Univariate Models and Transfer Function Analysis for Complex Economic Processes–Some Results

Abstract

This paper reports some results in economic time series analysis. Appropriate data were selected from the collection of economic time series published monthly by the Statistisches Bundesamt of the Federal Republic of Germany. These data were statistically analyzed and seasonal univariate models of the ARIMA-type were fitted to the time series. Then, transfer function models were built for pairs of related time series thus trying to enhance the model adequacy and prediction ability for the output series. Two examples may demonstrate merits and difficulties of this analysis. The effect of isolated identifiable events on the behaviour of a time series can be described by intervention models. This was investigated for retail sales data.