Tests for Dynamic Effects in Linear Panel Data Models

This paper proposes simple tests to detect dynamic and random effects in linear panel data models, in the form of lagged dependent variables and random effects. We use the analytical framework of Bera and Yoon (1993) to derive tests for the presence of random effects, lagged dependent variables, or both. All test statistics can be computed based on pooled OLS estimates, and hence can serve as a useful specification search tool to validate the adoption of a dynamic model.