The Method And Application Of Quantile Unit Root Test

Unit root test is an important method to analyze the stationarity of time series,and it is the basis of non-stationary time series.It is also the basis of co integration test,Granger causality test and impulse response analysis.The traditional method of unit root testing is mostly based on strict assumptions,.But the actual data is very complex,The quantile unit root test method comes into being,and the quantile unit root test method can test the stationarity of different quantile,The hypothesis that the quantile unit root test is satisfied is more relaxed than the traditional unit root test.This paper simulates the data generation process and analyzes the effectiveness of the unit root test method(power).By comparing the effectiveness of different unit root tests,the most suitable test method is found.,The results of Monte Carlo(Monte Carlo)simulation experiments show:(1)When the random error items are normal distribution,the ADF unit root test method,the PP unit root test method and the quantile unit root test method are consistent,The first two tests are better than the third methods.(2)When the random error term is the t distribution satisfying the degree of freedom of 2 and 3 the quantile unit root test method is superior to the ADF unit root test and the PP unit root test.In the part of empirical analysis,the consumer price index(CPI)and per capita GDP data of 31 provinces and cities in China from January1990 to March 2016 were selected,The result shows:(1)Under the traditional unit root test method,the economic and financial sequence data(CPI,per capita GDP)is either non-stationary or stationary,Using the quantile unit root test method,the time series data(CPI GDP per capita)showed different unit root tests at different,The CPI and per capita GDP data of six countries(USA,Japan,Australia,Sierra Leone,Niger and India)were selected.and the results of unit root test of CPI and GDP per capita in different countries were compared,Further more,it is proved that the application of this method is feasible,and the test result is more reliable and robust.In conclusion,the results of the test vary with the change of the sub loci,which reflects the distribution characteristics of the target variables more comprehensively,In order to better decision making recommendations,but also for subsequent quantile cointegration test to prepare for the analysis.