常见单位根检验方法对初始值都做了适当的约束,而经验研究中的数据往往由于各种冲击的存在无法满足相应的假定条件。所以,有必要讨论检验功效对初始值稳健的单位根检验方法。本文在研究初始值对单位根检验功效影响的基础上,基于Fisher统计量提出了检验功效关于初始值较稳健的组合p值单位根检验方法并研究了其小样本性质。并且,对我国CPI月环比时间序列的检验发现,随着我国宏观经济调控政策的完善,CPI逐渐趋于平稳。
All popular unit root tests make assumptions about the initial observation. They are, however, difficult to be satisfied because of various shocks to the empirical data. Therefore, it is necessary to develop a new test whose power is robust to the initial observation. After analyzing the effect of the initial observation on the power of unit root tests, we propose a combined p-value test for unit root based on the Fisher Statistic and consider its finite sample performance. With the application of our test to the CPI series, we find that the CPI series has been becoming stationary with the improvement of China's macroeconomic control policies.