Organisation for Migration.
Ghura, Dhaneswar. 1995. “Macro Policies, External Forces and Economic Growth in SubSaharan Africa.” Economic Development and Cultural Change, 43(4): 759-778.
Gillman, Max, Mark Harris and László Matyas. 2002. “Inflation and Growth: Some Theory and
Evidence.” Discussion Paper D5-1, 10th International Conferences on Panel Data.
Berlin: July 5-6.
Giovannini, Alberto and Marta de Melo. 1993. “Government Revenue from Financial
Repression,” American Economic Review. 83(4): 953-963.
Giuliano, Paola and Marta Ruiz-Arranz. 2009. “Remittances, Financial Development, and
Economic Growth.” Journal of Development Economics, 90(1): 144-152.
Glaeser, Ed, Rafael La Porta, Florencio de Silanes Lopez and Andrei Shleifer. 2004. “Do
Institutions Cause Growth?” NBER Working Paper 10568. Cambridge, MA: National
Bureau of Economic Research.
Glytsos, Nicholas P. 1997. “Remitting Behaviour of “Temporary” and “Permanent” Migrants:
The Case of Greeks in Germany and Australia.” Labour, 32(5): 468-496.
Glytsos, Nicholas P. 2002. “The Role of Migrant Remittances in Development: Evidence from
Mediterranean Countries.” International Migration, 40(1): 5-26.
Glytsos, Nicholas P. 2005. “The Contribution of Remittances to Growth: A Dynamic Approach
and Empirical Analysis.” Journal of Economic Studies, 11(3): 409-435.
391
Stellenbosch University http://scholar.sun.ac.za
Goldsmith, Raymond W. 1969. Financial Structure and Development. New Haven: Yale
University Press.
Gray, Cherly, Joel Hellman and Randi Ryterman. 2004. Anticorruption in Transition 2,
Corruption in Enterprise-State Interactions in Europe and Central Asia 1999-2002.
Washington, D.C.: The World Bank.
Greenaway, David, Wyn Morgan and Peter W. Wright. 2002. “Trade Liberalisation and Growth
in Developing Countries.” Journal of Development Economics, 67(1): 229-244.
Greene, William H. 2003. Econometric Analysis, 5th Edition. New Jersey: Prentice-Hall.
Greenwood, Jeremy and Boyan Jovanovic. 1990. “Financial Development, Growth and the
Distribution of Income.” Journal of Political Economy, 98(3): 1076-1107.
Greenwood, Jeremy and Bruce Smith. 1997. “Financial Markets in Development and the
Development of Financial Markets.” Journal of Political Economy, 98(5): 1076-1107.
de Gregorio, José. 1993. “Effects of Inflation on Economic Growth: Lessons from Latin
America.” European Economic Review, 36(April): 417-425.
de Gregorio, José and Pablo E. Guidotti. 1995. “Financial Development and Economic
Growth.” World Development, 23(3): 433-448.
Grier, Kevin B. and Gordon Tullock. 1989. “An Empirical Analysis of Cross-National Economic
Growth: 1951-80.” Journal of Monetary Economics, 24(2): 259-276.
Grossman, Gene M. and Elhanan Helpman. 1991. “Trade, Knowledge Spillovers, and Growth.”
European Economic Review, 56(2-3): 517-526.
Gubert, Flore. 2000. “Migration et Gestion Collective des risqués, L‟example de la Region de
Kayes (Mali).” Unpublished PhD Dissertation. Carnot: University of Clermont-Ferrand.
Gubert, Flore. 2002. “Do Migrants insure those who stay Behind? Evidence from the Kayes
Area (Western Mali).” Oxford Development Studies, 30(3): 267-287.
Gubert, Flore, Thomas Lassourd and Sandrine Mesplé-Somps. 2010. “Do Remittances Affect
Poverty and Inequality? Evidence from Mali.” Institut de Recherche pour le
Développment Document de Travail DT/2010-08. Paris: Université Paris Dauphine.
Guillaumont, Patrick and Lisa Chauvet. 2001. “Aid and Performance: A Re-Assessment.”
Journal of Development Studies, 37(6): 66-92.
Gujarati, Damdar N. 1970a. “Use of Dummy Variables …. A Generalisation.” American
Statistician, 24(5): 18-21.
Gujarati, Damdar N. 1970b. “Use of Dummy Variables in Testing for Equality between Sets of
Coefficients in Two Linear Regressions: A Note.” American Statistician, 24(1): 50-52.
392
Stellenbosch University http://scholar.sun.ac.za
Gupta, Kanhaya L. 1984. “Financial Liberalisation and Economic Growth: Some Simulation
Results.” Journal of Economic Development, 9(2): 25-43.
Gupta, Poonam. 2005. “Macroeconomic Determinants of Remittances: Evidence from India.”
IMF Working Paper 05/224. Washington, D.C.: International Monetary Fund.
Gupta, Sanjeev, Catherine A. Pattillo, and Smita Wagh. 2009. “Effect of Remittances on
Poverty and Financial Development in Sub-Saharan Africa.” World Development, 37(1):
104-115.
Gupta, Sanjeev, Robert Powell, and Yongzheng Yang. 2006. Macroeconomic Challenges of
Scaling Up Aid to Africa: A Checklist for Practitioners. Washington, D.C.: International
Monetary Fund.
Gurley, John G. and Edward S. Shaw. 1955. “Financial Aspects of Economic Development.”
American Economic Review, 45(4): 515-538.
Gurley, John G. and Edward S. Shaw. 1960. Money in a Theory of Finance. Washington, D.C.:
Brookings Institution.
Gustafsson, Björn and Negatu Makonnen. 1993. “Poverty and Remittances in Lesotho.”
Journal of African Economies, 2(1): 49-73.
Gyimah-Brempong, Kwabena and Elizabeth Asiedu. 2009. “Remittances and Poverty in
Ghana.” A Paper Prepared for Presentation at the 4th African Economic Conference,
November 11-13, 2009. Addis Ababa.
de Haan, Harjaan. 2000. “Migrants, Livelihoods, and Rights: The Relevance of Migration in
Development Policies.” DFID Social Development Working Paper 4. London:
Department for International Development.
de Haas, Hein. 2003. “Migration and Development in Southern Morocco. The Disparate SocioEconomic Impacts in the Todgha Oasis Valley.” Unpublished PhD Thesis. Nijmegen:
University of Nijmegen.
de Haas, Hein. 2007. “Remittances, Migration and Social Development: A Conceptual Review
of the Literature.” Social Policy and Development Programme Paper 34. Geneva:
United Nations Research Institute for Social Development.
Habibullah, Muzafa S. and Yoke-Kee Eng. 2006. “Does Financial Development cause
Economic Growth? A Panel Data Dynamic Analysis for the Asian Developing
Countries.” Journal of the Asia Pacific Economy, 11(4): 377-393.
Hadjimichael, Michael T. 1995. Sub-Saharan Africa: Growth, Savings, and Investment, 198693. Washington, D.C.: International Monetary Fund.
393
Stellenbosch University http://scholar.sun.ac.za
Hadjimichael, Michael T., Daneshwar Ghura Roger Nord, Martin Mühleisen and E. Murat Ucer.
1998. “Effects of Macroeconomic Stability on Growth, Savings, and Investment in SubSaharan Africa: An Empirical Investigation.” IMF Working Paper 97/98. Washington,
D.C.: International Monetary Fund.
Hadri, Kaddour. 2000. “Testing for Stationarity in Heterogeneous Panel Data.” Econometrics
Journal, 3(2): 148-161.
Hagen-Zanker, Jessica and Melissa Siegel. 2007. “The Determinants of Remittances: A
Review of the Literature.” MUGSG Working Paper 003. Maastricht: Maastricht
University Graduate School of Governance.
Hall, Robert E. and Charles I. Jones. 1999. “Why Do Some Countries Produce So Much Output
per Worker than Others?” Quarterly Journal of Economics, 114(1): 83-116.
Hansen, Henrik and Finn Tarp. 2000. “Aid Effectiveness Disputed.” Journal of International
Development, 12(3): 375-398.
Hansen, Henrik and Finn Tarp. 2001. “Aid and Growth Regressions.” Journal of Development
Economics, 46(4): 547-570.
Hansen, Lars P. 1982. “Large Sample Properties of Generalized Method of Moments
Estimators.” Econometrica, 50(3): 1029-1054.
Hanson, Gordon H. 2007. “Emigration, Remittances and Labour Force Participation in Mexico.”
Inter-American Development Bank INTAL-ITD Working Paper 28. Washington, DC:
Institute for the Integration of Latin America and the Caribbean / Integration, Trade and
Hemispheric Issues.
Hanson, Gordon H. and Christopher Woodruff. 2003. “Emigration and Educational Attainment
in Mexico.” Mimeo. San Diego: University of California.
Harrison, Ann E. 1996. “Openness and Growth: A Time Series Cross-Country Analysis for
Developing Countries.” Journal of Development Economics, 59(1): 125-154.
Harrison, Anne. 2003. “Working Abroad – the Benefits Flowing from Nationals Working in Other
Countries.” Roundtable on Sustainable Development. Paris: OECD.
Harrod, Roy E. 1948. Towards a Dynamic Economics: Some Recent Developments of
Economic Theory and their Applications to Policy. London: Macmillan.
Hartmann, Philipp, Florian Heider, Elias Papaioannou and Mario Lo Duca. 2007. “The Role of
Financial Markets and Innovation in Productivity and Growth in Europe.” European
Central Bank Occasional Paper 72/September 2007. Frankfurt am Main: European
Central Bank.
394
Stellenbosch University http://scholar.sun.ac.za
Hasanov, Fakhri. 2010. “Relationship between Inflation and Economic Growth in Azerbaijani
Economy: Is there any Threshold Effect?” Asia Journal of Business and Management
Sciences, 1(1); 1-11.
Hausman, Jerry A. 1978. “Specification Tests in Econometrics.” Econometrica, 46(6): 12511271.
Hayashi, Fumio. 2000. Econometrics. Princeton, New Jersey: Princeton University Press.
Hayes, Geoffrey. 1993. “Mirab Processes and Development on Small Pacific Islands: A Case
Study from the Southern Massim, Papua New Guinea.” Pacific Viewpoint, 34(1): 153178.
Heller, Peter S. 1975. “A Model of Public Fiscal Behaviour in Developing Countries: Aid,
Investment and Taxation.” American Economic Review, 65(3): 429-445.
Hermes, Niels. 1994. “Financial Development and Economic Growth: A Survey of the
Literature.” International Journal of Development Banking, 12(1): 3-22.
Hermes, Niels and Robert Lensink. 2001. “Capital Flight and the Uncertainty of Government
Policies.” Economic Letters, 71(3): 377-381.
Hicks, John. 1969. A Theory of Economic History. Oxford: Clarendon Press.
Higgins, Matthew L., Alketa Hysenbegasi and Susan Pozo. 2004. “Exchange Rate Uncertainty
and Workers‟ Remittances.” Applied Financial Economics, 14(6): 403-411.
Hildebrandt, Nicole and David J. McKenzie. 2005. “The Effects of Migration on Child Health in
Mexico.” Economia, 6(1): 257-289.
Hilferding, Rudolf. 1910. Das Finanzkapital: Eine Studie über die jüngste Entwick-lung des
Kapitalismus. Vienna: Wiener Volksbuchhandlung. [Finance Capital: A Study of the
Latest Phase of Capitalist Development, Morris Watnick and Sam Gordon (translators).
Tom Bottomore, (ed.). London: Routledge and Kegan Paul, 1981].
Hlouskova, Jaroslava and Martin Wagner. 2006. “The Performance of Panel Unit Root and
Stationarity Tests: Results from a Large Scale Simulation Study.” Econometric Reviews,
25(1): 85-116.
Holtz-Eakin, Douglas, Whitney K. Newey and Harvey S. Rosen. 1988. “Estimating Vector
Autoregressions with Panel Data.” Econometrica, 56(6): 1371-1395.
Hsiao, Frank S.T. and Mei-Chu W. Hsiao. 2006. “FDI, Exports, and Growth in East and
Southeast Asia: Evidence from Time Series and Panel Data Causality Analyses.” A
Paper Presented at 2006 International Conference on Korea and the World Economy V,
July 7-8, 2006. Seoul: Korea University.
395
Stellenbosch University http://scholar.sun.ac.za
Hsieh, Chang-Tai and Peter J. Klenow. 2009. “Misallocation and Manufacturing TFP in China
and India.” Quarterly Journal of Economics, 124(4): 1403-1448.
Hurlin, Christophe. 2004. “Testing Granger Causality in Heterogeneous Panel Data Models
with Fixed Coefficients.” Mimeo. Orléans: University of Orléans.
Hurlin, Christophe and Baptiste Venet. 2001. “Granger Causality Tests in Panel Data Models
with Fixed Coefficients.” Working Paper Eurisco 2001-09. Université Paris IX Dauphine.
Ilahi, Nadeem and Saqib Jafarey. 1999. “Guest Worker Migration, Remittances and the
Extended Family: Evidence from Pakistan.” Journal of Development Economics, 58(2):
485-512.
Im, Kyung S., M. Hashem Pesaran and Yoncheol Shin. 20003. “Testing for Unit Roots in
Heterogeneous Panels.” Journal of Econometrics, 115(11): 53-74.
IMF (International Monetary Fund). 1993. Balance of Payments Manual, 5th Edition.
Washington, D.C.: International Monetary Fund.
IMF (International Monetary Fund). 1995. Balance of Payments Compilation Guide.
Washington, D.C.: International Monetary Fund.
IMF (International Monetary Fund). 2005. World Economic Outlook, April 2005: Globalisation
and External Imbalances, World Economic and Financial Surveys. Washington, D.C.:
International Monetary Fund.
IMF (International Monetary Fund). 2008a. World Economic and Financial Surveys – Regional
Economic Outlook: Sub-Saharan Africa, April 2008. Washington, D.C.: International
Monetary Fund.
IMF (International Monetary Fund). 2008b. Global Financial Stability Report 2008, Autumn.
Washington, D.C.: International Monetary Fund.
IMF (International Monetary Fund). 2009a. Balance of Payments and International Investment
Position Manual, 6th Edition. Washington, D.C.: International Monetary Fund.
IMF (International Monetary Fund). 2009b. International Transactions in Remittances: Guide for
Compliers and Users. Washington, D.C.: International Monetary Fund.
IMF (International Monetary Fund). 2011a. Balance of Payments Statistics Yearbook.
Washington, D.C.: International Monetary Fund.
IMF (International Monetary Fund). 2011b. World Economic and Financial Surveys – Regional
Economic Outlook: Sub-Saharan Africa Recovery and New Risks, April 2011.
Washington, D.C.: International Monetary Fund.
396
Stellenbosch University http://scholar.sun.ac.za
IOM (International Organisation for Migration). 2010. World Migration Report 2010: The Future
of Migration – Building Capacities for Change. Geneva: International Organisation for
Migration.
Irwin, Douglas A. and Marko Trevio. 2002. “Does Trade raise Income? Evidence from the
Twentieth Century.” Journal of International Economics, 58(1): 1-18.
Islam, Nazrul. 1995. “Growth Empirics: A Panel Data Approach.” Quarterly Journal of
Economics, 110(4): 1127-1170.
Itzigsohn, Jose. 1995. “Migrant Remittances, Labour Markets, and Household Strategies: A
Comparative Analysis of Low-Income Household Strategies in the Caribbean Basin.”
Social Forces, 74(2): 633-655.
Jackman, Robert W. 1982. “Dependence on Foreign Investment and Economic Growth in the
Third World.” World Politics, 34(2): 175-196.
Jadhav, Narendra. 2003. “Maximizing Developmental Benefits of Migrant Remittances: The
Indian Experience.” Paper Presented at the Joint Conference of DFID-World Bank,
October 9-10, 2003. London: Department for International Development.
James, Kerry E. 1991. “Migration and Remittances: A Tongan Village Perspective.” Pacific
Viewpoint, 32(1): 1-23.
James, Kerry E. 1993. “Cash and Kin: Aspects of Migration and Remittances from the
Perspective of a Fishing Village in Vava‟u, Tonga.” In A World Perspective on Migration:
Australia, New Zealand and the USA, 359-374. Grant McCall and John Connell, (eds.).
Sydney: Centre for South Pacific Studies, University of New South Wales.
Jayaraman, TK, Chee-Keong Chong and Ronald Kumar. 2009. “Role of Remittances in
Economic Growth in Pacific Island Countries: A Study of Samoa.” Perspectives on
Global Development and Technology, 8(2009): 611-627.
Johnson, George E. and W.E. Whitelaw. 1974. “Urban-Rural Income Transfers in Kenya: An
Estimated Remittances Function.” Economic Development and Cultural Change, 22(3):
473-479.
Jongwanich, Juthathip. 2007. “Workers‟ Remittances, Economic Growth and Poverty in
Developing Asia and Pacific Countries.” UNESCAP Working Paper 07/01. Bangkok:
United Nations Economic and Social Commission for Asia and the Pacific.
Kagochi, John M, Ellene Kebede and David Summers. 2010. “Are Remittances a Source of
Development Capital? The Case of Sub-Saharan Africa.” International Research
Journal of Finance and Economics, 47(2010): 114-122.
397
Stellenbosch University http://scholar.sun.ac.za
Kalaitzidakis, Pantelis, Theofanis P. Mamuneas, Andreas Savvides and Thanasis Stengos.
2001. “Measures of Human Capital and Nonlinearities in Economic Growth.” Journal of
Economic Growth, 6(2001): 229-254.
Kalim, Rukhsana and Muhammad Shahbaz. 2009. “Remittances and Poverty Nexus: Evidence
from Pakistan.” International Research Journal of Finance and Economics, 29(2009):
46-59.
Kaminsky, Graciela L. and Sérgio L. Schmukler. 2003. “Short-Run Pain, Long-Run Gain: The
Effects of Financial Liberalisation.” NBER Working Paper 9787. Cambridge, MA:
National Bureau of Economic Research.
Kanbur, Ravi. 2000. “Aid, Conditionality and Debt in Africa,” In Foreign Aid and Development:
Lessons Learnt and Directions for the Future, 409-422. Finn Tarp, (ed.). New York:
Routledge.
Kao, Chihwa. 1999. “Spurious Regression and Residual-Based Tests for Cointegration in Panel
Data.” Journal of Econometrics, 90(1): 1-44.
Kapur, Basant K. 1983. “Optimal Financial and Foreign-Exchange Rate Liberalisation of Less
Developed Economies.” Quarterly Journal of Economics, 98(1): 41-62.
Kapur, Devesh. 2004. “Remittances: The New Development Mantra?” G-24 Discussion Paper
29. New York and Geneva: United Nations.
Kar, Muhsin and Eric J. Pentecost. 2000. “Financial Development and Economic Growth in
Turkey: Further Evidence on the Causality Issue.” Economic Research Paper 00/27.
Loughborough: Department of Economics, Loughborough University.
Karagöz, Kadir. 2009. “Workers‟ Remittances and Economic Growth: Evidence from Turkey.”
Journal of Yasar University, 4(13). 1891-1908.
Karras, Georgios. 2006. “Foreign Aid and Long-Run Economic Growth: Empirical Evidence for
a Panel of Developing Countries.” Journal of International Development, 18(1): 15-28.
Katseli, Louka T. and Nicholas P. Glytsos. 1986. “Theoretical and Empirical Determinants of
International Labour Mobility: A Greek-German Perspective.” CEPR Working Paper
148. London: Centre for Economic Policy Research.
Kaufmann, Daniel, Aart Kraay and Massimo Mastruzzi. 2003. “Governance Matters III:
Governance Indicators for 1996-2002.” World Bank Policy Research Working Paper
3106. Washington, D.C.: World Bank.
Kaufmann, Daniel, Aart Kraay and Massimo Mastruzzi. 2008. “Governance Matters VII:
Aggregated and Individual Governance Indicators for 1996-2007.” World Bank Policy
Research Working Paper 4654. Washington, D.C.: World Bank.
398
Stellenbosch University http://scholar.sun.ac.za
Kelly, Trish 1997. “Public Expenditures and Growth.” Journal of Development Studies, 34(1):
60-84.
Ketkar, Suhas and Dilip Ratha. 2001. “Development Finance during a Crisis: Securitisation of
Future Receivables.” World Bank Policy Research Working Paper 2582. Washington,
D.C.: World Bank.
Ketkar, Suhas and Dilip Ratha. 2004. “Recent Advances in Future-Flow Securitisation,” The
Financier, 11/12(2004-2005). [Online] Available: http://www.the-financier.com. Posted:
(unspecified). Accessed: May 10, 2010.
Keynes, John M. 1936. The General Theory of Employment, Interest and Money. London:
Macmillan.
Khatkhate, Deena R. 1988. “Assessing the Impact of Interest Rates in Less Developed
Countries.” World Development, 16(5): 577-588.
Kholdy, Shady. 1995. “Causality between Foreign Investment and Spillover Efficiency.” Applied
Economics, 27(8): 745-749.
Kim, D-H. 2011. “Trade, Growth and Income.” Journal of International Trade and Economic
Development, 20(5): 677-709.
Kim, Yoonbai. 2000. “Causes of Capital Flows in Developing Countries.” Journal of
International Money and Finance, 19(2): 235-253.
Kindleberger, Charles P. 1965. Europe‟s Postwar Growth: The Role of Labour Supply. New
York: Oxford University Press.
King, Robert G. and Ross Levine. 1993a. “Financial Intermediation and Economic
Development.” In Capital Markets and Financial Intermediation, 156-189. Colin Mayer
and Xavier Vives, (eds.). Cambridge, MA: Centre for Economic Policy Research.
King, Robert G. and Ross Levine. 1993b. “Finance and Growth: Schumpeter Might be Right.”
Quarterly Journal of Economics, 108(3): 277-284.
King, Robert C. and Ross Levine. 1993c. “Finance, Entrepreneurship, and Growth: Theory and
Evidence.” Journal of Monetary Economics, 32(3): 513-542.
King, Robert G. and Sergio T. Rebelo. 1990. “Public Policy and Economic Growth: Developing
Neoclassical Implications.” Journal of Political Economy, 98(5): 126-150.
Kiran, Burcu, Nilgün C. Yavus and Burak Güris. 2009. “Financial Development and Economic
Growth: A Panel Data Analysis of Emerging Countries.” International Research Journal
of Finance and Economics, 30(2009): 87-94.
399
Stellenbosch University http://scholar.sun.ac.za
Klenow, Peter J. and Andrés Rodríguez-Clare. 1997. “The Neoclassical Revival in Growth
Economics: Has it gone too Far?” In NBER Macroeconomics Annual: 1997 Vol. 12, 73114. Ben S. Bernanke and Julio Rotemberg, (eds.). Cambridge, MA: MIT Press.
Knack, Stephen. 2000. “Social Capital and the Quality of Governance: A Cross-Country
Empirical Analysis.” Policy Research Working Paper Series 2396. Washington, D.C.:
The World Bank.
Knack, Stephen. 2001. “Trust, Associational Life and Economic Performance in the OECD.” In
The Contribution of Human and Social Capital to Sustained Economic Growth and WellBeing: International Symposium Report, 172-202. John F. Helliwell, (ed.). Quebec:
Human Resources Development Canada and OECD.
Knack, Stephen and Philip Keefer. 1995. “Institutions and Economic Performance: CrossCountry Tests Using Alternative Institutional Measures.” Economics and Politics, 7(3):
207-227.
Knack, Stephen and Philip Keefer. 1997. “Does Social Capital have an Economic Payoff? A
Cross Country Investigation.” Quarterly Journal of Economics, 112(4): 1251-1288.
Knack, Stephen and Aminur Rahman. 2007. “Donor Fragmentation and Bureaucratic Quality in
Aid Recipients.” Journal of Development Economics, 83(1): 176-197.
Kneller, Richard A., C.W. Morgan and Sunti Kanchanahattakij. 2008. “Trade Liberalisation and
Economic Growth.” World Economy, 31(6): 701-719.
Knight, Malcolm and Julio A. Santaella. 1997. “Economic Determinants of IMF Financial
Arrangements.” Journal of Development Economics, 54(2): 405-436.
Knoop, Todd A. 1999. “Growth, Welfare, and the Size of Growth.” Journal of Economic Inquiry,
37(1): 103-119.
Knowles, Stephen and Clayton Weatherston. 2006. “Informal Institutions and Cross-Country
Income Differences.” CREDIT Research Paper 06/06. Nottingham: University of
Nottingham.
Kohsaka, Akira. 1984. “The High Interest Rate Policy under Financial Repression.” Developing
Economies, 22(4): 419-452.
Konica, Nevila and Randall K. Filer. 2005. “Albanian Emigration: Causes and Consequences.”
CERGE-EI Working Paper 181. Prague: The Centre for Economic Research and
Graduate Education – Economics Institute.
Koopmans, Tjalling C. 1965. “On the Concept of Optimal Economic Growth.” Pontificate
Academiae Scientiarum Scripta Varia, 28(1965): 225-300.
400
Stellenbosch University http://scholar.sun.ac.za
Kormendi, Roger C. and Philip G. Meguire. 1985. “Macroeconomic Determinants of Growth:
Cross-Country Evidence.” Journal of Monetary Economics, 16(1): 141-163.
Kozel, Valerie and Harold Alderman. 1990. “Factors Determining Work Participation and
Labour Supply Decisions in Pakistan‟s Urban Areas.” Pakistan Institute of Development
Review, 29(1): 1-18.
Krauss, Melvyn B. 1983. Development without Aid: Growth, Poverty and Government. New
York: McGraw-Hill.
Krueger, Alan B. and Mikael Lindahl. 2001. “Education for Growth: Why and For What?”
Journal of Economic Literature, 39(4): 1101-1136.
Krugman, Paul. (1994). “The Myth of Asia‟s Miracle.” Foreign Affairs, 73(6): 62-78.
Kumar, Chandra S. 2006. “Human Capital and Growth Empirics.” Journal of Developing Areas,
40(1): 153-179.
Laeven, Luc. 2003. “Does Financial Liberalisation Reduce Financing Constraints?” Financial
Management, 32(1): 5-34.
Landau, Daniel L. 1983. “Government Expenditure and Economic Growth: A Cross-Country
Study.” Southern Economics Journal. 49(3): 783-792.
Lartey, Emmanuel K.K. 2010. “Remittances, Institutions and Growth in Sub-Saharan Africa.”
Department of Economics Monograph. Fullerton: California State University.
Lartey, Emmanuel K.K., Federico S. Mandelman, and Pablo A. Acosta. 2008. “Remittances,
Exchange Rate Regimes, and the Dutch Disease: A Panel Data Analysis.” Federal
Reserve Bank of Atlanta Working Paper 2008-12. Fullerton: California State University.
Le, Thanh. 2008. “Trade, Remittances, Institutions, and Economic Growth.” MRG@UQ Paper
1833-4474. St. Lucia: University of Queensland Macroeconomic Research Group.
Lee, Ha Yan, Luca Antonio Ricci and Roberto Rigobon. 2004. “Once Again is Openness Good
for Growth.” Journal o Development Economics, 75(2): 299-328.
Lee, Jong-Wha. 1993. “International Trade, Distortions, and Long-Run Growth.” IMF Staff
Papers, 40(2): 299-328.
Lensink, Robert, Niels Hermes and Victor Murinde. 2000. “Capital Flight, Politics and Risk.”
Journal of International Money and Finance, 19(1): 73-92.
Leon-Ledesma, Miguel C. and Matloob Piracha. 2004. “International Migration and the Role of
Remittances in Eastern Europe.” International Migration, 42(4); 65-83.
de Leon-Manlagnit, Patrisha J.F. 2006. “The Role of Remittances in the Growth of Developing
Recipient Economies.” MA Project Work. Burnaby, BC: Department of Economics,
Simon Fraser University.
401
Stellenbosch University http://scholar.sun.ac.za
Levin, Andrew T., Chien-Fu Lin and Chia-Shang Chu. 2002. “Unit Root Tests in Panel Data:
Asymptotic and Finite Sample Propoerties.” Journal of Econometrics, 108(1): 1-24.
Levine, Ross. 1991. “Stock Markets, Growth, and Tax Policy.” The Journal of Finance, 46(4):
1445-1465.
Levine, Ross. 1993. “Financial Structures and Economic Development.” Revista de Análisis
Económico, 8(1): 113-129.
Levine, Ross. 1997. “Financial Development and Economic Growth: Views and Agenda.”
Journal of Economic Literature, 35(2): 688-726.
Levine, Ross, Norman Loayza and Thorsten Beck. 2000. “Financial Intermediation and
Economic Growth: Causality and Causes.” Journal of Monetary Economics, 46(1): 3177.
Levine, Ross and David Renelt. 1992. “A Sensitivity Analysis of Cross-Country Growth
Regressions.” American Economic Review, 82(4): 942-963.
Levine, Ross and Sara Zervos. 1998. “Stock Markets, Banks and Economic Growth.” American
Economic Review, 88(3): 536-558.
Levitt, Peggy. 2001. “Transnational Migration: Taking Stock and Future Directions.” Global
Networks, 1(3): 195-216.
Levy, Victor. 1988. “Aid and Growth in Sub-Saharan Africa: The Recent Experience.” European
Economic Review, 32(9): 1777-1795.
Lewis, Arthur W. 1954. “Economic Development with Unlimited Supplies of Labour.”
Manchester School of Economic and Social Studies, 22(May): 139-191.
Lewis, J.R. 1986. “International Labour Migration and Uneven Regional Development in Labour
Exporting Countries.” Tijdschrift voor Economische en Sociale Geografie, 77(1): 27-41.
Lianos, Theodore P. 1997. “Factors Determining Migrant Remittances: The case of Greece.”
International Migration Review, 31(1): 72-87.
Lipsey, Richard G., Douglas D. Purvis and Paul N. Courant. 1994. Microeconomics. New York:
Harper Collins College Publishers.
Lipton, Michael. 1980. “Migration from the Rural Areas of Poor Countries: The Impact on Rural
Productivity and Income Distribution.” World Development, 8(1): 1-24.
Loayza, Norman V., Klaus Schmidt-Hebbel and Luis Servén. 2000. “What drives Private Saving
across the World?” Review of Economics and Statistics, 82(2): 165-181.
Locke, John. 1695. Further Considerations Concerning Raising the Value of Money, 2nd
Edition. London: A. and J. Churchil.
402
Stellenbosch University http://scholar.sun.ac.za
Loomis, Terry. 1990. “Cook Island Remittances: Volumes, Determinants and Uses.” In
Migration and Development in the South Pacific, 61-81. John Connell, (ed.). Canberra:
National Centre for Development Studies, Australian National University.
López-Córdova, Ernesto. 2005. “Globalisation, Migration and Development: The Role of
Mexican Migrant Remittances.” Economia, 6(1): 217-256.
López-Córdova, Ernesto and Alexandra Olmedo. 2006. “International Remittances and
Development: Existing Evidence, Policies and Recommendations.” Intal-ITD Occasional
Paper 41. Buenos Aires: Institute for the Integration of Latin America and the
Caribbean, Inter-American Development Bank.
Lozano-Ascencio, Fernando. 1993. Bringing it Back Home: Remittances to Mexico from
Migrant Workers in the United States. San Deigo: Centre for US-Mexican Studies.
Lucas, Robert E.B. jr. 1988. “On Mechanics of Economic Growth.” Journal of Monetary
Economics, 22(1): 3-42.
Lucas, Robert E.B. jr. 1990. “Why Doesn‟t Capital Flow from Rich to Poor Countries?”
American Economic Review, 80(2): 92-96.
Lucas, Robert E.B. jr. 2005. International Migration and Economic Development, Expert Group
on Development Issues. Stockholm: Swedish Ministry of Foreign Affairs.
Lucas, Robert E.B. jr. and Oded Stark. 1985. “Motivations to Remit: Evidence from Botswana.”
Journal of Political Economy, 93(5): 901-918.
Lueth, Erik and Marta Ruiz-Arranz. 2007a. “Are Workers‟ Remittances a Hedge Against
Macroeconomic Shocks? The Case of Sri Lanka.” IMF Working Paper 07/22.
Washington, D.C.: International Monetary Fund.
Lueth, Erik and Marta Ruiz-Arranz. 2007b. “Macroeconomic Determinants of Workers‟
Remittances.” A Paper Presented at Policy Options and Challenges for Developing Asia
– Perspectives from the International Monetary Fund and Asia, 19-20 April, 2007.
Tokyo.
de Luna Martinez, J. 2005. “Workers‟ Remittances to Developing Countries: A Survey with
Central Banks on Selected Public Policy Issues.” World Bank Policy Research Working
Paper 3638. Washington, D.C.: World Bank.
Lynch, David. 1996. “Measuring Financial Sector Development: A Study of Selected AsiaPacific Countries.” The Developing Economies, XXXIV-1: 3-33.
Maddala, Gangadharrao S. 1971. “The Use of Variance Components Models in Pooling Cross
Section and Time Series Data.” Econometrica, 39(2): 341-358.
403
Stellenbosch University http://scholar.sun.ac.za
Maddala, Gangadharrao S. and Shaowen Wu. 1999. “A Comparative Study of Unit Root Tests
with Panel Data and a New Simple Test.” Oxford Bulletin of Economics and Statistics,
61(1, special issue): 631-652.
Maddala, Gangadharrao S., Shaowen Wu and P.C. Liu. 2000. “Do Panel Data Rescue
Purchasing Power Parity (PPP) Theory?” In Panel Data Econometrics: Future
Directions, 35-51. Krishnakumar, Jaya and Elazio Ronchetti, (eds.). Amsterdam: NorthHolland.
Mallik, Girijasankar and Anis Chowdhury. 2001. “Inflation and Economic Growth: Evidence from
South Asian Countries.” Asian Pacific Development Journal, 8(1): 123-135.
Mancini-Griffoli, Tommaso and Laurent L. Pauwels. 2006. “Did the Euro affect Trade? Answers
from End-of-Sample Instability Tests.” HEI Working Paper, Graduate Institute of
International Studies, Geneva, Economics Section.
Mankiw, N. Gregory, David Romer and David N. Weil. 1992. “A Contribution to the Empirics of
Economic Growth.” Quarterly Journal of Economics, 107(2): 407-437.
Manning, Richard and Koon-Lam Shea. 1989. “Perfectly Discriminatory Policy towards
International Capital Movements in a Dynamic World.” International Economic Review,
30(2): 329-348.
Maphosa, France. 2005. “The Impact of Remittances from Zimbabweans Working in South
Africa on Rural Livelihoods in the Southern Districts of Zimbabwe.” Forced Migration
Working Paper Series 14. Johannesburg: University of Witwatersrand.
Marshall, Monty G. and Keith Jaggers. 2011. Polity IV Project: Political Regime Characteristics
and Transitions, 1800-2009. Colorado: Societal-Systems Research Inc. and Colorado
State University.
Massey, Douglas S., Joaquín Arango, Graeme Hugo, Ali Kouaouci, Adela Pellegrino and J.
Edward Taylor. 1998. World in Motion; Understanding International Migration at the End
of the Millennium. Oxford: Clarendon Press.
Mathieson, Donald J. 1979. “Financial Reform and Capital Flows in a Developing Economy.”
International Monetary Fund Staff Papers, 26(3): 450-489.
Mathieu, Nicolas. 1998. Financial Sector Reform: A Review of World Bank Assistance.
Washington, D.C.: World Bank.
McCormick, Barry and Jackline Whaba. 2001. “Overseas Work Experience, Savings and
Entrepreneurship amongst Returnee Migrants in LDCs.” Scottish Journal of Political
Economy, 48(2, special issue): 164-178.
404
Stellenbosch University http://scholar.sun.ac.za
McKenzie, David J. and Hillel Rapoport. 2006. “Can Migration Reduce Educational Attainment?
Evidence from Mexico.” World Bank Policy Research Working Paper 3952. Washington,
D.C.: World Bank.
McKenzie, David J. and Hillel Rapoport. 2007. “Network Effects and the Dynamics of Migration
and Inequality: Theory and Evidence from Mexico.” Journal of Development Economics,
84(1): 1-24.
McKinnon, Ronald I. 1973. Money and Capital in Economic Development. Washington, D.C.:
Brookings Institution.
McKinnon, Ronald I. 1981. “Financial Repression and the Liberalisation Problem in LessDeveloped Countries.” In The World Economic Order: Past and Prospects, 365-386.
Seven Grassman and Erik Lundberg, (eds.). London: Macmillan.
McKinnon, Roland I. 1984. “Financial Liberalisation and the Debt Crisis in LDCs: The
International Misregulation of Commercial Banks.” CREG Memorandum 265. Standford:
Centre for Research in Economic Growth.
McKinnon, Roland I. 1991. “Monetary Stabilisation in LDCs.” In Liberalisation in the Process of
Economic Development, 319-336. Lawrence B. Krause and Kim Kilhwan, (eds.).
Berkley and Los Angeles: University of California Press.
Mehrhoff, Jens. 2009. A Solution to the Problem of Too Many Instruments in Dynamic Panel
Data GMM. Essen: Essen Fachbereich Wirtschaftswissenschaften.
de Mello, Luiz R. jr. 1999. “Foreign Direct Investment-led Growth: Evidence from Time Series
and Panel Data.” Oxford Economics Papers, 51(1999): 133-151.
Mencinger, Joże. 2003. “Does Foreign Direct Investment Always Enhance Economic Growth?”
Kyklos, 56(4): 491-508.
Mercan, Mehmet, Ismet Gocer, Sahin Bulut and Metin Dam. 2013. “The Effects of Openness
on Economic Growth for BRIC-T Countries: Panel Data Analysis.” Eurasian Journal of
Business Economics, 6(11): 1-14.
Merkle, Lucie and Klaus Zimmernann. 1992. “Savings, Remittances, and Return Migration.”
Economics Letters, 38(5): 77-81.
Miller, Merton H. 1998. “Financial Markets and Economic Growth.” Journal of Applied
Corporate Finance, 11(3): 8-14.
Migration Policy Institute. 2006. Remittances Profile: Sub-Saharan Africa. [Online] Available:
http://www.migrationpolicy.org/datahub. Posted: (unspecified). Accessed: February 09,
2010.
405
Stellenbosch University http://scholar.sun.ac.za
Mingat, Alain, Ramahatra Rakotomalala and Jee-Peng Tan (2002). Financing Education for All
by 2015: Simulations for 33 African Countries. Washington, D.C.: World Bank.
Miotti, Luis, El Mouhoub Mouhoub and Joel Oudinet. 2010. “Determinants and Uses of
Remittances to Southern and Eastern Mediterranean Countries: Insights from a New
Survey.” Development Studies Working Paper 288. Dauphine: Centro Studi Luca
D‟Agliano.
Moers, Luc. 1999. “How Important are Institutions for Growth in Transition Countries?”
Tinbergen Institute Discussion Paper 99-004/2. Amsterdam: Tinbergen Institute and
University of Amsterdam.
Montiel, Peter J. 1995. “Financial Policies and Economic Growth: Theory, Evidence, and
Country-Specific Experience from Sub-Saharan Africa.” AERC Special Paper 18.
Nairobi: African Economic Research Consortium.
Moore, Alvon and Kevin Greenidge. 2008. “Determinants and Volatility of Remittances in the
Caribbean.” A Paper Presented at the 29th Annual Review Seminar, 28-31 July, 2008.
Bridgetown: Central Bank of Barbados.
Morauta, Louise. 1985. “Urban Movement and Rural Identity: A Papua New Guinea Example,”
Mobility and Identity in the Island, Pacific Viewpoint, 26(special issue): 221-241.
Moreira, Sadrina B. 2005. “Evaluating the Impact of Foreign Aid on Economic Growth: A CrossCountry.” Journal of Economic Development, 30(2): 25-49.
Morisset, Jacques. 1993. “Does Financial Liberalisation Really Improve Private Investment in
Developing Countries?” Journal of Development Economics, 40(1): 133-150.
Morton, Jason, Priniti Panday and Maria Kula. 2010. “Remittances, Poverty and Growth.”
International Journal of Arts and Sciences, 3(2): 390-399.
Muliaina, Tolu. 2001. “Remittances, the Special System and Development in Samoa.” In
Current Trends in South Pacific Migration, Asia Pacific Migration Research Network
Working Paper No. 7: 20-40. Vijay Naidu, Ellie Vasta, and Charles Hawksley, (eds.).
Wollongong: Centre for Asia Pacific Social Transformation Studies.
Mundaca, Gabriela B. 2005. “Can Remittances enhance Economic Growth: The Role of
Financial Markets Development.” Mimeo. Oslo: Department of Economics, University of
Oslo.
Mundaca, Gabriela B. 2009. “Remittances, Financial Market Development, and Economic
Growth: The Case of Latin American and the Caribbean.” Review of Development
Economics, 13(2): 288-303.
406
Stellenbosch University http://scholar.sun.ac.za
Naceur, Samy Ben and Samir Ghanzouani. 2007.”Stock Markets, Banks, and Economic
Growth: Empirical Evidence from the MENA Region.” Research in International
Business and Finance, 21(2): 297-315.
Nell, Kevin S. 2000. “Is Low Inflation a Precondition for Faster Growth? The Case of South
Africa.” Keynes College Department of Economics Working Paper 1466-0814. Kent:
University of Kent.
Nelson, Richard R. and Edmund S. Phelps. 1966. “Investment in Humans, Technological
Diffusion, and Economic Growth.” American Economic Review, 56(1-2): 69-75.
NEPAD (New Partnership for Africa‟s Development). 2001. The New Partnership for Africa‟s
Development. Abuja: NEPAD Secretariat.
Nguyen, Viet C. 2008. “Do Foreign Remittances Matter to Poverty and Inequality? Evidence
from Vietnam.” Economics Bulletin, 15(1): 1-11.
Nicholas, David A. 1974. “Some Principles of Inflationary Finance.” Journal of Political
Economy, 82(2): 423-430.
Niimi, Yoko and Ҫağlar Özden. 2006. “Migration and Remittances: Causes and Linkages,”
World Bank Policy Research Working Paper 4087. Washington, D.C.: World Bank.
North, Douglass C. 1981. Structure and Change and Economic History. New York: W.W.
Norton.
North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance.
Cambridge, MA: Cambridge University Press.
OAU (Organisation for African Unity). 1985. Africa‟s Priority Programme for Economic
Recovery: 1986-1990. Addis Ababa: Organisation for African Unity.
Obstfeld, Maurice. 1994. “Risk, Diversification, and Growth.” American Economic Review,
84(5): 1310-1329.
Ogaki, Masao, Jonathan D. Ostry and Carmen M. Reinhart. 1996. “Saving Behaviour in Lowand Middle-Income Developing Countries: A Comparison.” International Monetary Fund
Staff Papers, 43(1): 38-71.
Orozco, Manuel. 2002. “Attracting Remittances: Practices to Reduce Costs and Enable a
Money Transfer Environment.” Multilateral Investment Fund of the Inter-American
Development Bank Research Paper. Washington, D.C.: Inter-American Dialogue.
Orozco, Manuel. 2003. Changes in the Atmosphere: Increase in Remittances, Price Decline
and New Challenges. Washington, D.C.: Inter-American Dialogue.
Orozco, Manuel. 2004. The Remittance Marketplace: Prices, Policies, and Financial
Institutions. Washington, D.C.: Pew Hispanic Centre.
407
Stellenbosch University http://scholar.sun.ac.za
Orrenius, Pia M., Madeline Zavodny, Jesús Caňas and Roberto Coronado. 2010. “Do
Remittances boost Economic Development? Evidence from Mexican States.” Research
Department Working Paper 1007. Dallas: Federal Reserve Bank of Dallas.
Orzell, Sarah. 2013. “Influence of Remittances on Poverty in Developing Nations.” Unpublished
Master of Public Policy Thesis. Washington, D.C.: Georgetown University.
Osaki, Keiko. 2003. “Migrant Remittances in Thailand: Economic Necessity or Social Norm?”
Journal of Population Research, 20(2): 203-222.
Oshikoya, Temitope W. 1992. “Interest Rate Liberalisation, Savings, Investment and Growth:
The Case of Kenya.” Savings and Development, 16(3): 305-320.
Osili, Una O. 2007. “Remittances and Savings from International Migration: Theory and
Evidence Using a Matched Sample.” Journal of Development Economics, 83(2): 446465.
Owens, Edgar. 1987. The Future of Freedom in the Developing World. New York: Pergamon
Press.
Özden, Ҫağlar and Maurice Schiff. 2005. International Migration, Remittances and the Brain
Drain. New York: World Bank and Palgrave Macmillan.
Pagano, Marco. 1993. “Financial Markets and Growth: An Overview.” European Economic
Review, 37(2-3): 613-622.
Papanek, Gustav F. 1973. “Aid, Foreign Private Investment, Savings and Growth in Less
Developing Countries.” The Journal of Political Economy, 81(1): 121-130.
Parsons, Christopher R., Ronald Skeldon, Terrie L. Walmsey and L. Alan Winters. 2007.
“Quantifying International Migration: A Database of Bilateral Migration Stocks.” In
International Migration, Economic Development, and Policy, 17-58. Çaglar Özden and
Maurice Schiff, (eds.). Washington, D.C.: Palgrave Macmillan and World Bank.
Patrick, Hugh T. 1966. “Financial Development and Economic Growth in Underdeveloped
Countries,” Economic Development and Cultural Change, XIV(2): 174-189.
Pendleton, Wade, Jonathan Crush, Eugene K. Campbell, Thuso Green, Hamilton Simelane,
Daniel Tevera and Fion de Vletter. 2006. “Migration, Remittances and Development in
Southern Africa.” Migration Policy Series 44. Cape Town: The Southern African
Migration Project.
Penninx, Rinus. 1982. “A Critical Review of Theory and Practice: The Case of Turkey.”
International Migration Review, 16(4): 781-818.
Perotti, Roberto. 1996. “Growth, Income Distribution and Democracy: What the Data Say.”
Journal of Economic Growth, 1(2): 149-187.
408
Stellenbosch University http://scholar.sun.ac.za
Persson, Torsten, Gerard Roland and Guido Tabellini. 2000. “Comparative Politics and Public
Finance.” Journal of Political Economy, 108(6): 1121-1161.
Persson, Torsten and Guido Tabellini. 2003. The Economic Effects of Constitutions.
Cambridge, MA: MIT Press.
Persson, Torsten and Guido Tabellini. 2004. “Constitutional Rules and Fiscal Policy Outcome.”
American Economic Review, 94(1): 25-46.
Pesaran, Hashem M. and Ronald Smith. 1995. “Estimating Long-Run Relationships from
Dynamic Heterogeneous Panels.” Journal of Econometrics, 68(1): 79-113.
Pessar, Patricia and Sherri Grasmuck. 1991. Between Two Islands: Dominican International
Migration. Berkeley: University of California Press.
Phillips, Peter C.B. and Hyungsik R. Moon. 1999. “Linear Regression Limit Theory for NonStationary Panel Data.” Econometrica, 67(5): 1057-1111.
Phillips, Peter C.B. and Hyungsik R. Moon. 2000. “Linear Non-Stationary Panel Data Analysis:
An Overview of Some Recent Developments.” Econometric Reviews, 19(3): 263-286.
Pinto, Brian, Vladimir Drebentsov and Alexander Morozov. 2000. “Dismantling Russia‟s Nonpayment Systems: Creating Conditions for Growth.” World Bank Technical Report No.
340. Washington, D.C.: World Bank.
Pleitez-Chávez, Rafael A. 2004. “Remittances as a Strategy to Cope with Systemic Risk: Panel
Results from Rural Households in El Salvador.” PhD Thesis. Columbus: Ohio State
University.
Poirine, Bernard. 1997. “A Theory of Remittances as an Implicit Family Loan Arrangement.”
World Development, 25(4): 589-611.
Politzer, Malia. 2008. “China and Africa: Stronger Economic Ties Mean More Migration.”
Migration Information Source, August: 2008.
Ponce, Juan, Iliana Olivié and Mercedes Onofa. 2008. “Remittances for Development? A Case
Study of the Impact of Remittances on Human Development in Ecuador.” Mimeo. Quito:
Facultad Latinoamericano dé Ciencias Sociales (FLASCO).
Pond, Bob and Barbara McPake. 2006. “The Health Migration Crisis: The Role of Four
Organisation for Economic Cooperation and Development Countries.” The Lancet,
367(9,520): 1,448-1,455.
Ponsel, Dorrit. 2001. “How Do Households Work? Migration, the Household and Remittance
Behaviour in South Africa.” Social Dynamics, 27(1): 165-189.
Portes, Luis S.V. 2009. “Remittances, Poverty and Inequality.” Journal of Economic
Development, 34(1): 127-140.
409
Stellenbosch University http://scholar.sun.ac.za
Pritchett, Lant. 2001. “Where has all the Education gone?” World Bank Economic Review,
15(3): 367-391.
Rajan, Raghuram G. and Arvind Subramanian. 2005. “What undermines Aid‟s Impact on
Growth?” NBER Working Paper 11657. Cambridge, MA: National Bureau of Economic
Research.
Rajan, Raghuram G. and Arvind Subramanian. 2007. “Does Aid Affect Governance?” American
Economic Review, 97(2): 322-327.
Rajan, Raghuram G. and Luigi Zingales. 1998. “Financial Dependence and Growth.” American
Economic Review, 88(3): 559-586.
Ramirez, Miguel D. and Hari Sharma. 2008. “Remittances and Growth in Latin America: A
Panel Unit Root and Panel Co-integration Analysis.” Yale Economics Working Paper
51. New Haven: University of Yale.
Ramsey, Frank P. 1928. “A Mathematical Theory of Saving.” Economic Journal, 38(152): 543559.
Rao, B. Bhaskara and Gazi Hassan. 2009. “Are the Direct and Indirect Growth Effects of
Remittances Significant?” MPRA Paper 18641. Available: http:www.mpra.ub.unimuenchen.de/18641. Posted: 15 November, 2009. Accessed: 28 July, 2011.
Rapoport, Hillel and Frédérick Docquier. 2006. “The Economics of Migrants‟ Remittances.” In
Handbook on the Economics of Reciprocity, Giving and Altruism, Vol. 2. Gerard V. Kolm
and J. Mercier-Ythier, (eds.). Amsterdam: North-Holland.
Ratha, Dilip. 2003. “Workers‟ Remittances: An Important and Stable Source of Development
Finance.” In Global Development Finance, 157-175. World Bank, (ed.). Washington,
D.C.: World Bank.
Ratha, Dilip and Sanket Mohapatra. 2007. “Increasing the Macroeconomic Impact of
Remittances on Development.” A Paper Prepared for G8 Outreach Event on
Remittances, 28-30 November, 2007. Berlin.
Ratha, Dilip and William Shaw. 2007. “South-South Migration and Remittances.” World Bank
Development Prospects Group Working Paper 102. Washington, D.C.: World Bank.
Rebelo, Sergio T. 1991. “Long-Run Policy Analysis and Long-Run Growth.” Journal of Political
Economy, 99(3): 500-521.
Reichert, Joshua S. 1981. “The Migrant Syndrome: Seasonal US Labour Migration and Rural
Development in Central Mexico.” Human Organisation, 40(1): 56-66.
410
Stellenbosch University http://scholar.sun.ac.za
Reinke, Jens. 2007. “Remittances in the Balance of Payments Framework: Current Problems
and Forthcoming Improvements.” A Paper Presented at Seminar on Remittance
Statistics, 26 February – 2 March, 2007. Ljubljana: Centre of Excellence in Finance.
Rensel, Jan. 1993. “The Fiji Connection: Migrant Involvement in the Economy of Rotuma.”
Pacific Viewpoint, 34(2): 215-240.
Resnick, Danielle and Regina Birner. 2006. “Does Good Governance Contribute to Pro-Poor
Growth? A Review of the Evidence from Cross-Country Studies.” DSCD Discussion
Paper 30. Washington, D.C.: International Food Policy Research Institute.
Rioja, Felix and Neven Valev. 2004a. “Finance and the Saviour of Growth at Various Stages of
Economic Development.” Economic Inquiry, 42(1): 127-140.
Rioja, Felix and Neven Valev. 2004b. “Does One Size Fit All: A Re-examination of the Finance
and Growth Relationship.” Journal of Development Economics, 74(2): 429-447.
Robinson, Joan. 1981. “The Generalisation of the General Theory and Other Essays.” The
Economic Journal, 91(364): 1041-1043.
Robinson, Rudi. 2004. “Globalisation, Immigrants‟ Transnational Agency and Economic
Development in their Homelands.” FOCAL Policy Paper FPP-04-9. Ottawa: Canadian
Foundation for the Americas.
Rodriguez, Edgar R. and Susan Horton. 1995. “International Return Migration and Remittances
in the Philippines.” Department of Economics Working Paper 95-01. Toronto: University
of Toronto.
Rodrik, Dani. 2004. “Globalisation and Growth – Looking in the Wrong Places.” Journal of
Policy Modelling, 26(4): 513-517.
Rodrik, Dani and Francisco Rodríguez. 2001. “Trade Policy and Economic Growth: A Skeptic‟s
Guide to the Cross-National Evidence.” In NBER Macroeconomics Annual 2000, Vol.
15, 216-325. Ben S. Bernanke and Kenneth S. Rogoff (eds.). Massachusetts: MIT
Press.
Rodrik, Dani, Arvind Subramanian and Francisco Rodriguez. 2001. “Institutions Rule: The
Primacy of Institutions over Geography and Integration in Economic Development.”
Journal of Economic Growth, 9(5): 131-165.
Romer, Paul M. 1986. “Increasing Returns and Long-Run Growth.” Journal of Political
Economy, 94(5), 1002-1037.
Romer, Paul M. 1990. “Endogenous Technological Change.” Journal of Political Economy,
98(5, special issue): 71-102.
411
Stellenbosch University http://scholar.sun.ac.za
Roodman, David M. 2009a. “How to Do xtabond2: An Introduction to Difference and System
GMM in Stata.” Stata Journal, 9(1): 86-136.
Roodman, David M. 2009b. “A Note on the Theme of Too Many Instruments.” Oxford Bulletin of
Economics and Statistics, 71(1): 135-158.
Rothgeb, John M. 1984. “The Effects of Foreign Investment on Overall and Sectoral Growth in
Third World States.” Journal of Peace Research, 21(1): 5-15.
Roubini, Nouriel and Xavier X. Sala-í-Martin. 1992. “Financial Repression and Economic
Growth.” Journal of Development Economics, 39(1): 5-30.
Rousseau, Peter L. and Paul Wachtel. 2000. “Equity Markets and Growth: Cross-Country
Evidence on Timing and Outcomes, 1980-1995.” Journal of Business and Finance,
24(12): 1933-1957.
Rubenstein, Hymie. 1992. “Migration, Development and Remittances in Rural Mexico.”
International Migration, 30(2): 127-153.
Russell, Sharon S. 1986a. “Migrant Remittances and Development.” International Migration,
XXX(3-4, special issue): 267-287.
Russell, Sharon S. 1986a. “Remittances from International Migration: A Review in Perspective.”
World Development, 14(6): 677-696.
Russell, Sharon S., Karen Jacobsen and William D. Stanley. 1990. “International Migration and
Development in Sub-Saharan Africa.” World Bank Discussion Paper 101. Washington,
D.C.: World Bank.
Ruud, Paul A. 2000. Classical Econometrics. New York: Oxford University Press.
Saaed, Afaf A.J. 2007. “Inflation and Economic Growth in Kuwait: 1985-2005 Evidence from
Co-Integration and Error-Correction Models.” Applied Econometrics and International
Development, 7(1): 143-155.
Sachs, Jeffery D. 2005a. The End of Poverty: Economic Possibilities of Our Time. New York:
Penguin.
Sachs, Jeffery D. 2005b. Investing in Development: A Practical Plan to Achieve the Millennium
Development Goals - United Nations Millennium Project. London: Earthscan.
Sachs, Jeffery D., John W. McArthur, Guido Schmidt-Traub, Margaret Kruk, Chandrika,
Bhadur, Michael Faye and Gordon McCord. 2004. “Ending Africa‟s Poverty Trap.”
Brookings Papers on Economic Activity, 2004(1): 117-216.
Sachs, Jeffery D. and Andrew M. Warner. 1995. “Economic Reform and the Process of Global
Integration.” Brookings Papers on Economic Activity, 1995(1): 1-118.
412
Stellenbosch University http://scholar.sun.ac.za
Sachs, Jeffrey D. and Andrew M. Warner. 1997. “Sources of Slow Growth in African
Economies.” Journal of African Economies, 6(3): 335-376.
Sahn, David E., Paul A. Dorosh and Stephen D. Younger. 1994. “Economic Reforms in Africa:
A Foundation for Poverty Alleviation.” A Paper Prepared for the Special Programme of
Assistance Donors Meeting, September 1994. Washington, D.C.: World Bank.
Sala-í-Martin, Xavier X. 1997. “I Just Ran Two Million Regressions.” American Economic
Review, 87(2): 178-183.
Salinas, Gonzalo and Ataman Aksoy. 2006. “Growth Before and After Trade Liberalisation.”
World Bank Policy Research Paper 4062. Washington, D.C.: World Bank.
Saltz, Ira S. 1992. “The Negative Correlation between Foreign Direct Investment and Economic
Growth in the Third World: Theory and Evidence.” Rivista Internazionale di Scienze
Economiché e Commerciali, 39(7): 617-633.
Sander, Cerstin and Samuel M. Maimbo. 2003. “Migrant Labour Remittances in Africa:
Reducing Obstacles to Developmental Contributions.” Africa Region Working Paper
Series 64. Washington, D.C.: World Bank.
Sargan, Thomas J. 1958. “The Estimation of Economic Relationships Using Instrumental
Variables.” Econometrica, 26(3): 393-415.
Sayan, Serdar. 2006. “Business Cycles and Workers‟ Remittances: How do Migrant Workers
Respond to Cyclical Movements of GDP at Home?” IMF Working Paper 06/52.
Washington, D.C.: International Monetary Fund.
Schiopu, Ioana and Nikolaus Siegfried. 2006. “Determinants of Workers‟ Remittances:
Evidence from the European Neighbouring Region.” ECB Working Paper 688. Frankfurt
am Main: European Central Bank.
Schneider, Friedrich and Dominik Enste. 2000. “Shadow Economies: Size, Causes, and
Consequences.” Journal of Economic Literature, 38(1): 77-114.
Schneider, Patricia H. 2005. “International Trade, Economic Growth and Intellectual Property
Rights: A Panel Data Study of Developed and Developing Countries.” Journal of
Development Economics, 78(2): 529-547.
Schrooten, Mechthild. 2005. “Bringing Home the Money: What Determines Remittances to
Transition Countries?” Discussion Paper A-466. Tokyo: Institute of Economic Research.
Schumpeter, Joseph A. 1912. Theorie der Wirtschaftlichen Entwicklung. Leipzig: Duncker and
Humbolt. [The Theory of Economic Development: An Inquiry into Profits, Capital, Credit,
Interest, and the Business Cycle, Redvers Opie, (translator). Cambridge, Mass: Harvard
University Press, 1934].
413
Stellenbosch University http://scholar.sun.ac.za
Scott, Gwen G. 2003. “Situating Fijian Transmigrants: Towards Racialised Transnational Social
Spaces of the Undocumented.” International Journal of Population Geography, 9(2):
181-198.
Sen, Amartya. 1999. “Development as Freedom.” Journal of Econometrics, 50(1-2): 15-29.
Sena, Augusto M. and Raimundo E.S. Fontenele. 2004. “The New Endogenous Growth
Theory: An Investigation on Growth Policy for Developing Countries.” A Paper for the
XLII Congress of the Brazilian Society of Economics and Rural Sociology. Cuiaba-MT,
Dynamic
Sectoral
and
Regional
Development.
http://www.sober.org.br/palestra/12/070079.pdf.
Posted:
[Online]
Available:
unspecified.
Accessed:
February 15, 2012.
Serino, Moises N.V. and Doughan Kim. 2011. “How Do International Remittances Affect
Poverty in Developing Countries? A Quantile Regression Analysis,” Journal of
Economic Development, 36(4): 17-40.
Shafiq, Muhammad, Ihtisham Ul Haq, Alam Khan and Shaista Khan. 2012. “The Role of
Foreign Remittances and Economic Growth in Poverty Alleviation: Time Series
Evidence from Pakistan.” World Applied Sciences Journal, 19(3): 366-369.
Shahbaz, Muhammad and Naveed Aamir. 2009. “Determinants of Workers‟ Remittances:
Implications for Poor People of Pakistan.” European Journal of Scientific Research,
25(1): 130-144.
Shahbaz, Muhammad, Muhammad N. Qureshi and Naveed Aamir. 2007. “Remittances and
Financial Sector Performance under Two Alternative Approaches for Pakistan.”
International Research Journal of Finance and Economics, 12(2007): 133-146.
Shan, Jordan and Qi Jianhong. 2006. “Does Financial Development „Lead‟ to Economic
Growth? The Case of China.” Annals of Economics and Finance,” 7(1): 197-216.
Shankman, Paul. 1976. Migration and Underdevelopment: The Case of Western Samoa.
Boulder: Westview Press.
Shaw, Edward S. 1973. Financial Deepening in Economic Development. London: Oxford
University Press.
Shelburne, Robert C. and Jose Palacin. 2007. “Remittances in CIS: Their Economic
Implications and the New Estimation Procedure.” ECE Discussion Paper Series, 2007.5
(November). Geneva: Economic Commission for Europe.
Shera, Adela and Dietmar Meyer. 2013. “Remittances and their Impact on Economic Growth.”
Social and Management Sciences, 21(1): 3-19.
414
Stellenbosch University http://scholar.sun.ac.za
Shrestha, Min B. 2005. “Financial Liberalisation in Nepal.” Unpublished PhD Thesis.
Wollongong: School of Economics, University of Wollongong.
Shrestha, Min B. and Khorshed Chowdhury. 2006. “Financial Liberalisation Index for Nepal.”
International Journal of Applied Econometrics and Quantitative Studies, 3(1): 41-54.
Siddique, Abu, E.A. Selvanathan and Saroja Selvanathan. 2010. “Remittances and Economic
Growth: Empirical Evidence from Bangladesh, India and Sri Lanka.” UWA Discussion
Paper 10.27. Perth: The University of Western Australia.
Sidrauski, Miguel. 1967. “Rational Choice and Patterns of Growth in a Monetary Economy.”
American Economic Review, 534-544.
Singer, David A. 2010. “Migrant Remittances and Exchanges Rate Regimes in the Developing
World.” American Political Science Review, 104(2): 307-323.
Singh, Raju J., Markus Haacker, Kyung-Woo Lee and Maëlan Le Goff. 2010. “Determinants
and Macroeconomic Impact of Remittances in Sub-Saharan Africa.” Journal of African
Economies, 20(3): 312-340.
Singh, Sunny K. and K.S. Hari. 2011. “International Migrant Remittances and its
Macroeconomic Impact on India.” IIMA Working Paper 2011-01-06 (January).
Ahmadabad: Indian Institute of Management.
Skeldon, Roland. 2002. “Migration and Poverty.” Asia-Pacific Population Journal, 17(4): 67-82.
Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. London:
W. Strahan and T. Cadell.
Snowden, P. Nicholas. 1987. “Financial Market Liberalisation in LDCs: The Incidence of Risk
Allocation Effects of Interest Rate Increases.” Journal of Development Studies, 24(1):
83-93.
Solimano, Andrés. 2003. “Workers‟ Remittances to the Andean Region: Mechanisms, Costs
and Development Impact.” Mimeo. Helsinki: United Nations Economic Commission for
Latin America and the Caribbean.
Solow, Robert M. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly Journal
of Economics, 70(1): 65-94.
Solow, Robert M. 1957. “Technical Change and the Aggregate Production Function.” Review of
Economics and Statistics, 39(3): 312-320.
Spatafora, Nikola. 2005. “Workers‟ Remittances and Economic Development.” In World
Economic Outlook: Globalisation and External Imbalances, 69-84. J. Edward Taylor,
(ed.). Washington, D.C.: International Monetary Fund.
415
Stellenbosch University http://scholar.sun.ac.za
Squalli, Jay and Kenneth T. Wilson. 2011. “A New Measure of Trade Openness.” The World
Economy, 34(10): 1745-1770.
Stahl, Charles W. 1982. “Labour Emigration and Economic Development.” International
Migration Review, 16(4): 868-899.
Stahl, Charles W. and Fred Arnold. 1986. “Overseas Workers‟ Remittances in Asian
Development.” International Migration Review, 20(4): 899-925.
Stark, Oded. 1978. Economic-Demographic Interactions in Agricultural Development: The Case
of Rural-to-Urban Migration. Rome: Food and Agricultural Organisation.
Stark, Oded. 1991. The Migration of Labour. Oxford: Basil Blackwell.
Stark, Oded and David E. Bloom. 1985. “The New Economics of Labour Migration.” American
Economic Review, 75(2), 173-178.
Stark, Oded and David Levhari. 1982. “On Migration and Risks in LDCs.” Economic
Development and Cultural Change, 31(1): 191-196.
Stark, Oded and Robert E.B. Lucas. 1988. “Migration, Remittances, and the Family.” Economic
Development and Cultural Change, 36(2): 465-481.
Stark, Oded, J. Edward Taylor and Shlomo Yitzhaki. 1986. “Remittances and Inequality.”
Economic Journal, 96(383): 722-740.
Steiner, Viktor and Johannes Velling. 1994. “Re-Migration Behaviour and Expected Duration of
Stay of Guest Workers in Germany.” In The Economic Consequences of Immigration to
Germany, 101-119. Gunter Steinmann and Ralf E. Ulrich, (eds.). Heidelberg: PhysicaVerlag.
Stiglitz, Joseph E. 1994. “The Role of the State in Financial Markets.” In Proceedings of the
World Bank Annual Bank Conference of Development Economics 1993, 19-52. Michael
Bruno and Bruno Pleskovic, (eds.). Washington D.C.: World Bank.
Stokey, Nancy L. 1991. “Human Capital, Product Quality and Economic Growth,” Quarterly of
Economics, 106(2): 587-616.
Stokey, Nancy L. and Sergio T. Rebelo. 1995. “Growth Effects of Flat-Rate Taxes,” Journal of
Political Economy, 103(3): 519-550.
Suki, Lenora 2007. “Competition and Remittances in Latin America: Lower Prices and More
Efficient Markets.” IADB-OECD Working Paper 2/2007. Washington, D.C.: InterAmerican Development Bank and OECD.
Swamy, Gurushri. 1981. “International Migrant Workers‟ Remittances: Issues and Prospects.”
World Bank Staff Working Paper 481. Washington, D.C.: World Bank.
416
Stellenbosch University http://scholar.sun.ac.za
Swan, Trevor W. 1956. “Economic Growth and Capital Accumulation.” Economic Record,
32(2): 334-361.
Tauchen, George. 1986. “Statistical Properties of Generalised Method-of-Moments Estimators
of Structural Parameters obtained from Financial Market Data.” Journal of Business and
Economic Statistics, 4(4): 397-416.
Taylor, Elizabeth. 1984. “Egyptian Migration and Peasant Wives.” Merip Reports, 124: 3-10.
Taylor, J. Edward. 1992. “Remittances and Inequality Reconsidered: Direct, Indirect and
Intertemporal Effects.” Journal of Policy Modelling, 14(2): 187-208.
Taylor, J. Edward. 1996. “International Migration and National Development.” Population Index,
62(2): 181-212.
Taylor, J. Edward. 1999. “The New Economics of Labour Migration and the Role of
Remittances in the Migration Process.” International Migration, 37(1): 63-88.
Taylor, Lance. 1983. Structuralist Macroeconomics: Applicable Models for the Third World.
New York: Basic Books.
Temple, Jonathan R.W. 1999a. “A Positive Effect of Human Capital on Growth.” Economics
Letters, 65(1): 131-134.
Temple, Jonathan R.W. 1999b. “The New Growth Evidence.” Journal of Economic Literature,
37(1): 112-156.
Terry, Donald F. and Steven Wilson. 2005. Beyond Small Change: Making Migrant
Remittances Count. Washington, D.C.: Inter-American Development Bank.
Thirlwall, Anthony P. 2011. Economics of Development – Theory and Evidence, 9th Edition.
London: Palgrave Macmillan.
Tobin, James. 1965. “Money and Economic Growth.” Econometrica, 33(4): 671-684.
Todaro, Michael P. and Stephen C. Smith. 2002. Economic Development, 8th Edition. Harlow:
Pearson.
Tongamoa,
Taiamoni.
1987.
“Migration,
Remittances
and
Development:
A
Tongan
Perspective.” Unpublished Master of Arts Thesis. Sydney: University of Sydney.
Toxopeus, Helen S. and Robert Lensink. 2007. “Remittances and Financial Inclusion in
Development.” UNU-WIDER Research Paper 2007/49. Helsinki: United Nations
University and World Institute for Development Economics Research.
Ulaşan, Bülent. 2012. “Openness to International Trade and Economic Growth: A CrossCountry Empirical Investigation.” Discussion Paper 2012-15. Ankara: Central Bank of
Turkey.
417
Stellenbosch University http://scholar.sun.ac.za
Umaru, Aminu and Anono A. Zubairu. 2012. “The Effect of Inflation on the Growth and
Development of the Nigerian Economy: An Empirical Analysis.” International Journal of
Business and Social Science, 3(10): 183-191.
UN (United Nations). 2002. International Migration Report 2002. New York: Department of
Economic and Social Affairs, Population Division, United Nations.
UN (United Nations). 2009. Human Development Report 2009: Overcoming Barriers –Human
Mobility and Development. New York: United Nations Development Programme.
UN (United Nations). 2010. Human Development Report 2010: The Real Wealth of Nations –
Pathways to Human Development. New York: United Nations Development
Programme.
UN (United Nations). 2011. Impact of Remittances on Poverty in Developing Countries. New
York: United Nations Development Programme.
UNDP (United Nations Development Programme). 2003. Human Development Report. New
York: United Nations Development Programme.
Unger, Kurt. 2005. “Regional Economic Development and Mexican Out-Migration.” NBER
Working Paper 11,432. Cambridge, MA: National Bureau of Economic Research.
Vamvakidis, Athanasios. 1999. “Regional Trade Agreements or Broad Liberalisation: Which
Path Leads to Growth?” IMF Staff Papers, 46(1): 42-68.
Vargas-Silva, Carlos and Peng Huang. 2006. “Macroeconomic Determinants of Workers‟
Remittances: Host vs. Home Country‟s Economic Conditions.” Journal of International
Trade and Economic Development, 15(1): 81-99.
Vertovec, Steven. 1999. “Conceiving and Researching Transnationalism.” Ethnic and Racial
Studies, 22(2): 447-462.
Vo, Han X. 2004. “Host Country Income Effects of Foreign Direct Investment: An Analytical
Framework.” Journal of Economics and Economic Education Research, 5(3): 81-100.
Wacziarg, Romain and Karen H. Welch. 2008. “Trade Liberalisation and Growth: New
Evidence.” World Bank Economic Review, 22(2): 187-231.
Wahba, Sadek. 1991. “What Determines Workers‟ Remittances?” Finance and Development,
28(4): 41-44.
Wai, U. Tun. 1959. “The Relation between Inflation and Economic Development: A Statistical
Inductive Study.” IMF Staff Papers, 7(2): 302-317.
Walker, Adrian M. and Richard P.C. Brown. 1995. “From Consumption to Savings? Interpreting
Tongan and Western Samoan Sample Survey Data on Remittances.” Asian and Pacific
Migration Journal, 4(1): 89-115.
418
Stellenbosch University http://scholar.sun.ac.za
van de Walle, Nicolas and Timothy A. Johnston. 1996. Improving Aid to Africa. Washington,
D.C.: Overseas Development Council.
White, Halbert. 1980. “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a
Direct Test for Heteroskedasticity.” Econometrica, 48(4): 817-838.
White, Howard and Richard Black. 2004. “Millennium Development Goals: A Drop in the
Ocean?” In Targeting Development: Critical Perspectives on the Millennium
Development Goals, 1-24. Richard Black and Howard White, (eds.). London:
Routledge.
Wiest, Raymond E. 2004. “External Dependency and the Perpetuation of Temporary Migration
to the United States.” In Patterns of Undocumented Migration: Mexico and the United
States, 110-135. Richard C. Jones, (ed.). Totowa, New Jersey: Rowman and Allanheld.
van Wijnbergen, Sweder. 1982. “Stagflationary Effects of Monetary Stabilisation Policies: A
Quantitative Analysis of South Korea.” Journal of Development Economics, 10(2): 133169.
van Wijnbergen, Sweder. 1983a. “Interest Rate Management in LDCs.” Journal of Monetary
Economics, 12(3): 433-452.
van Wijnbergen, Sweder. 1983b. “Credit Policy, Inflation and Growth in a Financially
Repressed Economy.” Journal of Development Economics, 18(2-3): 541-554.
Williamson, John and Molly Mahar. 1998. “A Survey of Financial Liberalisation.” Essays in
International Finance, No. 211 (November). Princeton: Department of Economics,
Princeton University.
Wooldridge, Jeffrey M. 2002. Econometric Analysis of Cross-Section and Panel Data.
Massachusetts: MIT Press.
World Bank. 1989. World Development Report 1989, New York: Oxford University Press.
World Bank. 1994. Adjustment in Africa: Reforms, Results, and the Road Ahead. New York:
Oxford University Press
World Bank. 2002. “Education for Dynamic Economies: Action Plan to Accelerate Progress
towards Education for All.” A Document Prepared by World Bank Staff for the Spring
Development Committee Meeting, (April). Washington, D.C.: World Bank.
World Bank. 2003. Supporting Sound Policies with Adequate and Appropriate Financing:
Implementing the Monterrey Consensus at the Country Level. Washington, D.C.: World
Bank.
World Bank. 2005. Financial Sector Assessment: A Handbook. Washington, D.C.: World Bank.
419
Stellenbosch University http://scholar.sun.ac.za
World Bank. 2006a. Global Economic Prospects: Economic Implications and Migration.
Washington, D.C.: World Bank.
World Bank 2006b. The Development Impact of Workers‟ Remittances in Latin America
Volume 2: Detailed Findings Report No. 37026. Washington, D.C.: World Bank.
World Bank. 2007. Migration and Remittances: Eastern Europe and the Former Soviet Union.
Washington, D.C.: World Bank.
World Bank. 2011a. World Development Indicators, April. Washington, D.C.: World Bank.
World Bank. 2011b. Global Development Finance, April. Washington, D.C.: World Bank.
World Bank. 2011c. Migration and Remittances Factbook 2011. Washington, D.C.: World Bank.
Wurgler, Jeffrey. 2000. “Financial Markets and the Allocation of Capital.” Journal of Financial
Economics, 58(1-2): 187-214.
Yang, Dean. 2007. “Coping with Disasters: The Impact of Hurricanes on International Financial
Flows, 1970-2002.” Unpublished Manuscript. Ann Arbor: University of Michigan.
Yang, Dean and HwaJung Choi. 2007. “Are Remittances Insurance? Evidence from Rainfall
Shocks in the Philippines.” World Bank Economic Review, 21(2): 219-248.
Yanikkaya, Halit. 2003. “Trade Openness and Economic Growth: A Cross-Country Empirical
Investigation.” Journal of Development Economics, 72(1): 57-89.
Zedillo, Ernesto, Abdulatif Y. Al-Hammad, David Bryer, Mary Chinery-Hesse, Jacques Delors,
Rebeca Grynspan, Alexander Y. Livshits, Majid A. Osman, Robert Rubin, Manmohan
Singh and Masayoshi Son. 2001. “Recommendations of the High-Level Panel on
Financing for Development.” A Paper Presented at UN General Assembly, (A/55/1000,
June 22, 2001). New York: United Nations.
Ziesemer, Thomas H.W. 2008. “Worker Remittances, Migration, Accumulation and Growth in
Poor Developing Countries.” UNU-MERIT Working Paper 2008-063. Maastricht: United
Nations University-Maastricht Economic and Social Research and Training Centre in
Innovation and Technology.
Ziesemer, Thomas H.W. 2009. “Worker Remittances and Growth: The Physical and Human
Capital Channel.” Journal of Economics and Statistics, 22(6): 743-773.
Ziliak, James P. 1997. “Efficient Estimation with Panel Data when Instruments are
Predetermined: An Empirical Comparison of Moment-Condition Estimators.” Journal of
Business and Economic Statistics, 15(4): 419-431.
420