A Hippocratic Oath for Bank Regulators: Do No Harm
Reform of banking regulation has been an active topic in the U.S. and Europe since the outbreak of the recent financial crisis. There is now broad agreement that, among other things, banks should improve their capital ratios by holding more and higher quality capital, but the practice of treating sovereign bonds and mortgages as higher quality assets appears to be counterproductive.
Haldane’s paper, “The Dog and the Frisbee”, sparked an informative debate about the need to simplify financial regulation that included contributions by Dolan, Kwak and Resti on this site. However, with the exception of the post by Resti, our fellow authors primarily focused on the complexity of financial institutions and the too-big-to-fail question. We would like instead to examine the current system of assigning risk weights to assets in more detail.
One of the findings by Haldane was that simple measures such as the ratio of market value of equity to unweighted assets have had much higher predictive power of financial trouble in the crisis than measures involving risk-weighted capital. Pushing it to the extreme, this could be interpreted as a suggestion to abandon the risk weights altogether and replace them with a plain leverage ratio. Resti in this respect claimed that „switching to un-weighted accounting data means that a possibly flawed set of risk-weights (the ones in Basel) would be replaced by a definitely flawed set of coefficients, implying that risk is constant across portfolios and counterparties”. To give a specific example, he claimed that withdrawing funds from the central bank to lend them to junk rated borrowers certainly alters a bank’s risk profile for the worse. While we agree that treating all assets on a bank’s balance sheet as if they had equal risk would be absurd, we would also claim that the current set of risk weights in the Basel framework is definitely flawed and could be harmful.
Residential mortgage exposures are assigned a 50% risk weight under Basel I and a 35% to 75% risk weight, mainly depending on the loan-to-value ratio, under the standardized approach in Basel II. There is, however, little evidence that losses from mortgage lending would be significantly lower than those from other types of lending. Although that was true in the 1990s in the US, according to the Federal Financial Institutions Examination Council (FFIEC) delinquency rates on residential real estate loans during last decade have been consistently higher than on commercial real estate loans and mostly even higher than the average delinquency rates on all loans. A frequently heard argument in support of lower weights for residential real estate loans is the existence of collateral. However, in times of crisis real estate became an illiquid asset, and the cost of foreclosure might account for a significant part of its value.
Moreover, overexposure of banks to the real estate sector has been a key cause of the current financial crisis in Spain, Ireland and the Baltic States, as well as the US. A high-level expert group on reforming the structure of the EU banking sector in a recent report also admitted that many systemic banking crises resulting in large commitments of public support have originated from excessive lending in real estate markets.
Residential mortgage claims are not the only category of exposures receiving preferential treatment in the Basel framework. Another privileged category is exposure to sovereign debt. National discretion to allow 0% risk weights for exposures to own sovereign in domestic currency is an extreme example and has definitely played a role in developing the current vicious link between troubled banks and troubled sovereigns, in particular, in Europe.
To give another example, in the standardized approach under Basel II exposures to sovereigns rated “BBB” are subject to a 50% risk weight while claims on corporates rated “BBB” have to be weighted at 100%. However, looking at a comparison of sovereign and corporate default rates in the table below we can see that, if anything, BBB-rated corporates have had lower default rates than sovereigns in the same rating category.
S&P ratings and default probabilities
Source: Damodaran (2010)
Thus, preferential treatment for sovereign debt doesn’t appear to be justified, unless losses from default differ radically by type of borrower. Estimates of default spreads by Damodaran, however, suggest that expected losses from exposures to sovereign and corporate debt in the same rating category are roughly the same as shown in the table below.
To paraphrase Resti, reducing lending to a BBB-rated corporate borrower and increasing lending to a BBB-rated sovereign does not improve a bank’s risk profile. However, the current Basel risk weights would motivate the management of a bank using the standardized approach to do exactly that, and there is evidence that European banks and, to some extent, U.S. banks have gravitated towards assets that carry lower risk weights during the last decade. A recent IMF working paper shows that for a sample of banks the ratio of RWAs to Total assets in 2011 was 57% in North America, 51% in Asia, but only 35% in Europe (Leslé and Avramova). While a part of that result might be caused by banks shifting from Basel I to Basel II and from the standardized approach to the internal ratings based approach under Basel II, at least a part of it is due to an increasing weight in bank balance sheets of the categories of exposures receiving preferential treatment. Goldstein provides evidence supporting this view for exposure to sovereign debt.
Under the internal ratings based (IRB) approach, banks determine the RWAs based on estimates of such parameters as probability of default (PD) and loss given default (LGD) for each exposure. The recent report of a high-level expert group on reforming the structure of the EU banking sector claims that the current levels of RWAs based on banks’ internal models and historical loss data tend to be quite low compared to the losses incurred in past real estate- driven crises. The group acknowledges the need for stricter measures regarding the treatment of real-estate lending within the capital requirement framework. More generally, the group also acknowledges that the RWAs calculated by individual banks’ internal models can be significantly different for similar risks. Thus, while the minimum leverage ratio in Basel III will provide a backstop to the risk-weighted capital requirement, there is also a need to improve the consistency of IRB models across banks.
Conclusion
Excessive exposure of banks to the real estate sector and sovereign debt is clearly among the causes of the recent financial crisis. Our analysis shows that it is difficult to justify preferential treatment of these risky categories under the standardized approach in the Basel II framework. Moreover, creating a regulatory framework that favors bank lending to the residential real estate sector and governments and then lamenting about banks not providing enough loans to corporate borrowers looks to us like a case of policymaking schizophrenia.
Although a minimum leverage ratio based on the unweighted assets can be a useful backstop, abandoning risk-based capital ratios altogether would not be desirable for the reasons indicated by Resti. Instead of having a complex matrix of risk weights based on both the category of exposure and its rating, the standard risk weights should be based on rating only, whereas all unrated exposures should have a uniform risk weight of 100%. If bank regulators seek a minimum standard, it should follow the Hippocratic Oath for physicians: “Do no harm”.
References
Damodaran, Aswath, “Into the Abyss: What if nothing is risk free?”, Stern School of Business, July 2010
Federal Financial Institutions Examination Council, Charge-Off and Delinquency Rates on Loans and Leases at Commercial Banks http://www.federalreserve.gov/releases/chargeoff/delallsa.htm
Goldstein, Morris, “The EU’s implementation of Basel III: A deeply flawed compromise”, VoxEU.org, May 27, 2012
High-level Expert Group on reforming the structure of the EU banking sector, Final Report, Brussels, 2 October 2012
Leslé, Vanessa Le and Sofiya Avramova, “Why Do RWAs Differ Across Countries and What Can Be Done About It?”, IMF Working Paper, March 2012
Resti, Andrea, “Bank Supervision: Bye Bye Basel?”, EconoMonitor, Se
7 Responses to “A Hippocratic Oath for Bank Regulators: Do No Harm”
EdDolan • October 10th, 2012 at 6:44 am
I completely agree that preferential treatment of real estate and sovereign loans has not worked out well. At the same time, I am skeptical of the idea that "Instead of having a complex matrix of risk weights based on both the category of exposure and its rating, the standard risk weights should be based on rating only"
The problem with using ratings is that they, like all other metrics, are subject to Goodwin's law. Ratings are already, notoriously, subject to potential inflation due to the desire of CRA's to please issuers in the hope of getting repeat business. One of the few things that keeps this tendency in check is market discipline from investors, who would like ratings to be accurate. However, when ratings are used as a regulatory parameter, regulated investors, too, have an incentive to push for inflated ratings, since buying overrated assets allows them to reduce the capital they must carry.
Unless someone figures out a way to isolate CRAs from such pressures, regulatory reliance on ratings remains suspect.
ThomasGrennes • October 10th, 2012 at 9:59 am
Bond ratings are definitely imperfect, but is there a better alternative? Ratings are publicly available, and errors in ratings are also publicly available. This cannot be said for some methods intended to improve bank safety. Currently the Fed is requiring information from
banks to carry out "stress tests" without revealing the model used to perform the tests to the banks or the public. How can the public evaluate the efficacy of a "secret test", which also may be subject to the same incentive problems faced by bond raters.
Andris_Strazds • October 10th, 2012 at 10:06 am
Ed, thank you for your comment! I agree completely that external ratings have their flaws and you mentioned just one of them. Another one is that rating agencies have not been free from political pressure. My personal favourite example is Greece and Estonia not that long time ago having identical BBB ratings. However, you would also have to admit that, for example, the default rates of AAA-rated borrowers have generally been much lower than those of BBB-rated borrowers and there is a significant correlation between ratings and default rates. Thus, what we are effectively proposing here is going from a clearly flawed system to a less flawed one, but we certainly haven't come up with a perfect solution. But then, is there a perfect solution at all?
EdDolan • October 10th, 2012 at 10:16 am
One alternative would be to try to reduce the biases in ratings, for example, by increasing competition to break the cartel of the big 3 NRSROs, or by requiring issuers to draw lots for random assignment of each issue to one of the NRSROs in order to remove the repeat business bias. If regulators continue to rely on ratings, perhaps some of these ideas should be tried.
ThomasGrennes • October 10th, 2012 at 2:05 pm
More competition in bond rating and better information would be welcome. Reform efforts have centered on increasing entry, limiting conflicts of interest, and greater transparency.
john d • October 10th, 2012 at 7:17 pm
Nothing more than another attack on the Sovereign State. Of course People backed paper is superior to junk produced by the "private sector" and is viewed as a problem for no other reason than economies have been trashed by bank fraud.
We have seen this in home mortgage. The banks created a seemingly hot and then a seemingly cold home real estate market. With many out of work, they want people to sell their their foreclosed homes at the auction into their cold market, thus taking more than $13Trillion out of the pockets of our citizens in the US.
It is the "private sector" that has been given all the rope it needed to hang itself. And, its time that the hangings begin.
ThomasGrennes • October 10th, 2012 at 8:41 pm
Private institutions, such as Countrywide, made many questionable transactions, including
loans to "friends of Angelo", that included Congressmen Frank and Dodd. Some of the worst transactions came from "crony capitalist" and "quasi-public" institutions, Fannie Mae
and Freddy Mac, and they managed to escape regulation from the Dodd-Frank reform law.
There is plenty of blame for the private sector, the public sector, and their interaction.
