Earnings growth dispersion contains information about trends in labor reallocation, unemployment change, and, ultimately, aggregate output. We find that initial macroeconomic estimates released by government statistical agencies do not fully incorporate this information. As a consequence, earnings growth dispersion predicts future restatements in nominal and real GDP growth (and unemployment change) both in the in-sample and out-ofsample tests. Further, when we adjust GDP estimates using the out-of-sample restatement predictions, we find statistically and economically significant effects for the monetary policy prescriptions (Taylor rule) and banking regulation (Basel III).