This paper considers model selection and model averaging in panel data models with a multifactor error structure. We investigate the limiting distribution of the common correlated effects estimator (Pesaran, 2006) in a local asymptotic framework and show that the trade-off between bias and variance remains in the asymptotic theory. We then propose a focused information criterion and a plug-in averaging estimator for large heterogeneous panels and examine their theoretical properties. The novel feature of the proposed method is that it aims to minimize the sample analog of the asymptotic mean squared error and can be applied to cases irrespective of whether the rank condition holds or not. Monte Carlo simulations show that both proposed selection and averaging methods generally achieve lower expected squared error than other methods. The proposed methods are applied to analyze the consumer response to gasoline taxes.

This study examines the convergence rate of mean reversion by estimating the half‐lives of sectoral real exchange rates using an extensive product price panel for Japan (with the USA as the numéraire). We find that the half‐lives of sectoral real exchange rates are remarkably distorted when the grouped half‐life is measured inappropriately and the cross‐sectional dependence and potential trend breaks are ignored. After taking account of these problems, the bias‐corrected half‐life for all goods is as low as 3.00 years, close to the bottom of the consensus view of 3 to 5 years. Moreover, the bias‐corrected half‐life of mean reversion is 2.40 years for traded goods, and only approximately half that for non‐traded goods. Finally, our findings also support the view that small‐sample bias correction is critical for half‐life estimations.