Pooling in Dynamic Panel-Data Models: An Application to Forecasting GDP Growth Rates
分析面板数据中合并模型的条件，发现小样本下合并预测更优，但随样本增大优势减弱；对18个OECD国家GDP增长率的实证表明，允许残差相关时，基于GLS的合并预测优于个体预测。
In this article, we analyze issues of pooling models for a given set of N individual units observed over T periods of time. When the parameters of the models are different but exhibit some similarity, pooling may lead to a reduction of the mean squared error of the estimates and forecasts. We investigate theoretically and through simulations the conditions that lead to improved performance of forecasts based on pooled estimates. We show that the superiority of pooled forecasts in small samples can deteriorate as the sample size grows. Empirical results for postwar international real gross domestic product growth rates of 18 Organization for Economic Cooperation and Development countries using a model put forward by Garcia-Ferrer, Highfield, Palm, and Zellner and Hong, among others illustrate these findings. When allowing for contemporaneous residual correlation across countries, pooling restrictions and criteria have to be rejected when formally tested, but generalized least squares (GLS)-based pooled forecasts are found to outperform GLS-based individual and ordinary least squares-based pooled and individual forecasts.