带异质线性趋势的二元选择面板模型偏误纠正估计

Bias-corrected Estimators for Binary Choice Panel Model with Heterogeneous Linear Trends

由于冗余参数问题,带异质线性趋势的二元选择面板模型参数的极大似然估计量存在严重偏误。为此,本文利用泰勒展开式推导了估计量偏误的期望,根据偏误的表达式构造了两种偏误纠正估计量。小样本模拟发现,对于较小的T,偏误纠正方法可以显著地减小参数的偏误。

The MLE estimators show severe bias because of incidental parameter problems in binary choice panel models with heterogeneous linear trends. This paper gives the leading terms of the bias for common parameters with Taylor expansion. By these results we obtain two kinds of bias-corrected estimators of common parameters. Monte Carlo simulations find that the bias-corrected estimators show great improvement to the MLE estimators of common parameters.