Linking bank regulatory capital buffer to business cycle fluctuations

Abstract

PurposeThe purpose of this paper is to assist bank regulators in Africa who are currently considering the implementation of Basel III countercyclical capital buffer (CCB) requirement.Design/methodology/approachUsing a panel data set of 129 commercial banks operating in 14 African countries over the period 2004–2014, this paper estimates the system generalized method of moments regression to examine the impact of business cycle on banks’ regulatory capital buffers and attempts to identify the influence of bank revenue diversification, market power and cost of funding (CF) on bank regulatory capital buffers. It further carries out some robustness analyses using a panel data set of 257 commercial banks in 23 African countries over the period 2004–2014.FindingsThe results show that higher regulatory capital buffers are associated with higher market power, higher revenue diversification and higher CF. Additionally, the results show significant evidence of procyclical behavior of bank capital buffers (BUFs) in the sampled countries.Practical implicationsThe results of this study suggest that African banking systems are not exposed to contagion and systemic risks arising from countercyclical movements of BUFs to the real economy. Therefore, this study does not support the implementation of the Basel III CCB requirement in the sampled African countries.Originality/valueConsidering that the results of existing studies on the cyclical behavior of BUFs are inconclusive, there is value in studying the cyclical movements of bank regulatory capital buffers in a set of countries that has not been analyzed before. Toward this direction, this is the first empirical study focusing on the cyclical behavior of bank regulatory capital buffers in Africa. Besides examining the cyclical behavior of bank regulatory capital buffers, this paper further investigates the effects of bank revenue diversification, market power and CF on bank regulatory capital buffers.