Armington Elasticities and the Third-Country Effects of Trade Wars

We revisit the Armington model developed by Feenstra, Luck, Obstfeld, and Russ (2018) that allows the elasticity of substitution between the domestic and foreign varieties to be different from that between alternative foreign varieties. We develop a novel two-stage gravity-based framework to estimate these two elasticities using data on bilateral trade flows, tariffs, and domestic production, exactly the same set of data used for counterfactual analysis. Our estimates suggest that the elasticity of substitution between the domestic and foreign varieties is 43% lower than that between alternative foreign varieties. We apply the model to quantify the global consequences of the US-China trade war starting from 2018. Counterfactual analysis suggests that imposing a uniform elasticity of substitution across all varieties would lead to (i) considerable underestimation of the third-country welfare gains from the US-China trade war, and (ii) overestimation of the noncooperative tariffs between the U.S. and China.