Comments on Basel Discussion Paper of July 2013

Douglas J. Elliott

The Basel Committee is to be commended for taking up an important and difficult set of issues in its July 2013 discussion paper “The regulatory framework: balancing risk sensitivity, simplicity, and comparability.” In the next few pages I provide some thoughts about how best to tackle this delicate balancing act. Please consider this a “comment” as requested by the committee. You will receive many very technical responses, but my intent here is to focus on the key, high-level issues.

Before doing so, let me briefly describe my credentials. I am a Fellow in Economic Studies at the Brookings Institution in Washington, a non-partisan public policy institute, often considered the world’s leading think tank. I was an investment banker for about two decades, focusing on financial institutions as clients, before joining Brookings about five years ago. My primary focus at Brookings is on global financial regulation and I have written extensively on bank capital issues. Two papers that may be of particular interest to the committee are the IMF report that I co-authored attempting to quantify the likely impact on credit pricing and availability of the global financial reforms, including the Basel III capital rules[1], and a paper I wrote summarizing why there are genuine social costs to raising capital requirements, which must be considered in tandem with the social benefits of financial stability[2].

Let me begin by strongly agreeing with the discussion paper that there is indeed a balancing act to be performed, because all three objectives matter and it is difficult to fully achieve one without harming one or both of the others. A highly tailored and detailed approach to risk analysis would be the likeliest to capture the relative riskiness of different portfolios and business models. However, it would surely fail on the simplicity and comparability scores. For its part, a very simple risk measurement, such as the level of assets on a bank’s balance sheet, manifestly fails the risk sensitivity objective and may well fail on comparability unless it is made more complex by mandating the exact accounting rules for what goes on a balance sheet.

Second, I believe that risk-sensitivity is the most important of the three objectives and should receive the most weight as trade-offs are made. The idea of capital and liquidity buffers is to ensure that there is adequate protection against the risks that banks face. Forced to choose between the objectives, I would prefer an approach that got those levels right even if it were difficult to understand them and to compare them across banks and jurisdictions. Of course, life is not that simple and there would be serious problems with an approach that was so opaque, including the grave difficulty of being sure that the levels were set correctly and then communicating to all the important constituencies that the levels were indeed right.

Because risk-sensitivity should be primary, I applaud the overall approach of the Basel Committee. This is to try to use a careful, risk-sensitive measure as the primary methodology for determining capital requirements, through the use of risk-weightings of assets. The secondary, but still important, objectives of simplicity and comparability are furthered through the use of a leverage ratio as a safety net. The simplicity and comparability of a leverage ratio in practice is often dramatically over-stated, given the very detailed measurement questions regarding what goes on a balance sheet, in what amounts, and to what extent off-balance sheet items should be added on to measure total exposures. Nonetheless, the leverage ratio has the virtue of relative simplicity and relative ease of explanation and comparison and therefore has an important role to play.

There is considerable room for improvement in Basel III, despite the desirability of the overall approach. For one, virtually everyone recognizes that there are problems with the use of internal bank models to generate risk-weightings, including that: banks have strong incentives to use approaches which result in low weightings; complicated models can fall into error, especially when there is insufficient history on which to judge; and banks can generate quite varying results for the same portfolios based on differences in their models. This variability would best be ironed out through a combination of more transparency in the models and their results, better supervisory oversight and more pressure to conform to industry standards, and the imposition of minimum risk weights in some cases.

It is also a problem that most sovereign debt is treated as being riskless, when it is quite apparent that some of it has significant risk. Given the high levels of sovereign debt held by virtually all banks, this encourages a tight linkage between the creditworthiness of banks and their national governments that would be lessened by a better recognition of the actual risk.

There are other weaknesses as well in Basel III’s approach to risk weightings. Taking them all together, there are those who argue for essentially abandoning the approach of risk weighting and moving to leverage ratios for their simplicity and comparability. As already noted, I believe these virtues are considerably overstated in regard to leverage ratios, which are much more complex and variable, especially for big banks, than generally advertised. Even more basically, a simple leverage ratio is essentially the same as a risk-weighted system in which every asset is weighted at 100%. It is extremely unlikely that a uniform risk weight for all assets is more accurate than the current risk weightings, no matter how pessimistic one is about the risk weightings. For example, sovereign debt of developed countries may well not merit a zero risk weighting, but that excessively low figure is still likely to be more accurate than treating such debt as being as risky as a long-term loan to a junk-rated company.

The risk-weighted approach is likely to be more accurate in measuring risk. This also means that it is less likely to create the wrong incentives. A straight leverage ratio with no exclusions would push banks away from holding cash or short-term government securities, since they will seldom earn a high enough return to cover their cost of capital on this basis. On the opposite end of the spectrum, it would encourage banks to hold highly risky assets that have consequently higher yields, since this produces the largest expected return on their total asset base and therefore their capital requirements. If, for political or other reasons, it becomes necessary to use a leverage ratio as the binding capital constraint, then I would support some crude adjustments to eliminate very low-risk assets from the calculation, in particular short-term treasury securities of highly creditworthy sovereigns and cash on hand at the central bank.

I do support giving greater weight to the leverage ratio than Basel III currently does, by raising the minimum level from 3%. The leverage ratio should be a safety net that becomes binding for banks whose risk weightings are decidedly low on average. This protects against an over-investment in assets that seem low-risk but may turn out to be riskier in reality and against deliberate gaming of the system by banks. We want to calibrate the minimum level so that it catches the outliers, without becoming the binding constraint in normal times for most banks. However, there is currently so large a gap between the amount of capital required under the risk-weighted approach and that required under the leverage ratio that it would be helpful to reduce the distance between them so that it has more chance of binding on the outliers.

A greater reliance on leverage ratios increases the importance of the Basel Committee’s current review of exposure measures for derivatives, securities financing agreements, and similar instruments. The existing approach substantially overstates exposures for the larger banks and, I am told, can understate it for many smaller banks. Greater accuracy in these measures would translate over into the level of balance-sheet equivalents for leverage ratio calculations under Basel rules.

An interesting potential change to the overall Basel approach is to move over time towards stress tests as the risk-sensitive approach or as a strong supplement to the current risk-weighted methodology. This would have significant advantages, since the degree of stress could be calibrated to real world concerns and the stress tests could have greater standardization across the banks in a given national system. On the other hand, it would lose some of the accuracy of the better internal models for measuring risk and may be less comparable across countries.

Use of stress tests in this manner needs considerably more analysis before it would make sense as the primary approach. Currently, stress tests often suffer from a number of vices, including: opacity; an ad hoc nature; great variability from year to year; and a lack of international standards. However, the potential advantages of this approach, combined with the fact that stress tests are currently the binding capital constraint in some countries and may become so in more, calls for the Basel Committee to make a major effort to consider the proper role of stress tests and how they might best be standardized by agreed guiding principles and best practices.

Thank you for your consideration of my views.