Infrastructure and Manufacturing Value Added in Sub-Saharan African Countries


 This paper examines the nexus between infrastructure and manufacturing value added (MVA) in sub-Saharan Africa (SSA). It employs panel data for 34 SSA countries spanning 2003 to 2018. The empirical results obtained from the static and dynamic panel estimation techniques applied suggest that infrastructure is essential for the improvement of manufacturing value added in SSA. Furthermore, our findings reveal that the infrastructure-MVA nexus varies by infrastructure types (electricity, transportation, information and communication technology (ICT) and water and sanitation) and across the different sub-regions that make up SSA. This study therefore posits that massive investment in infrastructure is a viable policy option for enhancing the growth and development of the manufacturing sector in SSA.


Introduction
Economic diversification has become a recurring policy objective of successive governments in many sub-Saharan African countries, especially the resource dependent ones. This is due to the increasing realisation that the highly concentrated structure of their production and exports hurt their global competitiveness and raise their vulnerability to the vagaries of the international market. To achieve the objective of economic diversification, it has been largely recognized that the manufacturing sector offers a leeway. Indeed, the economic success of the Asian Tigers is often cited as a testament to this assertion (Gulati, 1992;Asien, 2015).
The manufacturing sector also plays a key role in the industrialization and growth process of any country. This is because the sector offers unique opportunities for capital accumulation, promotes economies of scale by driving technological progress while providing spillover effects through linkages to other economic sectors, displays a higher level of productivity and has more capacity to generate employment compared to other sectors (Efobi and Osabuohien, 2016;Martorano, Sanfilippo and Haraguchi, 2017;Anyanwu, 2018). Again, by fostering productivity and sustainable economic growth, the manufacturing sector can also foster reduction in poverty and inequality (Ndulu, 2006;Lavopa and Szirmai, 2012). Despite the apparent importance of the manufacturing sector, particularly for SSA countries where structural and development indicators are appalling, the performance of the manufacturing sector has to a large extent been abysmal.
Boosting the performance of the manufacturing sector and consequently promoting industrializationwhich are integral aspects of development policy in many SSA countrieswould require the removal of major impediments to manufacturing value added growth. One of such bottlenecks is infrastructure deficit. Extant literature attests to the huge infrastructural gap in SSA compared to other developing regions (Yepes, Pierce and Foster, 2008;Foster and Briceno-Garmendia, 2009;Gutman, Sy andChattopadhyay, 2015, Kodongo andOjah, 2016;World Bank, 2017). Infrastructure is not just an input in the production process; it also complements other factor inputs (Kodongo and Ojah, 2016). Thus, it provides productivity enhancements.
Reliable infrastructure is crucial for powering businesses, lowering transactions costs, improving market access and the efficiency of other productive factors (Luo and Xu, 2018). In particular, energy infrastructure (electricity)the lifeblood of manufacturingis necessary for adding value to raw materials and intermediate products as they are being progressively transformed into final consumer products (Anyanwu, 2018). Transport infrastructure allows for movement of people and manufactured products in a cost efficient manner. Information and communication technology (ICT) aids production and exchange by easing the dissemination of information among economic agents (Ismail and Mahyideen, 2015). In sum, infrastructure can boost both the input and the output process in a production system (Efobi and Osabuohien, 2016) allowing for competitiveness in the production of industrial goods. All of these are germane for enhancing manufacturing value added and overall economic performance.
Within the African context, the bulk of the existing literature has focused on the growth effects of infrastructure and the role of infrastructure in manufacturing value added (MVA) in SSA has scarcely been accounted for. This is however, the kernel of this study. Examining the infrastructure-manufacturing value added nexus is fundamental to the achievement of the Sustainable Development Goal (SDG) 9 which is to build resilient infrastructure, promote inclusive and sustainable industrialization and foster innovation. For a comprehensive understanding of the effects of infrastructure on MVA in SSA, we consider the impact of various infrastructure types such as electricity, transport, ICT and water and sanitation on MVA. For robustness, we conduct the research on the effects of infrastructure on MVA in SSA for each of the sub-regions that make up SSA (East, West, Southern and Central Africa sub-regions). We also deployed both static and dynamic panel estimation techniques to implement our objective.
Our results show that infrastructure is indispensable to the growth of the manufacturing sector in SSA. The significant effects of infrastructure on MVA, however, vary across the regions of SSA and across the infrastructure type considered.
Given this introduction, the rest of the study proceeds as follows. Section 2 presents an overview of manufacturing value added in Africa. A brief review of the literature is documented in section 3 while section 4 contains the model specification and data sources and description. Section 5 presents the empirical results and section 6 contains the conclusion and recommendation.

Overview of Manufacturing Value Added in Africa
Overtime, the performance of the manufacturing sector in SSA has not shown any significant improvement. In fact, it has mostly followed a downward trajectory since the 1990s. For the period 2001-2005it reduced to 10.19% between 2006 and 2010 and further dipped to 9.88% between 2011 and 2015.

Source: Authors using data from World Development Indicators (2018)
The manufacturing sector in SSA also lags behind other developing regions of the world as shown in Figure 2

Literature Review
The seminal paper of Ashauer (1989) underscores the critical role of infrastructure for private sector productivity and output. Specifically, the study which triggered an extensive research on the (macro and micro) economic impact of infrastructure, examined the relationship between public capital and aggregate productivity in the US economy for the period 1949-1985 using a simple production function specification. The empirical findings suggest that the stock of public infrastructure capital is a significant determinant of total factor productivity. However, the economic significance of this result has been largely considered as implausible and not robust to the use of more sophisticated econometric techniques (Aaron, 1990;Tatom, 1991;Baltagi and Pinnoi, 1995). Hulten and Schwab (1991) find that public infrastructure has no effect on total factor productivity (TFP) growth in U.S. manufacturing at the regional level. However, the findings of Rietveld, Kameo, Schipper and Vlaanderen (1994) suggest that infrastructure is a significant variable in explaining the regional differences in the development of manufacturing industry in Central Java, Indonesia. Also, using time-series cross-section data from 1970from to 1993from , Stephan (1997 assessed the impact of road infrastructure on private production. The empirical results show a strong correlation between road infrastructure and output in German manufacturing at the regional level of the Bundeslander. Although over time, the explanatory power of road infrastructure in explaining the observed pattern in TFP growth over time is rather limited. The authors concluded that differences in road infrastructure seem to account for the productivity gap in manufacturing between East and West German regions. Nadiri and Mamuneas (1994) empirically investigates the extent to which public sector infrastructure and R&D capitals influence cost structure and productivity performance using data on twelve (12) two-digit U.S manufacturing industries. Their empirical findings reveal that both types of capital have significant productive effects although their effects on the cost structure vary across industries. Also, adopting a general equilibrium approach, the findings of Holtz-Eakin and Lovely (1995) lend credence to the positive impact of public capital on manufacturing variety proxied by the number of manufacturing establishments.
In the study on the effect of infrastructure on the Indian manufacturing sector, Sharma and Sehgal (2000) utilizing data spanning 1994-2006 find that infrastructure has a strong positive effect on total factor productivity, output and technical efficiency although its effect on labour productivity is significant but negligible. In the same vein, using time series data spanning 1965-1966 and 1998-1999, and estimating a cost function model for India's manufacturing sector, Goel (2002) shows that infrastructure enhances manufacturing sector productivity and lowers costs in India. Paul, Sahni and Biswal (2004) estimate a translog function which incorporates public capital infrastructure for 12 two-digit manufacturing industries. Their results obtained using annual time-series data for 1961-1995, reveal that there is strong evidence that public infrastructure plays an important role in the productivity of Canadian manufacturing industries and it is a substitute for private capital in the Canadian manufacturing sector. Also, employing a constant elasticity and substitution-translog cost model to determine the relationship between Canadian public infrastructure and private output, Brox and Fader (2005) finds similar resultsthe services of public capital enhance the productivity of private capital. Hulten and Isaksson (2007) posit that the relative importance of infrastructure in explaining differences in income and productivity levels depends on the stages of development. Following this, Isaksson (2010) examined whether energy infrastructure matters for cross-country differences in manufacturing levels and industrialization using data for 79 countries for the period 1970 to 2000. Additionally, the sample of countries was divided on the basis of income levels to account for the stages of development. The results suggest that energy infrastructure explains the differing rates of industrialization. Specifically, energy infrastructure is positive and significant across all income groups but its impact is greatest for the poorest economies and fastgrowing Asian tigers. Ahmed (2016) investigates the impact of social infrastructure on the manufacturing productivity using firm level data from Pakistan. Capturing the effect of regional disparities in investment in social goods and controlling for a potential selection bias from firms' location decisions, the study finds that social infrastructure indicatorshealth and educationpositively and significantly affect firm level productivity in Pakistan's manufacturing industries. However, this result only holds for urban districts. For rural regions, both health and education show a negative impact on firm productivity. Also, using data from Chinese manufacturing firms, Wan and Zhang (2017) find that infrastructure has a positive and significant effect on firm productivity.
The empirical findings of Mitra, Sharma and Véganzonès-Varoudakis (2016) from their assessment of the importance of infrastructure and information & communication technology (ICT) on total factor productivity (TFP) and technical efficiency (TE) of the Indian manufacturing sector using data from 1994-2010 indicate that infrastructure and ICT has a significant effect on the manufacturing productive performance, both in terms of total factor productivity (TFP) and technical efficiency. Anyanwu (2018) in his empirical assessment of the role of human capital in manufacturing value added development in Africa show that accessibility to ICT technology and infrastructure proxied by mobile phone and fixed phone subscriptions has no significant effect on manufacturing value added development in Africa. Furthermore, the disaggregated result on education mostly categorized as social infrastructure was mixed. Specifically, the empirical result indicated that primary education has an inverted U-shaped relationship with manufacturing value added; secondary education is negatively and significantly related to manufacturing value added while tertiary education is a significant positive driver of manufacturing value added development in Africa. Furthermore, Abokyi, Appiah-Konadu, Sikayena and Oteng-Abayie (2018) however conclude that electricity consumption has a negative effect on manufacturing output in Ghana. The results they attributed to the fact that while the electricity supply in Ghana may be improving, the share of industrial sector's consumption, on the average, has nosedived continuously.
Overall, while the central role of infrastructure development for accelerating manufacturing sector performance has been acknowledged in the literature, there are findings which suggest that infrastructure at worst has a negative impact or at best, no impact on manufacturing. Noticeably from the highlighted literature, empirical research on the nexus between infrastructure and MVA in SSA are remarkably thin.

Model Specification
The basic empirical model employed to examine the nexus between infrastructure and manufacturing value added is specified as follows: Where yit denotes manufacturing value added for country (i) at time (t). zit represents infrastructure development indicators and Xit are the set of control variables which include credit to private sector, gross fixed capital formation used to represent domestic investment, human capital proxied by primary school enrolment, foreign direct investment, money supply (monetary policy instrument), general government final consumption expenditure (fiscal policy instrument) as well as governance measures such as control of corruption, rule of law, socioeconomic conditions, government stability, democratic accountability and bureaucratic quality. vi is the individual country fixed effect parameter and vit is the error term which is assumed to be normally distributed with zero mean and constant variance.
Endogeneity problems arising from simultaneity may potentially bias the estimation results.
Indeed, it is logical to expect that infrastructure can spur manufacturing value added while the growth and development of manufacturing sector may call for new investment in infrastructural facilities. Although the fixed effects method can obviate the problem of heterogeneity, its usefulness in the presence of endogeneity challenge is limited. Therefore, this study also employs the Generalised Methods of Moment (GMM)a dynamic panel data estimation technique suitable for tackling the endogeneity issue.

Data Sources and Description
This study employs data for 34 SSA countries for the period 2003 to 2018. The data are culled from the African Development Bank (2019)  in SSA stood at 9.86% with minimum value of about 0.23% and maximum value of 35.22%.

Descriptive Statistics
This suggests that manufacturing value added in SSA for the period has been quite low. With regards to infrastructure in SSA, the aggregate infrastructure index (aidi_index) averaged about 19.05, ranging from about 0.37 to 94.32 with standard deviation of about 16.76. The implication of these figures is that infrastructural development in most SSA countries is largely on the same level although there are few countries with some level of progress in the provision of infrastructural facilities. Note: mva_gdp, aidi_index, transp_index, elec_index, ict_index, wss_index, rgdp, pse, man_emp, ggfce_gdp, bm_gdp, gfcf_gdp, fdi_gdp, kofgi and dcps_gdp represent manufacturing value added as a percentage of GDP, aggregate infrastructure index, transport composite index, electricity composite index, ICT composite index, water and sanitation composite index, real GDP, primary school enrolment, manufacturing employment, general government fixed consumption expenditure (% of GDP), broad money supply (% of GDP), gross fixed capital formation (% of GDP), foreign direct investment (% of GDP), KOF globalization index, and domestic credit to private sector (% of GDP) respectively. Also, coc, goveff, ps_abvt, regqual, ruloflaw, voc_acct, govindex represent control of corruption, government effectiveness, political stability and absence of violence/terrorism, regulatory quality, rule of law, voice and governance index respectively. Manufacturing employment, gross fixed capital formation and foreign direct investment are negatively and significantly associated with manufacturing value-added. A quick look at the correlation among the variables shows that there is absence of problem of multicollinearity as the coefficients of correlations among the most of the variables are relatively low.   (2010) Table   4).

Estimation Results
In the case of East Africa as reported in Table 5 Overall, as reported in Table 6 Table 7).
For robustness and to control for endogeneity as previously stated, we further examined the effects of infrastructure on MVA in SSA using the system Generalized Method of Moments (sGMM). We use lags of the dependent variable as well as those of the independent variables and other control variables as instruments. Specifically, we used lags of the dependent variable as an internal instrument while the lags of independent and other control variables are used as external instruments. The reliability of our instruments is determined using Sagan and Hansen's instrument over-identification tests. As shown by these two tests, the instruments employed are valid. Hence, there is no problem of instrument over-identification.
As reported in Table 8, our findings using the sGMM are more robust as we found significant positive effects of infrastructure on MVA in SSA. However, it is important to note that although transport infrastructure and ICT infrastructure have positive effects on MVA, the effects are not statistically significant. This suggests more investment is still required to spur transport and ICT infrastructure in SSA.
Finally, we present the results of the effect of other control variables on MVA using

Conclusion and Recommendation
The importance of the manufacturing sector as an engine of economic growth has been recognised in the literature. Nonetheless, manufacturing value added and output in SSA remains low. Several factors militate against the growth and development of manufacturing sector. One of which is the level and quality of infrastructure. It has been established that the availability of infrastructure facilities such as electricity, transportation, ICT as well as water and sanitation are good for industrial expansion and economic growth. In the light of this, this study investigated the effect of infrastructure development on manufacturing value-added in SSA. Specifically, the study examined the effects of aggregate infrastructure and that of electricity, transport, ICT and water and sanitation on MVA in SSA. To implement our objective, we deployed both static and dynamic panel estimation techniques.
Briefly, our results show that infrastructure development is indispensable to the growth and