The Quantitative Importance of News Shocks in Estimated DSGE Models

We estimate a dynamic stochastic general equilibrium (DSGE) model with several frictions and both unanticipated and news shocks, using quarterly U.S. data from 1954 to 2004 and Bayesian methods. We find that unanticipated shocks dominate news shocks in accounting for the unconditional variance of output, consumption, and investment growth, interest rate, and the relative price of investment. The unanticipated shock to the marginal efficiency of investment is the dominant shock, accounting for over 45% of the variance in output growth. News shocks account for less than 15% of the variance in output growth. Within the set of news shocks, nontechnology sources of news dominate technology news, with wage markup news shocks accounting for about 60% of the variance share of both hours and inflation. We find that in the estimated DSGE model (i) the presence of endogenous countercyclical price and wage markups due to nominal frictions substantially diminishes the importance of news shocks relative to a model without these frictions, and (ii) while there is little change in the estimated contributions of technology news when we restrict wealth effects on labor supply, the contributions of nontechnology news shocks are relatively more sensitive.