An Application Of A New Seasonal Unit Root Test For Trending And Breaking Series To Industrial Production Of The Brics

Seasonality is a quite dominant source of variation in many economic time series. Several empirical studies have shown that seasonality is not a stable component but rather of stochastic nature. In this spirit, literature has seen the development of several seasonal unit root tests by generalizing the frameworks of the conventional (non-seasonal) ones. In general, existing tests for seasonal unit roots are heavily dependent on some restricted specifications for the deterministic component or they are interested in a particular observation frequency. The seasonal unit root test of Hylleberg et al. (1990) [HEGY] is the most in empirical works. The strong point of this procedure is that it can distinguish between different seasonal unit roots. However, such a procedure does not take into account the breakpoints characterizing some macroeconomic data. In this study, we investigate the persistence property of the quarterly industrial production of Brazil, Russia, India, China and South Africa, covering the period of early 1990s till the last quarter of 2013, with the industrial production for BRICS characterized by strong seasonality and upward trend. For our purpose, we use the seasonal unit root test developed by Popp (2007), which controls for not only seasonal and trending behavior, but also for a break of unknown timing in seasonal means. Specifically, Popp’s (2007) test is based on an HEGY testing strategy with innovational outliers. It has been shown that this test has a stable size even in the presence of large breaks. Also, it can accurately identify the dates of small breaks. Our results indicate that, while there exist a non-seasonal unit root in the industrial production for Brazil, India and China, there is a strong evidence a seasonal unit root at Nyquist frequency for only China. In addition, there is a strong evidence of rejection of bi-annual unit roots in all BRICS industrial production. Using Jaccard (1902) coefficients for similarities, the greatest ones are recorded for Brazil and South Africa on the one hand, and Russia and South Africa on the other hand. By contrast, Russia and India have no similarity. Accordingly, some policy implications can be formulated for China as it is the only country, among the BRICS, which exhibits a seasonal unit root in its industrial production. Our results therefore, suggest that economists should examine the evolution of industrial production in China not only at the business cycle frequencies but also at the seasonal as well.