Title: A generalized focused information criterion for extremum estimators
Authors: Chu-An Liu - Academia Sinica (Taiwan) [presenting]
Xinyu Zhang - Academy of Mathematics and Systems Science, Chinese Academy of Sciences (China)
Abstract: Unlike the traditional model selection criterion, which picks a single model based on the global fit of the model, the Focused Information Criterion (FIC) is tailored to the parameter of interest and aims to select a model based on the parameter under focus. The aim is to develop FIC for different models in an unified theoretical framework and to propose a Generalized Focused Information Criterion (GFIC) for extremum estimators. To illustrate GFIC, we apply our theoretical results to several examples, including the maximum likelihood estimator, the nonlinear least squares estimator, the generalized method of moments estimator, and the minimum distance estimator.