Nonparametric and Semiparametric Methods for Econometric Analysis

Yale University, New Haven CT

Investigators

Abstract

This project consists of two sub-projects, each of which aims at developing new tools for econometric analysis. The first proposes a practical procedure for moment condition models. Moment condition models have been extensively studied over the last few decades, though relatively little is known as to how to estimate them using the Bayesian approach, which is convenient if one wishes to incorporate prior beliefs into her analysis. There exist at least two significant challenges associated with applications of the Bayesian approach to moment condition models. First, a flexible way to express one's beliefs needs to be used. Second, a general moment condition model induces complicated interdependence across unknown parameters that are being estimated. This project proposes a new approach to address them. It is applicable in many areas across diverse disciplines, since moment condition models are fundamental to general statistical analysis. The second sub-project proposes econometric tools for analyzing economic rationality while taking account of heterogeneity across individuals. In particular, it empirically examines the concept of stochastic rationality, which has been considered in various branches of economics and psychology. The empirical content of stochastic rationality is expressed in terms of a high dimensional system of inequalities stated in an indirect manner. This feature poses challenges both theoretically and computationally, and a set of novel econometric methods are proposed to overcome them. Additionally, these methods provide means to conduct policy analysis without imposing arbitrary assumptions. This has direct implications for evaluation of actual economic policies. Both sub-projects will produce computer programs for the proposed procedures, and they will be made available to the public at no costs. Also, the proposed activities are expected to provide educational benefits to students through research assistantships supported by the grant.