On the 10th April Alaistair Hall, a professor form the University of Manchester, will present the study on "GEL estimation based on group-specific moment conditions“.
This seminar will take place from 11.30 to 12.30 in Lecture Theatre 1 (Quelhas, 4th Floor).
Free admission.
Abstract:
In this paper, we consider the case where the population of interest is heterogeneous consisting of a number of homogeneous groups (sub-populations). Within each sub-population, the data is assumed to satisfy a set of population moment conditions indexed by a vector of model parameters, some of which are group-specific and some of which are common to more than one group. A group Generalized Empirical Likelihood (GGEL) estimator is proposed to estimate the parameters. The GGEL estimator is shown to be consistent and asymptotically normally distributed. The second-order bias is derived, and this formula is used to propose a bias-corrected version of GGEL. Inference methods are proposed for testing whether: (i) the overidentifying restrictions hold; (ii) the parameter vector satisfies a (potentially) nonlinear set of restrictions. In comparison to the analogous group Generalized Method of Moments estimator (GGMM), the GGEL is shown to have the same first-order asymptotic properties but to exhibit fewer sources of second-order bias.