本文研究面板数据空间误差分量模型(Spatial Error Components Model,SEC)的估计方法。为克服极大似然法在SEC模型估计中运算的困难,本文提出基于广义矩估计的可行广义最小二乘法(GMM—GLS),证明了估计量的一致性及有限样本下的有效性;并应用此模型,研究2000—2007年中国30个省(西藏除外)的物质资本存量、人力资本存量及能源消耗对实际GDP的影响,结果表明,采用SEC模型所得估计结果更为符合经济现实。
Given the computational difficulties of the maximum likelihood estimation for spatial error components (SEC) model, in this paper, we study and prove the consistency properties and effectiveness under finite sample of several moment-based generalized least squares (GMM-GLS) estimation methods. As an application of the SEC model, we estimate GDP output production function with China provincial panel data from 2000 to 2007. We found that the results of the SEC model are more realistic than those of previous findings in the literature using other estimation techniques.