The forecasting horizon of inflationary expectations and perceptions in the EU. Is it really 12 months?

The standard way today to obtain measures of inflationary expectations is to use questionnaires to ask a representative group of respondents about their beliefs of the future rate of inflation during the coming 12 months. This type of data on inflationary expectations as well as on inflationary perceptions has been collected in a unified way on an EU-wide basis for several years. By now, probably the largest database on inflationary expectations has been built up in this way.
We use this database to explore the forecasting horizons implicitly used by the respondents to questions about the expected rate of inflation during the coming 12 months. The analysis covers all EU member states that have relevant data. We examine the forecast errors, the mean error and the RMSEs, to study if the forecast horizon is truly 12 months as implied by the questionnaires. Our working hypothesis is that the forecast error has a U-shaped pattern, reaching its lowest value on the 12-month horizon. We also study the "backcast" error for inflationary perceptions in a similar way.
Our exploratory study reveals large differences across countries. For most countries, we get the expected U-shaped outcome for the forecast errors. The horizon implicitly used by respondents when answering the questions is not related to the explicit time horizon of the questionnaire. On average respondents use the same horizon when answering both questions, e.g. when respondents use a 12-month forecast horizon answering to the question on future inflation, they use the same forward looking horizon when answering to the question on past inflation. We suggest possible explanations for the differences observed.