Content uploaded by Brice Kamguia
Author content
All content in this area was uploaded by Brice Kamguia on Jul 22, 2023
Content may be subject to copyright.
Received: 29 August 2022
|
Accepted: 1 March 2023
DOI: 10.1111/1467-8268.12684
ORIGINAL ARTICLE
Does infrastructural development foster export upgrading
in Africa?
Brice Kamguia |Manuella Ndjakwa |Sosson Tadadjeu
The Dschang School of Economics and
Management, University of Dschang,
Dschang, Cameroon
Correspondence
Sosson Tadadjeu, The Dschang School of
Economics and Management, University
of Dschang, Dschang, Cameroon.
Email: stadadjeu@yahoo.fr
Abstract
A growing body of literature highlights the importance of export sophistication
for economic development. Given the newness of the literature on export
sophistication, its determinants are under‐exploited. As a pioneer study, this
paper attempts to fill this gap in the literature by examining the effect of
infrastructure development on export sophistication. Based on a panel of 45
African countries over the period 2003–2016, the results of the different
estimations show that infrastructure, including transport, electricity, ICT and
access to water and sanitation, improves export sophistication in Africa. Our
results also show that the effect of infrastructure varies at different intervals of
the export sophistication distribution. Therefore, improved infrastructure
would allow African countries to not only improve their export structure but
also achieve sustainable and durable growth.
KEYWORDS
Africa, development infrastructure, export sophistication, panel data
1|INTRODUCTION
The financial crisis of 2008 and the current health crisis have highlighted the need to identify new drivers of economic
growth. A recent promising strand of the literature suggests that export sophistication is a strong predictor of economic
growth and development (Hausmann et al., 2007; Hidalgo & Hausmann, 2009). The development economics literature
argues that growth in the industrial sector can help less developed countries achieve higher levels of economic growth
(Fan et al., 2018). In the absence of a large domestic market, growth in the industrial sector can also be achieved
through export growth. Weldemicael (2012) argues to this effect that stable and diversified exports promote faster
economic growth. However, export expansion alone, without structural shifts in export baskets, may not lead to
sustainable economic growth. This is particularly true if exports are dominated by primary and/or low‐productivity
commodities. The recent literature highlights the important role that structural transformation of export sectors, or
export diversification and sophistication, play in promoting faster and sustainable economic growth (Ajide, 2020;
Balland et al., 2022; Kamguia et al., 2022).
Export sophistication captures the overall productivity of a country's export basket. Since each product is identified
by a certain level of productivity, a country is considered a sophisticated exporter if its export basket is composed of
more products with higher productivity. There are many debates in the literature about the factors that influence
export sophistication. In classical models, a country's export sophistication is determined by its economic
fundamentals, such as physical and human capital, market size and globalization (Harding & Javorcik, 2012; Nguea
et al., 2022; Zhu & Fu, 2013). Hausmann et al. (2007) found that economic fundamentals account for a relatively small
Afr Dev Rev. 2023;1–16. wileyonlinelibrary.com/journal/afdr © 2023 African Development Bank.
|
1
proportion of the variation in export sophistication. Several authors have confirmed Hausmann et al.'s (2007)
assertions, which argue that export sophistication may also be influenced by certain idiosyncratic factors that induce
firms to engage in cost discovery in an economy. Other studies find that institutional quality promotes export
sophistication (Anand et al., 2012; Weldemicael, 2012).
As noted above, there is a growing body of literature on the determinants of export sophistication. The central
hypothesis of this study is that infrastructure contributes to the production and export of sophisticated products.
However, very little attention has been paid to the link between infrastructure development and export sophistication.
There is an extensive literature on the effects of infrastructure on development outcomes. Several works have examined
the effect of infrastructure on trade. Nordås and Piermartini (2004) examine the effect of infrastructure quality
on countries’trade performance. Their results show that infrastructure quality is an important determinant of
trade performance. Other studies examine the link between infrastructure stock and the level of growth (Agénor &
Moreno‐Dodson, 2006; Canning and Pedroni, 2004; Chakamera & Alagidede, 2018; Diandy & Seck, 2021; Ekeocha
et al., 2021). The main conclusion from these studies is that investment in network infrastructure helps to stimulate
economic growth in the long run. Several related studies show the positive effects of infrastructure on structural change
(Malah & Asongu, 2022), Income diversification (Leng et al., 2020), increase international trade (Lin, 2015),human
development and environmental sustainability (Nchofoung et al., 2022).
This article draws on the two strands of the economic literature discussed above. The first strand concerns the under‐
explored determinants of economic sophistication (Atasoy, 2020; Lectard & Rougier, 2018) and the second part concerns
the extensive literature related to the effects of development infrastructure on the performance of contemporary
economies (Ekeocha et al., 2021; Nchofoung et al., 2022). This article contributes to this double literature in several ways.
First, to the best of our knowledge, this is the first study to empirically examine the effect of development infrastructure
on export sophistication
1
. Moreover, this study uses a broader measure of infrastructure that captures the effects
of different types of infrastructure on export sophistication. Second, most studies on the determinants of export
sophistication suffer from endogeneity and generate biased results because they mainly use ordinary least squares (OLS).
To differentiate this study from previous ones, we use the system generalized method of moments (GMM) estimator,
which allows us to correct for the double causality of the assumed endogenous variables. In addition, in performing the
system GMM estimator, we incorporate Windmeijer finite‐sample correction for standard errors, which produces more
efficient estimators (Windmeijer, 2005). The third contribution is the use of three distinctive measures of export
sophistication in order to check the robustness of the results.
The rest of the document is organized as follows: the next section is a short literature review; Section 3discusses
methodology and data. The empirical results and the discussion are presented in Section 4. Finally, the last section
concludes.
2|ABRIEFSTATEOFTHEART
Very few studies have looked at the effects of infrastructure on export sophistication. The few studies that do exist
examine the effects of different infrastructures such as electricity, ICT or digitalization, and road infrastructure.
Atasoy (2020) discusses the effect of digitalization on three measures of export sophistication and finds that exports become
more sophisticated as digitalization increases. This result is corroborated by Lapatinas (2019), who also shows that internet
access favors exports of sophisticated products. Using open access data collected from the China Labor Dynamics Survey
Project, León et al. (2018) show that ICT adoption exerts a positive and statistically significant impact on diversification. This
result is shared by Gnangnon (2020), who uses a panel dataset containing 131 countries over 1995–2014 and shows that greater
Internet access is positively associated with diversification of service exports. Santoalha et al. (2021) examined, using panel data
on 142 European regions for the period 2006–2013, the effect of workforce skills associated with the use and development of
ICT technologies on green diversification. They found that digital skills endowment is a positive predictor of a region's ability to
specialize in new technology areas, and in particular for green specializations.
Rehman and Sohag (2022) examined the effect of transportation infrastructure on export sophistication and
diversification in G‐20 economies by applying the cross‐sectional autoregressive distributed lag dependence approach.
The empirical results showed that transportation infrastructure boosts export sophistication and diversification in both
the short and long run. Similarly, Shepherd (2010) shows that reducing transport costs improves export diversification
in developing countries. Another branch of the literature analyzes the effect of electricity infrastructure on export
diversification. Odeh and Watts (2019) show that electricity, solar and wind infrastructure improves diversification.
2
|
KAMGUIA ET AL.
As seen above, there is a growing body of literature on the determinants of export sophistication. However, very
little attention has been paid to the nexus between infrastructure development and export sophistication. This study
aims to fill the gap in the literature by analyzing the determinants of export sophistication with a special focus on the
effects of infrastructure development in Africa.
3|DATA AND METHODOLOGY
3.1 |Data
The purpose of this study is to examine the effect of infrastructure on export sophistication in a sample of 45 African
countries over the period 2003–2016.
2
The export data are taken from the BACI database at the six‐digit disaggregation
level of the Harmonized System (HS 92) version 1992 (Gaulier & Zignago, 2010). Infrastructure data are from the
African Development Bank (2018) Africa Infrastructure Development Index (AIDI). Data on control variables
are from the World Development Indicators. Data on institutional variables are from the World Bank, Worldwide
Governance Indicators.
3.1.1 |Dependent variable
The measure of export sophistication (denoted by EXPY) is defined as the average income associated with a
country's total exports. It can be assumed to be a proxy for the productivity of exports in a country. The idea is that
the products exported by richer countries have characteristics, such as high technology, that allow high‐wage
producers to compete in world markets (Lall et al., 2006). Therefore, the sophistication of a product is related to
the income of the countries that export the product. Thus, when a developing country manages to export products
similar to those of developed countries, its EXPY score increases (Weldemicael, 2012).
3
To reduce data skewness,
we take the natural logarithm of the main variables. For EXPY, its log transformation aims to interpret the
coefficients as elasticities (Saadi, 2020).
3.1.2 |Variable of interest
Our variable of interest is infrastructure development. Following the studies of Nchofoung et al. (2022) we use the
infrastructure development variables developed by the African Development Bank. These include the AIDI, which is
composed of four sub‐indicators: the Transport Infrastructure Composite Index (Transport), the Electricity
Infrastructure Composite Index (Electricity), the Information and Communication Technology Infrastructure
Composite Index (ICT) and the Water and Sanitation Infrastructure Composite Index (WSS). Figure 1, representing
the correlation between the EXPY and the different infrastructure variables, shows an ascending line that indicates a
positive relationship between these variables.
3.1.3 |Control variables
To guard against potential omitted variable bias, we consider more macroeconomic, environmental and institutional
controls. Specifically, for our baseline model, we consider foreign direct investment (FDI), education, institutional
quality, remittances, country size, per capita income and industrialization. These variables are chosen in accordance
with the recent related literature (Atasoy, 2020;Avometal.,2022; Gnangnon, 2020; Kamguia et al., 2022; Kenec‐k‐
Massil and Nvu‐h‐Njoya, 2021;Saadi,2020). Table 1presents the descriptive statistics for the variables used in this
study. All variables are used in logarithm (except for control of corruption) in order to facilitate interpretations of the
estimated coefficients in terms of elasticity.
The literature provides several channels through which FDI is likely to affect export sophistication. Indeed,
FDI by facilitating the transfer of knowledge, technology and managerial skills, can promote the export of more
sophisticated goods and services (Saadi, 2020). Education index is an average of mean years of schooling (of adults) and
KAMGUIA ET AL.
|
3
expected years of schooling (of children). Studies show that the skill level of the workforce and educational efforts are
expected to have a significant impact on each country's ability to improve their exports (Anand et al., 2012). In addition,
Vu (2022) shows the importance of institutions as a fundamental determinant of export sophistication. Thus, we also
expect a positive effect of control of corruption on export sophistication. Saadi (2020) highlights the role of remittances
in increasing the complexity of exports in developing countries. Thus, we also expect a positive effect of remittances on
export sophistication. We approximate country size by population density. Work shows that the size of the economy is
likely to influence the production and export of complex products because of the labor force and the demand it
represents (Lapatinas, 2019). It is accepted in the literature that the level of economic development positively
influences a country's ability to produce and export sophisticated goods (Hausmann et al., 2007). Therefore, we expect a
positive sign of income on export sophistication. Finally, industrialization is defined as a shift in employment from the
traditional sector to the advanced sector. Productivity gains associated with structural change explain the estimated
high return on investment (Opoku & Boachie, 2020). For robustness, we also include four additional variables, namely
natural resources, CO
2
emissions, financial development and urbanization.
3.2 |Methodology
The aim of this study is to assess the effect of development infrastructure on export sophistication in Africa. We assume
that infrastructure has a positive effect on economic sophistication. We apply two main empirical estimation
techniques to test this hypothesis. First, we implement an OLS regression of infrastructure on economic sophistication,
controlling for several determinants selected from the literature. Second, we perform a two‐step system GMM
specification to deal with potential endogeneity.
FIGURE 1 Relationship between export sophistication and infrastructure. [Color figure can be viewed at wileyonlinelibrary.com]
4
|
KAMGUIA ET AL.
First, we begin by applying the OLS method to estimate Equation (1) described as follows:
E
XPY αβInfrast βXε=+ + +
it it it it,1,2,, (1)
E
XPYit,is the export sophistication for country Iin period t.
Infrast
it,
is the development infrastructure. Xit,is the vector
of baseline control variables and εit,is the error term.
Although the pooled OLS is interesting, it does not control for some unobserved differences that can bias the parameter
estimates. Beyond the problems of unobserved heterogeneity and the failure of estimating time‐invariant factors, the fact
remains that OLS does not control for potential endogeneity problems that can arise from at least three sources. The first is
reverse causality: infrastructure may be endogenous, and so there is more likely to be a feedback effect from export
sophistication to infrastructure. The second is measurement error: measures of infrastructure or sophistication are more
likely to have measurement errors, especially in African countries. The third is omitted variables: there are important
variables (e.g. geographic, cultural or historical factors) that may be omitted from the regression models but are considered
crucial determinants of export sophistication and are correlated with some of the explanatory variables.
The common approach in the literature to deal the endogeneity is to use an instrumental variable approach (IV) or
a GMM. The IV using external instruments has been used to effectively solve the reverse causality problem. However, a
limitation of this approach is the difficulty of finding a purely exogenous external instrument that varies across
countries and over time, and this method also tends to ignore the endogeneity of other regressors. In addition, owing to
the difficulty of finding ideal instruments for our various infrastructure variables, we do not use an IV approach. In this
paper, the GMM approach is preferred. The GMM has the advantage of dealing with the endogeneity of all explanatory
variables by using internal instruments (Njangang et al., 2020). Therefore, we formulate the following dynamic model:
E
XPY αβEXPY βInfrast βXμγε=+ + + + ++
it it it it itit,1,−12,2,, (2)
where
E
XPYit,−
1
is the lagged value of
E
XPYit,,
μ
i
is the unobserved country‐specific effect and γ
t
is a time‐specific effect.
EXPY has also been treated with a lagged structure by Weldemicael (2012) and Lectard and Rougier (2018). A country's
TABLE 1 Descriptive statistics
Variables Observations Mean Standard deviation Minimum Maximum
EXPY 495 8.778 0.47 7.006 9.709
AIDI 630 2.678 0.816 −0.997 4.543
Transport 630 1.826 1.040 −0.968 4.073
Electricity 629 1.046 1.752 −4.51 4.605
ICT 614 −1.021 3.029 −6.908 4.191
WSS 628 3.783 0.49 1.799 4.595
FDI 599 0.984 1.329 −6.280 4.638
Education 627 −0.897 0.328 −2.071 −0.316
Control of corruption 630 −0.670 0.543 −1.887 0.853
Remittances 562 0.303 1.775 −8.605 3.082
Population density 630 3.853 1.196 1.020 6.434
Urbanization 630 15.086 1.534 10.653 18.321
Domestic credit 614 2.738 0.877 −0.303 5.076
Income 617 7.087 1.022 5.272 9.518
Industry 606 2.824 0.848 0.138 4.279
Natural resources 630 1.907 1.618 −6.749 4.218
CO
2
emissions 620 −1.171 0.6338 −2.936 0.670
Note: All the variables are in a natural log form (except for control of corruption).
KAMGUIA ET AL.
|
5
past export sophistication is likely to have an impact on its current export sophistication. Introducing lagged export
sophistication as an explanatory variable invalidates the standard static panel regression, due to “dynamic panel bias”
(Nickell, 1981). To address this dynamic panel bias, we use the system GMM estimator developed by Arellano and
Bover (1995) and Blundell and Bond (1998), which is an augmented version of the difference GMM. The use of system
GMM allows us to address the endogeneity problems of the right‐hand side variables. One attempt to mitigate
endogeneity is to use the lags of the right‐hand side variables as instruments. Specifically, all variables assumed to be
endogenous are instrumented by lags of order 2–5.
There are two main tests associated with the GMM: the over‐identification test (Hansen's test), which tests the
validity of the instruments used, in the sense that they must be correlated with the instrumented variables and not with
the error term; and the Arellano and Bond AR (2) error autocorrelation test, which tests the first‐order serial
correlation of the residuals in level, by testing the second‐order serial correlation of the errors in difference, since the
error terms expressed in first difference are correlated to the first order, owing to the construction of the system GMM
estimator. Thus, the validity of the system GMM estimator is conditioned by the quality of the instruments chosen
(Hansen test), as well as by the second‐order non‐autocorrelation of the errors in the difference equation AR(2).
4|RESULTS AND DISCUSSION
4.1 |Baseline results
Table 2presents the estimation results obtained from OLS. Column 1 shows the results when considering
the overall infrastructure development index. The coefficient associated with this variable is positive and
statistically significant with a magnitude, suggesting that an increase in the infrastructure index of 10% leads to an
improvement in sophistication of 1.56% points on average. This result can be justified by the fact that, by improving
income levels and human development (African Development Bank, 2018; Nchofoung et al., 2022), infrastructures
contribute to improve export sophistication. With respect to the control variables, the results show that FDI,
education, control of corruption, proxy for institutional quality, country size, income and industrialization have a
positive and statistically significant effect on export sophistication. This result is consistent with work that
examines the determinants of economic sophistication (Khan et al., 2020; Lectard & Rougier, 2018; Saadi, 2020;
Weldemicael, 2012;Zhu&Fu,2013).
Column 2 presents the results when we consider transport infrastructure. Thus, we find that the coefficient
associated with this variable is positive and statistically significant, with a magnitude suggesting that a 10% increase in
transport infrastructure leads to an improvement in economic sophistication of 0.39 percentage points on average. In
column 3, the results show that the coefficient associated with electricity infrastructure has a positive and statistically
significant effect. Thus, a 10% increase in electricity infrastructure leads to an improvement in sophistication of about
1.027% points. Theoretically, these results can be justified by the beneficial effects of transportation and energy
infrastructure on human capital formation, capital productivity and market access (Banerjee et al., 2020,2021). Thus,
by improving human capital, for example (Banerjee et al., 2021), energy infrastructure improves export sophistication.
Column 4 presents the results when we consider ICT infrastructure. It can be seen that the coefficient associated with
this variable is positive and statistically significant. This result implies that a 10% increase in ICT infrastructure
improves export sophistication by 0.175% on average. This result is consistent with those obtained by Lapatinas (2019)
and Atasoy (2020) who examine the effects of ICT on export sophistication. In the last column of the table, we consider
water and sanitation infrastructure. The results show that the coefficient associated with this variable is positive and
statistically significant with a magnitude suggesting that a 10% increase in water and sanitation infrastructure leads to
an improvement in sophistication of about 0.82%.
Although the OLS estimation provides results that allow us to assert that public infrastructure has a positive
and significant effect on the sophistication of economies, it is nevertheless true that these results suffer from
certain limitations, the most important of which is the failure to take endogeneity into account. To correct this bias, we
re‐estimated our model using the two‐step system GMM.
Table 3presents the results of the estimation of the effect of development infrastructure on export sophistication
with the system GMM. The lower part of this table reports the number of instruments used as well as the results of the
Hansen over‐identification and Arellano and Bond autocorrelation tests. Thus, the validity of the system GMM
estimator is conditioned by the quality of the instruments chosen (Hansen test), as well as the non‐second‐order
6
|
KAMGUIA ET AL.
autocorrelation of the errors in the difference equation AR(2). In all our specifications, the value of the Hansen test
(p‐values) is higher than the 10% level, which indicates that the null hypothesis of non‐correlation of the instrumental
variables with the error terms is verified. Therefore, the instruments used appear valid in practice, and the
GMM estimator is in a convergent system. This result is also supported by the acceptance of the null hypothesis of no
second‐order error autocorrelation, which can be inferred from the AR(2) test values (p‐values), which are above the
10% threshold in all our regressions.
TABLE 2 Infrastructure development and export sophistication (OLS‐FE)
Dependent variable: Export sophistication
Variables (1) (2) (3) (4) (5)
AIDI 0.1565***
(0.0480)
Transport 0.0395***
(0.0299)
Electricity 0.1027***
(0.0204)
ICT 0.1758***
(0.0319)
WSS 0.0821**
(0.0066)
Foreign direct investment 0.0522*** 0.0393*** 0.0297** 0.0413*** 0.0472***
(0.0122) (0.0120) (0.0118) (0.0117) (0.0123)
Education 0.3651*** 0.4698*** 0.2536** 0.3393*** 0.4229***
(0.0922) (0.0967) (0.0985) (0.0892) (0.0916)
Institutions 0.1489*** 0.1632*** 0.1577*** 0.1321*** 0.1556***
(0.0340) (0.0329) (0.0339) (0.0340) (0.0340)
Remittances 0.0163 0.0384** 0.0356** −0.0094 0.0262
(0.0185) (0.0170) (0.0164) (0.0182) (0.0180)
Population density 0.0711*** 0.0314*0.0399** 0.0614*** 0.0550***
(0.0175) (0.0172) (0.0159) (0.0161) (0.0170)
Income 0.1719*** 0.2211*** 0.1699*** 0.1044** 0.2065***
(0.0504) (0.0436) (0.0451) (0.0494) (0.0481)
Industry 0.1262** 0.0608 0.1493*** 0.1172** 0.0948*
(0.0509) (0.0567) (0.0508) (0.0468) (0.0519)
Constant 7.0864*** 6.9954*** 7.0697*** 8.6240*** 6.9954***
(0.4706) (0.4655) (0.4783) (0.5610) (0.4655)
Observations 448 448 447 448 448
R
2
0.4063 0.4911 0.4316 0.4417 0.4936
Time FE Yes Yes Yes Yes Yes
Note: Robust standard errors in parentheses; AIDI is the composite infrastructure development index; ICT is the information and communication technology
infrastructures; WSS is water and sanitation infrastructures; and Institutions is measured by the control of corruption index.
***p< .01.
**p< .05.
*p< .1.
KAMGUIA ET AL.
|
7
TABLE 3 Infrastructure development and export sophistication (GMM system)
Dependent variable: Export sophistication
Variables (1) (2) (3) (4) (5)
Lag dependent variable 0.2126*** 0.5403*** 0.5215*** 0.3177*** 0.4218***
(0.0469) (0.0149) (0.0181) (0.0500) (0.0445)
AIDI 0.2255***
(0.0542)
Transport 0.0770**
(0.0349)
Electricity 0.0875***
(0.0199)
ICT 0.0202***
(0.0050)
WSS 0.3521***
(0.0871)
Foreign direct investment 0.0938*** 0.0169** 0.0075 0.0617*** 0.0536***
(0.0081) (0.0069) (0.0088) (0.0117) (0.0074)
Education 0.1620 0.3248*** 0.0340 0.3515*** 0.1303
(0.2216) (0.0517) (0.0989) (0.1205) (0.1624)
Institutions 0.3633*** 0.1320** 0.3578*** 0.2837*** 0.2861***
(0.0676) (0.0496) (0.0696) (0.0414) (0.0725)
Remittances 0.0831*** 0.1000*** 0.0413** 0.0159 −0.0006
(0.0286) (0.0308) (0.0157) (0.0446) (0.0345)
Population density 0.1641 0.2327*** −0.1138 0.2142** 0.2302***
(0.0850) (0.0303) (0.0512) (0.0800) (0.0767)
Income 0.0248 0.0102 −0.0160 −0.0162 −0.0647
(0.0601) (0.0284) (0.0382) (0.0498) (0.0744)
Industry 0.1004*0.0393 0.0179 −0.0225 −0.0239
(0.0518) (0.0419) (0.0172) (0.0571) (0.0546)
Constant 5.5991*** 4.8497*** 4.8685*** 7.4317*** 3.6073***
(1.0378) (0.3064) (0.5781) (1.0590) (1.0899)
Time FE Yes Yes Yes Yes Yes
Observations 408 408 407 408 408
Number of countries 44 44 44 44 44
Instruments 42 43 43 38 42
AR (1) 0.016 0.007 0.009 0.013 0.007
AR (2) 0.723 0.455 0.257 0.941 0.652
Hansen 0.218 0.248 0.264 0.127 0.260
Note: Robust standard errors in parentheses.
***p< .01.
**p< .05.
*p< .1.
8
|
KAMGUIA ET AL.
As regards the results themselves, we find that irrespective of the infrastructure indicators selected, the coefficient
associated with the infrastructure variable is positive and statistically significant at the 1% level. This result indicates
the importance of infrastructure in the sophistication process of economies. This result is consistent with those
obtained in Table 2. When we look at the lagged dependent variable, we find that the coefficient associated with this
variable is positive and highly significant; this indicates the path dependence of export sophistication. Neglecting this
lagged dependent variable will worsen the effect of the other variables (Kočenda & Poghosyan, 2018).
4.2 |Robustness checks
To test the robustness of our baseline results, in this subsection, we perform two main sensitivity analyses. First, we
include additional control variables; second, we use two alternative measures of export sophistication.
We re‐estimate our model by introducing four additional control variables, namely, natural resources, CO
2
emissions, financial development and urbanization. The results of this estimation are contained in Table 4. From this
table, we see that regardless of the indicator used, infrastructure has a positive and statistically significant effect on
export sophistication. This result confirms those obtained in Tables 2and 3. In other words, we show that the effect of
infrastructure on export sophistication is robust to the introduction of additional control variables. With respect to the
additional control variables, we show that urbanization and financial development have a positive effect on economic
sophistication. Natural resources and CO
2
emissions have a negative effect. This result is consistent with the related
literature (Chu, 2020; Avom et al., 2022; Tadadjeu et al., 2022).
Second, we test the robustness of our results using two alternative dependent variables, namely the economic
complexity index and economic fitness.
4
In recent years, several studies have developed different indices to measure a
country's export sophistication. Each of these indices has its advantages and disadvantages (Zhu & Fu, 2013). Increases
in this index over time reflect to some extent the upgrading of a country's exports. The economic complexity index is
estimated from data linking countries to the products they export and measures the diversity and sophistication of a
country's export structure, adjusted for the difficulty of exporting each product (Balland et al., 2022). Taking into
account both a country's competitiveness and the complexity of a product, Tacchella et al. (2012) created the capability
index and argue that this index better explains the level of diversification of countries.
The results are contained in Table 5. Columns 1–5 present the results when considering the economic complexity
index as a measure of export sophistication. The results show that the use of these indicators does not change our
results. In other words, we find that infrastructure has a positive and statistically significant effect on economic
complexity. Columns 6–10 present the results of the effect of development infrastructure on economic fitness. From
this table, we see that the coefficients associated with the infrastructure variables are positive and statistically
significant, suggesting that an increase in infrastructure leads to an improvement in economic fitness. The results in
Table 5show that the effect of infrastructure on export sophistication is robust to the use of alternative dependent
variables.
4.3 |Further robustness checks: a non‐parametric approach
The results so far confirm a positive relationship between development infrastructure and export sophistication.
However, as suggested by Elburz et al. (2017), the effects of infrastructure on growth could vary across countries and
sectors and over time, and depending on the econometric method used. In this study, we came to the conclusion of a
positive effect between our variables using a parametric econometric approach based on OLS and GMM regressions.
However, the relevance of our analyses could be questioned owing to the fact that the OLS and GMM estimators focus
only on the average effect, and do not consider the effect that our infrastructure indicators could have on different
intervals of export sophistication.
To overcome this difficulty, the most widely used approach in the literature is quantile regression (QR). First
introduced by Koenker and Bassett (1978), QR is a non‐parametric approach widely used in the literature to take into
account the effect of one variable on another at different levels of its distribution. Thus, one of the contributions of this
study is to go beyond the approach that focuses on modeling the average effect of infrastructure on export
sophistication, but analyzes the effects of infrastructure at different intervals of the sophistication distribution. This
approach may be superior to the OLS approach at different points. “For example, unlike OLS, which can be inefficient
KAMGUIA ET AL.
|
9
TABLE 4 Results with additional control (system GMM)
Dependent variable: Export sophistication
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Lag dependent variables 0.2637*** 0.4128*** 0.5865*** 0.2386*** 0.5159*** 0.4789*** 0.2943*** 0.5553*** 0.3950*** 0.3207***
(0.0630) (0.0387) (0.0226) (0.0739) (0.0194) (0.0241) (0.0510) (0.0385) (0.0645) (0.0698)
AIDI 0.2714*** 0.2738***
(0.0908) (0.0934)
Transport 0.2231*** 0.4518***
(0.0624) (0.0843)
Electricity 0.1015*** 0.1704***
(0.0184) (0.0315)
ICT 0.0219*** 0.0184**
(0.0068) (0.0076)
WSS 0.7372*** 0.3950**
(0.1545) (0.1775)
Baseline controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Natural resources −0.113*** −0.0588** −0.0118 −0.2127*** −0.257***
(0.0395) (0.0225) (0.0159) (0.0325) (0.0353)
CO
2
emissions −0.642*** −0.1823** −0.1780*** −0.3801*** −0.3633*
(0.2047) (0.0823) (0.0616) (0.1400) (0.1927)
Financial development 0.0757*0.1163 0.0088 −0.0627 0.0822
(0.0440) (0.1117) (0.0382) (0.0384) (0.0663)
Urbanization 0.3048*** 0.4605*** 0.0243 0.1179** 0.1019
(0.1041) (0.1127) (0.0609) (0.0581) (0.1281)
Constant 3.8644*** 12.4202*** 4.8914*** 16.6707*** 3.7531*** 4.3847*** 7.3938*** 6.8241*** 5.6225*** 6.7509***
(1.2252) (1.8364) (0.5951) (2.2645) (0.5797) (1.1736) (0.8975) (1.1712) (1.5389) (1.9065)
Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 408 398 408 398 407 397 408 398 408 398
10
|
KAMGUIA ET AL.
TABLE 4 (Continued)
Dependent variable: Export sophistication
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Number of countries 44 44 44 44 44 44 44 44 44 44
Instruments 42 43 43 38 43 43 42 43 42 42
AR (1) 0.014 0.012 0.010 0.017 0.009 0.008 0.011 0.011 0.009 0.015
AR (2) 0.186 0.610 0.412 0.499 0.339 0.313 0.367 0.376 0.553 0.731
Hansen 0.234 0.335 0.161 0.315 0.282 0.230 0.213 0.205 0.222 0.181
Note: Robust standard errors in parentheses.
***p< .01.
**p< .05.
*p<.1
KAMGUIA ET AL.
|
11
if errors are highly non‐normal, QR is more robust to errors than OLS. QR is more robust to non‐normal errors and
outliers. Similarly, when the distribution of the dependent variable is broad, the mean can be highly variable in the
presence of strong heterogeneity in the sample”(Mignamissi & Malah Kuete, 2021, p. 7). This can be seen in Figure 2.
Thus, QRs address these limitations and provide a more accurate description of the distribution of a variable of interest
conditional on its determinants as opposed to a simple linear regression that focuses on the conditional mean.
The quantile estimator is obtained by solving the following optimization problem for the θth quantile (0 < θ< 1):
∈
∈≥ ∈≥
θyβxθyβxmin −+(1−)−
βR
iiyβx
ii
iiyβx
ii
{: } {: }
K
iiii
where yiis the export sophistication of country i,
β
is the vector of parameters to be estimated and
x
i
is the vector of
explanatory variables.
The results of quantile estimation compared with OLS are presented in Table 6. Column 1 presents the results of the
OLS estimation while columns 2–6 present estimates for the 10th, 25th, 50th, 75th and 95th quantiles. We observe that
the positive effect of development infrastructure varies throughout the distribution of export sophistication.
TABLE 5 Results with alternative dependent variables (system GMM)
Dependent variable: Economic complexity Dependent variable: Economic Fitness
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Lag dependent variables 0.4678*** 0.4872*** 0.8272*** 0.5023*** 0.659*** 0.7731*** 0.6849*** 0.8939*** 0.8738*** 0.8171***
(0.1072) (0.1100) (0.1894) (0.0762) (0.097) (0.1042) (0.0487) (0.0544) (0.0584) (0.0518)
AIDI 0.8771*** 0.2129***
(0.3143) (0.0573)
Transport 0.4813*** 0.1616***
(0.2509) (0.0510)
Electricity 0.4523*** 0.0439**
(0.1445) (0.0233)
ICT 0.0348*** 0.0084**
(0.0124) (0.0034)
WSS 0.781*** 0.1387**
(0.278) (0.0580)
Constant 13.7643*** −6.3450 7.3211 6.4855*** 3.8922 −2.7856** −3.363*** −1.4113** −0.4108 −1.904***
(4.9032) (4.3155) (5.1598) (1.4303) (2.6994) (1.1365) (0.9117) (0.6420) (1.0680) (0.7340)
Baseline controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 300 300 300 297 300 426 426 425 421 426
Number of countries 27 27 27 27 27 40 40 40 40 40
Instrument 24 24 21 26 26 25 30 30 24 27
AR(1) 0.070 0.002 0.002 0.000 0.000 0.006 0.003 0.004 0.001 0.001
AR(2) 0.263 0.442 0.274 0.205 0.197 0.786 0.939 0.890 0.938 0.773
Hansen 0.315 0.428 0.986 0.174 0.322 0.275 0.444 0.422 0.533 0.385
Note: Robust standard errors in parentheses.
***p< .01.
**p< .05.
*p< .1.
12
|
KAMGUIA ET AL.
TABLE 6 OLS vs. quantiles regression
Dependent variable: Export sophistication
OLS Q10 Q25 Q50 Q75 Q95
Variables (1) (2) (3) (4) (5) (6)
AIDI 0.1565*** 0.3441*** 0.2089** 0.1649*** 0.2140** 0.0257
(0.0480) (0.0795) (0.0810) (0.0614) (0.0937) (0.0773)
Baseline controls Yes Yes Yes Yes Yes Yes
Observations 448 448 448 448 448 448
Transport 0.0395*** 0.1134*−0.0020 0.1048*−0.0477 −0.0226
(0.0299) (0.0658) (0.0500) (0.0600) (0.0337) (0.0286)
Baseline controls Yes Yes Yes Yes Yes Yes
Observations 448 448 448 448 448 448
Electricity 0.1027*** 0.1902*** 0.1228*** 0.0928*** 0.0760*** 0.0168
(0.0204) (0.0392) (0.0286) (0.0288) (0.0264) (0.0255)
Baseline controls Yes Yes Yes Yes Yes Yes
Observations 447 447 447 447 447 447
ICT 0.1758*** 0.2912*** 0.1885*** 0.1816*** 0.1853*** 0.0443
(0.0319) (0.0309) (0.0400) (0.0426) (0.0662) (0.0562)
Baseline controls Yes Yes Yes Yes Yes Yes
Observations 448 448 448 448 448 448
WSS 0.0821** 0.3170** −0.0122 0.1329*0.1362 0.0305
(0.0066) (0.1571) (0.0860) (0.0779) (0.1182) (0.1033)
Baseline controls Yes Yes Yes Yes Yes Yes
Observations 448 448 448 448 448 448
Note: Robust standard errors in parentheses.
***p< .01.
**p< .05.
*p< .1.
FIGURE 2 Distribution of export sophistication index at different quantiles. [Color figure can be viewed at wileyonlinelibrary.com]
KAMGUIA ET AL.
|
13
Specifically, the results show that the positive effect of transport infrastructure on export sophistication is significant up
to the 50th quantiles. In other words, we find that the positive and significant effect of transport infrastructure on
export sophistication fades away when sophistication is high. For electricity and ICT infrastructure, the results show
that the effect is positive and statistically significant regardless of the quantile. In other words, the positive effect of
electricity and ICT infrastructure on sophistication is significant regardless of the distribution of export sophistication.
Finally, our results show that access to water and sanitation infrastructure does not have a statistically significant effect
at low levels of sophistication. This effect becomes significant from the 25th to the 75th quantiles.
5|CONCLUSION
The aim of this study was to analyze the effect of infrastructures on export sophistication in a sample of 45 African
countries. Using panel data covering the period 2003–2016 and the two‐step system GMM, we show that infrastructure
has, on average, a positive and significant effect on export sophistication. Moreover, the results show that different
types of infrastructure, including transportation, electricity, access to clean water and sanitation, and ICT, improve
export sophistication. Additional quantile analysis also confirms that the effect of different types of infrastructure
remains positive and heterogeneous.
Infrastructure has always been considered essential for development, notably through the improvement of human
capital and trade facilitation, among others. This study provides new results, suggesting that improved infrastructure
promotes export sophistication. Several recommendations are made based on these results. First, although some
countries have made significant progress in covering infrastructure related to the Sustainable Development Goals, such
as access to safe water and sanitation, and access to electricity, many African countries are lagging behind. Thus, we
suggest that African governments increase the budgets allocated to infrastructure. Second, we encourage government
initiatives to improve the efficiency of public investment in infrastructure. Finally, private and public initiatives to
promote rural electrification, water supply and sanitation, ICT development, and transportation infrastructure in all
regions of each country are strongly encouraged to improve export sophistication.
ORCID
Brice Kamguia http://orcid.org/0000-0001-8664-6600
Sosson Tadadjeu http://orcid.org/0000-0002-5451-0001
ENDNOTES
1
With the exception of the works of Lapatinas (2019) and Atasoy (2020), which examine only the effect of ICT on export sophistication.
2
The choice of this sample and time period is primarily dictated by data availability. Indeed, data on export sophistication are available for
the period 1995–2016. However, data on infrastructure are available for the period 2003–2018. Therefore, the period 2003–2016 appears to
be the best.
3
See Hausmann et al. (2007) for more details on the construction of this variable.
4
Data on economic complexity are available for 27 African countries while data on economic fitness are available for 40 countries.
REFERENCES
African Development Bank. (2018). The Africa infrastructure development index. Statistics Department. http://afdb.org
Agénor, P. R., & Moreno‐Dodson, B. (2006). Public infrastructure and growth: New channels and policy implications (Vol. 4064). World Bank
Publications.
Ajide, F. M. (2020). Infrastructure and entrepreneurship: Evidence from Africa. Journal of Developmental Entrepreneurship,25(03),
2050015.
Anand, R., Mishra, S., & Spatafora, N. (2012). Structural transformation and the sophistication of production (IMF Working Paper 12/59).
International Monetary Fund.
Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error‐components models. Journal of Econometrics,
68(1), 29–51.
Atasoy, B. S. (2020). The determinants of export sophistication: Does digitalization matter? International Journal of Finance & Economics,26,
5135–5159.
Avom, D., Keneck‐Massil, J., Njangang, H., & Nvuh‐Njoya, Y. (2022). Why are some resource‐rich countries more sophisticated than others?
The role of the regime type and political ideology. Resources Policy,79, 103067.
14
|
KAMGUIA ET AL.
Balland, P. A., Broekel, T., Diodato, D., Giuliani, E., Hausmann, R., O'Clery, N., & Rigby, D. (2022). The new paradigm of economic
complexity. Research Policy,51(3), 104450.
Banerjee, A., Duflo, E., & Qian, N. (2020). On the road: Access to transportation infrastructure and economic growth in China. Journal of
Development Economics,145, 102442.
Banerjee, R., Mishra, V., & Maruta, A. A. (2021). Energy poverty, health and education outcomes: Evidence from the developing world.
Energy Economics,101, 105447.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics,87(1),
115–143.
Canning, D., & Pedroni, P. (2004). The effect of infrastructure on long run economic growth. Harvard University,99(9), 1–30.
Chakamera, C., & Alagidede, P. (2018). The nexus between infrastructure (quantity and quality) and economic growth in Sub Saharan
Africa. International Review of Applied Economics,32(5), 641–672.
Chu, L. K. (2020). The effects of financial development on economic sophistication: Evidence from panel data. Applied economics letters,
27(15), 1260–1263.
Diandy, I. Y., & Seck, A. B. (2021). Infrastructures physiques et croissance économique en Afrique de l'Ouest: Le rôle des institutions.
African Development Review,33(1), 154–165.
Ekeocha, D. O., Ogbuabor, J. E., & Orji, A. (2021). Public infrastructural development and economic performance in Africa: A new evidence
from panel data analysis. Economic Change and Restructuring,55, 931–950.
Elburz, Z., Nijkamp, P., & Pels, E. (2017). Public infrastructure and regional growth: Lessons from meta‐analysis. Journal of Transport
Geography,58,1–8.
Fan, Z., Anwar, S., & Huang, S. (2018). Cultural diversity and export sophistication. International Review of Economics & Finance,58,
508–522.
Gaulier, G., & Zignago, S. (2010). Baci: International trade database at the product‐level (the 1994‐2007 version).
Gnangnon, S. K. (2020). Effect of the internet on services export diversification. Journal of Economic Integration,35(3), 519–558.
Harding, T., & Javorcik, B. S. (2012). Foreign direct investment and export upgrading. Review of Economics and Statistics,94(4), 964–980.
Hausmann, R., Hwang, J., & Rodrik, D. (2007). What you export matters. Journal of Economic Growth,12(1), 1–25.
Hidalgo, C. A., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences,
106(26), 10570–10575.
Kamguia, B., Tadadjeu, S., Miamo, C., & Njangang, H. (2022). Does foreign aid impede economic complexity in developing countries?
International Economics,169,71–88.
Kenec‐k‐Massil, J., & Nvu‐h‐Njoya, Y. (2021). Did colonisation matter for comparative economic complexity? Economics Letters,203, 109851.
Khan, H., Khan, U., & Khan, M. A. (2020). Causal nexus between economic complexity and FDI: Empirical evidence from time series
analysis. The Chinese Economy,53(5), 374–394.
Koenker, R., & Bassett, G., Jr. (1978). Regression quantiles. Econometrica,46,33–50.
Kočenda, E., & Poghosyan, K. (2018). Export sophistication: A dynamic panel data approach. Emerging Markets Finance and Trade,54(12),
2799–2814.
Lall, S., Weiss, J., & Zhang, J. (2006). The “sophistication”of exports: A new trade measure. World Development,34(2), 222–237.
Lapatinas, A. (2019). The effect of the Internet on economic sophistication: An empirical analysis. Economics Letters,174,35–38.
Lectard, P., & Rougier, E. (2018). Can developing countries gain from defying comparative advantage? Distance to comparative advantage,
export diversification and sophistication, and the dynamics of specialization. World development,102,90–110.
Leng, C., Ma, W., Tang, J., & Zhu, Z. (2020). ICT adoption and income diversification among rural households in China. Applied Economics,
52(33), 3614–3628.
León, O., Igartua, J. I., & Ganzarain, J. (2018). Performance and internationalization effects of the use of ICT in diversified companies. Small
Business International Review,2(1), 16–28.
Lin, F. (2015). Estimating the effect of the Internet on international trade. The Journal of International Trade & Economic Development,
24(3), 409–428.
Malah, Y. F., & Asongu, S. A. (2022). Infrastructure development as a prerequisite for structural change in Africa. Journal of the Knowledge
Economy,1–27. https://doi.org/10.1007/s13132-022-00989-w
Mignamissi, D., & Malah Kuete, Y. F. (2021). Resource rents and happiness on a global perspective: The resource curse revisited. Resources
Policy,71, 101994.
Nchofoung, T. N., Asongu, S. A., Njamen Kengdo, A. A., & Achuo, E. D. (2022). Linear and non‐linear effects of infrastructures on inclusive
human development in Africa. African Development Review,34(1), 81–96.
Nguea, S. M., Fotio, H. K., & Baida, L. A. (2022). Investigating the effects of globalization on economic sophistication in selected African
countries. African Development Review,34(3), 324–338.
Nickell, S. (1981). Biases in dynamic models with fixed effects. Econometrica,49, 1417–1426.
Njangang, H., Nembot, L. N., & Ngameni, J. P. (2020). Does financial development reduce the size of the informal economy in sub‐Saharan
African countries? African Development Review,32(3), 375–391.
Nordås, H. K., & Piermartini, R. (2004). Infrastructure and trade. World Trade Organization. Staff Working Paper ERSD‐, 2004–04.
Odeh, R. P., & Watts, D. (2019). Impacts of wind and solar spatial diversification on its market value: A case study of the Chilean electricity
market. Renewable and Sustainable Energy Reviews,111, 442–461.
KAMGUIA ET AL.
|
15
Opoku, E. E. O., & Boachie, M. K. (2020). The environmental impact of industrialization and foreign direct investment. Energy Policy,137,
111178.
Rehman, F. U., & Sohag, K. (2022). Does transport infrastructure spur export diversification and sophistication in the G‐20 economies? An
application of CS‐ARDL. Applied Economics Letters,1–5. https://doi.org/10.1007/s13132-022-00989-w
Saadi, M. (2020). Remittance inflows and export complexity: New evidence from developing and emerging countries. The Journal of
Development Studies,56(12), 2266–2292.
Santoalha, A., Consoli, D., & Castellacci, F. (2021). Digital skills, relatedness and green diversification: A study of European regions.
Research Policy,50(9), 104340.
Shepherd, B. (2010). Geographical diversification of developing country exports. World development,38(9), 1217–1228.
Tacchella, A., Cristelli, M., Caldarelli, G., Gabrielli, A., & Pietronero, L. (2012). A new metrics for countries' fitness and products' complexity.
Scientific Reports,2, 723.
Tadadjeu, S., Njangang, H., Ningaye, P., & Nourou, M. (2022). Oil dependence and access to water and sanitation in African countries: Does
the Extractive Industries Transparency Initiative matter? African Development Review,34(1), 54–67.
Vu, T. V. (2022). Does institutional quality foster economic complexity? The fundamental drivers of productive capabilities. Empirical
Economics,63, 1571–1604.
Weldemicael, E. (2012). Determinants of export sophistication. The University of Melbourne.
Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two‐step GMM estimators. Journal of Econometrics,
126(1), 25–51.
Zhu, S., & Fu, X. (2013). Drivers of export upgrading. World development,51, 221–233.
How to cite this article: Kamguia, B., Ndjakwa, M., & Tadadjeu, S. (2023). Does infrastructural development
foster export upgrading in Africa? African Development Review,1–16. https://doi.org/10.1111/1467-8268.12684
16
|
KAMGUIA ET AL.
