Full Program »
On the epistemic value of non-robust toy modelsSimple, highly idealized models can be found across the sciences. In this paper, I shed light on a so far underexplored aspect of these toy models: the fact that results derived from toy models are not robust to changes in the model assumptions. I argue that non-robust toy models can perform two important epistemic functions if one evaluates them indirectly in relation to target systems: they allow identifying a stylized problem with agenda-setting character or provide a template for viewing causal processes which informs model building practices. I illustrate these two functions via Dornbusch’s overshooting model of the exchange rate.
Philosophy, Logic, and Scientific Method
London School of Economics and Political Science