CHOICE BETWEEN DISAGGREGATE AND AGGREGATE SPECIFICATIONS ESTIMATED BYINSTRUMENTAL VARIABLES METHODS

A choice criterion is proposed for discriminating between disaggregate and aggregate models estimated by the instrumental variables method. The criterion, based on prediction errors, represents a generalization of criteria developed in the context of classical regression models. The article also derives general tests for aggregation bias in the ins trumental variables context. The criterion and the tests are applied i n an analysis of U.K. employment demand. It is shown that a model disa ggregated by 40 industries predicts aggregate employment better than a n aggregate model and that significant biases exist in estimates of th e long-run wage and output elasticities obtained from the aggregate mo del.