Modelling seasonal variation
The realisation among econometricians and applied economists that seasonal variation in many economic time series is often larger and less regular than often hitherto acknowledged, has led to an increased interest in seriously modelling seasonality. The relative size of the seasonal variation also means that such modelling is of major economic interest.
In the last few years important developments in modelling seasonality have occurred. The use of model based procedures and periodic models, the extension of integration and cointegration to the seasonal frequencies and the development of economic theories of seasonality are some of the most promising areas of research. However, the procedures applied by the official data-producing agencies have not been affected. In this paper we discuss the implications of these developments on the treatment of seasonality within the data-producing agencies.
Modelling seasonal variations has recently been subject to a number of studies of both econometricians and economics theorists. A major cause for this development is the realization that the seasonal components account for the major part of the variations in many economic time series, that the seasonal variation is far from regular but varying and changing, and that economic considerations are playing an important part in the analysis. Hence, the attitude of many economists expressed in the following quotation by Jevons is yielding at last. Jevons ( 1862, p. 4) wrote____________________