OR Forum: Design of Risk Weights

In the November-December 2014 issue of Operations Research we have chosen to highlight the paper “Design of Risk Weights” by Paul Glasserman and Wanmo Kang.  In this paper Glasserman and Kang explore a new approach for regulators to set minimum capital levels for banks.  The purpose of such capital level constraints is to limit the risk of collapse for large financial institutions if they take large losses in the assets they are holding and thus increase the stability of the financial system.  The current regulatory approach is to classify assets held by banks by their risk levels and assign weights to each asset category.  The risk weighted sum of the asset holdings of the bank then becomes the basis for setting capital requirements usually as a percentage. 

Dr. Jean-Charles Rochet is a Professor of Banking at the University of Zurich and SFI Senior Chair and Research Associate at IDEI Toulouse. (

)

Dr. Darrell Duffie is the Dean Witter Distinguished Professor of Finance and Shanahan Family Faculty Fellow at the Graduate School of Business,  Stanford University.(

)

Professor Rochet describes the contribution of the paper as follows:

"

This article sheds new light on the important debate about the choice of risk weights in regulatory capital ratios for banks. Several academics including Kim and Santomero (1988) and Rochet (1992), had criticized the first Basel Accord (1988) for being too coarse. Basically the message of these papers was that when regulatory risk weights differ from markets’ assessments of risks, there will be a distortion in the portfolio choices of banks. In some extreme cases, the risk of default of some banks might actually be increased (!) by regulation, which is of course exactly the opposite of what prudential regulation is supposed to do. Interestingly, Glasserman and Kang suggest that if the regulator’s objective is to minimize the risk in banks’ portfolios, the appropriate risk weights should be related to the expected profitability, not to the risk of the assets.

"

Professor Duffie states that: 

"

The results of Glasserman and Kang on the design of these risk weights are striking.

 

First, it is quite unexpected that any reasonable linear risk-weighting scheme for bank capital requirements would have such a clear conceptual foundation, given that the total risk of a portfolio of financial positions is obviously non-linear with respect to the amount of each asset in the portfolio. The authors show that a linear risk weighting scheme nevertheless arises naturally.

 

Second, the authors provide a risk-weighting scheme that not only ensures (at least in theory) that a bank has adequate capital, but also avoids causing demand distortion in asset markets, given the response of bank investment behavior to the risk weights.

"

Interestingly the two have opposing perspectives of the approach taken by Glasserman and Kang.  Rochet advocates for less complexity:

"

First, I do not subscribe anymore to the view that regulation should be complex enough to take care of the complexities of banks’ activities. I rather think that supervision (implying possible regulatory action) is more important and therefore that rules (regulation) should be simple. The Subprime Crisis has shown that complex regulations always fail because big banks are able to bypass them and supervisors cannot do much about this, even if they are well intentioned…..

 

Second, I am not in favor of adjusting regulatory weights too often, because it adds regulatory uncertainty to fundamental uncertainty, which is already huge. To make a parallel with monetary policy, it would not be a good idea to adjust inflation targets on a regular basis to take care of changes of the economic environment. The actual target is less important than the credibility that the monetary authorities will stick to it. This is why I do not believe that the adaptive method recommended by the authors at the end of the paper would really work. 

"

Duffie meanwhile believes that the system needs more sophisticated regulation.

"

Ironically, just as this research provides an improved approach to tuning risk weights to market risk premia so as to avoid causing asset market distortions, regulators are moving in the opposite direction! The new norm for bank capital requirements introduces a requirement that a bank’s total quantity of assets,irrespective of their riskiness, may not exceed a stated multiple of the bank’s capital.  In other words, for any bank to which this new “supplementary leverage requirement” is binding, the risk weights will all be equal! This new “leverage requirement” is simpler than the conventional risk-weighted capital requirement…..The simplicity of the new leverage requirement, which treats all assets as though equally risky, has thus promoted its heavy use in new capital rules, to the point that the capital levels of some banks, including some of the very largest U.S. banks, will be determined by the new equal-weighted leverage requirement rather than by risk-based capital requirements.”

 

“ It is indeed simpler to put a floor on a bank’s capital based on the total quantity of its assets, irrespective of their risks. The unintended consequence is that any bank constrained by this requirement can take more risk, without adding capital, merely by shifting from safer assets to riskier assets. This distorts incentives for portfolio choice, intermediation, and risk management, and in some cases could lead to excessive risk taking. This is exactly what the Glasserman-Kang approach avoids.” 

"

The authors respond: 

"

The comments highlight some of the tensions faced by regulators in establishing principles that should govern risk weights. In particular, Darrell Duffie warns against the dangers of an overly simple scheme, and Jean-Charles Rochet emphasizes the shortcomings of an overly complex scheme. This tradeoff touches on the broader debate on how risk-sensitive to make capital requirements.

Jean-Charles Rochet questions the feasibility of the adaptive scheme in the paper. The appropriate frequency of updates is indeed an important practical issue that we do not address. Nevertheless, we see value in building in some sort of feedback mechanism, given the inevitable weaknesses of any fixed set of rules. The references in the paper to work at the Bank of England and the Reserve Bank of India suggest that practical implementation may be possible.

"

Duffie summarizes the work well: 

"

Through improvements in risk-based capital requirements, we can push capital in the banking system up to safer levels without depending on rules that are blindfolded to the types of risk that banks take. The results of Glasserman and Kinger represent a very important milestone on this path.

"

Comments

Commenting has been disabled for this article.