Instrumental Variable Estimation with Discrete Endogenous Regressors

This paper develops a new objective function that can be used to estimate models with discrete endogenous regressors. The basis for the objective function is the objective function that one would have used if all covariates were assigned randomly. The paper than proposes a weighting function that uses instruments and ‘removes’ the endogeneity. Methods that assume that the endogenous regressor or the instrument are continuously distributed already exist and this paper deals

with endogenous variables that are discrete. Very preliminary.

For more information, contact Frank Schorfheide.