Working Paper: NBER ID: w25823
Authors: Xavier Dhaultfoeuille; Arnaud Maurel; Xiaoyun Qiu; Yichong Zhang
Abstract: This article presents the eqregsel command for implementing the estimation and bootstrap inference of sample selection models via extremal quantile regression. The command estimates a semiparametric sample selection model without instrument or large support regressor, and outputs the point estimates of the homogenous linear coefficients, their bootstrap standard errors, as well as the p-value for a specification test.
Keywords: eqregsel; sample selection models; extremal quantile regressions
JEL Codes: C21; C24; C87; J31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| eqregsel command (C22) | estimates parameters of sample selection models (C24) |
| eqregsel command (C22) | provides point estimates of homogeneous linear coefficients (C51) |
| eqregsel command (C22) | provides bootstrap standard errors and p-values for specification tests (C51) |
| eqregsel command (C22) | yields a more accurate estimation of the wage gap (J79) |
| traditional OLS regression (C29) | underestimates the differential due to selection bias (C20) |
| robustness check (C52) | does not reject the model at standard significance levels (C52) |
| black-white wage gaps (J31) | widening from 119 to 159 log points (C46) |