Robust Estimation of Parameter with One Half Rotating Panel

摘要

In this paper, based on stationary panel data model with one-way error component, FGLS estimation with one half rotating panel is investigated. It has been found that standard errors of FGLS estimators lead to extreme overconfidence. Therefore, based on FGLS estimation and PCSEs, a new estimation method is proposed to correct exorbitant overconfidence and make the estimation of parameters confident. Under the assumed stationary conditions, we prove the consistency and the asymptotic normality. Monte Carlo simulation is conducted to analyze finite sample properties. Compare to FGLS estimation, the method proposed in the paper can decrease the overconfidences on the premise of remaining A-Bias and MSE the same basically.