This paper studies macroeconomic risks in a panel of advanced economies based on a stochastic volatility model in which macro-financial conditions shape the predictive growth distribution. We find sizable time variation in the skewness of these distributions, conditional on the macro-financial environment. Tightening financial conditions signal increasing downside risk in the short term, but this link reverses at longer horizons. When forecasting downside risk, the proposed model, on average, outperforms existing approaches based on quantile regression and a GARCH model, especially at short horizons. In forecasting upside risk, it improves the average accuracy for several horizons up to four quarters ahead. The suggested approach can inform policymakers’ assessment of macro-financial vulnerabilities by providing a timely signal of shifting risks and a quantification of their magnitude.