AGGREGATE AND DISAGGREGATE SECTOR FORECASTING USING CONSUMER MEASURES
by Kumar, V., R. Leone and J. Gaskins
The usefulness of the combination of Katona's "ability and willingness to buy" framework and Bayesian vector auto regression for business forecasting was examined. Models were estimated with and without a measure of consumer confidence and with either a vector auto regression model with lag structure determined by a stepwise (VAR) or a Bayesian (BVAR) approach. Additionally, Kantona's monthly data were used in the operationalization of the framework. It was found that the BVAR, along with the consumer confidence measure, performed the best across three forecasting horizons and three performance measures. Any degradation of model performance at lower level of aggregation across expenditure categories was only modest. The findings suggest that the use of BVAR, with consumer confidence index as a predictor should be considered in business forecasting.
International Journal of Forecasting, Vol. 11 (Year 1995)
Printer Friendly Page
