Is it one break or ongoing permanent shocks that explains U.S. real GDP?

The relative importance of permanent versus cyclical shocks to GDP has been found to depend on the presence or absence of a single break in mean growth. We estimate unobserved components models conditional on a trend break having occurred in any specified quarter and use the Bayesian model averaging to combine the conditional estimates. We estimate a break occurred around 2006:1. Allowing for a break significantly reduces estimates of trend variance. However, enough spread remains in the posterior distribution to indicate that available data does not definitively settle the question of the relative importance of trend versus cycle.