The Real-Time Properties of the Bank of Canada's Staff Output Gap Estimates

摘要

We study the revision properties of the Bank of Canada's staff output gap estimates since the mid-1980s and show that the average revision has been significantly smaller since the early 2000s. Alternatively, revisions from econometric output gap estimates have not experienced a similar improvement. We show that the overestimation of potential output in real time following the 1991-92 recession explains the large revisions in the first half of the sample. Although Phillips-curve inflation forecasts slightly worsen when conditioned on real time instead of final gaps, their relative poor performance reflects the general lack of inflation predictability rather than real-time gap measurement issues.