Recovering Credible Trade Elasticities from Incredible Trade Reforms

This paper tackles a central measurement problem in international economics: the trade elasticity is inherently dynamic and expectation-dependent, yet the object policymakers and modelers want—the response to a "canonical" unanticipated, once-and-for-all tariff change—doesn’t show up anywhere in observed data. Using a dynamic heterogeneous-firm model calibrated to US trade liberalizations with China and Vietnam, we show that the true long-run response to an unanticipated, permanent tariff reform is roughly 15, about five times larger than typical empirical estimates. The figure shown here illustrates the core mechanism: panel (a) shows that tariff changes with different degrees of of persistence (ω) exhibit very different autocorrelation patterns, while panel (b) shows the implications for estimated trade elasticities. When tariffs are highly persistent (ω near 1), the trade elasticity is more than four times higher in the long run than in the short run value. But when tariffs are transitory (low ω, as is typical in the data), the short- and long-run elasticities are not materially different. This explains why reduced-form methods applied to typical tariff variation fail to uncover the large, gradual adjustment that would follow a canonical reform.