Reserve Bank of New Zealand
Estimating the Output Gap: a Kalman Filter Approach
Joint work with L. Christopher Plantier, RBNZ.
Real time estimates of the output gap undergo substantial revisions as more data become available and they are also subject to large uncertainty, i.e. they have large standard errors. In this paper, we estimate the output gap in an HPMV (Hodrick-Prescott Multivariate) context with the Kalman filter. The Kalman filter approach has a number of advantages over the initial approach proposed by Laxton and Tetlow (1992). The Kalman filter approach also allows us to relax/change the assumptions of the HPMV approach to see the differences. In this set up, the Hodrick-Prescott filter is augmented with additional structural equations to infer the unobserved potential output. We use the relationship between output and the capacity utilisation and output and unemployment to infer the unobserved potential output. We find that the real time revisions are much smaller with the HPMV approach compared to an HP filter, although the degree of improvement in real time estimates depends on how the equilibrium/trend values in structural equations are treated. In addition, real-time uncertainty with the HPMV filter is lower than that of the HP filter, but still remains high.