Brian Dombeck
Previous structural estimates have found that anticipated shocks are responsible for about half of observed fluctuations in macroeconomic data. But endogenous fluctuations in key macroeconomic variables caused by the arrival of news in rational expectations (RE) news-shock models are qualitatively similar to those generated by the adaptive learning process in models which do not feature news shocks but do allow for boundedly rational agents. This paper allows learning and news shocks to compete as explanations explanations for business cycles in a news-rich medium-scale DSGE model using likelihood-based classical and Bayesian methods to determine whether existing estimates of the relative importance of news shocks have been overstated due to model misspecification stemming directly from the assumption of rational expectations.