Macroeconomic risk assessments play an important role in the forecasts of many institutions. A risk forecast is related to the potential asymmetry of the forecast density. In this work, we investigate how the optimality of such risk forecasts can be tested. We find that the Pearson mode skewness outperforms the standard third-moment-based skewness as a measure of asymmetry. We consider problems of the tests likely to be encountered in practice and try to offer remedies where possible. In general, tests for macroeconomic risk forecast optimality tend to have at best moderate power given the empirically available small sample sizes.