We investigate what accounts for the different evidence between the aggregate and firm‐level data on the exchange rate elasticity of exports. The typical estimation of the macroeconomic export equations gives insignificant estimates for this elasticity compared to those from the recent firm‐level estimation. Using firm‐level data from Japan, we identify the sources of this discrepancy, and show that the failure to account for cost and demand factors as well as firm‐level productivity induces various kinds of biases for the aggregate estimate of the exchange rate elasticity of exports.