Infinite idealizations and approximate explanations in economics

If we take it that, at least in the social sciences, "realistic" implies "finite", then countless economic models involving infinitary assumptions must obviously be classified as unrealistic - for example, models with infinitely divisible goods, a continuum of traders, consumers optimizing over an infinite time horizon, or players optimizing over an infinite number of interactions. We argue that unrealistic models involving infinities can, in principle, supply explanations in economics. We develop a concept of approximate explanation based on the "method of decreasing abstraction", that is, the practice of approximating complex situations through a sequence of increasingly realistic models. Our account of approximate explanation renders the testing view of economic science compatible with scientific realism. However, compatibility does not extend to the folk theorems for infinitely repeated games, which are used widely in applied economics (e.g., in industrial economics) and beyond (e.g., spontaneous emergence of social order). Explanations based on these theorems are rejected by our criterion.