Africa Economic Brief

2018 l VOLUME 9 l ISSUE 5

VICE PRESIDENCY FOR ECONOMIC GOVERNANCE
AND KOWLEDGE MANAGEMENT

Regional Financial Integration and Economic

Activity in Africa
Chuku Chuku1

1| Introduction
The historical account of financial integration and cooperation integrated? Proponents have often argued that financial
in Africa is somewhat chequered and dates back to the colonial integration is the key to Africa’s accelerated development and
era. The earliest record of efforts towards regionalization was structural transformation (see Aryeetey and Nissanke (2005).
with the establishment of the Southern African Customs Union Overall, it is expected that financial regionalization would enable
(SACU) in 1910. Recent trends toward regionalization have the participating regional economies to take advantage of the
now moved beyond trade and customs unions to economic, “systemic scale economies" that accrue to larger financial
financial, and monetary integration. To fix ideas, regional systems. These scale effects emanate from several angles,
financial integration refers to a market- or institutionally-driven including the expansion of the spectrum of opportunities for
process of broadening and deepening the financial financial intermediation; the creation of larger markets, which
interrelationships within a region; whereas, financial develop- makes it more cost-effective to improve financial infrastructure;
ment refers to the process by which financial institutions and the efficiency effects that arise from the increase in the number
markets increase in size and influence on the rest of the of financial sector participants, which promotes diversification
economy. The process of financial integration involves several and healthy competition, and thus, eventually results in lower
activities: the elimination of cross-country investment barriers; prices for services; and lastly, the increased capacity to
equitable treatment of foreign and domestic investors; withstand financial crisis (see Wakeman-Linn, and Wagh, 2008).
harmonization of national policies, laws, and institutions;
synchronization of operational structures, such as technology The problem, however, is that despite the highlighted benefits
and information systems; and very importantly, the of financial integration to participating economies—a fact that
convergence of prices, returns, and risk assessments. has been established for other regions using stylized
information and empirical analysis (see, for example, Agenor,
Our aim in this brief is twofold: first, to determine the degree and 2003; and Edison et al., 2002)—it is not obvious that this is
timing of financial integration in selected sub-Saharan African also the case for Africa, especially because of the peculiar
stock markets using an unobserved latent variable; and second, idiosyncrasies of the existing regional financial architecture,
to understand the effect of regional financial integration on and the fact that there are hardly any studies that examine
economic activity in Africa. Perhaps the pertinent question is: this relationship beyond a correlation and descriptive statistics
why should markets, especially those in Africa, be regionally based framework.

1 Chuku Chuku is a member of the Macroeconomic Modelling Team at the African Development Bank and a Lecturer at the Department of Economics, University
of Uyo, Nigeria. Email. chukuachuku@gmail.com. He thankfully acknowledges funding (with Prof. Akpan Ekpo) for this project provided by the African Economic
Research Consortium (AERC) with Grant No. RC 13537. He benefited from review comments by Lemma Senbet, Isaac Otchere, Robert Lensink, Kalu Ojah,
Peter Quartly, and participants at the Plenary Session of the June 2016 AERC Biannual Workshop and various collaborative workshops. He would also like to
thank Manuel Arellano and participants at the 2015 Econometric Society Conference in Zambia for useful comments on research paper.

Disclaimer: The findings of this Brief reflect the opinions of the authors and not those of the African Development Bank,its Board of Directors or the
countries they represent.

1
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

2| Evidence on financial integration for the integration parameter to move towards zero indicates
in Africa sensitivity of the individual market to the influence of the
regional market, which is interpreted as regional integration;
To provide evidence on financial integration, one could whereas, the tendency for the integration parameter to lean
consider different dimensions of price-based indicators such towards unity indicates sensitivity to the global market; and

values of β_(i,t), on the other hand, indicate that the domestic

as price convergence, price sensitivity, price co-movements, hence, can be interpreted as financial globalization. Negative
cycle synchronisation, or return correlations. We use Haldane
and Hall’s (1991) framework to measure the level of financial market diverges from both the regional and global markets.
integration in Africa, this method is based on a Kalman filter
estimation of a dynamic linear regression.2 We use two In Figure 1, we observe that stock markets in six countries:
alternative benchmarks: first, we assume that the South Ghana, Kenya, Namibia, Nigeria, South Africa, and Swaziland
African stock market is the dominant regional market in Africa; are more responsive to the regional stock market benchmark
and second, we construct a weighted regional average as an relative to the global (or US) benchmark in the period between
alternative measure.3 The dominant market in the rest of the 1995 and 2005. This conclusion is informed by the estimated
world is proxied by the capitalization of the US equity market. values of the time-varying parameters, which are contained
In Figures 1 and 2, we present the estimated time-varying within the range -0.5 to 0.5, a result we interpret as evidence
parameters of the model, which is a measure of the of financial convergence and, hence, integration during the
unobserved trajectory of financial integration for the case period. The values for the stock markets of Tanzania and
where we use a weighted regional index as the dominant Botswana are closer to unity; also, for Nigeria, after 2005, the
market (Figure 1) and the case where we use the South African values begin to lean toward unity. This pattern indicates that
stock market as the dominant regional market (Figure 2). these stock markets are increasingly becoming more sensitive
to changes in the global stock market relative to the regional
The interpretation of the time-varying parameter is somewhat market. The negative values obtained for Ghana, especially
intuitive. When there is convergence between a country’s during the early 1990’s, indicates that the Ghanaian stock
equity market and the regional benchmark, the value of the market diverged from the regional and global markets during
time-varying parameter (β_(i,t)) would approach zero that period. However, the more recent trend for Ghana
infinitestimally. Conversely, if country i’s market is rather indicates sensitivity toward the regional benchmark. The
converging to the global dominant market (here the US impact of the financial crisis of 2007-2010, which is highlighted
market), the time-varying parameter would approach one in the figure, does not seem to change the observed pattern
(unity). Therefore, with this model construction, the tendency of the integration process in any noteworthy way.

Figure 1 Sensitivity and convergence

of stock markets to a regional benchmark

2 See detailed specification of the model in Ekpo and Chuku (2017).

3 Data on average annual market value of outstanding shares for selected countries between 1990 and 2014 are retrieved from the Standard and Poor’s (S & P)
database. The selected stock markets include Botswana, Ghana, Kenya, Namibia, Nigeria, South Africa, Swaziland, Tanzania and Zambia. The criteria for
including countries in the sample was mainly driven by the availability of stock market data for a reasonable amount of time.

2
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

In Figure 2, we repeat the analysis, but this time assuming that has high variability; with a standard deviation of 36.56 and a
the South African stock market is the dominant regional mean value of 20.91. The between- and within-sample
market. We observe that only the Ghanaian and Namibian properties (i.e., the standard deviation of 20.52 and 30.62
stock markets are strongly sensitive to the South African respectively) suggest that there is significant variability in two
regional benchmark relative to the global benchmark. Although dimensions: the extent of financial regionalization between
it is not obvious why the Ghanaian and Namibian stock countries, and the progress over time towards financial
markets are more sensitive to the South African stock market, regionalization within countries.
one plausible explanation, at least for Ghana, is the effect of
the listing of Ashanti Gold stocks on the South African Stock In Table 1, we report the main results from the SGMM
Exchange.4 Overall, the results indicate that equity markets in regressions. In Models 2, 3, 4, and 6, the financial regionaliza-
frontier African economies seem to have been moving away tion variable is negatively and significantly related to growth.7
from the dominant regional benchmark toward the global The results generally show evidence that an increase in
benchmark since 2005. Bearing in mind that there is financial regionalization is associated with higher GDP per
considerable variation in the pattern of the sensitivity by country, capita (GDPpc) growth rates. However, because statistical
the main insight from this exercise is that African stock markets and economic significance does not always coincide, we
are increasingly becoming more sensitive to the global stock measure the economic and quantitative extent of the impact
market dynamics than they are to regional dynamics. of financial regionalization on per capita GDP growth by
multiplying the coefficient on financial regionalization by its
standard deviation of 36.56. We do this for the results in
3| Methodology and key findings Model 2 and obtain 0.88. This result implies that a one
standard deviation decline in financial regionalization will, on
We use parametric and non-parametric regression analysis average, will result in an increase in GDP growth by 0.88 of a
to examine the relationship between regional financial percentage point. Moreover, given that the coefficient on
integration and economic activity in sub-Saharan Africa.5 In financial development has a negative sign in all the six
particular, we use system generalized methods of moments regressions, the results on the financial development variable
(SGMM) and local-linear least squares nonparametric do not support the finance leading hypothesis of growth. To
regressions to uncover the relationship between financial ascertain the quantitative impact of financial development on
integration and economic activity in sub-Saharan Africa. An growth, we multiply the coefficient by the standard deviation
unbalanced panel dataset was collected for 17 sub-Saharan (i.e., -0.025 × 25.13), and find that a one standard deviation
African countries from 1980 to 2011.6 The descriptive increase in the level of financial development leads to 0.62 of
statistics show that our measure of financial regionalization a percentage point decline in the growth rate.

Figure 2 Sensitivity and convergence of stock

markets to the South African benchmark

4 We thank an anonymous reviewer at the Journal of African Economies for pointing us to this potential explanation for the Ghana case.
5 See detailed specifications in Ekpo and Chuku (2017).
6 The list of countries include Benin, Botswana, Burkina Faso, Cameroon, Cote d’Ivoire, Ethiopia, Gabon, The Gambia, Ghana, Kenya, Malawi, Nigeria, Senegal,

Sierra Leone, South Africa, Togo, and Zambia.

7 Most of the discussion on the results are based on the fully specified Model 2.

3
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

As expected, by introducing interaction terms, the results In Figure 3, we depict the conditional mean surface plot of
provide more insight. From Model 3 in Table 1, the coefficient economic growth along with financial integration and financial
of the interaction term between financial regionalization and development, holding all other variables fixed at their
financial development is positive and statistically significant respective means. The plot indicates that the relationship
(0.031). This result can be interpreted from two perspectives between growth and the financial variables in Africa are clearly
depending on which variable we hold constant as the non-linear. This is particularly so in the direction of financial
moderating variable. If we moderate the interaction based on development. Further, the partial regression in the direction of
the financial development variable, then the result implies that each of the regressors does not appear to change very much
the growth-enhancing effects of financial regionalization are as the other predictor varies. What we see, however, is that
greater in countries with higher levels of financial develop- improving financial integration could have mild and
ment. On the other hand, by moderating on the financial monotonically increasing effects on growth.
regionalization variable, the result implies that the growth-
inhibiting effects of financial development are mitigated by Moreover, Figure 3 also shows that low levels of financial
financial regionalization. This result supports the theoretical integration (i.e., financial segmentation), when interacted
predictions by Boyd and Smith (1997), which predict that with low levels of financial development is consistent with
international financial integration only has positive effects on negative growth rates. The steep slope that characterises
economic performance in countries with well-developed initial improvements in the financial sector, especially during
financial systems. the early financial development stages, reveals the strong
potential for financial development to spur growth even
Although quite revealing, the results from the SGMM analysis when there is financial segmentation. The observed
give us reason to suspect that some underlying non-linearities inverted-U shape along the financial development axis also
and complementarities may be driving the nature of the implies that there is a threshold level of financial
relationship between financial integration on economic activity development that is consistent with growth in a financially
in Africa. To unpack the intricacies of the relationship, we also segmented economy. Furthermore, there is evidence of
conduct a non-parametric regression analysis, which helps complementarities, as higher levels of financial development
to deal with potential complementarities, interactions, and combined with more financial integration return higher levels
trade-offs between financial integration and other identified of growth.
significant drivers of economic activity in Africa—all of which
were previously masked in the parametric analysis. We do this The contour maps are a cross-sectional representation of the
by examining the conditional mean plots of the growth surface three-dimensional (3D) surface plots. They map specific
charts, along with the financial integration variable and some points of the financial integration and financial development
important covariates. variable to the same predicted level of growth; in this case,

Table 1 GMM Growth Regressions: Dependent variable per capita GDP growth

Model1 Model2 Model3 Model4 Model5 Model6

GDPpc Growth(-1) -0.120*** -0.146*** -0.163* 0.006 -0.162*** 0.004
(0.043) (0.056) (0.093) (0.054) (0.054) (0.056)
Financial Regionalization -0.014 -0.024** -0.361*** 0.513** -0.032 -0.067*
(0.012) (0.012) (0.098) (0.221) (0.034) (0.04)
Fin. Depth -0.01 -0.025*** -0.705*** -0.002 -0.031*** -0.004
(0.008) (0.01) (0.185) (0.011) (0.01) (0.008)
Human Capital Index 0.268 6.588***
(0.678) (1.869)
Inflation 0.046 -0.029
(0.049) (0.085)
Polity Stability 0.296*** 0.317***
(0.047) (0.073)
Openness 0.017
(0.015)
Fin.Reg*Fin. Depth 0.031***
(0.008)
Fin.Reg*Human Capital -0.272**
(0.117)
Fin.Reg*Polity 0.001
(0.005)
Fin.Reg*Inflation 0.004*
(0.002)
Constant 4.207*** 2.709* 14.784*** -9.268** 5.201*** 3.879***
(0.403) (1.421) (3.168) (3.675) (0.902) (1.369)
Observations 465 334 341 341 334 341

4
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

they can be referred to as the iso-growth maps. In Figure 4, where growth is high with moderate levels of financial
we plot the iso-growth maps for financial integration and development and integration, and second is the scenario
development in SSA. The maps show that there are at least where high growth rates are attainable with some level of
two scenarios consistent with the highest attainable levels of regional financial segmentation and moderate financial
growth—around 9 and 9.5 percent. First is the scenario development.

Figure 3 Fitted surface plot of GDP growth

on financial integration and development

Figure 4 Iso-growth maps for financial integration and development

Figure 5 Fitted surface plot for financial integration and governance

5
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

In a similar vein, we examine how financial integration and The results for the joint effect of trade openness and financial
governance interact to drive economic activity in Africa, integration on growth when other regressors are held
holding all the other variables constant at their respective constant at their respective means are plotted in Figures 7
means. The results are plotted in Figure 5, which shows that and 8. From the surface plot in Figure 7, we observe that
the slope of the plot in the direction of governance is relatively openness is monotonically associated with higher levels of
flat, indicating that the positive effect of governance on growth growth, and this association is steep, indicating that even
is relatively minuscule. Further, with good governance, we see small improvements in the level of openness could potentially
that the effect of improvements in financial integration has result in significant growth improvements. As expected, the
quick and accelerated positive effects on growth. This is highest level of growth is attainable when there is better
inferred from the steep slope of the surface plot in the financial integration and more trade openness, the reason for
direction of financial integration. The picture gets clearer when this conclusion becomes obvious when we examine the
we examine the associated contour maps in Figure 6, where, contour maps in Figure 8, which indicates that the peaks
at the lowest level of negative growth, i.e., around -6.5, (mountain) of the surface plot in Figure 7 is around 8.5 percent
governance indicators are low and financial integration is also and is consistent with higher levels of openness and financial
low; whereas, at the highest level of growth, somewhere integration.
around 7 percent on the contour plot, governance indicators
are generally higher and financial integration is also high.

Figure 6 Iso-growth maps for financial integration and governance

Figure 7 Fitted surface plot for financial integration and openness

6
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

Figure 8 Iso-growth maps for financial integration and openness

4| Lessons for reforms function of monetary policy regimes in the different

economies. To induce convergence in interest rate spreads,
The evidence indicates that the regional financial convergence monetary authorities in the region can coordinate policies
process seemed to have stagnated around 2005, with more and perhaps set acceptable bands for monetary policy rates
contemporary evidence indicating a tilting towards financial in the region. This could also help to synchronise liquidity
globalisation. Therefore, this can be interpreted as a first-round positions in the region, so that excess liquidity in some
indicator of the heightening exposure to contagion effects from countries could be mopped-up by banks in other countries
global financial bubbles and busts. Indeed, the evidence shows with deficit liquidity.
that the 2008 global financial crisis had moderate effects on
selected African stock markets. Policy reforms that seek to
mitigate and cushion contagion effects from global financial 5| Concluding remarks
markets are necessary now more than ever.
Over the past three decades, there has been a considerable
Secondly, because our measure of financial integration is effort towards regional financial integration in Africa. This brief
inversely related to economic activity, which implies that endeavoured to trace the degree and evolution of stock
tighter interest rate spreads in credit markets enhance growth market integration using time-varying parameters estimated
in the region, it follows that by strengthening competition in by Kalman filtering techniques, and attempted to answer
regional banking, tighter spreads could be achieved, so that questions about the growth effects of regional financial
further growth can be realised and sustained in the region. A integration using parametric and nonparametric regression
specific reform action along this line would be to lighten as techniques. Results from the empirical investigation are used
much as possible restrictions on cross-border banking to rationalise potential reforms for fast-tracking the integration
establishments and operations. Further, an abatement of the process in Africa.
restrictions on the requirements for banks from the rest of the
world (America and Europe) to operate in Africa might be Overall, the results show that significant progress was made
used to further intensify competition and hence, achieve towards regional financial integration of stock markets during
narrower interest rate spreads in the region. A caveat in this the 1990s and 2000s, but recently the evidence indicates that
regard is, however, instructive. Because of the opaque nature financial markets in SSA are becoming more sensitive to
of the business of deposit money banking, and the fact that global benchmarks. Also, the results indicate that
an easing of restrictions could lead to an increased tendency improvements in the level of financial integration are
for adverse selection and moral hazard in the system. This associated with higher levels of economic activity. And the
policy option would have to be corroborated with growth-enhancing effects of regional financial integration are
strengthening the capacity of supervisory authorities to have greater in countries with higher levels of financial
cross-border oversight functions on banking operations. development, although there is a threshold beyond which it
begins to diminish. When using investment growth as a
It is also important to strengthen monetary policy measure of economic activity, countries that are more
coordination in the region since interest rate spreads are a regionally integrated enjoy higher levels of capital investments.

7
 AEB 2018 l VOLUME 9 l ISSUE 5 l VICE PRESIDENCY FOR ECONOMIC GOVERNANCE AND KOWLEDGE MANAGEMENT

References

Aryeetey, E. and M. Nissanke (2005). Financial integration and development: Liberalization and reform in sub-Saharan Africa.
Routledge, London.

Ekpo, A. and C. Chuku (2017). Regional Financial Integration and Economic Activity in Africa. Journal of African Economies,
26(suppl_2), ii40-ii75.

Haldane, A. G. and S. G. Hall (1991). Sterling's Relationship with the Dollar and the Deutschmark: 1976-89. The Economic
Journal, 101(406), 436-443.

Wakeman-Linn, J. and S. Wagh (2008). Regional financial integration: Its potential contribution to financial sector growth and
development in sub-Saharan Africa. In International Monetary Fund seminar “African Finance for the 21st Century,” Tunis, March
(pp. 4-5).

DESIGN & LAYOUT AfDB PCER