Bias-Corrected Common Correlated Effects Pooled Estimation in Dynamic Panels
针对同质动态面板，改进了共同相关效应混合估计量，提出偏差校正版本，在时间跨度固定或增长时均一致，蒙特卡洛实验显示偏差和方差显著改善，并应用于估计温度冲击对总产出增长的动态影响。
This article extends the common correlated effects pooled (CCEP) estimator to homogenous dynamic panels. In this setting, CCEP suffers from a large bias when the time span (<i>T</i>) of the dataset is fixed. We develop a bias-corrected CCEP estimator that is consistent as the number of cross-sectional units (<i>N</i>) tends to infinity, for <i>T</i> fixed or growing large, provided that the specification is augmented with a sufficient number of cross-sectional averages, and lags thereof. Monte Carlo experiments show that the correction offers strong improvements in terms of bias and variance. We apply our approach to estimate the dynamic impact of temperature shocks on aggregate output growth.