中文摘要：

本文在对邹至庄检验和递归最小二乘法剖析基础上，综合有序聚类法和虚拟变量方法，提出了一种新的结构稳定性检验方法一有序聚类虚拟变量法，克服chow检验和递归最小二乘法等在系数稳定性检验中的缺陷，利用蒙特卡罗模拟方法比较了这三种方法的优劣。并利用这三种方法对从沪深300成份股中按πPS方法抽取出来的30只股票进行β系数稳定性检验比较。研究表明，有序聚类虚拟变量法在检验系数稳定性比邹至庄检验方法和递归最小二乘法更为有效。

英文摘要：

This article give a new method-ordinal cluster dummy variable to test coefficients stationarity, overcoming the defect of the Chow test and recursive least squares method. Monte Carlo simulation was used to compare the new method with Chow test and recursive least squares. The findings show that ordinal cluster dummy variable is better than other tow methods. At last, the method is applied to test the beta coefficient of Chinese stock market.