Empirical Likelihood: Improved Inference within Dynamic Panel Data Models

This paper proposes and analyses an hybrid of Owen's (1988, 1990, 1991) Empirical Likelihood (EL) and bootstrap, EL-bootstrap, as an alternative to the General Method of Moments (GMM) within dynamic panel data models. We concentrate on the finite-sample size properties of their overidentification tests. Our results show that EL-bootstrap may be a good alternative to GMM estimation within this setting. The practical usefulness of our findings is illustrated via application on an AR(1) univariate panel data model with individual effects using the cash-flow series of 174 firms in the United States.