Research On Corporate Governance Of Southeast Asian Corporations

Unit root test is an important method to determine the time series stationary, and stability is the premise of a lot of time series modeling, so the unit root test work has become a necessary time series modeling before. The traditional unit root test method builds on many more ideal assumptions, such as a linear trend, normally distributed residuals, residuals autocorrelation does not exist, no structural mutation hypothesis. However, as the data sources and constantly enrich and more complex data structures, the traditional unit root test methods can not meet the many complex situations unit root test requirements, there has been a lot of testing the efficacy of low or even pseudo-examine the case.In view of this, it is necessary to further the factors affecting the unit root test results were discussed, the unit root test efficacy data generated in different mechanisms to define, thus providing data based on the method of choice for the unit root test. By Monte Carlo simulation method discussed quantitatively test the effectiveness of several unit root test data generation methods under different mechanisms. The simulation results show a linear time series, ADF and PP test test test efficacy are good; for nonlinear time series trend, KPSS test has better efficacy testing; for time series containing the mutation, KPSS test has better test efficacy. For small sample data, PP test the effectiveness of the best test; when the sample size is increasing, and test the effectiveness of a linear sequence converge to 1; correlation coefficient close to 1, ADF tests and KPSS test has good recognition results.