We contribute to the recent literature on the economic effects of those weather conditions that deviate from their regular seasonal pattern. To this end we use local temperature and snow measurements across Germany to analyze their impact on German monthly industrial and construction-sector production. We find noticeable effects of the various (linear and nonlinear, contemporaneous and dynamic) weather regressors, which in the –seasonally adjusted– construction sector growth data imply an extra explanatory power of more than 35% of the variation, compared to benchmark predictive regressions with usual leading indicators. As expected, the impact is quite a bit less in industrial production, in the single-digit range. From our estimates we obtain (seasonally as well as) weather adjusted production series. We also assess the forecasting contribution in pseudo real time settings where publication lags are taken into account, and we apply model averaging techniques to analyze the relevance of abnormal weather for recession forecasting in probit models.
