Re-anchoring countercyclical capital buffers: Bayesian estimates and alternatives focusing on credit growth

• Published in: International journal of forecasting Vol. 33; no. 4; pp. 1007 - 1024
• Main Authors: Gonzalez, Rodrigo Barbone, Marinho, Leonardo Sousa Gomes, Lima, Joaquim Ignacio Alves de Vasconcellos e
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2017
• Subjects: Banking crisis; Bayesian STM; Countercyclical capital buffer (CCB); Financial cycle; Noise-to-signal (NS); Noise-to-signal (NS); Banking crisis; Countercyclical capital buffer (CCB); Financial cycle; Bayesian STM
• ISSN: 0169-2070; 1872-8200
• DOI: 10.1016/j.ijforecast.2017.04.006

We re-evaluate the Basel Committee on Banking Supervision’s (BCBS) proposed framework with the credit-to-GDP gap as an anchor indicator relative to the countercyclical capital buffer (CCB), and propose an alternative approach that focuses on credit-to-GDP growth. We estimate Bayesian structured time series models (STM) fully and recursively for 34 countries and evaluate whether these state components and related indicators can anticipate crises over the following three years. Using an early warning framework similar to the original BCBS one, we find leading indicators that outperform the credit-to-GDP gap in anticipating banking crises, with lower noise-to-signal (NS) ratios and similar sensitivities to threshold variation (assessed using receiver operational characteristics (ROC)). Moreover, the credit-to-GDP gap fails an exercise using limited information, suggesting that the 10% anchor on this indicator that was put forward by the BCBS can be misleading in countries with short credit series. Finally, we present an illustrative panel of CCB use with our leading indicators.