Random Coefficient Autoregressive Unit Root Tests And Their Application

Since the 1980s, it has become one of most important issues the unit test to infer the time series integration in analysis of the time series. Although, in big sample, the traditional unit root test of the time series (for example, DF test, ADF test and PP test) has a better power. But, as to small sample data, their test power is very low. Also, the power of the unit root test (namely, the reliability of the unit root test) plays a significance role in how to propose macro-economic policy. In order to improve the power of the unit root test, many people start to improve the traditional unit root test from two aspects that enlarge the object of the unit root test (namely, data structure) and develop the unit root test method in time series.This article makes a comparative analysis on many kinds of the random coefficient unit root test. At the same time, tests the stationary of 16 macroeconomic variables using the time series random coefficient unit root test. We find that 9 in 16 are fixed coefficient unit root process, or random coefficient unit root process. Therefore, the impact of the macroeconomic (for example, economical stationary policy) has a lasting influence to the standard of the macroeconomic. Also, it is bigger and deeper to actual macroeconomic than to name macroeconomic. According, this article supplies some theory evidence and efficient analysis frame to our country macroeconomic policy.