Partially Adaptive Estimation of Truncated Regression Models: A Comparison with Several Semiparametric Estimators

• Activity Number: 411
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Patrick Ansel Turley*+ and James B McDonald
• Companies: Brigham Young University and Brigham Young University
• Keywords: truncated regression ; partially adaptive ; flexible distributions ; semi-parametric methods ; Monte Carlo simulations

Abstract

Applying traditional regression methods or parametric methods (such as OLS or a Tobit-like estimator) to truncated regression models leads to biased and inconsistent estimators when the error distribution is misspecified. This paper proposes using partially adaptive estimators based on flexible error distributions to account for possibly skewed or leptokurtotic errors. Monte Carlo simulations and empirical applications are used to compare the performance of these estimators to several semi-parametric estimators. Preliminary results show improved performance of partially adaptive estimators over the other estimators considered based on the sample RMSE for the data considered.