Forecasting and Seasonal Adjustment
There have been many studies and papers written about the effects of seasonal adjustment on the relationships among variables. However, there has been a dearth of studies about the effects of seasonal adjustment on the problem of forecasting. Since the development of time series models often has forecasting as a major product, it is essential to study the effects of seasonal adjustment on forecasting in these models. In this paper, we present an application of mu1tivariate time series forecasting applied to five economic time series, in which we compare forecasts developed from seasonally adjusted data with forecasts from seasonally not-adjusted data. The results of this exercise are mixed. For some forecasting situations, using not-seasonally adjusted data provides better forecasts for most of the variables In this study. However, in other instances, using seasonally adjusted data provides better forecasts for most of the variables in this study. The results appear to depend on the length of the forecast period. A1 so, it appears that the best solution in some instances might be to develop model s for both seasonal1y adjusted data and not-seasonally adjusted data.
Suggested citation: Bagshaw, Michael L., 1985. “Forecasting and Seasonal Adjustment,” Federal Reserve Bank of Cleveland, Working Paper no. 85-07.