Robust Variable Selection of Linear Panel Data Models with Fixed Effects
This paper provides a robust variable selection process for fixed effect panel data models by combining composite quantile regression and adaptive lasso penalty method.The fixed effect is eliminated by the forward orthogonal deviation transformation,and then the objective function of penalty composite quantile regression is constructed by the adaptive Lasso,which can estimate regression coefficients and select important variables simultaneously.The orcale property is proved under some regular conditions.The proposed method not only eliminates the fixed effect on the estimator,but also has robustness.Monte Carlo simulation is used to study the finite sample properties of the proposed method,and it is applied to real data analysis.
万方数据
维普
