Why Diagnostic Expectations Cannot Replace REH

Gennaioli and Shleifer (GS) have proposed diagnostic expectations (DE) as an empirically-based approach to specifying participants’ expectations, which, like REH, can be applied in every model. Beyond its supposedly general applicability, GS’s formalization of DE implies that participants systematically and predictably overreact to news. Here, we present a formal argument that Kahneman and Tversky’s compelling empirical findings, and those of other behavioral economists, do not provide a basis for a general approach to specifying participants’ “predictable errors.” We also show that the overreaction of participants’ expectations is not a regularity, but rather an artifact of GS’s particular specification of DE.