中文摘要：

运用面板数据建模分析时,在考虑截面异质性的情形下存在参数过多的问题。提出一种基于面板数据关系结构的聚类方法,能有效解决模型估计时参数过多的问题;提出内距离与外距离概念,有效解决了聚类分析时定量确定分类数的问题。将此方法运用于动态面板数据的建模分析,统计模拟结果显示有较好的小样本性质。基于理论模型,采用中国1996-2012年的省级面板数据,实证分析了金融发展对房地产业发展的动态影响,分析效果与现实经济发展较吻合,证明该方法有较好的应用性。

英文摘要：

By using panel data modeling and analysis,consider the case of a cross-sectional heterogeneity,there are too many parameters of the problem.This paper,using the principle of clustering,puts forward a clustering method based on panel data relationship structure,can effectively solve the problems of the model to estimate the parameters of too much.Puts forward the concept of internal distance and external distance to solve the problem of clustering analysis in determining the classification number.This method is applied to the article modeling of dynamic panel data analysis,Monte Carlo simulation the estimated results of small sample properties.Based on the theoretical model,using the Chinese provincial panel data from 1996 to 2012,empirical analysis on the dynamic effect of financial development of the real estate industry development,the property of application is good.