Empirical Productivity Distributions and International Trade

We use firm-level data for 15 countries and 13 manufacturing sectors to estimate firm-level productivity parameters and to establish representative country-sector-specific empirical productivity distributions. We use these distributions against the backdrop of multi-sector versions of the models of Eaton and Kortum (2002) and Melitz (2003) to quantify the role of technology in shaping international trade flows. We find that, on average, absolute advantage measured as productivity differences across countries within sectors explains 15% and 21% of the total variation in bilateral trade shares in the models of Eaton and Kortum (2002) and Melitz (2003), respectively. In contrast, on average, comparative advantage measured as productivity differences across sectors within countries explains 39% and 47% of the variation in trade flows in these two models. We also demonstrate that empirical productivity distributions entail quantitatively important micro-to-macro implications for marginal responses of trade flows to changes in trade costs, for gravity-type estimation of trade models, and for comparative statics isomorphism between the customarily parameterized models of international trade. We confirm the predictions of the two aforementioned models under empirical productivity distributions in the data.