A Monte Carlo approach is suggested for correctly sized backtesting of Value-at-Risk estimates by means of the dynamic quantile test and a Portmanteau statistic. The latter shows preferable power features but fails in case of unconditional VaR misspecification.

We test for causality between inflation and its associated uncertainty by means of both in-sample and out-of-sample modelling. Our findings indicate that the impact of inflation on inflation uncertainty is more pronounced than the reverse causal effect.