Evaluating the European Central Bank's uncertainty forecasts

While several studies have examined the point forecasts by central banks, few have focused on interval forecasts. This study examines the interval forecasts for inflation and growth made by the European Central Bank (ECB). The ECB states that each of its interval forecasts will cover the corresponding realization with a probability of 57.5%. I show that their actual coverage is always larger, and that the differences from the nominal coverage of 57.5% often turn out to be significant. Yet, tests concerning the dynamics of the actual coverage do not reject the null hypothesis of independence, such that correct conditional coverage cannot be rejected in most cases. Moreover, I find that, in general, the actual coverage of the interval forecasts cannot be predicted by the state of economic uncertainty, but the actual coverage of the 7-month-ahead interval forecast for growth declines if financial market uncertainty increases.