Ongoing Research

In this study, we focus on the estimation of the model of an individual cross-section unit by utilizing the whole panel. We allow for misspecification in the heterogeneity of the panel data model. Our aim is to obtain efficient estimators for the model of the targeted unit by taking into account the heterogeneity.

In this study, we focus on the effects of hurricane damages on the inwards and outwards migration. We use county level data from the U.S. The idea is to estimate a panel data model that can account for unobserved heterogeneity between counties and that can model the dynamics of change across time.

This ongoing work studies the effects of temperature changes on economic productivity. We estimate a country level panel data model that takes into account the nonlinear effects of temperature changes on productivity. The estimation method we use for this purpose allows for cross-sectional dependence. Cross-sectional dependence arises due to global climate and economic shocks that affect each country to a different extent. In our models these shocks include any common factor that has an influence on economic productivity and temperatures of each country. It has been shown that if these factors are ignored, the estimators will be inaccurate. In this study we revisit the existing literature on the empirical question of the effects of temperature changes on economic productivity and investigate the effects of ignored cross-sectional dependence.

This paper considers the estimation of long run effects in dynamic heterogeneous panel models with I(1) variables and cross-sectional dependence. In particular, we focus on a dynamic cointegrated panel data model with unobserved common factors. Simple estimators for the individual long-run parameters and a mean group estimator for the mean of the long-run parameters in conditional error correction models are proposed, and their properties are evaluated. The estimator can be viewed as belonging to the class of common correlated effect estimators, as initially proposed by Pesaran (2006), that makes use of cross-section averages to control for cross-sectional dependence. Monte Carlo simulation results are provided to suggest that the estimators perform well in small samples.

This paper proposes Lagrange-multiplier type tests for weak exogeneity in panel error correction models with unobserved common factors. We consider a test for the absence of error correcting behaviour from the marginal model for individual units, as well as a joint panel test. We use the common-correlated effects method to approximate the unobserved common factors. We derive the asymptotic distributions of individual specific statistics and propose Fischer and group-mean type panel tests. We show that the test statistics are not affected by the approximation method we use under certain conditions. The finite sample properties of the tests are assessed in a Monte Carlo simulation.