In the presence of principal-agent problems, published macroeconomic forecasts by professional economists may not measure expectations. Forecasters may use their forecasts in order to manipulate beliefs about their ability. I test a cross-sectional implication of models of reputation and information-revelation. I find that as forecasters become older and more established, they produce more radical forecasts. Since these more radical forecasts are in general less accurate, ex post forecast accuracy grows significantly worse as forecasters become older and more established. These findings are consistent with reputational factors at work in professional macroeconomic forecasts.