统计视角下的结构突变检验设计及其应用  

A Test for Structural Break from the Perspective of Combination of Nonparametric Statistics and Its Application

作　　者：杨林涛[1] 韩兆洲[1] 

机构地区：[1]暨南大学经济学院统计学系,广东广州510632

出　　处：《数理统计与管理》2016年第2期286-298,共13页Journal of Applied Statistics and Management

基　　金：国家社会科学基金重点项目(15ATJ001);全国统计科学研究重点项目(2015LZ35)

摘　　要：在结构突变检验中,针对参数统计检验需要依赖于严格假定的局限,本文创新性地提出了一套切实可行的非参数统计检验方法和检验程序。蒙特卡罗模拟表明,这套非参数统计组合设计具有较高的检验效力;结构突变前后样本量大小并不影响对截距与斜率双突变以及截距不变、斜率突变进行检验的检验效力,而对截距突变、斜率不变以及无结构突变检验影响较大。Parametric statistical tests needed to rely on strict assumptions,so the paper presented practical methods and test procedures of nonparametric statistical test.Monte Carlo Stimulation showed that：The test power of the combination design was very high;the sample size before and after structural break did not affect the test power of the double mutant of intercepts and slopes and the single mutant slopes,and it has had a significant impact on the single mutant intercepts and non structure mutation.

关 键 词：结构突变 非参数统计 蒙特卡罗模拟 

分 类 号：O212[理学—概率论与数理统计]