Modelling inflation: different goals call for different solutions

Analysing prices behavior is not a unambiguous matter, it requires different methodologies depending on the aim of the study, on the period of time considered and on the available data. In what follows, we explore three different ways to model prices dynamics responding to alternative aims. We start from considering sector-level inflation indexes identifying the method still used in the Banca d'Italia to conduct the NIPE exercise; secondly, we consider individual prices in order to assess the degree of price stickiness in France, Germany and Italy using a factor model able to identify the effects of different kinds of shocks on prices at different level of aggregation; finally, we test the ability of the factor model in forecasting the overall inflation index of the same three Euro countries finding a significant forecasting power in the unobservable factors. Our results confirm that the data and the period of time considered can lead to quite different outcomes; moreover, different models can alternatively be the most precise in foreseeing inflation depending on the horizon of prediction or the country considered. In conclusion, the best way to analyse prices behavior is peculiar to the aim of the study: different methodologies, in fact, can be the most appropriate for different exercises.