From small to large: A systematic comparison of gradually more complex econometric models

This paper treats the problem of validating macroeconometric models. A set of nine models are estimated on the same data base and subjected to the same simulation tests. They range from the simple IS model to a relatively realistic model of the Swedish economy in the postwar period. The main conclusion is that minor specification changes may drastically alter the dynamic characteristics of models, despite the fact that the models are theoretically plausible. A second purpose of the paper is to try to bridge the gap between the simple IS or IS-LM models used in the classroom and the much larger and more complex models that are actually used.