A Case Study in Prediction: The Market for Watermelons
Wilford L. L'Esperance
This paper discusses two forecasting experiments involving models of the water-melon market. The first experiment compares the forecasts of an interdependent model estimated by limited information, single equation with those of a model using least squares reduced form. The second experiment compares the forecasts of the interdependent model with those of a causal chain model. It is found that the forecasts of the interdependent model are generally better than those of the alternative models.