Spatial panel data models with structural change

时间:2021-03-29         阅读:

光华讲坛——社会名流与企业家论坛第5663期


主题Spatial panel data models with structural change

主讲人首都经济贸易大学 李鲲鹏教授

主持人统计学院 常晋源教授

时间2021年3月30日(周二)下午2:00-3:00

直播平台及会议ID腾讯会议,268 404 446

主办单位:数据科学与商业智能联合实验室 统计学院 科研处

主讲人简介:

李鲲鹏现为首都经济贸易大学教授,国际经管学院院长。研究方向为大数据计量经济学,在国内外经济学、统计学、管理学期刊发表论文30篇,包括Annals of Statistics, Journal of Business & Economic Statistics, Journal of Econometrics, Management Science, Review of Economics and Statistics等。现在担任Journal of Business & Economic Statistics期刊编委。

内容提要:

Spatial panel data models are widely used in the social science, especially in the economics discipline. The existing studies on spatial models usually assume parameters stabilities. Such an assumption may be restrictive given that the presence of structural changes in the relationship of economic variables has been well documented in the literature. This paper proposes and studies spatial panel data models with structural change. Our basic model is a static spatial autoregressive panel data one, and we consider using the quasi maximum likelihood (QML) method to estimate it. The asymptotic theory of the QML estimators including consistency, convergence rates and limiting distributions, are established under large-$N$ and large-$T$ setup. We next extend our theory in two directions: dynamic model and large-$N$ and fixed-$T$ setup. We also study the hypothesis testing for the presence of structural change. The three super-type statistics are proposed. We run simulations to investigate the performance of the QML estimators and find that the QML estimators behave well in our simulations.

空间面板数据模型在社会科学中被广泛使用,尤其是在经济学中。现有的关于空间模型的研究通常假设参数的稳定性。鉴于文献中已充分描述了经济变量关系的结构性变化的存在性,因此,这种假设可能具有局限性。本文提出并研究了具有结构变化的空间面板数据模型。本文的基本模型是基于静态空间自回归面板数据,并采用最大拟似然(QML)方法对模型进行估计。同时,在N和T较大的情形下,给出了QML估计量的相合性、收敛速度和极限分布等渐近性质。在之后的研究中,本文将理论拓展到两个方向:动态模型和N足够大T固定的情形。本文还进行了结构变化是否存在的假设检验分析,并提出三种超类型的检验统计量。通过数值模拟,本文研究了QML估计量的性能,并发现该方法在有限样本的模拟研究中表现良好。

最新信息