Ecomomic Development
Ecomomic Development
I. INTRODUCTION
The pioneering contributions of Goldsmith (1969), Mckinnon (1973) and
Shaw (1973) regarding the relationship between financial development and
economic growth has remained an important issue of debate in developing
economies. The theoretical argument for linking financial development to growth is
that a well-developed financial system performs several critical functions to enhance
the efficiency of intermediation by reducing information, transaction, and monitoring
costs. A modern financial system promotes investment by identifying and funding
good business opportunities, mobilises savings, monitors the performance of
managers, enables the trading, hedging, and diversification of risk, and facilitates the
exchange of goods and services. These functions result in a more efficient allocation
of resources, in a more rapid accumulation of physical and human capital, and in
faster technological progress, which in turn feed economic growth [Creane, et al.
(2004)].
Most of the literature has mainly focused on the role of macroeconomic
stability, inequality, income and wealth, institutional development, ethnic and religious
diversity and financial market imperfections.1 Among these factors the role of financial
Muhammad Arshad Khan is Associate Professor, Government Postgraduate College,
Muzaffarabad (Azad Kashmir). Abdul Qayyum is Associate Professor, Pakistan Institute of Development
Economics, Islamabad. Saeed Ahmed Sheikh is Former Director Finance and Planning, University of
Azad Jammu and Kashmir, Muzaffarabad (Azad Kashmir).
Authors’ Note: We are grateful to Mr Omer Siddique, Prof. Rafique Ahmed, and other
participants of the PSDE conference for useful comments and suggestions.
1
The terms “financial development” and “financial intermediation” are used interchangeably in
this paper. Financial development, however, should be thought of as a broader concept that also includes
financial innovations that occur outside the banking system. Because of the lack of data regarding non-
bank financial innovation in developing countries like Pakistan, the level of financial intermediation
effectively measures the degree of financial development by the banking system. For a comprehensive
survey of recent evidence, see Levine (1997).
820 Khan, Qayyum, and Sheikh
markets in the growth process has received considerable attention. In this framework,
financial development is considered by many economists to be of paramount
importance for output growth. Particularly, government restrictions on the banking
system such as, interest rate ceiling, high reserve requirements and directed credit
programmes hinder financial development and reduce output growth [Mckinnon
(1973) and Shaw (1973)]. The early contributions due to Mckinnon (1973) and Shaw
(1973) postulate that the government intervention in the pricing and allocation of
loanable funds impedes financial repression mainly depressing real interest rates.
Governments are faced with only limited options such as inflationary financing, thus
even further deteriorating the real interest rate. Mckinnon emphasises that the order
and appropriate sequencing of financial reforms in the financial sector would be much
more effective once price stabilisation has taken place. In fact, financial development is
not lonely a function of liberalising monetary instruments but “consistent
macroeconomic policy package comprising a range of policies, including temporary
financial market supervision in order to monitor credit worthiness of borrowers and to
avoid distortions such as moral hazards and adverse selection”.2
The endogenous growth literature stresses the influence of financial markets
on economic growth.3 Benhabib and Spiegel (2000) argue that a positive
relationship is expected between financial development and total factor productivity
growth and investment. However, their results are very sensitive to model
specification. Furthermore, Beck, et al. (2000) find that financial development has a
large and positive impact on total factor productivity, which feeds through to overall
GDP growth [Neusser and Kugler (1998)]. A number of theorists have emphasised
the role of financial development in better identifying investment opportunities,
reducing investment in liquid but unproductive assets, mobilising savings, boosting
technological innovation, and improving risk taking.4
The problem with the previous studies is that a positive relationship between
financial development and output growth can exist for different reasons. As output
increases the demand for financial services increases too, this in turn has a positive effect
on financial development. Robinson (1952) argues that “by and large, it seems to be the
case that where enterprise leads finance flows”. Kuznets (1955) states that financial
market begins to grow as the economy approaches the intermediate stage of the growth
process and develop once the economy becomes mature. Lucas (1988) states that “the
importance of financial matters is very badly overstressed” while Chandavarkar (1992)
notes that “none of the pioneers of development economics… Even list finance as a factor
of development” [Luintel and Khan (1999)]. Thus the demand for the particular types of
financial services generated by economic development.
2
See for example, Galbis (1993), Kapur (1992) and Hanson and Neal (1985).
3
For further detail, see among others Bencivenga, et al. (1995), Greenwood and Smith (1997) and
Obstfeld (1994).
4
See Levine (1997) for further detail.
Financial Development and Economic Growth 821
-5
RDR1
-10
-15
-20
1970 1975 1980 1985 1990 1995 2000 2004
Years
Table 1
Structure of the Financial Sector in 1990
(Shares in % and Amount in Billion Rupees)
Assets Advances Investment
Number Amount Share Amount Share Amount Share
Banks 24 425.6 61.5 218.5 48.7 111.3 89.0
State-owned 7 392.3 56.7 201.2 44.8 104.1 83.2
Private – – – – – – –
Foreign 17 33.4 4.8 17.3 3.9 7.3 5.8
NBFIs1 36 133.9 19.4 98.3 21.9 13.7 11.0
State-owned 13 124.3 18.0 94.7 21.1 13.3 10.6
Private 23 9.6 1.4 3.6 0.8 0.4 0.3
CDNS 1 131.9 19.1 131.9 29.4 – –
Equity Markets2 2 90.0 – – – – –
Total 63 691.5 100.0 448.7 100.0 125.1 100.0
Source: Pakistan: Financial Sector Assessment 1990-2000 (SBP), p.13.
1
NBFIs also include four specialised banks and HBFC.
2
Market capitalisation of KSE in lieu of assets, not added in total.
Financial Development and Economic Growth 823
Table 2
Indicators of Financial Deepening (in Percent)
1961– 1971– 1981–
Indicators 70 80 90 1990 2000 2001 2002 2003 2004
Financial Depth1 36.14 41.76 41.25 39.20 36.90 36.70 39.90 43.10 49.20
2
Financial Depth – 35.00 32.36 27.91 37.51 33.23 36.03 40.32 44.16
Currency/M2 – – 32.14 37.6 25.4 24.6 24.7 23.8 23.3
Currency/GDP – 13.53 13.29 14.7 9.4 9.0 9.9 10.3 10.6
Private Sector
Credit/Total Credit – – – 51.5 53.3 55.5 54.3 61.3 93.4
State-owned Bank
Assets/Total Assets – – – 92.2 66.6 64.1 70.5 70.1 71.0
Source: IFS CD-ROM and Pakistan: Financial Sector Assessment 1990-2000, 2001-2002 (Published by
SBP).
Note: 1Financial depth is measured as broad money (money + quasi money) divided by nominal GDP
lagged by one year. Broad money includes the sum of currency outside the banks plus demand,
time, savings and foreign currency deposits of residents other than the central government.
2
Financial depth is also measured as liquid liabilities minus currency in circulation, divided by
nominal GDP lagged by one year. Demetriades and Luintel (1996) argue that without deducting
currency in circulation, we are left with primarily a measure of monetisation, not financial depth
(p. 360).
Financial Development and Economic Growth 825
Table 2 shows that financial depth (i.e., M2/GDP) increased steadily. It should
be noted that a large ratio represents a more developed and efficient financial sector.
In 1990 the average monetary assets were around 39.20 percent of GDP, while it was
reached to 49.2 percent of the GDP in 2004. This ratio has recorded a gradual
growth, showing an improvement in the financial sector. An alternative measure of
financial depth, which is frequently used, is the ratio of bank deposit liabilities to
GDP. This ratio assesses the degree of monetisation in the economy. A steady
growth in this ratio over the period of study also indicate an improvement in the
financial sector. Both indicators of financial depth can be depicted in Figure 2.
0.60
0.55
0.50 RM2
0.45
0.40
1970 1975 1980 1985 1990 1995 2000 2004
Years
0.45
0.40
0.35
RBDL
0.30
0.25
0.20
1970 1975 1980 1985 1990 1995 2000 2004
Years
826 Khan, Qayyum, and Sheikh
Where γ is real output, F is a measure of financial depth, r is the real deposit rate, S
is the share of investment and u is an error term. To capture the effect of financial
sector reforms introduced by the government of Pakistan in the late 1980s, we have
introduced a dummy variable (D90).8 Except real deposit rate, all the variables are
expressed in logarithmic form.
The present study is based on annual data covering the period from 1971
through 2004. Financial depth (F) is calculated by taking the difference between total
liquid liabilities minus currency in circulation divided by one period lagged nominal
8
We introduced a dummy variable D90 assigning zero for 1971-1989 and one for 1990-2004.
Financial Development and Economic Growth 827
GDP.9 y is the logarithm of real GDP measured as a ratio of GDP to Consumer Price
Index (CPI 2000=100). S is the share of investment proxied by the gross fixed
capital formation to nominal GDP. The data on these variables has been taken from
IFS CD-ROM. Real deposit rate is calculated by taking the difference between the
nominal deposit rate and inflation rate. The variable inflation rate is computed as the
log-difference of CPI. The data on deposit rate is obtained from the various issues of
the State Bank of Pakistan’s Quarterly Bulletins and Annual Reports.
Where y is real GDP, t is time trend and x is a vector of explanatory variables (i.e.
F, r, S). Investigation of the presence of a long run relationship amongst the
variables of Equation (1) is tested by means of bounds testing procedure of
Pesaran, et al. (2001). The bounds testing procedure is based on the F-stat or Wald
9
The standard measure of F used in the literature is the ratio of broad money⎯usually M2⎯to the
level of nominal GDP [World Bank (1989)]. However, this ratio measures the extent of monetisation
rather than of financial depth. In developing countries, monetisation can be increasing without financial
development; therefore, it is not an entirely satisfactory indicator of financial depth. We, therefore, define
financial depth as a ratio of total bank deposit liabilities to one period lagged nominal GDP.
828 Khan, Qayyum, and Sheikh
statistics and is first stage of the ARDL cointegration method. Accordingly, a joint
significance test that implies no cointegration, ( H 0 : β 2 = β3 = β 4 = 0 ), should be
performed for Equation (2). The F-test used for this procedure has a non-standard
distribution. Thus, two sets of critical values are computed by Pesaran, et al. for a
given significance level. One set assumes that all variables are I (0) and other set
assumes that they are all I (1). If the computed F-statistic exceeds the upper critical
bounds value, then the H0 is rejected. If the F-statistic fall into the bounds then the
test becomes inconclusive. If the F-statistic lies below the lower critical bounds
value, it implies no cointegration.10
Once a long run relationship is established, then the long run and error
correction estimates of the ARDL model can be obtained from Equation (2). At the
second stage of the ARDL cointegration method, it is also possible to perform a
parameter stability test for the appropriately selected ARDL representation of the
error correction model. A general error correction representation of Equation (2) is
formulated as follows:
k k k k
∆y t = β0 + ∑ β1i ∆yt −i + ∑ β2i ∆Ft −i + ∑ β3i ∆rt −i + ∑ β4i ∆St −i + λECt −1 + ηt (3)
i =1 i =0 i =0 i =0
Where λ is the speed of adjustment parameter and EC is the residual that is obtained
from the estimated cointegration model of Equation (1).
The two-step ARDL cointegration procedure is implemented in estimation
of Equation (1) for Pakistan using annual data over the period 1971–2004. In the
first stage, the order of lags on the first-differenced variables for Equation (2) is
usually obtained from unrestricted vector autoregression (VAR) by mean of
Akaike Information Criterion (AIC).11 Given the limited number of observations,
we experimented up to 2 years on the first-difference of each variable and
computed F-statistics for the joint significance of lagged levels of variables in
Equation (2). The computed F-test statistic for each order of lags is presented in
Table 3.
Table 3
Statistics for Selecting Lag Order and the Existence of Long-run Relationship
Order of Lag AIC SBC χ2SC(1) F-statistics
1 73.2168 64.4223 0.3470E-4 27.8486*
2 71.1001 61.0622 0.10636 5.1943**
10
This similar to the Johansen and Juselius multivariate cointegration procedure, which has five
alternative cases for long run.
11
Bahmani-Oskooee and Bohl (2000) and Bahmani-Oskooee and Ng (2002) argued that the
results of this stage are sensitive to the order of VAR.
Financial Development and Economic Growth 829
Based on the minimum value of AIC, the lag length of order 2 is selected.
When 2 lags are imposed, there is strong evidence for cointegration because the
calculated F-statistic is 5.1943, which is greater than the critical value of the
upper level of the bound (i.e. 4.01) at the 5 percent level of significant. This
result gives an indication for the existence of a long run relationship among y, F,
r and S.12
Given the existence of a long run relationship, in the next step we used the
ARDL cointegration method to estimate the parameters of Equation (1) with
maximum order of lag set to 2 to minimise the loss of degrees of freedom. In search
of finding the optimal length of the level variables of the long run coefficients, lag
selection criteria such as the AIC is utilised. The long run results of Equation (1)
based on AIC are reported in panel A of Table 4 along with their appropriate ARDL
model. The diagnostic test results of Equation (1) for short run estimates are also
displayed in panel B of Table 4.
Table 4
ARDL Estimates
Dependent Variable yt
Regressor Coefficient t-values p-values
Panel A: The Long-run Results
Ft 3.3663 2.2558 0.035
rt 0.1792 3.5074 0.002
St 0.3550 0.3517 0.729
D90 0.4840 2.4429 0.024
INPT 14.9318 4.6622 0.000
Panel B: The Short-run Diagnostic Test Statistics
χ2SC(1) 0.3470E-4
χ2FF(1) 2.4203
χ 2
NO(2) 0.8787
χ 2
Het(1) 0.0047338
Note: ARDL (1, 2, 2, and 2) selected on the basis of AIC. The full table of the short run estimates are
available from the author. χ2SC, χ2FF χ2NO and χ2Het are Lagrange multiplier statistics for test of
residual correlation, functional from mis-specification, non-normal errors and heteroskedasticity,
respectively. These statistics are distributed as Chi-square values with degree of freedom in
parentheses. INPT is the intercept term.
12
At lag 2, the residuals are white noise as indicated by the Lagrange Multiplier test of serial
correlation. i.e. χ2SC(1).
830 Khan, Qayyum, and Sheikh
As can be seen from Table 4 that the estimates possessed expected signs
and apart from the share investment, all other coefficients are statistically
significant at the 5 percent level of significance. The results suggest that
financial depth and the real deposit rate are particular important factors
contributing to economic growth in Pakistan in the long run. The coefficient of
financial depth indicates that in the long run a 1 percent increase in financial
depth increases real output by 3.37 percent. While the coefficient of real deposit
rate also suggests that a 1 percent rise in real deposit rate will increases real
output by 0.18 percent in the long run. Although, the coefficient of the share of
investment is positive, but statistically insignificant. Finally, the financial
reforms exert positive and significant impact on real output over the period of
investigation as indicated by the coefficient of the dummy variable (D90).
The ECM output corresponding to the ARDL (1, 2, 2, and 2) is given in
Table 5.
Table 5
Error Correction Representation of ARDL Model
Dependent Variable: ∆yt
Regressor Coefficient t-value p-value
∆Ft 0.0956 1.3001 0.206
∆yt–1 0.1191 1.7090 0.101
∆rt 0.0072 4.8381 0.000
∆rt–1 –0.0041 –2.8802 0.008
∆St 0.3559 6.1964 0.000
∆St–1 0.1267 2.1010 0.047
∆D90 0.0273 1.3971 0.176
∆INPT 0.8419 3.0675 0.005
ECt–1 –0.0564 –2.3790 0.026
2 2
R 0.79 R adj 0.69
S.E. Regression 0.02 F-stat 9.5432
R.S.S 0.009 AIC 73.2168
Equation-LL 85.2168 DW-stat 1.9869
Note: ARDL (1, 2, 2, and 2) selected on the basis of AIC. R.S.S, LL, AIC and DW are respectively
residual sum of squares, log likelihood, Akaike’s Information Criteria and Durbin Watson stat.
The estimated lagged error correction term ( ECt −1 ) is negative and highly
significant. This result supporting the cointegration among the variables represented
by Equation (1). The feedback coefficient is –0.06, which suggests a slow adjustment
process. Nearly 6 percent of the disequilibria of the previous period’s shock adjust
back to the long run equilibrium in the current year. The results further suggest that
in the short run financial depth exerted positive impact on the economic growth.
However, in the short run, the coefficients on the changes in financial depth
( ∆Ft , ∆Ft −1 ) are hardly significant at the 20 percent and 10 percent level of
significance. Although, the short run response of real deposit rate is significant but
very small, suggesting that there is a need for further liberalisation of interest rate.
Furthermore, the changes in the share of investment exerted positive and significant
impact on changes in real output in the short run.
We also performed the CUSUM and CUSUMSQ stability test for estimated
error correction model. Figure 4 plots the CUSUM and CUSUMSQ.
10
–5
–10
–15
1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2004
1.5
Plot of Cumulative Sum of Squares of Recursive Residuals
1.0
0.5
0.0
–-0.5
1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2004
It can be seen from the Figure 4 that the plots of CUSUM and CUSUMSQ
statistics are well within the critical bounds implying that all the coefficients in the
error correction model are stable.
V. CONCLUDING REMARKS
This paper has examined the empirical relationship between financial
development and economic growth in Pakistan over the period 1971–2004, using
Autoregressive Distributed Lag (ARDL) approach. The results show that, in the long
run financial depth and real interest exerted positive impact on economic growth.
While the share of investment is although positively correlated to real income, but
remained insignificant. Furthermore, in the short run economic growth is positively
and significantly affected by changes in the share of investment. Moreover, changes
in real interest rate exerted positive (negative)13 impact on growth. However, the
response of real interest rate is very small in the short run. The feed back coefficient
is negative and significant, suggesting about 0.06 percent disequilibrium in the
previous period is corrected in the current year. We find a stable long run
relationship between economic growth and financial depth, as indicated by the
CUSUM and CUSUMSQ stability tests. Unlike Ireland (1994) and Demetriades and
Hussein (1996), our findings are consistent with the view that economic growth is an
outcome of the financial development.
Based on the above findings we can derive some important policy implication.
• If policy-makers want to promote growth, then attention should be focused
on long run policies, for example, the creation of modern financial
institutions, in the banking sector and the stock markets.
• The financial markets affect the cost of external finance to the firm and,
therefore, their effects should be materialise through facilitating the
investment process.
• Unless conditions for low-cost investment are created, long run growth is
impossible.
REFERENCES
Arestis, P., and Panicos Demetriades (1997) Financial Development and Economic
Growth: Assessing the Evidence. The Economic Journal 107, 783–799.
Benhabib, J., and M. M. Spiegel (2000) The Role of Financial Development in
Growth and Investment. Journal of Economic Growth 5, 341–360.
Beck, T., R. Levine, and N. Loyaza (2000) Finance and the Sources of Growth.
Journal of Financial Economics 58, 261–300.
13
In the current period the short run impact of real interest rate is positive, while next period the
impact of real interest rate is negative on economic growth.
Financial Development and Economic Growth 833
can use stock market variables for Pakistan to test for the hypothesis that
financial development leads to higher economic growth.
Finally, the empirical evidence on finance-growth nexus is still far from
being conclusive. Although evidence for robust association between financial
factors and growth has been increasing over time, but the direction of causality
has been subject to controversy. The authors in the introductory part of the paper
state that the previous studies are problematic in the sense that positive
relationship between financial development and economic growth could exist
for different reasons since an increase in output could lead to increase in demand
for financial services. However, it is not clear from the paper how they have
overcome this problem. They themselves have found positive relationship
between financial development variables and economic growth in Pakistan. In
this regard, it is interesting to note that in the case of Pakistan economic growth
was higher in the pre-reform period than in the 1990s.
Omer Siddique
Pakistan Institute of Development Economics,
Islamabad.
REFERENCES
King, Robert G., and Ross Levine (1993) Finance and Growth: Schumpeter
Might be Right. Quarterly Journal of Economics 108:3, 717–737.
Levine, Ross, and David Renelts (1992) The Sensitivity of the Economic
Growth Regressions. American Economic Review 82:4, 942–953.
Siddiqui, Rehana (2003) Energy and Economic Growth in Pakistan. The
Pakistan Development Review 43:2, 175–200.