Abstract:

The implications of stationarity and periodic components in macroeconomic data are profound. Whether do periodic components have relation to stationarity of time series? Nonstationarity of time series is often due to the unit root. It is important for unit root tests to select the appropriate number of lags. The sample spectrum is widely used to investigate time series’ periodic components and the cycle frequency. We integrate the autocorrelation function (ACF) and the sample spectrum with unit root tests to examine time series’ stationarity in this paper. In our empirical analysis for two returns series of Chinese stock markets’ indexes in 2007, nonstationarity is not rejected for two returns series when we select the number of lags of unit root test in light of results of the ACF and the sample spectrum. Therefore, there seems to be some evidence that periodic components have relation to stationarity of two returns series.