A copula-based method for accommodating feedback in random effects panel models

Abstract This paper proposes a copula-based framework for accommodating dynamic feedback in random effects panel data models. The method offers two advantages compared to the canonical existing approach. First, it connects more naturally to economic theory and therefore might offer improved fits to real-life data. Second, it does not impose restrictions on correlations between unobservables, which again might offer improved fits in settings where large correlations are expected to exist. An empirical illustration examines the effect of employment on poverty.