Fundamental Uncertainty in Model Predictions: Analysis of Modern Macroeconomic Models from the Perspective of Friedman's Instrumentalism

Abstract

In 1966, M. Friedman published an essay on the methodology of positive economics, in which he emphasized the role of predictions as a decisive criterion for accepting macroeconomic models. This paper analyzes to what extent modern macroeconomics in 2023 is guided by these ideas. The first part of this paper deals with the role of predictions in macroeconomic models and shows that a large part of the models currently used do not contain any predictions at all. The next part of the paper explains the problem of Lucas' critique and the resulting complications in making predictions. The following section shows how modern dynamic stochastic general equilibrium (DSGE) models are attempting to deal with Lucas' criticism. The last part of the text analyzes further fundamental complications in making macroeconomic predictions, such as the issue of free human decision-making, the butterfly effect and the normative form of the questions being studied. The paper shows that a portion of modern macroeconomic models that do not contain predictions would undoubtedly fail Friedman's test, and models created for the purpose of prediction still have not completely dealt with serious methodological complications. Due to the nature of the subject matter, macroeconomic models will likely never be able to provide completely reliable predictions.