Data Dependent Endogeneity Correction in Cointegrated Panels*
通过蒙特卡洛模拟，评估了几种信息准则在协整面板回归中用于数据依赖内生性修正的小样本表现，发现不同准则存在实际差异，其中小样本表现最佳的准则也对应着最优的估计量。
Abstract This paper examines the small‐sample performance of several information based criteria that can be employed to facilitate data dependent endogeneity correction in estimation of cointegrated panel regressions. The Monte Carlo evidence suggests that the criteria generally perform well but that there are differences of practical importance. In particular, the evidence suggests that, although the estimators of the cointegration vectors generally perform well, the criterion with best small‐sample performance also leads to the best performing estimator.