Forecasting state retail sales: Econometric vs. time series models

Abstract

The volume of retail sales is an important indicator of state economic activity and forms the base of the percentage sales tax. Accurate forecasting of the variable is of interest to fiscal authorities and private analysts as well. This paper compares the performance of two techniques in forecasting net taxable retail sales: the ARIMA time series model and a structural model which is representative of the econometric approach. Four measures of forecast accuracy are calculated and the relative merits of the techniques are discussed. The ARIMA model was found to perform better than the structural model during the evaluation period used in the analysis. However, it is also shown that a predictor formed from combining the ARIMA and structural model predictors may be superior to exclusive use of the ARIMA.