Measuring Output Gap Nowcast Uncertainty

We propose a methodology to gauge the uncertainty in output gap nowcasts across a large number of commonly deployed vector autoregressions in US inflation and various measures of the output gap. Our approach constructs ensemble nowcast densities using a linear opinion pool. This yields well-calibrated nowcasts for US inflation in real time from 1991q2 to 2010q1, in contrast to those from a univariate autoregressive benchmark. The ensemble nowcast densities for the output gap are considerably more complex than for a single VAR specification. They cannot be described adequately by the first two moments of the forecast densities. To illustrate the usefulness of our approach, we calculate the probability of a negative output gap at around 45 percent between 2004 and 2007. Despite the Greenspan policy regime, and some large point estimates of the output gap, there remained a substantial risk that output was below potential in real time. Our ensemble approach also facilitates probabilistic assessments of “alternative scenarios”. A “dove” scenario (based on distinct output gap measurements) typically raises substantially the probability of a negative output gap (including 2004 through 2007) but has little impact in slumps, in our illustrative example