Professional Documents
Culture Documents
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
College of Business, Tennessee State University is collaborating with JSTOR to digitize, preserve and
extend access to The Journal of Developing Areas
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
ANALYSING THE SOURCES OF ECONOMIC
GROWTH IN AFRICA USING GROWTH A
ACCOUNTING AND A PANEL VAR APPROACH
Boopen Seetanah
University of Mauritius, Mauritius
Sawkut Rojid
The World Bank'
ABSTRACT
Using growth accounting framework and panel data analysis, while accounting for
endogeneity issues in growth modelling, this paper analyses the sources of growt
COMES A member countries. The growth accounting suggests a moderate level of TFP
contribution of capital and labour has been more or less the same. The resul
econometrics study are also along the same line with capital accumulation, openness
observed to be among the most important ingredient of growth. The econometric
shown the importance of some other variables like political and institutional stabil
development, IMF assistance funds and spill over effects as potential factors that
growth.
INTRODUCTION
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
368
1998,sound macro policies (Fischer, 1993; Levine and Renelt, 1992 and Barro, 1995),
financial markets (Levine and Zervos, 1993 and King and Levine, 1993), sound
government policies (Barro, 1995; Hall and Jones, 1997), Foreign Direct Investment
(Borenstein, Gregorio and Lee, 1998; Bende-Nabende and Ford, 1998 and Li and Liu,
2005 among others). The list is not exhaustive. Few works from the African region
following the above avenue can be found from Nyatepe-Coo (1998) Fosu (1999), Tahari,
Ghura, Akitoby, and Aka (2004), Akinlo (2004) and Khadaroo and Seetanah (2007,
2008).
However, the overwhelming number of research has focused on developed and
newly industrialised countries (NIC) cases (and samples) with very few studies on
developing countries, particularly for Eastern and Southern African region and trading
blocks. Even then the issue of dynamics and endogeneity in growth modelling has been
largely ignored until only recently. Empirical results and policy implications from the
existing literature based on panel sets of developed countries and NIC are not readily
applicable to developing countries given the own specificities and constraints of such
countries. This study aims at investigating the sources of growth for the African region,
particularly the COMES A block. COMES A, the 19-nation African trade bloc, in the
context of globalisation of markets, launched its FTA on October 31, 2000 with nine of
its members initially participating in the project, which dismantled trade barriers and
guaranteed free movement of goods and services in the region . It provides a good case
study as COMESA includes the major African nation in the Eastern and Southern Africa
and most of the countries in COMESA have experienced an annual GDP constant growth
rate of at least 5 % in 2006-2007, outperforming the rest of the continent. It is pushing
ahead with its plan to create a Common External Tariff and Custom Union, to adopt a
common currency for regional members by 2025 and to ultimately promote growth
within the region. Moreover, at a broad highly aggregated level it can be said that
COMESA members are of a homogeneous type when it comes to the composition of the
economy.
In the first instance, we employ a growth accounting analysis to decompose the
sources of growth for 14 countries using yearly data over the period 1978-2007. In a
subsequent analysis, to account for the dynamic and endogenous nature of growth,
Generalised Methods of Moments in a Cobb Douglas and Panel VAR methods in a
Solow Growth model are also used to identify econometrically the potential ingredients
of growth for the a sample of COMESA members, based on data availability. Results
from the above analysis are believed to supplement the scant literature on African growth
and can provide insightful policy implications for governments.
The rest of the paper is organised as follows. Section 2 discusses the growth accounting
method (which is also augmented to account for human capital) and analyses the results
thereof. Section 3 probes into the econometric specifications and estimation of the
potential determinants of growth using both Cobb Douglas and Solow Growth
specifications and Section 4 concludes.
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
369
Yt = AtK^;a (1)
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
370
that a has a simple interpretation under the additional assumption that there is perfect
competition among firms in the economy. To see this, consider the profit maximization
problem of a firm using production function (1) that rents capital at rental rate r, and hires
labour at wage wt:
max {4 K£r-rtK,-WtLt }
K,,L,
The first-order conditions for this problem require that the marginal product of each
factor equals its price:
C J Y"a
Lt
rt =aAt (3)
yK.j
f r \a
Lt
w, = (1 - a) A, (4)
Given these expressions, we can compute the capital and labour shares for the economy
(the fraction of output used to pay capital and labour, respectively) as:
rtKt _
Capital Share = y ^ (5)
*t
WtLt _i
Labour Share= y~ (6)
*t
Therefore, under the assumption of perfect competition, the capital share is a measure of
the parameter a. Once the capital and labour shares have been found, equation 2 can be
used to compute productivity values for any given year.
To proceed to the goal of decomposing economic growth into separate contributions from
capital, labour, and productivity, we argue that growth rates can be computed as (natural)
log-differences; thus the growth rate of output in a given year, for example, can be
computed as \ogYt+\ - log Yt. Taking logs of the production function (1) gives:
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
371
log Yt+l - log Yt= log 4+1 - log At + a(log Kt+] - log Kt)
+ (\-a)(\ogLt+l - log Lt) (8)
Thus, the growth rate of output equals the growth rate of product
times the growth rate of the capital stock, plus 1 - a the growth rate of t
Intuitively, the impact of any given factor of production on output growth
to its share in output. Assume, for example, that the capital share is one
labour share is two-thirds. If the capital stock increased by three percent
this growth would translate into just one percent growth in output. If ins
force were to grow by three percent, the resulting output growth would b
Changes in the labour force have a bigger impact on output, since la
important than capital in production.
A central element in the above process is to estimate the appropria
capital and labour. The early literature drew on the national accounts of s
countries and set the parameter for the capital share between 0.3 and 0.4 (w
share varying correspondingly between 0.6 and 0.7). Thus the parame
technology, is assumed to be the same across countries and this is qu
general, authors conduct some sensitivity analysis on the value of the para
of them tests the assumption of identical technologies across countries. It
that Senhadji (2000) relaxed the assumption of identical technologies acros
estimating separate production functions for 88 countries and foun
differences across countries.
In fact estimating the capital shares assumes the traditional constant return to
scale Cobb-Douglas production function in per capita form and to measure the relative
contribution of factor accumulation and productivity the following regression can be
estimated (see Senhadji, 2000).
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
372
by total labour force and both proxies are from the World Bank World Development
Indicators 2008 and from the IMF World Economic Outlook (WEO). The capital stock
series are constructed using the perpetual inventory methodology (PIM). It is assumed
that the initial capital-output ratio in 1980 was 3 and the depreciation rate was set at 5
percent (following Connell and Ndulu, 2000 and Tahari, Ghura, Akitoby, and Aka,
2004). The results from growth accounting exercises for individual countries and for
COMES A as a group over the period 1980-2007 are provided in Table 1.
Table 1 presents growth rates of GDP and the contributions of the various factors to
economic growth for selected COMESA countries as well as for the aggregate sample.
Burundi 1980-90 3.5 2.65 1.4 -0.6 Mau 1980 6.35 3 2 1.35
ritius 1990
Egypt 1980-90 4.9 4.7 1.2 -2 Ugan 1980 3.5 2.7 2.5
da 1990 1.7
1991-00 4.4 1.7 1.5 1.1 1991 6.2 2.5 2.5 1.2
2000
Ethi 1980-90 2.6 -0.2 2.2 0.6 Zam 1980 1.3 1.3 1.5 -1.5
opia bia 1990
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
373
2001 5.2 2.3 2.7 0.2 2001 - 5.2 2.4 2.2 0.6
2007 2007
Kenya 1980 4.6 2 1.5 1.1 Zimb 1980 5 2.7 1.8 0.4
1990 abwe 1990
2001 3.3 2.0 1.5 -0.2 2001 4.53 2.1 1.4 0.95
2007 2007 3 5
1980 1.9 1.1 2.0 -1.1 1980 3.35 1.5 1.6 0.28
2007 2007
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
374
TFP growth for the region during the period and even negative productivity for the period
1980-2000. Collins and Bosworth (1996) also reported negative growth rate (-0.42) of
TFP for sample of SSA over the period 1960-1994 and Bosworth and Collins (2003)2
confirmed a negative or no TFP growth for a study period 1960-2000. More than half of
the countries in the sample experienced declines or no increase in TFP on average during
1980-2007. Mauritius and Uganda experienced an average TFP growth of more than 1
percent during the period. The increased TFP growth during the latter period are those
countries which benefited from the efficiency gains from the shifts in the policy regime
beginning around 1980, the ongoing process of liberalization and opening up of the
economy and from the implementation of macroeconomic and structural reforms they
embarked on. The associated increases in reliance on markets and reductions in the role
of government would be expected to result in improved economic efficiency.
Although COMESA members appears on the average to outperform fellow SSA
countries at large in TFP, the average TFP growth for COMESA is remains below the
world average and could explain the relatively poor overall growth performance of the
region. It is noteworthy that nearly the East Asian countries and China which are among
the best performers worldwide have TFP contribution of around 1.2 with China alone
reporting around 2.5 (for the period 1960-2005) Bosworth, Collins , Virmani (2006)
recently reported TFP of 1.7 for the case of India over the period 1980-2004.
It should be stressed that Real GDP growth was driven largely by factor accumulation,
and in fact nearly all of the output growth during the first period is associated with
increases in factor inputs, particularly labour. With time capital is also observed to
establish itself as an important driver of growth outpacing the contribution of during the
last decade. This is reflected probably by the massive investment in capital infrastructure
from the part of government coupled by an overall increase in domestic and foreign direct
investment. Over the 1980-2007 period the contribution of labour and capital appears to
be the same. As a robustness check the contribution of the various factors to GDP growth
is computed using the same alpha coefficient (ie the sample alpha Of 0.5) for all countries
and the findings were found robust to that alternative value (the latter corresponds to the
COMESA average) The growth accounting framework can be extended to assess the role
of human capital for the determination of output and economic growth. Up to this point,
potential human capital differences across countries or over time were subsumed in the
productivity term. If information on the stock of human capital is available, we can
incorporate human capital into the production function in order to measure its
contribution to growth. Thus extending the Cobb Douglas function with human capital
term takes the following form:
Yt = AtK?H?Lx;a-p
Thus, there are now three inputs, physical capital Kt, the size of the labour force
Lt, and human capital Ht and we assume that the production function have constant
returns to all three inputs combined. To proceed to the goal of decomposing economic
growth into separate contributions from capital, labour, and productivity, we argue that
growth rates can be computed as (natural) log-differences; thus the growth rate of output
in a given year, for example, can be computed as logTM-1 - log Yt. The growth rate of
output can therefore be expressed as:
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
375
log Yt+l - log Yt = log At+X - log 4 + a(log Kt+X - log Kt)
Thus, the growth rate of output equals the growth rate of productivity, plus a
times the growth rate of the capital stock, plus fi times the growth rate of human capital,
plus 1 - a~ P the growth rate of the labour force. The measurement of data are as above
and human capital is measured here as the average years of schooling. However given
that such information is available for very few countries as reported below.
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
376
From the above, including human capital reveals interesting results and shows that this
factor is interestingly explaining part of the growth as well (maybe captured in labour
mainly from the previously study). This is particularly true for the case of Mauritius and
Kenya.
Bosworth and Collins (2003) have argued that the growth accounting framework
is a useful tool to understand growth experiences across countries. The same authors
have, however, noted the limitations of this methodology. A key weakness relates to the
interpretation that the measured residual from the growth accounting exercise represents
TFP growth. In practice, in addition to providing a measure of gains in economic
efficiency, the residual may also reflect a number of other factors, including political
disturbances and conflicts, institutional changes, droughts, external shocks, changes in
government policies, and measurement errors. This limitation is particularly important for
sub-Saharan African countries mired in conflicts and subject to significant drought
related and external shocks. Also, the results from growth accounting exercise should not
be misconstrued as providing the fundamental causes of growth (rather than the
proximate sources of growth).
Y,=4K,«I>> (io)
The Cobb-Douglas function is both hom
economy's output, A the shift in the pro
which is assumed to be risk neutral, K th
time. The Cobb-Douglas production fu
Parameters (31 and (32 measure the elasti
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
377
(11)
(12)
The lowercase variables are the natural log of the respective uppercase variable. The
disturbance term Ot is a deviation from the production function relationship. When the
output and inputs time series are unit root processes but the disturbance term is a
stationary process, equation (3) becomes a cointegrating relation implying a long-run
relationship between output, capital stock labour. The slope coefficients /?/, and fi2 are
then interpreted as long-run elasticities and are greater than zero.
A Cobb-Douglas production function necessitates inputs including capital stock and
labour. Labour has been proxied by employment and output by real Gross Domestic
Product at constant price. The data source and definition is the same as for the growth
accounting case in the previous section.
where Y is total output, PRI is the private investment ratio of the country, OPEN proxies
the level of openness of a country, EDU accounts for the quality of labour, FD is
financial development, INF is as measure of inflation and FDI one of Foreign Direct
Investment, and IMF is the amount of IMF financial assistance.
In the growth literature there exist a unanimous consensus (see Delong and Summers,
1990, 1994; Reinhart, 1989 and more recently Arin, 2004) of the role of private
investment (PRI) in promoting economic performance, possibly because technological
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
378
change is embodied in recent vintages of capital. The gross fixed capital formation as a
percent of real GDP is used as a proxy for investment in physical capital.
OPEN, which proxies for the level of openness of the country is also included in the
economic model following the work of Dollar (1992), Sachs and Warner (1995) and
Edwards (1998). These authors supported the idea that increased trade openness raised
economic growth through access for a country to the advances of technological
knowledge of its trade partners, access bigger markets and encouraging the development
of R&D through increasing returns to innovation and also through providing developing
countries with access to investment and intermediate goods that are vital to their
development processes. In addition, they are likely to have a greater division of labour
and production processes that are more consistent with their comparative advantages,
which enable them to grow faster. Following previous work, this is captured by the ratio
of total trade (Export + Import) to GDP.
We include a measure of education (EDU) to account for the quality of labour.
This follows the arguments and empirical evidences of Mankiw, Romer, and Weil
(1992), Barro (1998) and more recently Temple (2001). Human capital can be thought of
as affecting economic growth in the sense that workers with higher levels of education or
skills should, ceteris paribus, be more productive and more inventive and innovative.
Higher levels of human capital may also encourage capital accumulation, or may raise the
rate of technological catch-up for follower countries (Temple, 2001). The secondary
enrolment ratio is used to proxy for the above. The relationship between inflation and
GDP growth is twofold. In the short run, high inflation can be associated with high
growth in high activity cycles. In the long run, however, high inflation is associated with
macroeconomic policy mis-management and is expected to have a negative impact on
growth (Fischer (1993)3. Fischer and Modigliani (1978) argued that firms and workers
devote productive resources to deal with inflation. They further note that inflation
uncertainty reduces efficiency by discouraging long-term contracts and increasing
relative price variability. A high and unpredictable rate of inflation generally results in
poor performance of businesses and households. While most authors find growth and
inflation to be inversely related (see Fischer 1993, Levine and Renelt, 1992; Levine and
Zervos, 1993 and Barro, 1995) with the implication that inflation is quite costly, there are
exceptions (Sala-I-Martin, 1991). The consumer price index (CPI) has been used to
capture the role of inflation.
The financial system is also known to affect the level of economic growth in a
country. According to Levine and Zervos (1993), economies with more developed and
more efficient financial systems will be able to more effectively allocate savings to the
best investments, which in turn leads to increased productivity, potentially higher savings
rates, and faster growth. Financial development is directly linked with money supply
growth. FD is a proxy of financial development and of money supply growth and is
measured as the ratio of M3 to the country's Gross Domestic Product (GDP). It a typical
measure of 'Financial Depth' and has been widely used (King and Levine, 1993).
A particularity of most COMESA member states is that their budgets are dependent to a
large extent on IMF funds and assistance to accompany them development. For instance
Congo DR assistance amount to about 30% of its budget. To account for this fact, we
have included a measure of the IMF assistance to respective countries and this is
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
379
measured by the ratio of IMF funds received to GDP (IMF).Data was obtained from the
IMF data base.
Other alternative growth and policy variables have also been included in the model
specification (at times in lieu of some existing ones due to the limited amount of
observations and in a bid to get efficient and consistent estimates). These are discussed
below and represented by model specification 3. Barro (1995) argued that government
policies play a very crucial role in determining where an economy will go in the long run.
For example, favourable public policies including better maintenance of law, fewer
distortions of private markets, less non-productive government consumption and greater
public investment in high-return areas - lead, in the long run, to higher levels of real per
capita GDP. Hall and Jones (1997) believe that differences in levels of economic success
across countries are driven mainly by the institutions and government policies and
political stability that frame the economic environment. We thus include (POL), namely,
the political risk rating as provided by the International Country Risk Guide (ICRG
2007)4. The rating awards the highest value to the lowest risk and the lowest value to the
highest risk and provides a mean of assessing the political and institutional framework of
the countries (see ICRG 1999). SPILL is the spatial spill-over effects of the member
countries on the host country. This is introduced to take into account of the growth
impact of countries within the block on the host country economic progress (spill over
effects). This is particularly true given that most countries in the sample share both
physical and trade barriers. Such effects are measured by computing the average growth
rates of member countries in the block (excluding the growth rate of the host country).
Data Sources
The dependent variable output was proxied by the Gross Domestic Product at const
price (OUTPUT) and was generated from IFS. All other variables have been colle
from the World Bank World Development Indicators (various issues) and also from
IMF International Financial Statistics (various issues) and the Penn World Tabl
(updated). The sample size5 and the period of study (1988-2007) have been select
per data availability and it is a known fact that data has often been an important const
for African nations.
For the econometric analysis, these model specifications can be expressed as a log-li
regression, where lowercase variables are the natural log of the respective uppe
variables:
Model Specification 1
y» =«,, +PA +Pih +MU (14)
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
380
Model Specification 2
Model Specification 3
y„ = a„ +AM, +P2openjt + P3eduit + /3,gMjt + pjmf +
P6Pol + P7spUl + nit (16)
We use i to index countries (for the case of panel data)
error term. An often ignored issue before making the ap
the variables are stationary or not. We thus carry o
dependent and independent variables in all the abov
approach of Im, Pesaran, and Shin (IPS) (1995) who deve
the joint null hypothesis that every time series in the
reject a unit root in favor of stationarity (the results we
ADF and Fisher-PP panel unit root tests) at the 5 percen
deemed safe to continue with the panel data estim
specifications.
It is also likely that there may theoretically exists bi-causal (reverse causation) and
indirect effects together with dynamic feedbacks among the majority of variables in a
growth function. Including the above issues are essential to the modelling of our
hypotheses and this has been crucially ignored in the literature. Moreover, there might
still be the possibility of the loss of dynamic information even in panel data framework as
the dependent variable may have something to do in explaining itself as well (Levine et
al, 2000). To account for the existence of the above links related to endogeneity and
causality issues, we use vector autoregressions (VAR) on panel data6 which enable us to
consider the complex relationship between the various growth determinants and output
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
381
level. Moreover Panel VAR also allows for a firm-specific unobserved heterogeneity in
the levels of the variables. Panel-data vector autoregression combines the traditional
VAR approach, which treats all the variables in the system as endogenous, with the
panel-data approach, which allows for unobserved individual heterogeneity. We specify a
first order VAR model as follows:
Diagnosis tests
Sargan Test of Overidentifying restrictions prob>chi2=0.23
Arellano-Bond test of 1" order autocorrelation
Arellano-Bond test of 2"d order autocorrelation prob>chi2=0.52
Prob>chi2= 0.57
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
382
We estimate the coefficients of the system given in (2) after the fixed effects have been
removed and Table 5 report the results of the model.
Response Constant
yt-i pri,-, open,.} •4 inft-1 fdt-i imf.
to
l
(1.98)*
pri 0.43 0.23 0.58 0.09 0.12 (1.86)* (2.11)*
(1.86)* (2.15)** (2.21)** (1.01) (1.71)* -0.11 0.30 0.08
open 0.54 0.11 0.07 0.66 0.04 -
(1.82)*
(1.91)* (2.18)** (1.13) (1.99)* (1.78)* (1.89)* (1.98)*
edu 1.12 0.14 0.09 0.04 0.56 -0.11 0.05 0.07
(1.01) (1.11)
(2.02)* 0.45 0.08
0.12 (2.15)**(1.11)
(1.99)* 0.13 0.64
(1.15)
(2.01)*
Obs 220
No of 12
countries
The VAR model is estimated by GMM and fixed effects are removed prior to estimation. Reported
numbers show the coefficients of regressing the row variables on lags of the column variables.
Heteroskedasticity adjusted /-statistics are in parentheses. *** indicates significance at 1% level, **
at 5% and *** at 10% respectively. The small letters denotes variables in natural logarithmic and t
values are in parentheses.
The augmented Solow Growth Model (Row 1) reveals that the private
investment remains the major driver of growth with openness level, education and
financial development playing promising roles as well. The key role of investment in
promoting growth is in line with earlier studies of Beddies (1999) for the case of Gambia,
Ghura and Hadjimichael (1996), Rodrik (1998), Calamitsis, Basu and Ghura (1999) for a
sample of Sub Saharan countries and Fosu (2001) for a larger sample of African states.
Noteworthy is that openness is also observed to be another important ingredient of
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
383
growth in COMES A states. This confirms the results of Sachs and Warner (1997),
Sacerdoti, Brunschwig and Tang (1998), Calamitsis, Basu and Ghura (1999) and Fosu
(2001) and te Velde (2006) for sample of African countries.
As expected, a significant and positive relationship is derived from education,
suggesting its vital role for COMES A. Knight, Loayza and Villanueva (1992) and Barro
(1991, 1995) also obtained positive elasticity for large panel data sets. Beddies (1999),
for the Gambian case, Khadaroo and Seetanah (2007) for Mauritius and Calamitsis, Basu
and Ghura (1999), Sacerdoti, Brunschwig and Tang (1998) and Rodrik 1998 also confirm
such results for the African context. The positive and non negligible impact of FDI on
COMESA performance concur with the findings with that of Nyatepe-Coo (1998),
Bosworth and Collins (1999) and Assanie and Singletone (2002) for the case of
developing countries which include African panel data set. For countries case Obwona
(1999) for Uganda, Akinlo (2004) for Nigeria and Subramanian and Roy (2001) for
Mauritius obtained similar positive effects.
Financial depth is also interestingly observed to have a positive, although a
relatively lower impact on growth of COMESA countries. Allen and Ndikumana (1998)
for the case of the Southern African Development Community (SADC) found some
evidence of a positive correlation between financial development and growth of real per
capita GDP. Such results are confirmed by O'Connell and Ndulu (2000) and Seetanah
and Padachi (2007) for the case of a sample of African countries. As expected IMF
assistance to respective countries appears to have been a non negligible support to these
countries development as judged by its significance and magnitude of the coefficient. As
per theoretical prediction, inflation is seen to have negative influence on economic
performance. The adverse effect of inflation is consistent with pionneering work from
Barro (1995). Te vclde (2006) and Ghura and Hadjimichael (1996) also reported negative
association of inflation for SSA region.
Interestingly the positive and significant coefficient of yt_] also suggests that
lagged income of the countries in the sample contributes positively towards the current
level of y confirming the existence of dynamism and endogeineity in the modeling
framework. This is consistent with recent works from Bende-Nabende, Ford, Sen and
Slater (2000) and Choe (2003) and Li and Liu (2005). In fact the value of the coefficient
of the lagged income is 0.53 implying a coefficient of partial adjustment a of 0.47. This
means that y in one year is 47 percent of the difference between the optimal and the
current level of y. The other explanatory variables are also confirmed to be important
ingredients in explaining growth pattern in these countries.
Panel VaR analysis not only allows us to detect the presence of dynamism in
growth modeling but to also control and investigate important endogeineity issues. From
the table above, the row 2 can for instance be seen as an investment function, row 3 as an
openness equation, row 4 as a human capital function and so on. The presence of these
estimated functions allows us to establish the direction of causation and the presence of
reinforcing effects. Thus from the table it can easily be seen that investment, openness
and trade, education, financial development, inflation and IMF- assisted funds not only
explain growth but vice versa as well ,in other words there are bi-causal (reverse)
relationships with reinforcing effects.
Moreover there also exists indirect effects on growth for example, referring to
the investment equation, it can be seen that the control variables have all significant
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
384
effects on investment level and since investment in return is a major element of growth,
thus these variables also work indirectly on growth. Similar interpretation can be
obtained for other equations in the system.
Model 3: Augmented Solow Growth Model (Political Stability and Spill over effects)
As far as the third econometric equation is concerned (full results have not been reported)
the additional variable pol, which is a proxy the political and institutional framework of
the countries, turned out to be positive and significant, in line with theoretical predictions
and empirical works of Gynimah-Brempong and Traynor (1999) and Rodrik (1998) for
Africa. Moreover the variable SPILL, the spatial spill-over effects of the member
countries on the host country, is reported with a positive and significant parameter and
this suggest that neighbouring countries do matter in explaining growth of host country.
Broadly the same overall results for the other variables in the model were obtained.
Particularly, political stability and spillover effects are also seen to have reverse causation
on economic growth. Orthogonalised impulse-response functions analysis and variance
decompositions overall produced equivalent results. Details are available from the
corresponding author upon request.
CONCLUSION
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
385
ENDNOTES
1 The views of the authors are not necessarily the views of the affiliated instit
2 Bosworth and Collins (2003) provide a comparison of the growth perf
various sub-regions in the world during 1960-2000 (see appendix 5)
3 Macroeconomic performance plays an important role for growth sustainab
(1993) has shown that growth is negatively associated with inflation, large bu
and distorted foreign exchange markets
4 See International Country Risk Guide (1999, 2007), Brief Guide to the Rati
5 The COMESA countries included in the sample are: Burundi, RD Con
Ethiopia, Kenya, Madagascar, Malawi, Mauritius, Rwanda, Uganda, Zam
Zimbabwe.
6 We also attempted the use of Instrumental Variable (IV) methodology and obtained
similar results on the overall.
7 In our case the model is "just identified", i.e. the number of regressors equals the
number of instruments, therefore system GMM is numerically equivalent to equation-by
equation 2SLS.
8 The constant term reported in Table 2 differs markedly between the OLS and FIML
estimations. Using OLS, the constant term is 1,008.64 for all observations. Using FIML,
the constant terms are adjusted for unobserved heterogeneity; the value in Table 2 (
133.96) refers to the birthweight factor of unobservable type 1 (wy).
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
386
APPENDICES
Memorandum item:
Ail low-income
countries in the world 2/ 3.S 2,1 1.3 0.4
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
387
East Asia, excluding China (7) 6.7 3.9 2.3 0.5 1.0
REFERENCES
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
388
Bende-Nabende A., and Ford, J.L. "FDI, Policy Adjustments and Endogenous
Growth: Multiplier Effects from a Small Dynamic Model for Taiwan, 1959-1995", World
Development, 1998, Vol. 26 No. 7, pp. 1315-1330
Bende-Nabende, A., J. Ford, B. Santoso, and S. Sen. "The Interaction between
FDI, Output and the Spillover Variables: Cointegration and VAR Analyses for APEC",
Applied Economics Letters, 2003,10 (3): 165.72.
Bosworth, B., and Susan M. "The Empirics of Growth: An Update", Brookings
Papers on Economic Activity, 2003, no.2, pp. 113-206.
Bosworth, B., Susan M., and Yu-chin C. "Accounting for Differences in
Economic Growth", Brookings Discussion Papers in International Economics, 1995, No.
115 October, pp. 1-63.
Bosworth, B, Susan M. C, and A Virmani. "Sources of Growth in the Indian
Economy", India Policy Forum 3, 2007, pp. 1-50.
Brou Aka E., Akitoby., B, Ghura., D & Tahari A. "Sources of Growth in Sub
Saharan Africa", IMF Working Papers , 2004, 04/176
Calamitsis, E. A., A Basu, and D Ghura. "Adjustment and Growth in Sub-.
Saharan Africa", IMF Working Paper, 1999, WP/99/51
Choe, J.I. "Do foreign direct investment and gross domestic investment promote
economic growth?", Review of Development Economics, 2003, 7(1), 44-57
De Gregorio, J. 1996. "Financial Development and Economic growth", IMF
Working Paper WP92/101., 1996
De Long, B, J. and Summers, L.H. "Equipment Investment and Economics
Growth", Quarterly Journal of Economics, 1990, 106, May.
DeLong ,B,J. and L H. Summers. "Equipment Investment and Economic
Growth: Reply", Quarterly Journal of Economics, 1994, 109:3 (August), pp. 803-807.
Diamond, D.W. "Financial Intermediation and Delegated Monitoring", Review
of Economic Studies, 1984, 51(3), pp.393-414.
Dollar, D. "Outward-Oriented Developing Economies Really Do Grow More
Rapidly: Evidence from 95 LDCs, 1976-1985", Economic Development and Cultural
Change, 1992,40:523-544.
Easterly, W. The Elusive Quest for Growth, 2001, Cambridge, MA. MIT Press.
Edwards, S. Crisis and Reform in Latin America, 1998, New York: Oxford University
Press.
Fosu, A, K... "Economic Fluctuations and Growth in Sub-Saharan Africa: The
Importance of Import Instability", The Journal of Development Studies, 2001, vol. 37(3),
pages 71-85.
Fosu, A. K.. "The External Debt and Economic Growth in the 1980s: Evidence
from Sub-Saharan Africa", Canadian Journal of Development Studies, 1999, 20(2): 307
18.
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
389
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms
390
2003.
Pereira, A. M. and O Roca-Sagales. "Spillover effects of public capital
formation: evidence from the Spanish Regions", Journal of Public Economics, 2003, Vol.
53 pp 238-256
Reinhart, C M. "Private Investment and Economic Growth in Developing
Countries". IMF Working Paper, No. 89/60, 1989, Available at SSRN:
http://ssrn.com/abstract=884880
Rodrik, D.. "Trade Policy and Economic Performance in Sub-Saharan Africa",
NBER 1998, Working Paper 6562 (Cambridge)
Romer, P.M. "Increasing returns and Long-Run Growth", Journal of Political
Economics, 1986, 94, (5), October.
Roubini, N. and Sala-i-Martin, X., "Financial Development, the Trade Regime
and Economic Growth", NBER Working paper series, 1991, No. 3876.
Sacerdoti, E, S Brunschwig, and J Tang., The Impact of Human Capital on
Growth: Evidence from West Africa, IMF Working Paper 98/162 ,1998
(Washington international Monetary Fund).
Sachs, J and Warner, A. "Natural Resource Abundance and Economic Growth",
NBER Working Paper, 1995, No. W5398
Schultz, T.W. "Investment in Human capital", American Economic Review, 1961, LI(I),
March..
Seetanah B and A J Khadaroo."Assessing the contribution of land, sea and air
transport capital to the economic performance of the small island state of Mauritius",
Applied Economics Letters, 2007, Vol 14, pp 1151-1155.
Seetanah B and Padachi D and Durbarry R. "Financial development and
economic growth in Africa: A dynamic Panel Data Analysis , International Journal of
Business Research, 2007, VII (1), 145-152
Seetanah B and Khadaroo J. "Transport and economic performance: The case of
Mauritius", Journal of Transport Economics and Policy, 2008, 42(2), 1-13
Senhadji, A. "Sources of Economic Growth: An Extensive Growth Accounting
Exercise", Staff Papers, International Monetary Fund, 2000, Vol. 47, No.l, pp. 129-57.
Singh, H. and Jun, K. W. "Some New Evidence on Determinants of Foreign
Direct Investment in Developing Countries", World Bank Policy Research Working
Paper, 1995, no. 1531
Solow, R.M.. "A Contribution to the theory of Economic Growth", Quarterly
Journal of Economics, 1956, pp 70-82
Temple, J. R. W. "Generalizations that aren't? Evidence on education and
growth", European Economic Review, 2001, 45(4-6), 905-918.
Velde, D.W. te., 200. "Measuring state-business relations in sub-Saharan
Africa". IPPG Discussion, 2006, Paper 4 available online,
http://www.ippg.org.uk/publications.html
World Bank "The East Asian Miracle: Public Policy and Economic Gro
1993, New York: Oxford University Press.
This content downloaded from 132.203.227.63 on Thu, 27 Jul 2017 16:40:03 UTC
All use subject to http://about.jstor.org/terms