Regime‐Dependent Nowcasting of the Austrian Economy

Abstract

We nowcast and forecast economic activity in Austria, namely, real gross domestic product (GDP), consumption, and investment, which are available at a quarterly frequency, using a preselected number of monthly indicators based on a combination of statistical procedures. We consider regime‐dependent and non–regime‐dependent mixed data sampling approaches and compare their forecast and nowcast accuracies in terms of the root mean square error and the mean absolute error. We are particularly interested in whether explicitly considering different regimes improves the nowcast. We examine business cycle‐related regimes (good/bad economic times) and financial uncertainty regimes (high/low uncertainty) and compare regime‐dependent and non–regime‐dependent models applying, among others, forecast combination methods. We find strong evidence that taking explicit account of regimes improves nowcasting and that only a handful of variables are important for nowcasting. In addition, different variables are important in different regimes. We observe, for example, that for GDP, in bad times, real industrial production matters more than its survey counterpart, namely, production expectations, and the other way around. The most important predictor for consumption in both regimes is bank loans to households, while for investment labor market indicators are most relevant. For all target variables, industrial production is more important in bad times than in good times.