On Seasonality in the East Seto Area BSI Series

This paper analyzes the nature of seasonality in quarterly observations for the East Seto Area BSI series. We begin with quantitative mesures of seasonality. Most series have at least 30 percent of their non-trend variation mopped up by seasonal dummy variables alone. We turn to tests of the order of integration. Unit root tests are applied to determine whether the seasonal component in each variable exhibits stochastic nonstationality. 36 series are found to have a seasonal unit root only and the remaining series are almost stationary. This implies that the the ΔΔ(4) transform, frequently implicitly embodied in any seasonal adjustment program, leads to overdifferencing.

論説 (Article)