Implications of Diagnostic Expectations : Theory and Applications

A large psychology literature argues that decision-makers' forecasts of their future circumstances appear overly influenced by their perception of the new information embedded in their current circumstances. We adopt the diagnostic expectations (DE) paradigm (Bordalo et al., 2018) to capture this feature of belief formation and develop the micro-foundations for applying DE to a broad class of macroeconomic models. In this environment, DE apply to both exogenous and endogenous variables. We derive three theoretical properties of DE in the presence of endogenous variables: (i) endogenous predictability, (ii) endogenous non-stochasticity, and (iii) the failure of the law of iterated expectations under distant memory. We show that these properties imply (i) a joint determination of actions and DE; (ii) the possibility of silencing DE by policy actions; (iii) the possibility of time-inconsistency. We analyze two approaches to deal with the issue of time inconsistency: naivete and sophistication. We illustrate our analysis' relevance in two applications. First, we provide a portable solution algorithm to incorporate DE into recursive linear models. In an RBC model, DE generate rich and novel propagation dynamics and a boom-bust cycle. Second, a Fisherian model shows that policy makers' behavior has pervasive macroeconomic effects by activating or silencing DE