Files
Abstract
The forecast performances of the fixed coefficient demand model are
compared with those of spline function and the Cooley-Prescott varying
parameter demand models using consumption and price data for beef, pork,
chicken and turkey. In general, the varying parameter models outperformed
the fixed coefficient model and the spline function varying
parameter model appears to be slightly superior to the Cooley-Prescott
model. However, no single model was consistently superior over all the
commodities in the capacity to predict either the turning points or commodity
levels. Apparently, the explicit specification of structural
change using spline rather than random coefficient model offers some
improvement in commodity forecasting.