Improving macroeconomic model credibility: Reducing reliance on frictions through observed inflation expectations

Abstract

This paper reassesses the credibility of rational expectations in dynamic stochastic general equilibrium (DSGE) models by examining their limitations in capturing real-world macroeconomic dynamics. Using the Smets and Wouters model as a baseline, we incorporate various survey-based measures of inflation expectations to evaluate their effects on model fit and forecasting accuracy. The framework is extended to include financial frictions, unemployment, time-varying inflation targets, and news shocks. Results show that incorporating observed inflation expectations improves model fit, reduces posterior uncertainty around shocks, and lowers reliance on ad hoc rigidities such as indexation and investment adjustment costs, particularly in case of U.S. data. However, these expectations do not consistently improve the model’s ability to replicate key macroeconomic moments. The magnitude and direction of improvements vary across model variants and between the U.S. and the Euro area. Robustness checks suggest that observed expectations help mitigate the effects of structural breaks. Overall, the findings enhance model predictions and challenge the sufficiency of relying solely on rational expectations in DSGE models.