Optimism bias? The elasticity puzzle in international economics revisited

The elasticity of substitution between domestic and imported goods is a central parameter in macroeconomic models, but after decades of empirical studies there is no consensus on its magnitude. Earlier literature using time series arrives at low values, while more recent studies using panel-based econometric methods on disaggregated data find higher values. We examine the econometric methodology of this more recent literature, which follows the seminal work by Feenstra (1994), looking in more detail at the effect on the results of the non-linear mapping between reduced-form and structural parameters. Our main contribution is the use of bootstrap methods, which offer more insight into the Feenstra method and can explain why researchers applying it may tend to find high estimates. The bootstrap not only allows us to obtain considerably less biased estimates of the structural elasticity parameter, but also to better characterize their accuracy, a point vastly overlooked by the literature.