MICRO VERSUS MACRO ACREAGE RESPONSE MODELS: DOES SITE-SPECIFIC INFORMATION MATTER?

Because requisite micro data frequently are unavailable, it is common practice to use aggregate data to estimate economic relationships representing the behavior of individual agents. A substantial body of literature has examined conditions under which inferences between micro and aggregate specifications can be made. Less attention has been focused on the relative accuracy of predictions for each scale of model. In an empirical application, we compare the goodness-of-fit measures of eight sets of acreage response models, varying in aggregation from field- (micro-) level to regional- (macro-) level models. Results suggest aggregate models are superior to the micro model in predicting acreage response, even thought the micro models contain substantially more data on site-specific characteristics.