CIRJE-J-135. Kunitomo, Naoto, "Seasonality and Seasonal Switching Time Series Models", July 2005.

In the recent X-12-ARIMA program developed by the United States Census Bureau for seasonal adjustments, the RegARIMA modeling has been extensively utilized. We shall discuss some problems in the RegARIMA modeling when the time series are realizations of non-stationary integrated stochastic processes with fixed regressors. We propose to use the seasonal switching autoregressive moving average (SSARMA) model and the regression SSARMA (RegSSARMA) model to cope with seasonality commonly observed in many economic time series. We investigate the basic properties of the SSAR (seasonal switching autoregressive) models. We argue that the phenomenon called "spurious seasonal unit roots" could be an explanation for a good fit of the seasonal ARIMA models to actual data. Some results of economic data analyses are reported.