Controlling for Exporter-Year Factors when Estimating Import-Demand Elasticities

In seminal studies, Feenstra (1994) and Broda and Weinstein (2006) develop an econometric method to quantify the welfare gains from trade associated with increases in product varieties. While it is the most widely respected structural approach, it relies on a contentious assumption: the orthogonality of import demand and supply residuals. This paper demonstrates theoretically that including additional controls for exporter-year factors increases the likelihood that the orthogonality condition is satisfied in practice. In particular, the extended model adresses the potential correlation between product quality (demand shocks) and exporter characteristics (supply shocks). The paper then implements the benchmark and the extended model in international trade data. Three main results emerge from the analysis. First, F-tests support the extended model over the constraint (benchmark) model. Second, the mean and standard deviation of the estimated elasticities are both smaller in the extended model compared to the bench- mark model. Third, the correlation between the two sets of estimates is positive (as expected) but small, at 0.33. Overall, these empirical results suggest that it is important to control for exporter-year factors when estimating import-demand elasticities.