Do data revisions matter for DSGE estimation?

This paper checks whether the coefficient estimates of a famous DSGE model are robust to macroeconomic data revisions. The effects of revisions are captured by rerunning the estimation on a real-time data set compiled using the latest time series available each quarter from 1997 through 2015. Results show that point estimates of the structural parameters are generally robust to changes in the data that have occurred over the past twenty years. By comparison, estimates of the standard errors are relatively more sensitive to revisions. The latter implies that judgements about the statistical significance of certain parameters depend on which data vintage is used for estimation.