Regional forecasting models typically involve sets of explanatory variables that are highly collinear. As a result, the specifications and the forecasts themselves are highly sensitive to minor changes in specification and additions or deletions to the sample observations. This present paper illustrates how persistent effects derived from regional production relations can be incorporated in estimating regional forecasting models. Inclusion of these effects is shown to reduce problems of multicollinearity and to improve the structural representation and the forecasting performance of regional econometric models.
ASJC Scopus subject areas
- Economics and Econometrics