存在测量误差的面板自回归模型的工具变量估计

On the IV Estimation of Auto-regression Model for Panel Data with Measurement Errors

对于存在测量误差的面板数据自回归模型,首先讨论了POLS(Pooling OLS)和LSDV(least square of dummy variable)估计存在向零的衰减偏差及其非一致性,其次对于混合自回归模型和个体固定效应自回归模型给出了工具变量应满足的条件。研究发现这时工具变量的选择是十分困难的。

In this paper, it is found firstly that POLS and LSDV of auto- regression model are attenuation bias towards zero and inconsistent for panel data with measurement errors. Secondly, the conditions of instrument variables （Ⅳ） in the models are put forward. And it is very difficult to select Ⅳ in this case.