|
|
2009-11-25 Measuring Output Gap Uncertainty
Anthony Garratt
(Birkbeck College)
James Mitchell
(NIESR)
Shaun P. Vahey
(Melbourne Business School)
We propose a methodology for producing density forecasts for the output gapin real time using a large number of vector autoregessions in inflation and outputgap measures. Density combination utilizes a linear mixture of experts frameworkto produce potentially non-Gaussian ensemble densities for the unobserved outputgap. In our application, we show that data revisions alter substantially our probabilisticassessments of the output gap using a variety of output gap measures derivedfrom univariate detrending filters. The resulting ensemble produces well-calibratedforecast densities for US inflation in real time, in contrast to those from simple univariateautoregressions which ignore the contribution of the output gap. Combiningevidence from both linear trends and more flexible univariate detrending filters inducesstrong multi-modality in the predictive densities for the unobserved outputgap. The peaks associated with these two detrending methodologies indicate outputgaps of opposite sign for some observations, reflecting the pervasive nature of modeluncertainty in our US data.
|
|
|
