Weighing up the credit-to-GDP gap : a cautionary note

It has been argued that credit-to-GDP gaps (credit gap) are useful early warning indicators for banking crises. In addition, the Basel Committee on Banking Supervision has also advocated using these gaps - estimated using a one-sided Hodrick-Prescott filter with a smoothing parameter of 400,000 - to inform policy on the appropriate counter-cyclical capital buffer. We use the weighted average representation of the same filter and show that it attaches high weights to observations from the past, including the distant past: up to 40 lags (10 years) of past data are used in the calculation of the one-sided trend/permanent component of the credit-to-GDP ratio. We show how past data that belongs to the 'old-regime' prior to the crises continue to influence the estimates of the trend for years to come. By using narrative evidence from a number of countries that experienced deep financial crises, we show that this leads to some undesirable influence on the trend estimates that is at odds with the post-crisis environment.