Capital Adequacy and Hidden Risk

by Mike Mariathasan and Ouarda Merrouche

Originally published 29 June 2013 at

The regulation of bank capital has recently come under renewed scrutiny. This column argues that the way we implement capital regulation needs to be reconsidered because banks under-report risk, thereby escaping government intervention and maintaining market access. One possible way forward, something already implemented under Basel III, is to ask banks to satisfy a capital requirement relative to total (rather than risk-weighted) assets. Overall, simple, transparent, workable rules are what we should be aiming for.

The regulation of bank capital has recently come under renewed scrutiny. While some commentators argue for higher requirements (e.g. Admati and Hellwig 2012), others – and the banking industry in particular – are quoting the risk of reduced credit and the corresponding costs for the economy. At the same time, there have also been frequent reports suggesting that banks were hiding risks during the Crisis, in order to escape governmental intervention and to maintain market access (Hume 2012, Comfort 2012).

Because the Basel accords define the minimum capital requirement as a requirement on the ratio of eligible capital over risk-weighted assets, these two observations are related. If banks are able to under-report risk, their risk weights will be lower. For a given level of eligible capital a misreporting bank will therefore appear to be better capitalised than a bank that reports truthfully. Under the premise that equity funding is more expensive for the bank than debt funding, this implies that under-reporting risk allows banks to reduce funding costs.

It follows that the question of how to implement capital regulation in practice needs to be reconsidered if current procedures enable banks to hide risk in this way – even if a desirable level of capitalisation could be agreed upon in theory. In a recent paper, we discuss the evidence, suggesting that under-reporting of risk does indeed occur (Mariathasan and Merrouche 2013), and here we make suggestions on how capital regulation could be improved going forward.

Risk-weighted Assets under Basel II and III

While the latest revision of the Basel accords has left the issue of calculating risk weights largely untouched, it constituted the focus of the revision from Basel I to Basel II. The concern at the time was that the risk categories of Basel I were too crude, and that this would impair the efficiency of the regulation. The solution was to allow for more sophisticated procedures for calculating risk weights, and in particular to use the banks’ internal credit risk models for regulatory purposes. Since Basel II, risk weights can therefore be calculated in two ways:1

  • The ‘Standardised Approach’ uses external ratings to allocate assets into six risk categories.
  • The ‘Internal Ratings-Based Approach’ uses banks’ internal credit risk models to calculate input parameters for a regulatory Value-at-Risk model.

Under the Foundational Internal Ratings-Based Approach, banks calculate the probability of default themselves, while the set of parameters also includes the loss given default, the exposure at default, and the residual maturity under the Advanced Internal Ratings-Based Approach.

In particular, the latter approach was designed to improve the risk-sensitivity of capital charges. It is aimed at enhancing efficiency of the banking sector by allowing safer banks to be more leveraged, while also containing regulatory arbitrage among risky banks. As previously indicated, however, they also introduce the issue of truthful risk reporting; especially when limited data makes it difficult to verify underlying distributional assumptions.

Limited Risk-sensitivity of Internal Ratings-based Weights

The Internal Ratings-Based Approach was designed to improve the risk-sensitivity of capital charges. Yet, there is plenty of recent evidence suggesting that risk-sensitivity continues to be limited. In February 2013, the European Banking Authority published an interim report on the determinants of risk-weighted assets. Comparing regulatory risk weights across a sample of 89 banks, the report concludes that almost 30% of the variation in risk weights cannot be explained by ‘fundamentals’. The report only discusses preliminary results, but its conclusions align with the more general insight that Internal Ratings-Based Approach weights capture risk only insufficiently. Andy Haldane and Vasileios Madouros of the Bank of England, for example, had raised the issue in their 2012 speech at the Jackson Hole Symposium (Haldane and Madouros 2012), the Financial Services Authority had indicated it already in 2010 (FSA 2010), and Vallascas and Hagendorff (2013) recently confirmed it for market measures of risk. At the same time, there is survey evidence from the banking industry as well, revealing that financial managers themselves do not trust risk weights, and that the mandatory disclosure requirements of Basel II do not help transparency in the way they were intended to (Samuels et al. 2012).

What Determines Risk Weights?

If risk weights are not risk-sensitive, the question arises as to what drives them instead. One possibility is that the banks’ internal models are sophisticated, but flawed and that the missing 30% are modelling mistakes. There is, however, also the possibility that banks use the discretion of the Internal Ratings-Based Approach to minimise funding costs by reducing capital charges, for example, via convenient distributional assumptions. Blum (2008) provides a simple model of the corresponding incentives, and there is plenty of recent narrative evidence suggesting that banks do use model complexity and non-verifiable assumptions towards their benefit. The Economist (2012), for example, has referred to the capital adequacy ratio under the Internal Ratings-Based Approach as ‘DIY capital’. The problem with such a systematic underestimation of risk is that it makes it harder to distinguish safe from unsafe banks, and thus to take precautionary measures.

Figure 1 illustrates this point, and is taken from our recent working paper (Mariathasan and Merrouche 2013). Separating banks that were resolved during 2007-2010 from banks that were not resolved, Figure 1 shows that – looking at risk-weighted assets – it is much harder to distinguish safe from risky banks after they have been approved for using the Internal Ratings-Based Approach. This figure is consistent with the concerns raised by Haldane, and confirmed in our empirical analysis.

Figure 1. Risk-weighted asset densities (%) before and after implementation of advanced approach; resolved versus not resolvedSource: Mariathasan and Merrouche 2013.
Note: This chart is obtained taking as threshold the quarter by which the law required full implementation of the advanced approaches.
Click to enlarge

In our paper, we analyse the role of Internal Ratings-Based Approach approval for risk weights and make a first step towards distinguishing between accidental underestimation of risk, and strategic risk-modelling. Using a cross-sectional sample of 115 Internal Ratings-Based Approach banks from 21 OECD countries to explain the evolution of the ratio of risk-weighted assets over total assets between 2004 and 2010, we document that approval for the Internal Ratings-Based Approach corresponds to lower average risk weights. We also document that the decline in risk weights is particularly prevalent among weakly capitalised banks (which have higher incentives for risk-weight manipulation in the model of Blum 2008), and that not all of the effect can be explained by a reallocation of the banks’ assets. To distinguish between accidental miscalculation of risk and intentional manipulation, we show that banks with lower average risk weights do not necessarily make loans of lower quality (i.e. they are able to identify credit risk), and that they behave less prudently in general (they pay more dividends, for example). We also show that the decline in risk weights is muted in environments where supervisory scrutiny is higher; more specifically, when the legal liabilities of external auditors are higher, and when the supervisor assumes responsibility for fewer banks. For resolved banks, we find that risk weights increase prior to resolution, but only if the bank was not yet approved for the Internal Ratings-Based Approach; if the bank is weakly capitalised and has received its Internal Ratings-Based Approach approval, risk weights are even lower prior to resolution (although not significantly).

Regulatory Lessons

Our work suggests that the Internal Ratings-Based Approach enables banks to reduce capital charges strategically and that they make use of this option in particular when they are weakly capitalised. Because this introduces uncertainty into the financial sector, and leaves banks to be undercapitalised, the question becomes whether, and how, regulation should respond.

One possible way forward that has already been implemented under Basel III is to ask banks to satisfy a capital requirement relative to total (rather than risk-weighted) assets. Not only would this limit the degree of possible manipulation, according to the model of Blum (2008), it would also improve incentives of the banks to report risk weights accurately. This pertains to the more general insight regarding the value of simple and transparent rules in regulation. While more sophistication may improve risk sensitivity (and thus efficiency) in theory, it seems to be the case that it increases opacity in practice. If one does not want to sacrifice the benefits of more sophisticated risk-modelling, it may be worth debating whether it should not be the regulator who calculates the model’s parameters. While this would certainly be costly, it would eliminate the moral-hazard problem that we have identified and it would also restore the consistency between the regulatory model and the models that are underlying the calculation of the input parameters. A final, more preliminary, thought might be to introduce leverage-dependent capital charges in order to mitigate the higher incentives for manipulation.



Admati, Anat and Martin Hellwig (2013), The Bankers’ New Clothes: What’s Wrong With Banking and What to Do about it?, Princeton University Press.
Blum, Jürg M (2008), “Why Basel II may need a leverage ratio restriction”, Journal of Banking and Finance 32(8), 1699–1707.
Comfort, Nicholas (2012), “Deutsche Bank Accused by Ex-Employee of Hiding Securities Losses”, Bloomberg, 6 December.
EBA (2013), “Interim results of the EBA review of the consistency of risk-weighted assets. Top-down assessment of the banking book”, European Banking Authority.
Financial Services Authority (2010), “Results of 2009 Hypothetical Portfolio Exercise for Sovereigns, Banks and Large Corporations”, March.
Haldane, Andrew G and Vasileios Madouros (2012), “The dog and the frisbee”, speech given at the Jackson Hole Symposium, Wyoming, 31 August.
Hume, Neil (2012), “Reflections of a UK banks analyst”, Financial Times, Alphaville, 9 June.
Mariathasan, Mike and Ouarda Merrouche (2013), “The Manipulation of Basel Risk Weights”, CEPR Discussion Paper 9494.
Samuels, Simon, Mike Harrison, and Nimish Rajkotia (2012), “Bye bye Basel? Making Basel more relevant”, Barclays Equity Research, 23 May.
The Economist (2012), “DIY capital. An edifice of modern banking regulation comes under scrutiny”, The Economist, 8 December.
Vallascas, Francesco and Jens Hagendorff (2013), “The Risk Sensitivity of Capital Requirements: Evidence from an International Sample of Large Banks”, Review of Finance, published online 12 January.

1. Counting the Foundational and the Advanced Internal Ratings-Based Approach as one.


Share this Econintersect Article:
  • Print
  • Digg
  • Facebook
  • Yahoo! Buzz
  • Twitter
  • Google Bookmarks
  • LinkedIn
  • Wikio
  • email
  • RSS
This entry was posted in Banking News, macroeconomics, money, money and banking and tagged , , , , , , , , , , . Bookmark the permalink.

Make a Comment

Econintersect wants your comments, data and opinion on the articles posted. You can also comment using Facebook directly using he comment block below.