摘要
对于内生突变情形下的单位根检验,突变点的确定方法会影响到单位根检验的功效,不同方法在确定突变点位置时的表现也不尽相同。本文首先评述了几种常用的突变点确定方法及相应的单位根检验,然后对基于各类回归式残差平方和最小值确定突变点的方法进行了比较分析,本文所设数据生成过程有别于已有研究,并首次考虑了依据可行广义最小二乘(FGLS)估计来确定突变点。在此基础上,还对比分析了几种不同突变点确定方法下的单位根检验功效和实际检验水平。结论显示,依据FGLS残差平方和最小值得到准确突变点的频率最高,且在AO模型下据此进行Perron检验具有较高的功效且不会发生较大的水平扭曲。
For the unit root test of endogenous structural break,the determination method would influence the power of unit root test,and the results of different methods are not all the same in the process of estimating the break point.In the paper,above all we describe and evaluate the several common methods for determinate the break point and corresponding unit root test,and then,comparatively analyze the determination methods of break point based on the square sum of residual of different regression equations.The data generating process in this paper is different from the existing research,and estimating the break point according the FGLS procedure is firstly put forward.At last,discuss the power and actual size of unit root tests based on several different determination method of break point.The result reveals that the frequency of estimating break point correctly is the highest based on the minimizing the residual square sum of FGLS regression,and based on the estimated break point,the Perron test for the AO model have better power and little size distortion.
出处
《统计研究》
CSSCI
北大核心
2010年第5期101-109,共9页
Statistical Research
关键词
结构突变
单位根检验
突变点
残差平方和
可行广义最小二乘估计
Structural break
Unit root test
Break point
Square sum of residual
Feasible Generalized Least Squares Estimate