Trade and Income: Using Civil Wars in Transit Countries As External Shocks on Trade

Mathieu Couttenier (Paris 1 Sorbonne)
Vincent Vicard (Banque de France)

Abstract: The causal relationship between trade openness and economic growth/income has been widely studied in the literature. The main challenge is to disentangle causation from correlation because of reverse causality and omitted variable bias such as geography or institutions. One way to circumvent the endogeneity issue is to identify events in third countries that affect national trade but do not directly affect domestic growth. In this paper, we propose to identify the trade / growth relationship by using civil war in third countries as an exogenous shock to trade. A civil war in a neighboring country is likely to impact domestic growth in two ways. First, directly, through negative security spillover and migration of refugees, and second, indirectly, through reduced trade. We use this indirect effect to instrument for trade openness in a growth equation. This second effect is particularly relevant for landlocked countries, whose neighbors serve as transit countries for a large share of their exports. In this case, a civil war in a transit country will reduce bilateral trade but also overseas trade since the access to international port is less secure. Civil wars in transit countries can be considered as a good time varying exogenous instrument for trade openness in a growth equation, since we are able to isolate the impact of a civil war in a neighboring country from the additional impact on multilateral trade of a civil war in a transit country. We estimate the impact of civil wars on trade within a gravity framework. We find that civil war in neighboring countries have a negative effect on bilateral trade, and an additional negative effect on overseas trade when the civil war takes place in a transit country of a landlocked country. We then use this shock to instrument trade openness in a growth regression and show that trade openness has a positive and robust effect on GDP per capita.