Assessment criteria for output gap estimates

This paper assesses the statistical reliability of different measures of the output gap for the Euro-11 area and the US using output, inflation and unemployment systems. In order to assess the reliability of an output gap estimate three criteria are adopted. Firstly, the estimate should have forecasting power over inflation. Secondly, the ex post statistical revisions of the output gap should not differ significantly from previously computed measures. Thirdly, the estimate of the output gap should be positively correlated with standard measures of capacity utilization. We find that under our multivariate specification, unobservable components (UC) type models of the output gap show temporal consistency between sequential and final estimates and are consistent with known cyclical indicators. On the other hand, our UC models for the output gap have limited forecasting power for inflation, since they underperform an arbitrary autoregressive model. (C) 2002 Elsevier Science B.V. All rights reserved.