Does Africa trade less than it should and why?

Africa’s share in world exports has declined sharply over time. This raises the question whether the observed pattern of exports from Africa is consistent with the predicted or expected level of trade. This paper addresses the question whether or not Africa is an under-trading continent. We answer this question using a much improved dataset for obtaining predicted trade and by employing methods that correct for bias in estimates of under-trading. Our results indicate that globally Africa is an under-exporter in our preferred Heckman specification. This result is robust to addition of different controls and application of different variants of the gravity model of trade. We further ask the question what could explain Africa’s under-trading. We find that accounting for transport and communication infrastructure reduces the under-trading effect for Africa and in some specifications of the gravity model, the under-trading effect vanishes altogether. In order to assess the impact of infrastructure on trade, we employ a semi parametric variant of the gravity model. This model allows for unknown nonlinear impact of infrastructure on trade and also complementarity among several infrastructure variables. Results from the semi-parametric model provide evidence for significant non-linear impacts of infrastructure where the effects for a large number of African countries are significant and compare favorably with marginal effects of infrastructure in countries in other continents being in comparable income brackets. This model also finds evidence for complementarity across transport and communication infrastructure implying that much greater impacts will be likely if the infrastructure are developed jointly rather than in isolation.