Uncertainty and Stationarity in Financial and Macroeconomic Time Series—Evidence from Fourier Approximated Structural Changes

Abstract

The idea that many macroeconomic variables are unit root processes serves voluminously as a preliminary result in empirical works, but it is just a result of misspecification or weak identification with respect to the structural breaks. This contribution raises the size or power in tests of a null of a stationary process/unit root by Fourier approximation which converts the estimation of location and style of breaks into the problem of appropriate frequency selection. An examination of China’s 15 representative macroeconomic series indicates that only the financial series have good reason to be regarded as unit root processes; most of others are better regarded as trend stationary with smooth transitions. The inference based on these results affirms that China’s real business cycles are indeed fluctuations around different deterministic trends, and it is not the noise component rather the historical events corresponding to the breaks that have persistent effects. The results also support that large government-initiated shocks aimed at improving fundamentals are indeed capable of positive effects on the balanced growth path.