摘要
扰动项的空间过程会影响空间计量模型的估计效果。通过探究空间关联误差效应下空间动态面板数据(SDPD)模型拟极大似然估计(QML)的有限样本性质,发现含空间自回归误差项的SDPD模型大样本性质较好;其估计结果优于不含空间自回归误差项的模型;较强的误差项空间相关性对参数估计精度的影响程度较大;误差项分布偏离正态性会影响模型的估计结果,但模型总体估计的稳健性良好。总体蒙特卡洛结果与本文的理论分析一致。
The disturbance of spatial processes will influence the estimate effects of spa- tial model. We explore the finite sample properties of spatial dynamic panel data (SDPD) model of quasi-maximum likelihood estimation (QML) in the presence of spatial error effects in this paper. Monte Carlo simulation shows that the large-sample properties of SDPD model containing spatial autoregressive error term are better. Its estimation results are better than those free from spatial autoregressive error term. The accuracy of parameter estimation will be greatly influenced when there presences strong spatial correlation in error term. Then the deviation from normality of the error term will affect the model estimation results, but the overall estimate of the robustness of the model is good. The overall results of Monte Carlo are consistent with the theoretical analysis in this paper.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2016年第4期126-144,共19页
Journal of Quantitative & Technological Economics
基金
国家自然科学基金项目(71073073
71273122
71473109
41461025)
江西省研究生创新专项基金项目(YC2015-B048)的资助