Countercyclical Capital Buffers: bayesian estimates and alternatives focusing on credit growth

We re-evaluate the proposed framework of the Basel Committee on Banking Supervision (BCBS) to look into the credit-to-GDP gap as a leading indicator related to the Countercyclical Capital Buffer (CCB) and propose an alternative approach focusing at credit-to-GDP growth. We follow earlier work that the Hodrick-Prescott (HP) filter, especially with the proposed smoothing factor calibration, HP(400k), could possibly create spurious cycles. Moreover, it would not properly fit short credit series. With that in mind, we estimate Bayesian STMs for 34 countries and evaluate on-line (one-sided) estimates of their state components as well as other variables derived from their joint posterior distributions to anticipate crisis. The probabilities associated with the slope of the credit-to-GDP estimated using a one-sided STM have lower noise-to-signal ratios (NS) than the credit-to-GDP gap, especially considering a robustness exercise comprise of short series. The slope of the one-sided HP(150), which is simpler but closely related to our STM in its gain function, also performs better in anticipating crisis both in short and long series when compared to the credit-to-GDP gap. Finally, we put forward an exercise of CCB using the last available data point and our five leading indicators in all 34 countries