Assessing the Uncertainty of the German NAIRU in a State Space Framework Using Different MSE Approximations

This paper investigates uncertainty around point estimates of the German NAIRU in a state space framework. The relative accuracy of alternative MSE approximations for the states is compared using Monte Carlo simulations. A bootstrap method yields confidence intervals with the highest coverage probability compared with simulation-based and delta method MSE approximations. The degree of uncertainty of the German NAIRU is estimated with different models. Whereas a univariate model of unemployment renders the NAIRU almost uninformative, extending it to a trivariate model consisting of unemployment, inflation and output helps to decrease the uncertainty considerably. However, because there are long periods of time where differences between the NAIRU and observed employment are not significantly different, the overall uncertainty remains high