This study provides estimates of the real effects of macro-uncertainty de- composed into fundamental and overconfidence bias components. Crucially, overconfidence biases lower ex-ante measures of uncertainty, while fundamen- tal uncertainty raises both ex-ante and ex-post measures. This distinction is useful since the estimates on the real effects of the overconfidence component of uncertainty mitigate endogeneity concerns. I first document evidence for overconfidence biases from survey density forecasts in the US survey of pro- fessional forecasters. Then, using a sign and zero restrictions identification scheme in a vector autoregression (VAR), I find that increases in fundamental uncertainty and declines in overconfidence tend to lower real activity.