See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/312044644

Manufacturing and Economic Growth in Africa: A

Panel Test of Kaldor's First Growth Law

Article · December 2016

CITATION READS

1 101

2 authors, including:

Olumuyiwa Olamade
Caleb University Lagos
21 PUBLICATIONS 73 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Technology Enabled Customer Relationship Management in Aluminium Industry in Nigeria View project

All content following this page was uploaded by Olumuyiwa Olamade on 04 January 2017.

The user has requested enhancement of the downloaded file.

Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

Manufacturing and Economic Growth in Africa: A Panel Test of

Kaldor’s First Growth Law

Olumuyiwa Olamade* Oluwasola Oni

Department of Economics, Caleb University. PMB 21238, Ikeja, Lagos, Nigeria

Abstract
This paper examines the importance of the manufacturing sector for economic growth in African countries.
Although many African countries have posted impressive growth performance in last one decade. A notable fact
of this growth is the declining share of manufacturing in the gross domestic product (GDP). Will the contraction
of the manufacturing sector hurt African economic growth in the long-run? We approach this question by testing
Kaldor’s first law of economic growth using panel data for a sample of 28 African countries over the period
1981-2015. Results obtained from pooled Ordinary Least Squares, Fixed Effects, and System Generalized
Method of Moments provides current evidence to support manufacturing as the engine of growth in Africa. The
Fagerberg-Verspagen (1999) criteria show that despite the falling share of manufacturing in the GDP, the
difference between the coefficient of manufacturing output growth and share of manufacturing in GDP is
positive and significant. We conclude that de-industrialisation will adversely affect both the growth rate of the
non-manufacturing sectors and of the whole economy in African countries.
Keywords: Economic growth, manufacturing, non-manufacturing, productivity, value added

1. Introduction
Since Clark (1941) a wide range of literature, especially those associated with the post-Keynesian and
evolutionary economists, have empirically confirmed that structural change is the necessary process of economic
growth. The expansion of the manufacturing sector is generally viewed as the most significant engine of the
growth process. Kaldor (1966, 1967) posits a strong positive causal relationship between the growth of
manufacturing output and that of the GDP. This relationship rests on certain special characteristics of the
manufacturing sector, which makes it the engine of GDP growth and of living standards. First, manufacturing is
characterised by both static and dynamic increasing returns to scale, while other non-manufacturing activities are
subject to diminishing returns. Secondly, manufacturing output growth draws labour from non-manufacturing
activities where there are diminishing returns resulting in productivity growth in these activities because the
average product of labour is above the marginal product. The manufacturing sector’s characteristics with regard
to GDP growth is the foundation of what now is known as the Kaldor’s growth laws. The laws state that: (1)
manufacturing is the engine of GDP growth (2) the productivity of the manufacturing sector is positively related
to it's on growth (also known as the Verdoorn’s law), and (3) the productivity of the non-manufacturing sector is
positively related to the growth of the manufacturing sector. Thus, the post-Keynesian development paradigm
based on Kaldor’s ‘engine of growth’ hypothesis advance the strengthening of the manufacturing sector, even if
the sector offers no comparative advantage in the initial stage of development. The special characteristics of the
manufacturing sector will enhance its competitiveness over time and spread positive externalities to other
important sectors of the economy (Cantore, Clara and Soare, 2014).

126
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

Developments in structural transformation paths of some developing countries in the last two decades
appear to question the role of manufacturing as engine of growth or the sole engine of growth. Dasgupta and
Singh (2006) outline certain long-term structural tendencies observed in some developing countries which prima
facie challenges the Kaldor’s theses. These tendencies include the onset of de-industrialisation at a much lower
level of per capita income than historically observed in the advanced countries during their period of
industrialisation; the phenomenon of jobless growth in the formal manufacturing sector in both slow-growing
and fast-growing economies; and a faster long-term growth of services than manufacturing. Drawing from the
experience of India, Amirapu and Subramanian (2015) posit that any sector of the economy could lead growth if
it features a high level of productivity, dynamic productivity growth, extensive backward and forward linkages,
exportability, and comparative advantage for the home country. Thus, services or any other sector or branches
thereof could displace manufacturing as engine of growth.
With developments in information and communications technology (ICT) and the emergence of
services as the leading growth sector in many developing countries the debate on the best way to stimulate
growth in developing countries remains unsettled. The question then arises as to whether African countries need
to industrialize in order to grow and prosper. Put differently, should manufacturing remain the focus of
industrial policy in developing countries, and Africa in particular? This paper attempts to answer this question by
examining the GDP-manufacturing growth relationship. If we find significant evidence that, the influence of
manufacturing output expansion on economic growth transcends the percentage share of manufacturing in the
GDP, we conclude that the manufacturing as engine of growth hypothesis is still relevant to Africa. The
manufacturing sector in Africa offers opportunities for economic growth through economy-wide diffusion of
technological progress believed to originate principally from the manufacturing sector. The rest of the paper is
organized as follows; review of relevant literature comes up section 1, while the description of the data used and
the source is in section 2. The theoretical foundation of the work, the regression equations, and estimation
methodologies are set out in section 3. We report the findings and conclude in sections 4 and 5, respectively.

2. Review of literature
There appear to be two major definitive consensuses in the empirical literature regarding the pattern of growth in
developed and developing economies. First is that the major sources of economic growth for developed and
developing economies are completely different. Secondly, that the linear structural change from agriculture to
manufacturing, and then from manufacturing to services that characterized the economic transformation of the
developed economies may not generally apply to developing economies. In addressing the economic growth
concerns of developing economies, two key and somehow divergent strands of literature are identifiable. On the
one hand is the neoclassical paradigm which emphasises that countries ought to specialise in those sectors they
have comparative costs advantage than competitors. On the other hand, the post-Keynesian and evolutionary
economists argue that countries ought to specialise in those strategic sectors that can stimulate economy-wide
productivity and innovation, even if such sectors confers no comparative advantage at the initial stages of
development (Cantore et al, 2014).
While both paradigms are appropriate for developed economies largely driven by rapid technological
changes based on efficient accumulation of physical and human capital, they pose different development

127
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

challenges for low-income and middle-income developing countries. According to the comparative advantage
paradigm, developing countries ought to focus on agriculture and successively transit from agriculture to the
modern sectors following the linear structural transformation path. Su and Yao (2016) identified the challenge
posed by the neoclassical paradigm for low-income developing countries as the lack of a driving force to transfer
resources from agriculture to the modern sectors. Where the driving force is available and activated, research has
empirically shown that the structural change bonus resulting from such transfer has been a major source of
growth in developing countries (Timmer and Szirmai, 2000; Temple and Woessmann, 2006; Rodrik, 2009 and
Timmer and de Vries, 2007). For the middle-income countries, the challenge is how to transit from middle-
income to high-income countries or avoid the supposed ‘middle-income trap’ (Agenor, Canuto and Jelenic,
2012). For countries experiencing income growth slowdowns, Krugman (1994) identified labour productivity
growth as a veritable new source of long-term economic growth. Thus, breaking out of income trap and catching
up with developed economies is actually a process of eliminating the productivity gap. Eichengreen, Park and
Shin in a 2011 paper also advanced the productivity argument as a fundamental development challenge of
developing countries. To Eichengreen et al, slowdown in the rate of total factor productivity (TFP) growth may
result in prolonged slow output growth, and freezes income growth. Which route should developing economies
take to ensure sustained long-term income growth? According to the neoclassical growth model, it is the way of
efficient accumulation of physical and human capital. The post-Keynesians and evolutionary economists point to
the way of continual productivity growth.
The essential contribution of the Kaldor’s engine of growth hypothesis is the proposition of a theoretical
foundation for a development strategy, which locates manufacturing output growth as the fulcrum for both
efficient physical and human capital accumulation (neoclassical theses) , and factor productivity growth (post-
Keynesians prescription). If productivity growth in both the manufacturing sector and non-manufacturing sector
of the economy is positively related to output growth in the manufacturing sector as proposed by Kaldor (1966,
1967), then a transfer of resources from other sectors to manufacturing will result in more rapid aggregate
growth (Szirmai, 2011; Cantore et al, 2014). Evidence of this dynamic shift effect in developing countries is
unambiguous as productivity growth in manufacturing has been more rapid than in the primary sector (Szirmai,
2011). Further, the manufacturing sector compared to other sectors has a higher demand for capital and
investment thereby providing opportunities for capital accumulation and increase in the private saving ratio (Su
and Yao, 2016). The neoclassical growth theory regards savings as one the most important factors for long-run
economic growth.
The manufacturing sector, more than other sectors, offer superior opportunities for embodied and
disembodied technological progress crucial for the development of developing countries. Rapid capital
accumulation puts into operation in firms new machines that incorporate the latest technological advances that
drive productivity growth in firms and in the economy. Greenwood, Hercrwitz, and Krusell (1997) estimated that
60% of labour productivity growth is directly attributable to embodied technological progress. It is logical that if
new machines embody technology that is more productive than that of older machines, then a sustained
investment in new machines should lead to an increase in TFP growth. The positive effects of embodied
technological progress are also positive for advanced economies. Stiroh (2001) inquiring into recent changes in
the US economy attributes accelerated aggregate productivity growth to a combination of accelerating technical

128
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

progress in high-tech industries and corresponding investment and capital deepening. Sakellaris and Wilson
(2002) evaluated the impact of embodied technological change on US manufacturing productivity growth
between 1972 and 1996, and concluded that the role of investment-specific technological change as an engine of
growth is even larger than previously estimated.
The manufacturing sector allows for faster growth rate in both embodied and disembodied
technological progress. Cornwall (1977) argued that embodied and disembodied technological progress largely
originates in the manufacturing sector and diffused to other sectors therefore making manufacturing the locus of
technological progress in the economy. Manufacturing generates more extensive and stronger linkages with, and
spillovers into, the economy than nonmanufacturing activities (Herzer, 2007). While linkages can create
economies of scale, spillovers create an environment for new product and process technologies ideas resulting in
further expansion of both the manufacturing and nonmanufacturing sectors.
Empirical economic growth literature using different econometric models had tested and confirmed the
validity of manufacturing as engine of growth. Studies at national and regional levels largely agreed that output
growth in the manufacturing sector is uniquely important to the process of national economic growth as
aggregate economic growth positively relates to both output growth and productivity growth in the
manufacturing sector. At the level of individual countries, the U.S. Department of Commerce (1995) finds the
manufacturing sector a powerful source and a principal arena of growth and change. Other country level studies
include the works done for the United Kingdom (Stoneman, 1979), Australia (Whiteman, 1987), Greece
(Drakopoulos and Theodossiou, 1991) and Turkey (Bairam, 1991).
Wells and Thirlwall (2003) evaluated Kaldor’s law across African countries. Dasgupta and Singh
(2005) evaluated the engine of growth hypothesis for 30 developing countries. In a 2006 work, Dasgupta and
Singh analysed a sample of 48 developing countries. Szirmai (2009) worked with a panel of 63 developing
countries and 16 advanced countries for the period 1950-2005. Lavopa and Szirmai (2012) used a sample of 92
countries for the period 1960 – 2010. Libanio and Moro (2013) applied panel data for 11 largest economies in
Latin America during the period 1980-2006. All these studies find strong evidence to support the engine of
growth hypothesis for manufacturing. Pacheco-Lopez and Thirlwall (2013) reinterpreted Kaldor´s first law.
Starting with a premise that the model in its original form is essentially ‘closed-economy’ hypothesized that
export growth drives GDP growth, and export growth itself is a positive function of manufacturing output
growth. Using a dataset comprising 89 open developing economies for the period 1990-2011, the authors show
that manufacturing growth translates into economic growth through international trade.
Eguez (2014) applied Kaldor’s first law to a world panel of 119 countries over the period 1990 – 2011.
The study confirmed that manufacturing continues to be an engine of growth in both low and middle-income
countries. Manufacturing activities with higher technology component generate more space for technological
progress, human capital development and productivity increase, which ultimately contribute positively to a faster
economic growth. For a sample of 80 countries, Cantore et al (2014) decomposed manufacturing sector growth
into intensive and extensive industrialisation. They affirm the validity of Kaldor’s law of manufacturing as
engine of growth while concluding that intensive rather than extensive industrialisation more closely relates to
GDP growth. In a more recent study, Su and Yao (2016) in a sample of 180 middle-income countries for the
period 1950-2013 found that compared with other sectors, manufacturing development can better utilise human

129
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

capital and economic institutions, improve the incentives of savings, enhance the technological accumulation.
In conclusion, empirical literature undeniably is divided on the engine of growth hypothesis. While
some supports the engine of growth hypothesis for manufacturing others argued that the recent surge in service
sector expansion in some developing countries and early de-industrialisation experienced by others appears to
suggest that manufacturing is not the only engine of growth. For instance, Eguez (2014) found that
manufacturing and services both turn out to be engines of growth for middle-income countries, though
manufacturing is the stronger source of growth. The result suggests that for these countries manufacturing is not
the only route to achieving economic growth. While strongly validating the manufacturing as engine of growth
hypothesis, Cantore et al (2014) showed that not every dollar of additional manufacturing value added
contributes to growth.

3. Data description and source

The variable representing manufacturing output growth is the growth rate of the manufacturing value added. The
non-manufacturing sector is made up of the valued added of both the agriculture and services sectors. This
follows the practice well established in the literature. As the influence of manufacturing value added growth rate
on economic growth might not be significantly visible in a single year, we work with a 5-year average of the
growth rate of real GDP, manufacturing value added growth rate, and non-manufacturing value added growth
rate. Based on data constraint we select a sample of 28 African countries for which continuous data are available
for the period 1981 – 2015. All the data used are from the 2016 edition of the World Bank’s World Development
Indicators (WDI). All computations were done on Eviews 9.5.
4 Empirics
4.1 The equations
Kaldor’s first law of growth often referred as the ‘engine of growth hypothesis’ posits that the growth rate of
GDP is positively related to the growth rate of manufacturing output. Kaldor expressed the hypothesis as:
q = a1 + a2m, (1)
where q and m represent growth of GDP and manufacturing output, respectively. The regression
coefficient (a2) should be positive and less than a unity suggesting that the overall growth rate of the economy is
associated with the excess of the growth rate of manufacturing output over the growth rate of non-manufacturing
output.
Kaldor premised the explanation for the correlation between the growth rate of manufacturing output
and the aggregate economic performance on two possible reasons. The first relates to the fact that the expansion
of manufacturing output leads to the transfer of labour to the manufacturing sector from the low productivity
non-manufacturing sector. The result is an increasing economy-wide productivity with little or no negative
impact on the output of the non-manufacturing sector, given the existence of surplus labour. The second reason
relates to the existence of static and dynamic increasing returns in the manufacturing sector. While static returns
relate essentially to economies of scale internal to the firm, dynamic returns refer to increasing productivity
derived from learning by doing, ‘induced’ technological change, and external economies in production.
As regards equation (1) Kaldor made the important point that the correlation between q and m is not
only due to manufacturing output constituting a large component of GDP, rather that high economic growth rates

130
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

is positively related to the excess of manufacturing output growth over non-manufacturing output growth. He
demonstrated that countries that exhibit GDP growth rates over 3% a year, present a manufacturing growth rate
output higher than the growth rate of the non-manufacturing sector. Kaldor expressed this claim in equation form
as:
q = a3 + a4(m – nm) (2)
where nm refers to the growth of non-manufacturing output, and (m – nm) the excess of manufacturing
output growth over non-manufacturing output growth. We will simply denote this excess as (λ) and rewrite
equation (2) as: q = a3 + a4(z). To further support his first law of growth Kaldor showed that non-manufacturing
output growth also responds positively to the growth of manufacturing output resulting in overall economic
performance growth. This he expressed as:
nm = a5 + a6m (3)
Equation (1) essentially is the culmination of the manufacturing as engine of growth hypothesis. The
other two equations offer support for the hypothesis and abate the endogeneity problem, which characterises
equation (1). We will subject equations (1) – (3) to pooled OLS and Fixed Effects (FE) regression. The system
generalised method of moments (system GMM) will test only equation (1). For each of the estimation
techniques, appropriate transformation of the three original equations is made to suit the assumptions of the
technique.

4.2 Empirical methodology

Empirical research has shown that the traditional ordinary least squares (OLS) estimation (the technique used by
Kaldor) is incapable of handling the problems of highly correlated regressors, country heterogeneity, reverse
causality, etc often associated with empirical growth regression analyses. Kaldor’s own results for his first law
suffer from endogeneity bias, which may arise from the independent variable (m) being correlated with the error
term, omitted variables in the regression, and simultaneity bias. In the relations between GDP growth (q) and
manufacturing growth (m) both variables could be reciprocally correlated. In the face of reciprocal causality, the
OLS technique produces biased estimates. To circumvent this concern researchers have employed alternative
econometric models to test the engine of growth hypothesis. In this paper, we employed the pooled OLS, FE
model and the system GMM. The pooled OLS is simply the conventional least squares method fitted to a panel
sample. In the face of heterogeneity across countries and the likelihood that the country-specific effects may be
correlated to the regressors, the FE model provides improvements to pooled OLS results and thus address the
concern of spurious correlation. We chose system GMM over the difference GMM because it offers better
results. It improves difference GMM by introducing instruments in differences for equations in levels. We
therefore expect that our results would progressively improve as we proceed from pooled OLS to FE, and to
system GMM.
4.2.1 Pooled OLS
For the pooled OLS regression equation (1) which describes the relationship between GDP growth and
manufacturing output expansion is transformed as:
qit = b1t + b2i(mit) + λ t + ɛit, b2 > 0 (4)
where λ is the time-specific effects introduced to check if the influence of manufacturing growth on GDP growth

131
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

changes during the study period. Similarly, equation (2) which predicts the GDP growth rate by the difference
between the growth rates of manufacturing value added and non-manufacturing value added is transformed as:
qit = b3t + b4i(zit ) + λt + ɛit,b4 > 0 (5)
In the same vein, equation (3) which posits that non-manufacturing value added growth positively responds to
growth of manufacturing value added is transformed for pooled OLS regression as
(nmit)= b5t + b6i(mit) + λt + ɛit, b6 > 0 (6)
The pooled OLS model estimates equations (4) – (6) without regard to country-specific income effects.
We conduct heteroscedasticity and autocorrelation tests to check the validity and reliability of the
estimates obtained by pooled OLS. Tests of the null hypotheses of no cross-section random effect, and of the
combined cross-section and time random effect conducted with the Breusch-Pagan, Honda, King-Wu,
Standardized Honda and Standardized King-Wu statistics overwhelmingly supports the acceptance of the null at
5%. However, the Pesaran scaled LM, Biased-corrected scale LM, and Pesaran CD all fail to accept the null of
no cross-section dependence in residuals.
4.2.2 Fixed Effects model
The main assumption of the FE model is that the error term is divided into two distinct components as: ɛit = fi +
µ it. Where µ it is the conventional idiosyncratic random error and f the country-specific effects. In this model, f
depends on the regressors and is therefore a random variable. Like in the Pooled OLS, we estimate the FE model
without regard to country-specific effects often captured by the introduction of income category dummies. To
implement the FE model, the following transformation of equations (1) – (3) apply:
qit = c1t + c2i(mit) + λt + fi + µ it. (7)
qit = c3t + c4i(zit) + λt + fi + µ it. (8)
(nmit)= c5t + c6i(mit) + λ t + fi + µ it. (9)
Combinations of the Pesaran scaled LM, Biased-corrected scale LM, and Pesaran CD tests provide
strong support to accept the absence of residuals correlation at 5% in the FE model. We therefore, assume the
validity and reliability of the estimates of the three equations under the pooled OLS and FE model.
4.2.3 System GMM
Researchers have also estimated equation (1) by including the lagged values of the dependent variable (q) as a
way to capture possible autocorrelations as in Holland et al (2012).
There is a consensus that when such a dynamic model is regressed with either the OLS or FE techniques the
coefficient of the lagged dependent variable may bias upwards in case of individual-specific effects or
downwards in the case of FE (Nickell, 1981). In this situation, the GMM is helpful to correct the bias. The GMM
is a modern econometric technique to deal with endogeneity bias with or without lagged dependent variable.
There are different alternatives to perform the GMM technique. Arellano and Bond (1991) proposed the
difference GMM estimator that transforms the regressors, usually by differencing them in order to remove
country specific FE, which are the source of endogeneity. Then the first difference of the dependent variable is
instrumented with lagged values of the regressor in level. Nevertheless, the past values in levels may turn out to
be poor instruments for first differences (Blundell and bond 1998). Consequently, Arellano and Bover (1995),
and Blundell and Bond (1998) proposed the system GMM estimator that builds a system using the original
equation with the dependent variable in first difference and the transformed equation. In this model, the

132
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

transformed equation has a dependent variable in levels that is instrumented with suitable lags of their own first
differences based on the assumption that the first differences of instrument variables are uncorrelated with
country FE. According to Roodman (2006), this process allows the introduction of more instruments, and can
dramatically improve efficiency. Equation (1) is expressed for system GMM regression as follows:
qit = α1t + φ(qit-1) + α2i(mit) + λt + fi + µ it (10)
For consistency and asymptotic normality, GMM relies on the correct specification of the model and,
given the model, on the specification of correct moment conditions (Andrews and Lu, 2001). Thus, when fitting
a model by GMM, it is required that a check is made to see whether instruments used are uncorrelated with the
errors (orthogonality condition). In GMM estimation, the Hansen’s J statistic is the most common test statistic.
The test statistic has a χ2 distribution under the null hypothesis that the instruments are valid. If the equation,
excluding suspect instruments, is exactly identified the J-statistic will be zero. A J-statistic of less than 0.1 is
acceptable to satisfy the orthogonality condition (Benchimol, 2013).
Finally, we apply the Fagerberg-Verspagen criteria to the coefficient estimate of the system GMM to
obtain additional evidence in support or against the manufacturing as engine of growth hypothesis. The criteria
test whether the coefficient of manufacturing value added growth is positive, and if positive, whether it is
significantly greater than the share of manufacturing value added in the GDP. If the difference is positive and
significant, it is interpreted as support for the engine of growth hypothesis.

4. Empirical findings and discussion

The results of pooled OLS, FE and system GMM regressions are presented in Tables 1, 2 and 3 respectively. The
pooled OLS and FE model returned the same result for equation (1) which tests the direct relationship between
GDP growth (q) and manufacturing output growth (m). In the two models, the constant terms (b1 and c2) and the
coefficient of manufacturing output growth (b2 and c2) are (0.023) and (0.31) respectively, and are significant.
Under the pooled OLS and FE regressions the constant terms vary within the range 0.22 and 0.34. The
implication is that when manufacturing valued added remain unchanged, on the average, GDP grows within the
range captured by the constant term. The coefficients of manufacturing output growth (b2 and c2) and of the
excess of manufacturing output growth over non-manufacturing output growth (b4 and c4) lie between 0.23 and
0.31. This means that when manufacturing output and the excess of manufacturing output over other sectors
output grows by 1%, the GDP growth rate increases within the range of 0.23% and 0.31%. It is notable that these
coefficient values are considerably less than one, which suggests that the greater the manufacturing output
growth and the excess of manufacturing growth over other sectors, the greater the GDP growth. Also, we found
that the lowest coefficient of the manufacturing output growth (27% in the pooled OLS regression) is higher than
the average share of manufacturing value added in the GDP (13.05%) computed for the 28 countries for the
period covered in the analysis. Results obtained for equations (4) and (5) with the pooled OLS and FE supports
manufacturing as engine of economic growth for Africa. As manufacturing output increases and in excess of
output growth in the non-manufacturing sector, it drives growth in the GDP via growth of output in other sectors
of the economy. Equation (6) which shows the effect of manufacturing output growth on the growth performance
in non-manufacturing sector produced a coefficient value of 1.15 suggesting that a 1% increase in manufacturing
output induces more than a 1% increase in non-manufacturing output.

133
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

Table 1: Pooled OLS results

Equation 4 Equation 5 Equation 6
(b2) (b4) (b6)
Coefficient 0.31 0.27 1.15
(0.0000) (0.0000) (0.0000)
Constant 0.023 0.034 0.022
(0.0000) (0.0000) (0.0000)
R2 0.35 0.09 0.66
DW 1.24 1.35 1.71
LM tests for random effects (cross-section)
Null hypothesis: No cross-section effect
Breusch-Pagan 4.56 0.86 0.12
(0.0328) (0.3543)* (0.7282)*
Honda 2.13 0.93 -0.347542*
(0.0164) (0.1771)*
King-Wu 2.13 0.93 -0.347542*
(0.0164) (0.1771)*
Standardized Honda 2.29 1.08 -0.230429*
(0.0111) (0.1404)*
Standardized King-Wu 2.29 1.08 -0.230429*
(0.0111) (0.1404)*
LM tests for random effects (cross-section and time)
Null hypothesis: No cross-section and time effects
Breusch-Pagan 16.07 5.83 3.56
(0.0001) (0.0158)* (0.0591)*
Honda 3.91 2.23 1.07
(0.0000) (0.0128)* (0.1432)*
King-Wu 3.98 2.41 1.53
(0.0000) (0.0079) (0.0630)*
Standardized Honda -0.01922* -1.83625* -3.108506*
Standardized King-Wu 0.92 -0.809213* -1.782791*
(0.1797)*
Residual cross-section dependence test
Null hypothesis: No cross-section dependence (correlation) in residuals
Breusch-Pagan LM 537.71 489.84 473.23
(0.0000) (0.0001) (0.0006)
Pesaran scaled LM 4.79 3.04 2.45
(0.0000) (0.0023) (0.0145)
Pesaran CD 7.05 5.19 2.65
(0.0000) (0.0000) (0.0081)
Period included :7, Cross-section included: 28, Total panel (balanced) observations: 196

134
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

The result shows that the non-manufacturing sector in African countries responds positively to output
growth in the manufacturing sector. More importantly, the result reinforces the commanding role of the
manufacturing sector as the locus of technological progress because of its more extensive and stronger linkages
with, and spillovers into, the economy. Manufacturing spillovers create an environment for new product and
process technologies ideas resulting in further expansion of both the manufacturing and nonmanufacturing
sectors (Cornwall, 1977; Herzer, 2007). Thus, manufacturing drives economic growth both by its own output
expansion and by the induced output growth in other sectors of the economy. The manufacturing sector drives
GDP growth both by its own output growth and by the induced growth of output in other sectors. Consequently,
there appears a great latitude for manufacturing sector expansion to drive growth in Africa.
Table 2: Fixed effects regression results
Fixed effects (cross-section and period fixed)
Equation 7 Equation 8 Equation 9
(c2) (c4) (c6)
Coefficient 0.31 0.23 1.15
(0.0000) (0.0002) (0.0000)
Constant 0.023 0.034 0.022
R2 0.56 0.32 0.72
DW 1.68 1.61 1.99
Residual cross-section dependence test
Null: No cross-section dependence (correlation) in residuals
Pesaran scaled LM 1.84 2.53 5.88
(0.0656)* (0.0114) (0.0000)
Bias-corrected scaled -0.49 0.19 3.55
LM (0.6225)* (0.8429)* (0.0004)
Pesaran CD -1.76 -1.59 0.74
(0.0788)* (0.1107)* (0.4616)*
Period included :7, Cross-section included: 28, Total panel (balanced) observations: 196
The diagnostics in terms of cross-section heteroscedasticity and combined cross-section and time
heteroscedasticity conducted for the pooled OLS are all satisfactory at 5% when testing the positive relationship
between excess of manufacturing growth over non-manufacturing output growth as a predictor of GDP growth
(equation 5). We therefore accept the null of no heteroscedasticity effect. We cannot accept at 5% the null of no
effect on the test of the positive relationship between manufacturing output growth and GDP growth (equation
4). Also, the OLS regression failed the test of no correlation in residuals. Expectedly, we can accept the null of
no correlation in residuals at 5% for FE regression. The four test statistics largely supports the acceptance of the
null at 5%. We therefore hold our regression coefficient values are valid and reliable to predict the growth rate of
GDP.

135
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

Table 3: System GMM regression and Fagerberg-Verspagen test results

GMM Fagerberg-Verspagen test
Equation 10 Share of m in Difference Wald test P value
(α2) GDP (Ƞ) (α2 - Ƞ) statistic value
Coefficient 0.56 0.1303 0. 43 t-sta. 1.4771 0.1439
(0.0584)* F-sta. 2.1817 0.1439
Chi-sq. 2.1817 0.1397
J-Statistic 0.00 - - - -
(0.9212)*
GMM: Period included 3, Cross-section included: 28, Total panel (balanced) observations: 84. White period
instrument weighting matrix, white period standard errors & covariance (d.f corrected). Instrument rank 11.

We apply the system GMM to test the principal model of the engine of growth hypothesis represented
by equation (1). The system GMM provides a higher value of manufacturing output growth coefficient than the
pooled OLS and FE. A 1% rise in manufacturing output growth increases the growth rate of the GDP by 0.56%
as compared to a maximum 0.31% obtained in pooled OLS and FE. The higher value of (m) under the GMM
estimation compared to OLS and FE implies that the traditional techniques, which do not properly handle the
problem of endogeneity, bias the coefficient estimate of manufacturing value added downwards. This result is
consistent with other studies like Acevedo (2009), Cantore et al (2014), Eguez (2014), and Su and Yao (2016).
In the analysis of the pooled OLS and FE regressions, we found ample support for the claim of
manufacturing as the engine of growth by simply looking at the sign and magnitude of the coefficient of the
regressors. A further test of the hypothesis is the criteria of Fagerberg and Verspagen (1999). The criteria test
whether the coefficient of the manufacturing output growth (m) is positive and significantly higher than the share
of manufacturing output in GDP (Ƞ). Essentially, we test the null hypothesis (α2 = Ƞ) against the alternative (α2 >
Ƞ). The result of the Wald test reported in Table 3 showed the value of each of the test statistic greater than (α2 -
Ƞ) and having significant p-values. There is therefore no reason to accept the null hypothesis of no difference
between α2 and Ƞ. There is thus empirical evidence to suggest that the positive influence of manufacturing on
economic growth in Africa is not due to manufacturing output share in total GDP but principally resulting from
the excess of manufacturing growth rate over non-manufacturing activities.

5 Conclusion
This paper analysed the relation between manufacturing output growth and economic growth in 28 African
countries during 1981-2015 from the perspective of the Kaldor’s first law of economic growth. We analysed the
relation between manufacturing output growth and GDP growth by examining the effects of manufacturing value
added growth on overall economic performance, and the output growth of non-manufacturing sector. The results
presented uphold the “manufacturing as engine of growth” hypothesis, and suggest that the economic growth
process of African countries is in a significant way positively correlated to the growth of the manufacturing
sector. This result agrees with similar studies for low and middle-income developing countries. We therefore
conclude that a prolonged contraction of manufacturing output (de-industrialisation) will be harmful to the

136
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

economic growth of African countries.

The Economic development of African countries has suffered severally because of their vulnerability to
low agriculture terms of trade and resource price volatility. Sustained economic growth in African countries can
result from a process of growth-enhancing structural change. Ocampo (2005) recommends two key mechanisms
to drive this process: a shift toward high-productivity manufacturing, and the creation of new inter-sectoral
linkages that leads to a more intensely integrated production structure. Structural change towards higher
productivity and sophisticated manufacturing activities can lower the vulnerability of African countries to
external shocks and bring benefits from the positive externalities that the manufacturing sector transmits in the
rest of the economy. The task for policy makers is to decide the type of manufacturing activities that will trigger
sustainable economic growth in Africa. Evidence shows that manufacturing activities with higher technology
component generate more opportunity for technological progress, human capital development and productivity
increase, which ultimately contribute positively to a faster growth. However, it is unlikely that developing
countries with limited technological capabilities can initiate an advanced industrialisation process (Egüez, 2014).
Thus, African countries could start by developing basic and labour intensive industries that optimally exploits
the comparative advantage in agriculture, and progressively move up to medium and advanced technology
manufacturing, which demands the inputs of specialised services. With industrialisation as the focus of economic
development, well-articulated and integrated industrial policies, effective macroeconomic management, and
strong commitment to policy implementation will ensure that productivity growth and technological change
originating from the manufacturing sector engender spillover mechanisms that benefit other sectors of the
economy.

Reference
Acevedo, A., Mold, A. and Caldentey, E. (2009). The analysis of leading sectors: A long term view of 18 Latin
American economies’. Paper No. 15017, Munich Personal RePEC Archive (MPRA),
http://mpra.ub.umi-muenchen.de/15017.
Agenor, P., Canuto, O., and Jelenic, M. (2012). Avoiding middle-income growth traps. Economic Premise, 98.
The World Bank. Retrieved from
http://www.siteresources.worldbank.org/EXTPREMNET/Resources/EP98.pdf
Amirapu, A., and Subramanian (2015). Manufacturing or services? An Indian illustration of a development
dilemma. Center for Global Development Working Paper 409. Washington DC: Centre for Global
Development.
Andrews, D. W. K. and Lu, B. (2001). Consistent model and moment selection procedures for GMM estimation
with application to dynamic panel data models. Journal of Econometrics, 101, 123-164
Arellano, M. and Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an
application to employment equations. The Review of Economic Studies (Oxford University Press) 58(2),
277-297.
Arellano, M. and Bover, O. (1995). Another look at the instrumental-variable estimation of error components
models. Journal of Econometrics 68, 29-52.
Bairam, E. (1991). Economic growth and Kaldor’s law: The case of Turkey 1925-1978. Applied Economics 23,

137
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

277-280.
Benchimol, J. (2013). Memento on Eviews output. Available at
http://www.jonathanbenchimol.com/data/teaching/eviews-training/memento.pdf
Blundell, R. and Bond, S. (1998). GMM estimation with persistent panel data: An application to production
functions. Working Paper No.W99/4, The Institution for Fiscal Studies.
Cantore, N., Clara, M., and Soare, C. (2014). Manufacturing as an engine of growth; which is the best fuel?
Working Paper 01/2014. Development Policy, Statistics and Research Branch. UNIDO, Vienna.
Retrieved from http://www.unido.org//fileadmin/user_media/services/PSD/WP/_eBook.pdf
Clark, C. (1941). The conditions of economic progress. London: Macmillan.
Cornwall, J. (1977). Modern Capitalism: Its Growth and Transformation. London: Martin Robertson.
Dasguptan, S. and Singh, A. (2005). Will services be the new engine of Indian economic growth? Development
and Change 36(6), 1035-56.
Dasguptan, S. and Singh, A. (2006). Manufacturing, services and premature deindustrialization in developing
countries: a Kaldorian analysis. UNU-WIDER Research Paper No. 2006/49
Drakopoulos, S.A. and Theodossiou, I. (1991). Kaldor’s approach to Greek economic growth. Applied
Economics 23, 1683-1689.
Egüez, M. E. A. (2014). Manufacturing the only engine of growth? An extension of Kaldor´s first law.
Universitat Wien. Available at repositorio.educacionsuperior.gob.ec/bitstream/28000/.../1/T-
SENESCYT-00632.pdf
Eichengreen, B., Park, D., and Shin, K. (2011). When fast economies slow down: international evidence and
implications for China. Working Paper No.16919, National Bureau of Economic Research, Cambridge,
MA.
Fagerberg, J. and Verspagen, B. (1999). Modern Capitalism' in the 1970s and 1980s. Modern Capitalism in the
1970s and 1980s. Working Paper No.1999002, Centre for Technology, Innovation and Culture,
University of Oslo.
Greenwood, J., Hercowitz, Z., and Krusell, P. (1997). Long-run implication of investment-specific technological
change. American Economic Review 87, 342-362
Herzer, D. (2007). How does trade composition affect productivity? Evidence for Chile. Applied Economics
Letters, 14, 909-912.
Holland, M., Vieira, F., da Silva, C. and Bottecchia, L. (2012). Growth and exchange rate volatility: a panel data
analysis. Applied Economics, 45, 3733-3741
Kaldor, N. (1966). Causes of the slow rate of economic growth of the United Kingdom. An Inaugural Lecture.
Cambridge University Press, Cambridge.
Kaldor, N. (1967). Strategic factors in economic development. Cornell University Press, Ithaca.
Krugman, P. (1994). The myth of Asia’s miracle. Foreign Affairs. November/December, 62-78
Lavopa, A., and Szrmai, A. (2012). Manufacturing growth, manufacturing exports and economic development
1960-2010. 14th ISEE Conference, Brisbane,
http://www.aomevevents.com/media/files/ISS%202012/ISS%20SEssion%209/Lavopa%281%29.pdf
Libanio, G. and Moro, S (2013). Manufacturing industry and economic growth in Latin America: A Kaldorian

138
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

approach. Minas Gerais: Federal University of Minas Gerais.

Nickell, S. (1981). ‘Biasses in dynamic models with fixed effects’. Econometrica, 49, 1417-1426noy
Ocampo. J. A. (Ed.) (2005). Beyond Reforms – Structural Dynamics and Macroeconomic Vulnerability. Stanford
University Press and the World Bank.
Pacheco-Lopez., and Thirlwall, A. P. (2013). A new interpretation of Kaldor’s first law for open developing
economies. ftp://ftp.ukc.ac.uk/pub/ejr/RePEc/ukc/ukcedp/1312.pdf
Rodrik, D. (2009). Growth after the crisis. Harvard Kennedy School, Cambridge, MA.
Roodman, D. (2006). How to do xtabond2: An Introduction to “Difference” and “System” GMM in Stata.
Working Paper No. 103, Center for Global Development.
Sakellaris, P. and Wilson, D. J. (2002). Quantifying embodied technological change. European Central Bank.
Working Paper No. 158.
Stiroh, K. J. (2001). What drives productivity? FRBNY Review, March
Stoneman, P. (1979). Kaldor´s law and British economic growth. Applied Economics 11, 309-319
Su, D. & Yao, Y. (2016). Manufacturing as the key engine of economic growth for middle-income economies.
ADBI Working Paper Series No. 573. Retrieved from
http://www.adb.org/sites/default/files/publication/184350/adbi-wp573.pdf
Szirmai, A. (2011). Industrialisation as an engine of growth in developing countries, 1950-2005. Structural
Change and Economic Dynamics, 23, 406-420. DOI:10.1016/j.strueco.2011.01.005.
Temple, J. and Woessmann. L. (2006). Dualism and cross-country growth regressions. Journal of Economic
Growth 11(3), 187-228
Timmer, M. P. and de Vries, G. J. (2007). ‘Structural change and economic accelerations in Asia and Latin
America: a new sectoral data set’. Cliometrica, 3(2), 165-190.
Timmer, M. P. and Szirmai, A. (2000). Productivity in Asian Manufacturing: the structural bonus hypothesis
examined. Structural Change and Economic Dynamics 11(4), 371-392.
U. S. Department of Commerce (1995). Engines of growth: manufacturing industries in the U. S. economy.
Economics and Statistics Administration Office of Business and Industrial Analysis. Available at
www.esa.doc.gov
Wells, H. and Thirlwall, A. P. (2003). Testing Kaldor’s growth laws across the countries of Africa, African
Development Review, 15 (2/3): 89-105.
Whiteman, J. L. (1987). Productivity and growth in Australian manufacturing industry. Journal of Post
Keynesian Economics 9, 576-592.

Appendix
List of countries used in the analysis covering the period 1981 - 2015
5. Algeria
6. Benin
7. Botswana
8. Burkina Faso
9. Cameroon

139
 Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.7, No.22, 2016

10. Central African Republic

11. Comoros Island
12. Congo D. R.
13. Congo Republic
14. Gabon
15. Gambia
16. Kenya
17. Lesotho
18. Malawi
19. Mauritius
20. Morocco
21. Namibia
22. Nigeria
23. Rwanda
24. Senegal
25. Seychelles
26. South Africa
27. Sudan
28. Swaziland
29. Togo
30. Tunisia
31. Zambia
32. Zimbabwe

140

View publication stats