Quantile Regression under Limited Dependent Variable

TL;DR

This paper introduces ldvqreg, a Stata command for quantile regression with censored or binary dependent variables, using a smoothed objective function, improving estimation accuracy in such cases.

Contribution

It presents a novel estimation method for quantile regression with censored and binary data, implemented in a new Stata command, enhancing existing techniques.

Findings

Accurately estimates parameters in censored data scenarios.

Outperforms existing quantile regression methods with censoring.

Applied successfully to women's labor supply data in Uruguay.

Abstract

A new Stata command, ldvqreg, is developed to estimate quantile regression models for the cases of censored (with lower and/or upper censoring) and binary dependent variables. The estimators are implemented using a smoothed version of the quantile regression objective function. Simulation exercises show that it correctly estimates the parameters and it should be implemented instead of the available quantile regression methods when censoring is present. An empirical application to women's labor supply in Uruguay is considered.