Lucas and Artificial Worlds

Because an economy is a complex system, economists encounter a gap between theory and application. To do justice to the complexity of the economy, more and more comprehensive models were built. This development was not justified by its results: more comprehensiveness did not lead to better predictions, which was seen as essential for policy evaluation. To deal with this difficulty, two strategies are possible. First, the project of builidng more and more comprehensive models may be eliminated. Second, a new criterium to assess models may be proposed, namely to verify whether they mimic the relevant characteristics of an economy. This new test criterion implies a new methodology for model building. This article explores the development of the second strategy and its methodology, which is influenced by the insights of artificial intelligence into both model building and tests