Estimators and inference in a censored regression model with endogenous covariates

摘要: This paper derives the asymptotic distribution theory for censored regression models with endogenous covariates under no parametric assumptions on disturbance distribution, extending modeling framework of Powell (1986). While it is well known that some restrictions use reduced form residuals will lead to consistent estimators structural parameters, derivation essential inference and usual hypothesis testing not obvious this model. The problem arises because model generates a set moment are discontinuous in making standard methods inapplicable. illustrates techniques Pakes Pollard (1989) can be adapted by treating multi-stage as simultaneously satisfying joint vector population restrictions, partitioning covariance matrix appropriately. A Monte Carlo study high practical power estimators, shows they provide useful alternative depend Gaussian or other specified distributions.