On the Identification of Diagnostic Expectations: Econometric Insights from DSGE Models

This paper provides the first econometric evidence for diagnostic expectations (DE) in DSGE models. Using the identification framework of Qu and Tkachenko (2017), I show that DE generate dynamics unreplicable under rational expectations (RE), with no RE parameterization capable of matching the autocovariance implied by DE. Consequently, DE are not observationally equivalent to RE and constitute an endogenous source of macroeconomic fluctuations, distinct from both structural frictions and exogenous shocks. From an econometric perspective, DE preserve overall model identification but weaken the identification of shock variances. To ensure robust conclusions across estimation methods and equilibrium conditions, I extend Bayesian estimation with Sequential Monte Carlo sampling to the indeterminacy domain. These findings advance the econometric study of expectations and highlight the macroeconomic relevance of diagnostic beliefs.