Abstract:
Probit models are employed to evaluate leading indicators for Germany's recessions. The predictive power of leading indicators is found to be lower than assumed in previous studies. Although, monetary variables provide the best predictive power for recessions, survey data and order inflows show a lag rather than a lead to the recession time series. US interest rates have also some information content with respect to the German cycle. Constructing a model with a set of variables to predict recessions does not help to improve the forecasts. The out-of-sample performance of the indicators appeals to be even worse.
