Revisiting the Effectiveness of African Economic Integration : A Meta-Analytic Review and Comparative Estimation Methods

The gravity model has extensively been used in estimating the effectiveness of a number of RTAs in the world. However, many previous studies that assess the effectiveness of African RTAs using gravity model produce contrasting results and are characterize by two main shortcomings. Firstly, these studies failed to account for multilateral resistance term (MRT). The omission of the MRT contributes to biased estimates for standard variables in the gravity model. This bias is reflected in parameters of these previous studies. Secondly, there is significant proportion of zero flows in developing countries' trade measurement; however, these studies fail in dealing with them properly. The zero flows are more endemic when one considers only African bilateral trade, which has over 50% zero flows. In an attempt to correct this anomaly, previous studies rely on the Tobit model or replacing zero flows with small values. However this strategy has been labeled as infeasible and producing inconsistent parameters. In this study, we conduct a meta-analysis of previous empirical studies to explain the potential heterogeneity in the studies and compare the different estimation methods of the gravity model to Poisson Pseudo Maximum Likelihood (PPML). Using panel data on trade flows from 1980 to 2006 for 47 African countries, we estimate the gravity model for the five major RTAs on the Africa continent. We find that although there is general positive impact of African RTAs, impacts are highly sensitive to different estimation methods and they tend to be significantly overestimated when zero flows are not properly dealt with. Comparative assessment of the five major RTAs indicates highly uneven performance