Mean-Variance Tradeoff in Trade Distortions: Analytical Results and Evidence from China

Trade models typically assume trade distortions (such as tariffs or trade costs) are constant across firms. We consider distortions that vary across firms and measure the gains from reducing dispersion in trade distortions. First we use a standard model to solve for the reduction in the mean that is welfareequivalent to any given reduction in variance, and show that the mean-variance tradeoff is linear. Second we show how changes in the variance of trade distortions affect real income using a two step extension of the procedure from Arkolakis, Costinot and Rodriguez-Clare (2012). We apply these results to Chinese firm-level data and solve for equivalent mean reductions and for welfare gains in three counterfactuals: 1) reducing variance among all firms to that of private firms in China, 2) reducing variance to the level of French firms, and 3) removing all variance. JEL: F12, F13, F14, O24