Inference in Incomplete Econometric Models

Princeton University, Princeton NJ

Investigators

Abstract

This project develops methods that deal with estimation of economic models while imposing minimum plausible assumptions. In particular, the class of models that the project examines are defined in terms of inequality restrictions. These inequality restrictions arise in models with incomplete data (data that are censored or top-coded), in models with multiple equilibria, or in models with measurement error, among others. The estimation techniques that are developed are applied first in estimation problems with interval data with endogeneity. Second, the methods are used to obtain more robust inference in English auction models with bids data. Third, the methods are applied to nonlinear models with measurement error. The project will also examine the issue of constructing valid tests that enables confidence regions for the estimated components of these models.