Comparing Macroeconometric Models: A Review Article

Abstract

no effective methods for deciding which conclusion is correct. In consequence, contradictory hypotheses continue to coexist sometimes for decades or more.' This is not a good reason for abandoning applied economics altogether, however, for then there would be 'almost no way of selecting from among a plethora of possible explanations the one that best explains economic events'. For Blaug, the only hope is to improve both theoretical and applied econometrics, and specifically the workaday practices of the latter; and he endorses a number of suggestions by Mayer (1980) to strengthen the claims of economics as a 'hard science'. Given that econometric results can seldom be treated as evidence from a single 'crucial experiment', but rather that evidence accumulates from a sequence of separate studies, 'then it would be most useful to have, from time to time, surveys that pull the evidence together, in the sense not just of describing what various studies show, but by resolving any contradictions between them. At present on too many questions we are buried in an inchoate mass of seemingly contradictory evidence' (Mayer 1980, p. 173). In commenting on a survey of this kind, Diamond (1985) wishes that 'we had more neutral evaluations of empirical results to complement existing debates in the literature', while also being disappointed that the author of the survey in question did not divulge his own views on the credibility of the results surveyed. Having undertaken such a survey himself, Atkinson (1988) notes that 'all too few authors make a serious attempt to fit their findings into their place in the literature, indicating how the difference in models, methods, or data lead the results to be different'. Macroeconometric modelling is the area where most comparative work has been done, and there is a model comparison literature extending back to the 1950s. This article is a review of the latest contribution to that literature,* in the light of what has gone before, and in the light of the different responses that have emerged to the various criticisms in the preceding paragraph. As Klein notes at the outset, the main world centre for modelling work has been the United States, and, correspondingly, most of the model com