A Sparse Decomposition of Democracy and Trade
with In Song Kim and Marc Ratkovic; 2016 Asian Political Methology Conference
Do trading partners’ political institutions matter in international trade? In answering this question, researchers typically relate aggregate measures of institutional structure with total volumes of bilateral trade, while pairs that do not trade are excluded from the analysis. This common approach blurs important distinctions between institutional impacts on the selection of trade partners (the extensive margin) and the volume of trade (the intensive margin), while it engenders selection bias. This paper overcomes these problems. First, it decomposes the widely used Polity IV measure into its components to identify political institutions that are trade promoting. Second, we propose a two-stage Bayesian LASSO estimator to deal with the twin problems of selection and multicollinearity that are pervasive in the use of gravity equations in empirical studies of international trade. We find that political institutions matter more in the extensive margin than in the intensive margin of trade. We also find that democratic political institutions are not always promoting. Specifically, we find that while trade appears to be fostered by political stability, it may actually be a means of circumventing the kleptocratic tendencies of hereditary rule. We identify heterogeneous effects of both the standard gravity variables and the polity variables varying across industries and time, with a notably recent tendency for some aspects of the longstanding democracies: party competition and institutional durability, to inhibit trade.