A running issue with regard to banking regulation is the extent to which regulators should move away from risk-sensitive capital requirements towards simpler requirements, such as the leverage ratio. The Basel Committee has released new proposals on the internal ratings-based approach for credit (the IRB) which build on the thinking that risk-based capital requirements are less important (BCBS 2016a). The Committee suggest simpler methods of setting capital requirements for some portfolios (to replace current modelling of the risks) and more use of floors beneath modelled parameters. Also, although thinking has shifted during the debate to leverage being a backstop to the risk-based requirements, this is in practice changing with leverage becoming a front stop for a number of banks (EBA 2016) and will shift more if higher leverage ratios are introduced for globally significant financial institutions (GSIFIs) (BCBS 2016b). In this column, I look at the evidence that has influenced the debate on simple versus complex credit risk requirements and shows that none of the papers to date has tested the risk-based credit requirements of Basel II against leverage. The column sets out two new tests that do test Basel II and produce a different result from the earlier papers, highlighting the importance of risk sensitivity.
A number of papers have influenced the debate and were taken as indicating that the Basel II risk-based credit requirements agreed in 2004 were misguided. Demirguc-Kunt et al. (2010) look at the relationship between capital ratios pre-Crisis (both the Basel capital ratio and the leverage ratio) and stock market returns during the crisis and find for the largest banks, it is the leverage ratio that is most significant. Haldane and Madouros (2012) take a sample of 116 large banks and looks at the relationship of both the Basel capital ratio and the leverage ratio pre crisis with subsequent failure/survival during the crisis using a logit approach. What is not well understood is that for the major banks the risk-based capital requirements of Basel II were not introduced until 2008 and were not used in the US; the comparison in these papers is therefore between the leverage ratio and the quasi leverage ratio of Basel I – it had a fixed capital requirement of 8% for most private sector credit exposures. Blundell-Wignall and Roulet (2013) take 94 US and EU banks and look at the relationship between the distance to default in the period 2004-2011 and both the Basel capital ratio and the leverage ratio, finding for the latter. The four years from 2004 to end-2008 were Basel I, and throughout the period the US banks were not on Basel II credit risk requirements. For all the papers, therefore, the results are reflecting or largely reflecting Basel I requirements.
This raises the question why the leverage ratio is outperforming the quasi leverage ratio of Basel I. Re-running the Haldane and Madouros logit regressions but including the loan leverage ratio as well as the leverage ratio as explanatory variables for later failure or survival shows the former is not significant. The results are therefore dominated by the trading books of the banks. The dominant factor here was the regulatory arbitrage enabled by the simple Basel I requirements. Illiquid credit exposures such as loans being warehoused to go into RMBS structures could be placed in the trading book requirements which assumed short holding periods. The Basel Committee has announced changes to deal with this gap.
Given the timing of the introduction of Basel II, the actual failures during the Crisis cannot be used to test the Basel II requirements against the leverage ratio; this requires new tests. One approach is to use simulated bank failures. The combination of the 2014 asset quality review (AQR) in Europe with a severe macroeconomic stress test mimics a crisis (ESRB 2014). The AQR, which forced additional provisioning for existing weak credits, mirrors the winnowing of borrowers in a severe economic environment and the stress test assessed the effect of severe economic conditions on new arrears, feeding into further provisions and write-offs and the write-down of the trading book. Using 104 European banks that took part in the stress test – of which 20 failed the test – it is possible to construct a logit analysis of the predictors of survival or failure – Basel II risk -based requirements or the leverage ratio. Figures 1 and 2 show the clustering of banks with a low core Tier 1 ratio that failed the test and no pattern for the leverage ratio.
Figure 1: Risk-based core Tier 1 capital ratios of EU banks participating in the 2014 EBA stress test, end-2012
Figure 2: Core Tier 1 leverage ratio of EU banks participating in the 2014 EBA stress test, end-2012
Note: Figures based on a sample of 104 banks, 84 who successfully passed the stress test and 20 who were considered by the ECB to have failed.
Source: Published accounts.
A logit regression shows that the core Tier 1 capital ratio is significant at the 1% level, whereas the leverage ratio is not significant. This is of course a more stylised test than a full blown crisis, but it is to a significant degree focused on credit quality, which is what the Basel II risk-based credit requirements are trying to measure.
Another way to test the Basel II requirements against the leverage ratio is look at which of the two is an important explanatory variable for CDS spreads. Taking a panel dataset covering 30 large banks globally (but excluding the US banks which did not implement Basel II credit risk requirements) for the period 2008 to 2015 (i.e. the period of Basel II), and regressing CDS spreads in Q2 against the earlier end-year Tier 1 capital ratios and leverage ratios and a measure of the sovereign debt crisis in Q1, the Tier 1 ratio is significant at the 1% level whereas the leverage ratio is not significant.
This highlights the importance of tests of capital requirements distinguishing between the Basel I period, which ran up to end 2007, and the Basel II period from 2008 onwards. Basel I was a quasi-leverage ratio for credit risk – with little risk differentiation. It was with Basel II that the risk-based credit risk requirements came into force. If the analysis is focused on Basel II against leverage, using two new tests, then Basel II comes out as the more important indicator of solvency risk.
Author's note: This column is based on "Simpler capital Requirements versus Risk-Based - The Evidence", in SUERF Conference Proceedings 2016/2: Banking Reform.
Basel Committee on Banking Supervision (BCBS) (2016a), "Reducing variation in credit risk-weighted assets –constraints on the use of internal model approaches", Consultative Document.
Basel Committee on Banking Supervision (BCBS) (2016b), "Revisions to the Basel III leverage ratio framework", Consultative document.
Blundell-Wignall, A. and C. Roulet (2013), “Business models of banks, leverage and the distance-to-default”, OECD Journal: Financial market trends 103.
Demirguc-Kunt, A. E. Detragiache and O. Merrouche (2010), “Bank Capital: Lessons from the Financial Crisis”, Journal of Money, Credit and Banking 45, 1147–1164.
European Banking Authority (EBA) (2016), CRD IV- CRR/Basel III monitoring exercise – results based on data as of 30 June 2015.
European Systemic Risk Board (ESRB) (2014), EBA/SSM stress test: The macroeconomic adverse scenario.
Haldane, A. and V. Madouros (2012), “The dog and the Frisbee”, speech given at the Federal Reserve Board of Kansas City’s 36th Economic Policy Symposium, Jackson Hole, 31 August.