Systemic risk and macroeconomic activity: International evidence

In this paper, we propose an easily implementable combination forecast method to examine whether systemic risk can predict macroeconomic activity. Our method uses various weighting schemes to combine common individual systemic risk measures. Empirical results demonstrate that our novel forecasting strategy provides better and more stable out-of-sample performance than a set of established methods. Furthermore, the forecast combination puzzle identified by Stock and Watson (2004) is also present in systemic risk. Our findings further reveal that institution-specific risk measures, such as and , consistently provide reliable information to inform predictions of future macroeconomic downturns across most regions. The strength and direction of those asymmetric relationships are closely associated with the risk levels (i.e., quantiles of independent variables) of systemic risk indicators. • We propose a novel forecast combination with three types of combining weights. • The forecast combination puzzle presents in the systemic risk. • This paper exploits the heterogeneity in developed and developing economies. • The strength and direction of the asymmetric relationships closely associate with the risk-levels of systemic risk indicators.