Chameleon models in economics: A note

In a recent article Pfleiderer (2020) argues that models should be put through a 'reality filter' before they should be taken seriously as a basis for empirical testing or policy analysis. He regards models that make dubious assumptions as 'chameleons', because while as theory they may give insights, their empirical performance is artificial and disguises their true nature as impossible beasts. In this note we argue that the correct response to such models is not to reject them ex ante based on a 'reality filter' but to subject them to rigorous empirical testing wherever possible. We describe a methodology in macroeconomics that can be used for this purpose, and we show how this approach avoids several problems that arise from using a 'reality filter' without further scrutiny.