Empirical Strategy

Table of Contents:

The basic empirical model that underpins this relationship takes its clue from the empirical model Sekkat and Varoudakis (2000) that was applied in Meon and Sekkat (2008), which assumes that the exports of manufacturers are explained by the following relationship:


Where the main explained variable is the ratio of exports to GDP for the relevant year (Xit), while the explanatory variables include the real effective exchange rate (Eit) that captures the countries' currency appreciation/depreciation. The other variables include the GDP growth rate of country 'i's' partners (RYP) and the lag of investment in the relevant sector over GDP (Iit-1).

Noting that the empirical model of Sekkat and Varoudakis (2000) closely relates to the thesis of this study, we apply this model in this study by including some of their covariates and our main variables of interest. In this study, we are interested in the infrastructural provision and the interactive variable with institutions. The covariates—from Sekkat and Varoudakis' model—that is relevant to this study are the exchange rate and investment. The main reason for the choice of these covariates include the fact that exchange rate will reflect the relevant price for trade as an increase in the exchange rate will mean an appreciation of the exporter's currency and this will have a negative effect on trade. Likewise, the inclusion of investment variable is based on the grounded assumption that investment will improve manufacturing output and consequently, trade (Liu et al. 2001, 2002; Makki and Somwaru 2004).

Therefore, the empirical model for this study is presented as:


It is expected that infrastructure 'Infras' and institution 'Inst' will have a positive effect on manufacturing export. This—in no gainsay—is expected due to the role of infrastructure and institution on export (Meon and Sekkat 2008; Cissokho et al. 2013). The main focus of this study is the behavior of the interactive variable 'Infras x Instit ', which presents the multiplicative between institutions and infra-structural provision. A positive variable connote that the improvement of institution will improve infrastructural provision that affects growth. In essence, the complimentary effect is being portrayed by a positive sign. On the contrary, a negative sign connote that our argument is flawed and the relationship between the variables is substitutive.

4.1 Variable Definition and Source

The variables that was included in the model [Eq. (2)] are defined in Table 7 and the sources were also presented.

4.2 Method of Analysis

To ensure that the estimated results are not spurious, alternative econometric methods was applied in the estimation. The Ordinary Least Square (OLS) regression will be applied in the estimation. Noting the issues—like heteroscedasticity

Table 7 Variables definition and source





Manufacturing export


Manufacturing export, measured as percentage of merchandise export


Exchange rate


Real exchange rate




We applied the growth rate of the manufacturing value added as a proxy for the extent of investment in the manufacturing sector. Gross fixed capital formation as a percentage of GDP would have being used but this is more generic




Measured as the average of internet users per 100 persons, mobile and fixed line telephone subscribers per 100 persons, and telephone users per 100 persons




Corruption (CC) is the extent of corruption and the extent to which public offices are misused for private gains; Government Effectiveness (GE) captures the quality of government policies and the commitment of the government to such policies


Note: The institutional variables are standardized on a scale from -2.5 (weakest institutions) to +2.5 (strongest institutions). WDI World Development Indicators, WGI World Governance Indicators

and autocorrelations—related with the OLS technique, the Feasible Generalised Least Square (FGLS) technique was also applied because it allows for the presence of heteroscedasticity across the sampled countries and autocorrelation within the panels. This provides for panel-corrected standard errors. These two approaches will be relevant for sensitivity checks. As a matter of importance, the Systems type of GMM estimation technique, which has been favoured by some studies like Asiedu and Lien (2011); Asongu (2014). The uniqueness of the SGMM technique is that it uses internally generated instruments to addresses issues of endogeneity (Blundell and Bond 1998, 2000). For the SGMM technique to be relied upon, it is expected that the test for autocorrelation AR (2) and the Sargan test for instrument

over-identification must be ≥ 0.05.

The SGMM equation type for Eq. (2) is as follows:


The other variables are as earlier defined and the lag of the explained variable has 'α' coefficient. The variable 'η' is the unobserved country-specific effects and the error term is 'εi'.

4.3 Sample

The 15 ECOWAS countries were included for the period 2000-2012. The sampled countries include: Benin, Burkina Faso, Cape Verde, Cote D'Ivoire, Gambia, Ghana, Guinea, Guinea Bissau, Liberia, Mali, Niger, Nigeria, Senegal, Sierra Leone, and Togo. The period chosen was based on data availability for the chosen variables.

< Prev   CONTENTS   Next >