Inflation expectations in Phillips Curves models for the euro area

This paper takes a fresh look at the use of the Phillips curve and various in ation expectation proxies for tracking euro area in ation dynamics in the aftermath of the global nancial crisis of 2008. Because in ation expectations can be measured in a multitude of alternative ways and the Phillips curve model itself is subject to many potential speci cation choices, we employ a novel thick modelling perspective that is data and model-agnostic and estimate a large number of di erent Phillips curve models using di erent data series for di erent components of our models. We nd that Phillips curve models without any forward-looking expectational terms are uniformly the worst predictors of euro area in ation rates after 2013, when measured for the RMSE criterion across all models and speci cations. This result underlines the importance of in ation expectations in tracking the recent dynamics of euro area in ation and shows that in ation persistence alone or in combination with di erent slack and cost push terms cannot satisfactorily explain the euro area in ation story during the period of missing in ation after 2012. We also illustrate the usefulness of the thick modelling approach for practical modelling and forecasting of the euro area in ation series.