Although most studies look at the role of either the capital ratio (CR) or the leverage ratio (LR) in past credit crises, the relationship between the CR and LR is more useful for future policy because changes to the risk-weights categories affect the correlation patterns between the ratios. An appropriate CR can be set individually versus one-size-fits-all ratios.
The capital requirements set by the Basel Committee on Banking Supervision (BCBS) have failed to prevent the credit crises experienced in 1990–1991 and in 2007–2009, even though banks were overcapitalized prior to the 2007–09 crisis. Although many focus on either the capital ratio (CR) or the leverage ratio (LR) when considering capitalization, the authors examine the relationship between the two ratios and how the correlation fluctuates as regulators change the number of risk-weights categories.
How Is This Research Useful to Practitioners?
Regulations designed to maintain banking stability affect the behavior of banks and the assets they hold. Even though the CR and LR differ only in their denominators, banks will behave differently depending on which ratio is the binding constraint. Prior research studied the effects of capital requirements either by analyzing the impact on lending growth or by focusing on risk incentives; most of these studies focused on the 1990–91 crisis and study either the CR or the LR. The authors complement prior research by looking at the introduction of new risk-weights categories and how the number of categories affects the relationship between the two ratios.
Key findings include that all banks went bankrupt if they failed either the Tier 1 CR or a 3% LR pre-crisis. Also, if a bank failed Tier 1, then it also failed the LR requirement of 4% during all three phases of the crunch. More banks failed because of the minimum 3% LR than the CR requirement, suggesting that a 3% requirement could be overly conservative. The authors find that Basel I played a role in the first crunch but that Basel II cannot be blamed for the second crunch and that the LR had a hand in triggering the second crunch. They find a mathematical relationship in the sensitivity of the CR to a change in the risk weights and suggest setting a CR that takes into account changes in risk weights as well as the implications for the LR.
How Did the Authors Conduct This Research?
First, the authors conduct a bank failure analysis looking at CR and LR requirements and find that banks’ survival is determined more heavily by Basel requirements than was found by prior research by Avery and Berger (Journal of Banking & Finance 1991). The authors then propose that the CRs can be combined to set more appropriate and less heuristic targets for the CR and LR. Studying the correlation changes between the LR and CR and how the changes relate to lending and GDP implies a relationship that could serve as an economic signal for regulators, which is applied against the two crunches for verification.
The authors study the effects of the standards coupled with the leverage requirements in three stages: pre-, mid-, and end of crisis. The dataset used is based on call reports from the Federal Deposit Insurance Corporation for the crunch periods of Q1 1991 to Q2 1991 and Q3 2007 to Q2 2009. Risk-weights data are only available as of 1990, when the recession started, instead of 1988; thus, the dataset is constrained for the second crunch to control for bias. The CR and LR are interconnected to Tier 1 capital by definition. The CR is defined as K/RWA, where K represents Tier 1 capital and RWA represents risk-weighted assets. The LR is defined as K/A, where K represents Tier 1 capital and A represents assets. The subsamples of the dataset are based around the distribution of the CR.
With two credit crises within a 20-year time frame, this research provides insight into how such crises can be prevented going forward. Because the economics of CRs is more nuanced than what theory might suggest, regulators should consider policies that address the behavioral aspect of the banks with regard to changing risk-weights categories and correlation patterns of the CR and LR.