The merit of the journey: a counterpoint to RAROC critics

This article presents an extract of a larger chapter on RAROC contained in Credit Risk Management, a new book by Brian Ransom Ranson says he wanted to write a book on credit risk for today's managers but "without the math that so often takes a simple idea and makes it incomprehensible."

A basic premise of finance is that capital should only be invested if the probable future return on that capital will exceed its cost. Any potential investment that entails using existing capital or generating incremental capital, then, should be required to fulfill this principle.

Risk-adjusted return on capital (RAROC) is a relatively new tool for applying this premise in the lending and credit risk management context. Also known as return on risk-adjusted capital (RORAC) or risk-adjusted return on risk-adjusted capital (RARORAC), the concept boils down to the opening sentence of this article--nothing more, nothing less.

Bankers Trust pioneered the use of RAROC models in its lending process more than 20 years ago. Today, many banks have built such models, and some use them as decision-making tools at the heart of their lending processes.

Of course, RAROC critics are vocal about its limitations. And there are quite a few. It is not a solution in itself, nor is it more than a measure at a point in time (albeit a critical point in time). But, on balance, most banks that have introduced a RAROC model have retained the concept and built on the model, indicating that the perceived advantages outweigh the disadvantages. Few banks would return to the pre-model processes even if their model were no longer a traditional RAROC structure.

Ultimately, the journey toward developing a model offers tremendous value to those willing to see beyond its limitations.

Criticisms of RAROC

The fact that RAROC is a temporary measurement is just one criticism of its use. So, yes, it is flawed; however, many of the criticisms seem designed to assail rather than contrast with an alternative. Four of the most common criticisms, cited in a recent work on credit risk by Algorithmics, are as follows:

1. RAROC is a no-arbitrage technique. It does not reconcile the prices of loans with those of similar securities available in the market (e.g., bonds, other loans, and credit derivatives). Hence, it cannot assess comparative business opportunities and arbitrage-like situations arising from relative price mismatches. (1)

This statement makes the assumption that a RAROC model should but does not provide a market-relative price to the user but does not do so. It is an accurate statement in that a RAROC model is derived from an inward focus because it uses an individual lender's overhead and tax levels, correlations from an individual balance sheet, and judgmental assessments of risk. That internally acceptable price may not relate to the external market but certainly must be compared.

It also assumes that the operation of a RAROC model implies that the contemplated deal may be at an inappropriate price relative to market alternatives. This is increasingly unlikely in the U.S. syndicated loan market, which has become dominated by relatively few very large banks that originate and syndicate. These banks are well aware of the capital markets price for the credit risk of their clients, and they assume that their other clients (the buyers of the loan) also have that information. Thus, the idea that there remains a wondrous undiscovered arbitrage between corporate loans and bonds is simply not true. Loans have different structures, recovery rates, terms, documentation, and more, and the price should (and most probably does) reflect all those factors.

Credit derivatives are a somewhat different issue, but statistics on credit-derivative pricing remain unreliable because of the problems of data, documentation, and liquidity that had significant impact in the early years of that market.

2. RAROC neglects the "stated consistency of many loan cash flows." (2) Two issues are raised here: 1) Loans have an indefinite maturity date because of prepayment options; and 2) the exposure at default is very difficult to estimate. These are necessary assumptions in a calculation.

The problem of loan prepayment options remains a major thorn that is often ignored in the RAROC calculation because it is a "conservative" assumption. Early prepayment, if known, would have the effect of reducing term risk and improving RAROC. The problem of estimating exposure amounts is insoluble. There is some new data on revolving loan usage levels, but, for the moment, RAROC and other estimation models will require judgmental input on these issues.

3. RAROC takes a static view of credit risk. (3)

For banks that have used these models for some time, this is not the case. Most RAROC users perform the calculation to decide whether to proceed with the transaction. From that point, there is little doubt that the RAROC will change because the risk will change. But as time passes, the bank's calculation of RAROC should also change. It may be because risk changes; it may be from a drawdown on a revolving line or from prepayment. Good loan systems and reviews will then automatically recalculate the RAROC. RAROC should not be a static view. It should be a continuous view, albeit one structured around the model rather than the market.

4. RAROC considers a fixed-horizon pricing. (4)

This is an accurate statement; however, any loan or bond does precisely the same at origination. A 10-year corporate bond is quoted over the government 10-year rate (notwithstanding the probable five-year call option generally available). As argued in another portion of Credit Risk Management, a risk assessment should be forward looking. The market price of a bond contains information on the risk of that asset based on the assessment of its future prospects. Within that price, however, especially for investment-grade issuers, lies interest rate risk as well as credit risk. Moreover, the higher the quality of the issuer, the more the interest rate risk dominates.

A paper by Jean Dermine of Insead (5) also contains a number of RAROC criticisms. The first is that instead of using a single hurdle rate, RAROC should use a rate that is sensitive to the yield curve and to the industry in which the borrower operates. It is difficult to agree with these conclusions. First, bank loans and bank borrowing are predominantly undertaken at floating rates, and therefore the revenue calculation is invariably carried out based upon the spread from each bank's cost of funds. The question of setting a hurdle rate is a thorny one; however, there seems to be no argument that would favor multiple rates. The intention is to obtain a return over a bank's cost of capital, and there seems to be no valid reason why any bank should compromise with a borrower from any particular industry group. A certain industry group may assist in the diversification of the bank's portfolio; however, that is a benefit that should accrue to the bank, not the borrower.

A somewhat more virulent criticism of RAROC states:

   RAROC becomes the antithesis of
   sound enterprise risk management.
   The practical implementation
   of RAROC does not promote
   personal responsibility. It
   suggest that management can
   place a company on auto pilot
   and let a computer make decisions.
   It denies that risk management
   is about people and
   suggests it is about mathematical
   formulas. For example, RAROC
   does not acknowledge Nat individuals
   make a difference in the
   profitability of a line of business..
   or that ... individuals make
   a difference in the riskiness of a
   line of business. (6)

Wow. I cite this because it demonstrates at least that there are passionate views about all models when used, and RAROC models can engender greater passions than most. I will, I hope, demonstrate that this diatribe is simply that, but would repeat the quote of John Box, "All models are wrong, but some are more useful than others."

The more informed criticisms of RAROC are not invalid, but they tend to come from a perspective that assumes greater centrality of the calculations. They also tend to come in part from analysts reaching the conclusion that the calculation is likely to be inaccurate or at least imprecise.

Well, yes, they are imprecise. It is very easy to point to any RAROC calculation and find issues that could change the calculation with valid adjustments or could apply a different methodology to reach either the income or the capital measure. But this is not the point. The idea is to make decisions using a measure of both risk and reward. Those seeking true accuracy are cursed to be eternally disappointed. Those designing RAROC models with that as the objective are also likely to be disappointed. That does not mean the pursuit is without merit; rather, it means that the results need expert judgment before their final use is decided.

Advantages of Using RAROC

The advantage that can be provided by a RAROC model lies mainly in the discipline it can bring to lending decisions. The model itself is not the objective because it will only be as good as its builders. RAROC is not an end in itself. Its advantages are more in the way that it ensures that risk and reward remain linked and in the consistency of decision making that it forces. Its critics tend to focus on what it does not do. Yet, in most cases, they find things that it does not purport to do. It remains a good idea, but, like any model, it needs intelligent users.

Perhaps it was most succinctly put by Ralph C. Kimball, an economist with the Federal Reserve Bank of Boston:

   By forcing line managers to
   include the opportunity cost of
   equity when making investment
   and operating derisions, banks
   expect to elicit better decision
   making by managers [my
   emphasis]. By implementing
   performance measurement and
   incentive systems driven by economic
   profit and allocated equity
   capital, senior managers also
   hope to align managerial behavior
   more closely with the interests
   of shareholders. (7)

Having a calculated RAROC for a transaction does not obviate the need for a careful review of all new credits and a senior screening (whether by committee or some other means) of deals that bring a lot of incremental risk. However, a RAROC model provides a number of advantages to a discriminating user, including the following:

* It provides a platform on which to calculate both risk and return and thereby remove the bias from one objective or the other. Credit officers in committees can lapse into a defensive mode looking for flaws in the agreement or in the borrower's financial affairs. Relationship lenders are trying to meet the business plan and reach a deal that the customer will accept, with the credit committee a hurdle to be overcome. A RAROC calculation can bring an added dimension by showing the use and return on capital.

* If provided to all commercial/corporate lenders and used appropriately, a RAROC model can almost ensure that decisions made in different locations, at different times, and with different relationship managers will be made using the same principles and calculation methodology. This is perhaps the reason so many financial institutions have adopted such models. Banks have many decision makers in the lending business, and their negotiation skills can vary substantially. Sometimes the best salespeople measured by volume achieve that result through inappropriate pricing. Sometimes certain risk officers achieve a reputation for being "tough" by insisting on higher risk ratings than their peers. A RAROC model tends to level the playing field and give all staff the chance for a common comparison of their transactions.

* A RAROC model emphasizes that risk must be compensated fur, while ensuring that the risk is both measured and appropriately considered through the enforced completion of the calculation.

* A RAROC model can provide a "what if" capability to the user. In most cases, the relationship manager or the credit officer can solve for the price or the risk and rebalance by adjusting one or the other. This can provide a valuable sales resource to the discriminating client relationship manager. It is not unheard of for a relationship manager to take a client through the calculation to help negotiate the structure to the price level or the term that the client is seeking. At the same time, the opportunity comes to the discerning credit officer to negotiate more safety in the deal.

Although these benefits can provide some improvement to the traditional credit process, it must be repeated that the RAROC calculation is not an end in itself. The weather in Britain (as well as in Texas) provokes the quip, "If you don't like it, don't worry, it will change in a minute." Unfortunately, one of the realities of RAROC is that the calculation is certain to change because risk changes and time passes.

Is It Worth the Effort?

In many traditionally run and long-standing restitutions, change has a difficult burden of proof. Bringing in the mathematics and analytical approach of a RAROC model is often resisted because the model will certainly be imperfect. Opponents can raise several imperfections to discredit the model and its implementation, many of which have been identified already in this article:

* It is a temporary measure.

* It can be manipulated because it needs judgment-based inputs.

* It struggles with longer time horizons.

* It can produce results at odds with market pricing.

* It is not client-friendly.

This is an impressive list of problems, but the critical question is not whether a RAROC model is the answer to a banker's prayer. The critical question is whether its introduction will improve the existing lending process, the decision-making ability, and the performance of corporate lending. The answer is specific to each institution. Lending or making promises to lend is a process akin to investing, and lenders must think like investors.

For many banks, lending is a subsidization of other business; for those that believe that such a strategy is logical, a RAROC model will not help.

For some banks, financial institutions, and fund managers, corporate credit risk assets are part of a pure investment process, and so the only price is the market price. A RAROC model makes little sense for these buyers of loans.

For many banks, the calculation of expected loss and economic capital is not on the agenda, and so a RAROC model would similarly be rejected.

Many others have such poorly developed information systems that it cannot be successfully integrated for the moment.

Eric Falkenstein has commented that "RAROC is not an off-the-shelf technology one can apply, but a complicated set of rules that needs to be calibrated for each bank's unique set of products, incentive compensation plans, pricing models, and, most importantly, information systems." (8) It is not for everyone.

Those who argue that a RAROC model is the answer tend to gloss over the problems of the output. Those who build them, however, tend to learn a great deal about their management of loan assets. The models tend to improve on the institution's rating system, they put more consistency into structuring and pricing, and they often force institutions to upgrade their management information systems.

This is why the production of a RAROC model can be a rewarding journey to a somewhat disappointing destination. I would not discourage any credit risk manager from taking the journey.

Notes

(1) Aguais, Forest, and Rosen, "Building a Credit Risk Valuation Framework for Credit Instruments," Enterprise Credit Risk Using Mark-to-Future, www.algorithmics.com.

(2) Ibid.

(3) Ibid.

(4) Ibid.

(5) Dermine, Jean, "Pitfalls in the Application of RAROC in Loan Management," The Arbitrageur, Volume 1, No. 1, 1998.

(6) See article by Glyn Holton, consultant, www.contingencyanalysis.com.

(7) Kimball, Ralph C., "Economic Profit and Performance Measurement in Banking," The New England Economic Review, July/August 1998.

(8) "Capital Priorities: Practival Advice on Implementing RAROC," The Journal of Lending & Credit Risk Management, May 1999.

Ranson can be reached by e-mail at Brian.Ranson@bmo.com. To order Credit Risk Management, visit www.sheshunoff.com or call 1-800-456-2340.

Brian Ranson is executive managing director at BMO Monegy, a division of Bank of Montreal. This discussion was reprinted from B.Ranson, Credit Risk Management (Sheshunoff Information Services [c] 2003). Reprinted with permission of the publisher and the sole copyright owner. All rights reserved. For more information about this book, call 800.456.2340 or visit www.sheshunoff.com.

COPYRIGHT 2003 The Risk Management Association
COPYRIGHT 2005 Gale Group