Macroeconometric modeling as a "photographic description of reality" or as an "engine for the discovery of concrete truth" ? Friedman and Klein on statistical illusions

This paper discusses a longstanding debate between two empirical approaches to macroeconomics: the econometrics program represented by Lawrence R. Klein, and the statistical economics program represented by Milton Friedman. I argue that the differences between these two approaches do not consist in the use of different statistical methods, economic theories or political ideas. Rather, these differences are deeply rooted in methodological principles and modeling strategies inspired by the works of Léon Walras and Alfred Marshall, which go further than the standard opposition of general vs. partial equilibrium. While Klein's Walrasian approach necessarily considers the economy as a whole, despite the economist's inability to observe or understand the system in all its complexity, Friedman's Marshallian approach takes into account this inability and considers that economic models should be perceived as a way to construct systems of thought based on the observation of specific and smaller parts of the economy.