ABSTRACT
Large macroeconometric models have been widely used by government and business for about 25 years and the number and types of model in use is still increasing.1
Their ability to survive and breed is something of a puzzle. Given the widespread criticism to which they have been subjected, one might have expected them to be heading for extinction. The models have been subject to repeated attacks by theorists who argue that they do not represent best practice understanding of the economy and econometricians who argue that they do not represent the data, particularly as systems. In addition, their poor record at ex ante forecasting has been the subject of general scorn. These criticisms are not new. Pesaran and Smith (1985, 1995) discuss how they arose, and they have continued. On the basis of these criticisms many have concluded that the whole enterprise of macroeconometric modelling is inherently flawed, since as the Bank of England (February 1996 Inflation Report, p. 46) put it ‘the economy is too complex and rapidly changing for its behaviour to be captured in any fixed set of equations or “model” of the economy’.
