单位根KPSS检验趋势项检验量的改进  

Improvement of Trend Statistics for Unit Root KPSS Test

作　　者：江海峰 胡根华 Jiang Haifeng;Hu Genhua(School of Business,Anhui University of Technology,Maanshan Anhui 243032,China)

机构地区：[1]安徽工业大学商学院,安徽马鞍山243032

出　　处：《统计与决策》2024年第3期34-39,共6页Statistics & Decision

基　　金：国家社会科学基金资助项目(21BJY213)。

摘　　要：为完善KPSS检验流程,借鉴单位根右侧检验理论,将递归估计方法引入KPSS趋势项检验中,构造两种递归估计上确界趋势项检验量和一种非递归估计趋势项检验量。理论研究表明,所有趋势项检验量在大样本下都收敛于维纳过程的泛函,但与已有检验量的分布不同,需要重新模拟以获取临界值。蒙特卡洛模拟显示,和非递归估计趋势项检验量相比,递归估计上确界趋势项检验量具有满意的检验水平,但临界值更大。在三种突变结构模型中,所有检验量在全局结构突变模型中检验功效最高,且符合理论研究结果。在三种检验量中,狭义递归估计趋势项检验量检验功效总体占优,说明递归估计上确界趋势项检验量具有优越性。In order to improve the KPSS test process,this paper uses the unit root right-side test theory for reference,introduces the recursive estimation method into KPSS trend term test,and constructs two kinds of recursive estimation upper bound trend term test and one kind of non-recursive estimation trend term test.Theoretical research shows that all trend statistics converge to the fonctionelle of Wiener process in large samples,but different from the distribution of existing statistics,needing re-simulation to obtain critical values.Monte Carlo simulation shows that compared with the non-recursive estimation trend statistics,the trend term of recursive estimation upper bound has a satisfactory test level,but the critical value is larger.Among the three structural mutation models,all statistics have the highest efficiency in the global mutation model,and are in line with the theoretical research results.Among the three kinds of trend statistics,the test efficiency of the narrow recursive estimation statistics is generally dominant,indicating the superiority of recursive estimation trend term statistics.

关 键 词：单位根过程 KPSS检验量 趋势项检验量 递归估计 蒙特卡洛模拟 

分 类 号：F224.0[经济管理—国民经济] O212.1[理学—概率论与数理统计]