Good Predictions and Bad Policies

Relatively little has been said on economic policy by participants in the debate on the realisticness of assumptions in economic models. What has been said is that a `Friedmanian´ methodology which accepts unrealistic assumptions and is only concerned with correct predictions is appropriate from the perspective of a practical economist who is in charge of designing policy. This paper tries to show that this is not true. Even if a model provides very accurate predictions of an event, its ability to provide valid explanations is determined by the realisticness of its underlying assumptions. Different assumptions yield different explanations and unrealistic assumptions tend to provide no explanation at all. There is a strong relation between the way a phenomenon is explained and understood and the actions that are consequently recommended. Therefore, a model based on unrealistic assumptions is not a reliable source of advice on policy.