The reliability of output gap estimates in real time

Compared to its central role in policy discussions in the United States and most other developed countries, the reliability of the measurement of the output gap has attracted relatively little academic study. Furthermore, both the academic literature and the debate among practitioners have tended to neglect a key factor. Although in a policy setting it is necessary to estimate the current (i.e. end-of-sample) output gap without the benefit of knowing the future, most studies concentrate on measurement that employs data that only become available later. In this paper we examine the reliability of alternative output detrending methods, with special attention to the accuracy of real-time estimates. We show that ex post revisions of the output gap are of the same order of magnitude as the output gap itself, that these ex post revisions are highly persistent and that real-time estimates tend to be severely biased around business cycle turning points, when the cost of policy induced errors due to incorrect measurement is at its greatest. We investigate the reasons for these ex post revisions, and find that, although important, the ex post revision of published data is not the primary source of revisions in output gap measurements. The bulk of the problem is due to the pervasive unreliability of end-of-sample estimates of the trend in output.