Credit Risk Rating at Large U.S. Banks - United States dollar · PDF fileCredit Risk Rating at...

Credit Risk Rating at Large U.S. Banks

William F. Treacy, of the Board’s Division of BankingSupervision and Regulation, and Mark S. Carey, ofthe Board’s Division of Research and Statistics, pre-pared this article.

Internal credit ratings are becoming increasingly im-portant in credit risk management at large U.S. banks.Banks’ internal ratings are somewhat like ratingsproduced by Moody’s, Standard & Poor’s, and otherpublic rating agencies in that they summarize the riskof loss due to failure by a given borrower to pay aspromised.1 However, banks’ rating systems differsignificantly from those of the agencies (and fromeach other) in architecture and operating design aswell as in the uses to which ratings are put. Onereason for these differences is that banks’ ratings areassigned by bank personnel and are usually notrevealed to outsiders.2

For large banks, whose commercial borrowers maynumber in the tens of thousands, internal ratings arean essential ingredient in effective credit risk manage-ment.3 Without the distillation of information thatratings represent, any comparison of the risk posedby such a large number of borrowers would beextremely difficult because of the need to simulta-

neously consider many risk factors for each of themany borrowers. Most large banks use ratings in oneor more key areas of risk management that involvecredit, such as guiding the loan origination process,portfolio monitoring and management reporting,analysis of the adequacy of loan loss reserves orcapital, profitability and loan pricing analysis, and asinputs to formal portfolio risk management models.Banks typically produce ratings only for business andinstitutional loans and counterparties, not for con-sumer loans or other assets.

In short, risk ratings are the primary summaryindicator of risk for banks’ individual credit expo-sures. They both shape and reflect the nature of creditdecisions that banks make daily. Understanding howrating systems are conceptualized, designed, oper-ated, and used in risk management is thus essential tounderstanding how banks perform their businesslending function and how they choose to control riskexposures.4

The specifics of internal rating system architectureand operation differ substantially across banks. Thenumber of grades and the risk associated witheach grade vary across institutions, as do decisionsabout who assigns ratings and about the manner inwhich rating assignments are reviewed. In general,in designing rating systems, bank management mustweigh numerous considerations, including cost, effi-ciency of information gathering, consistency of rat-ings produced, staff incentives, the nature of thebank’s business, and the uses to be made of internalratings.

A central theme of this article is that, to a consider-able extent, variations across banks are an example ofform following function. There does not appear to beone ‘‘correct’’ rating system. Instead, ‘‘correctness’’depends on how the system is used. For example, abank that uses ratings mainly to identify deterioratingor problem loans to ensure proper monitoring mayfind that a rating scale with relatively few grades isadequate. In contrast, if ratings are used in computing

1. For example, bonds rated Aaa on Moody’s scale or AAA onStandard & Poor’s scale pose negligible risk of loss in the short tomedium term, whereas those rated Caa or CCC are quite risky.

2. For additional information about the internal rating systems oflarge and smaller banks, see Thomas F. Brady, William B. English,and William R. Nelson, ‘‘Recent Changes to the Federal Reserve’sSurvey of Terms of Business Lending,’’Federal Reserve Bulletin,vol. 84 (August 1998), pp. 604–15; see also William B. English andWilliam R. Nelson, ‘‘Bank Risk Rating of Business Loans’’ (Board ofGovernors of the Federal Reserve System, April 1998).

For information about the rating systems of large banks and aboutcredit risk management practices in general, see Robert Morris Asso-ciates and First Manhattan Consulting Group,Winning the CreditCycle Game: A Roadmap for Adding Shareholding Value ThroughCredit Portfolio Management(1997).

For a survey of the academic literature on ratings and credit risk,see Edward I. Altman and Anthony Saunders, ‘‘Credit Risk Measure-ment: Developments over the Last 20 Years,’’Journal of Banking andFinance,vol. 21 (December 1997), pp. 1721–42.

3. See the Federal Reserve’s Supervision and Regulation LetterSR 98-25, ‘‘Sound Credit Risk Management and the Use of InternalCredit Risk Ratings at Large Banking Organizations’’ (September 21,1998), which stresses the importance of risk rating systems forlarge banks and describes elements of such systems that are ‘‘nec-essary to support sophisticated credit risk management’’ (p. 1).SR Letters are available on the Federal Reserve Board’s web site,http://www.federalreserve.gov.

4. Credit risk can arise from a loan already extended, loan commit-ments that have not yet been drawn, letters of credit, or obligationsunder other contracts such as financial derivatives. This article followsindustry usage by referring to individual loans or commitments as‘‘facilities’’ and overall credit risk arising from such transactions as‘‘exposure.’’

internal profitability measures, a scale with a rela-tively large number of grades may be required toachieve fine distinctions of credit risk.

As with the decision to extend credit, the ratingprocess almost always involves the exercise of humanjudgment because the factors considered in assigninga rating and the weight given each factor can differsignificantly across borrowers. Given the substantialrole of judgment, banks must pay careful attentionto the internal incentives they create and to internalrating review and control systems to avoid introduc-ing bias. The direction of such bias tends to be relatedto the functions that ratings are asked to perform inthe bank’s risk management process. For example, atbanks that use ratings in computing profitability mea-sures, establishing pricing guidelines, or setting loansize limits, the staff may be tempted to assign ratingsthat are more favorable than warranted.

Many banks use statistical models as an element ofthe rating process, but banks generally believe thatthe limitations of statistical models are such thatproperly managed judgmental rating systems delivermore accurate estimates of risk. Especially for largeexposures, the benefits of such accuracy may out-weigh the higher costs of judgmental systems. Incontrast, statistical credit scores are often the primarybasis for credit decisions for small lending exposures,such as consumer credit.

Although form generally follows function in thesystems used to rate business loans, our impression isthat in some cases the two are not closely aligned.For example, because of the rapid pace of change inthe risk management practices of large banks, theirrating systems are increasingly being used for pur-poses for which they were not originally designed.When a bank applies ratings in a new way, such as inrisk-sensitive analysis of business line profitability,the existing ratings and rating system are often usedas-is. It may become clear only over time that thenew function has imposed new stresses on the ratingsystem and that changes in the system are needed.

Several conditions appear to magnify such stresseson bank rating systems. The conceptual meaning ofratings may be somewhat unclear, rating criteria maybe largely or wholly maintained as a matter of culturerather than formal written policy, and corporate data-bases may not support analysis of the relationshipbetween grade assignments and historical loss experi-ence. Such circumstances make ratings more difficultto review and audit and also require loan review unitsin effect to define, maintain, and fine-tune ratingstandards in a dynamic fashion.

This article describes internal rating systemsat large U.S. banks, focusing on the relationship

between form and function, the stresses that are evi-dent, and the current conceptual and practical barriersto achieving accurate, consistent ratings. We hope topromote understanding of this critical element of riskmanagement—among the industry, supervisors, aca-demics, and other interested parties—and therebypromote further enhancements to risk management.

This article is based on information from internalreports and credit policy documents for the fiftylargest U.S. bank holding companies, from interviewswith senior bankers and others at more than fifteenmajor holding companies and other relevant institu-tions, and from conversations with Federal Reservebank examiners. The institutions we interviewedcover the spectrum of size and practice among thefifty largest banks, but a disproportionate share of thebanks we interviewed have relatively advanced inter-nal rating systems.5

THE ARCHITECTURE OFBANK INTERNALRATING SYSTEMS

In choosing the architecture of its rating system, abank must decide which loss concepts to employ, thenumber and meaning of grades on the rating scalecorresponding to each loss concept, and whetherto include ‘‘watch’’ and ‘‘regulatory’’ grades on suchscales. The choices made and the reasons for themvary widely, but on the whole, the primary determi-nants of bank rating system architecture appear to bethe bank’s mix of large and smaller borrowers andthe extent to which the bank uses quantitative sys-tems for credit risk management and profitabilityanalysis. In principle, banks must also decide whetherto grade borrowers according to their current con-dition or their expected condition under stress.Although the rating agencies employ the latter,‘‘through the cycle,’’ philosophy, almost all bankshave chosen to grade to current condition (seethe box ‘‘Point-in-Time vs. Through-the-CycleGrading’’).

Loss Concepts and Their Implementation

The credit risk of a loan or other exposure over agiven period involves both theprobability of default(PD) and the fraction of the loan’s value that is likelyto be lost in the event of default(LIED). LIED isalways specific to a given facility because it depends

5. Internal rating systems are typically used throughout U.S. bank-ing organizations. For brevity, we use the term ‘‘bank’’ to refer toconsolidated banking organizations, not just the chartered bank.

898 Federal Reserve Bulletin November 1998

on the structure of the facility. PD, however, is gener-ally associated with the borrower, the presumptionbeing that a borrower will default on all obligations ifit defaults on any.6 The product of PD and LIED istheexpected loss(EL) on the exposure in a statisticalsense. It represents an estimate of the average per-centage loss rate over time on a group of loans allhaving the given expected loss. A positive expectedloss isnot,however, a forecast that losses will in factoccur on any individual loan.

The banks at which we conducted interviews fallinto two categories with regard to loss concept. About60 percent have one-dimensional rating systems,in which ratings are assigned to facilities. In suchsystems, ratings approximate EL. The remaining

40 percent have two-dimensional systems, in whichthe borrower’s general creditworthiness (approxi-mately PD) is appraised on one scale while the riskposed by individual exposures (approximately EL) isappraised on another; invariably the two scales havethe same number of rating categories.7

A number of banks would no doubt dispute ourcharacterization of their single-scale systems as mea-suring EL; in interviews, several maintained thattheir ratings primarily reflect the borrower’s PD.However, collateral and loan structure play a role ingrading at such banks both in practical terms and inthe definitions of grades. Moreover, certain specialtyloans—such as cash-collateralized loans, those eli-gible for government guarantees, and asset-basedloans—can receive relatively low risk grades, a dis-tinction reflecting the fact that the EL for such loans

6. Admittedly, PD might differ across transactions with the sameborrower. For example, a borrower may attempt to force a favorablerestructuring of its term loan by halting payment on the loan whilecontinuing to honor the terms of a foreign exchange swap with thesame bank. However, for practical purposes, estimating a singleprobability of any default by a borrower is usually sufficient.

7. The policy documents of banks we did not interview indicatethat they also have one- or two-dimensional rating systems, and ourimpression is that the discussion of loss concepts above appliesequally well to these banks.

Point-in-Time vs. Through-the-Cycle Grading

A common way of implementing a long-horizon, through-the-cycle rating philosophy involves estimating the borrow-er’s condition at the worst point in an economic or industrycycle and grading according to the risk posed at that point.Although ‘‘downside’’ or ‘‘borrower stress’’ scenarios arean element of many banks’ underwriting decisions, everybank we interviewed bases risk ratings on the borrower’scurrent condition. Rating the current condition is consistentwith the fact that rating criteria at banks do not seem to beupdated to take account of the current phase of the businesscycle. Banks we interviewed do vary somewhat in the timeperiod they have in mind when producing ratings, withabout 25 percent rating the borrower’s risk over a one-yearperiod, 25 percent rating over a longer period such as thelife of the loan, and the remaining 50 percent having nospecific period in mind. How closely raters adhere to timehorizon guidelines at banks that have them is not clear.

In contrast to bank practice, both Moody’s and S&P ratethrough the cycle. They analyze the borrower’s currentcondition at least partly to obtain an anchor point fordetermining the severity of the downside scenario. Theborrower’s projected condition in the event the downsidescenario occurs is the primary determinant of the rating.Only borrowers that are very weak at the time of theanalysis are rated primarily according to current condition.Under this philosophy, the migration of borrowers’ ratingsup and down the scale as the overall economic cycleprogresses will be muted: Ratings will change mainly forthose firms that experience good or bad shocks that affectlong-term condition or financial strategy and for those

whose original downside scenario was too optimistic. Theagencies’ through-the-cycle philosophy probably accountsfor their considerable emphasis on a borrower’s industryand its position within the industry. For many firms, indus-try supply and demand cycles are as important or moreimportant than the overall business cycle in determiningcash flow.

In interviews, we did not discuss the reasons that banksrate to current condition, but two possibilities are the greaterdifficulty of the agency method and differences in theinvestment horizon of banks relative to that of users ofagency ratings. Consistency of ratings across a wide varietyof credits may be easier to achieve when the basis is therelatively easy-to-observe current condition. Also, greaterdifficulty means through-the-cycle grading entails greaterexpense, and for many middle-market credits the extraexpense might render such lending unprofitable for banks.

Regarding investment horizon, the rating agencies’ phi-losophy may reflect the historical preponderance of long-term, buy-and-hold investors among users of ratings. Suchusers are naturally most interested in estimates of long-termcredit risk. That banks should naturally have a short-termorientation is not clear, especially as the maturity of bankloan commitments has increased steadily over the pastdecade or two. If it were not for the considerations offeasibility and cost, as well as the fact that many banks useratings to guide the intensity of monitoring of borrowers,the banks’ choice of point-in-time grading would be moredebatable.

Credit Risk Rating at Large U.S. Banks 899

is far less than for an ‘‘ordinary’’ loan to the sameborrower. Such single-grade systems might be mostaccurately characterized as having an ambiguous ormixed conceptual basis rather than as clearly measur-ing either PD or EL. Although an ambiguous basismay pose no problems when ratings are used mainlyfor administrative and reporting purposes and whenthe nature of the bank’s business is fairly stable overtime, a clear conceptual foundation becomes moreimportant as quantitative models of portfolio risk andprofitability are used more heavily and during periodsof rapid change.

In two-dimensional systems, one grade typicallyreflects PD and the other EL. Banks with such sys-tems usually first determine the borrower’s grade (itsPD) and then set the facility grade equal to theborrower grade unless the structure of the facility issuch that LIED is substantially better or worse than‘‘normal.’’ Implicitly, grades on the facility scalemeasure EL as the PD associated with the borrower

grade multiplied by a standard or average LIED(table 1). In this way, a two-dimensional system canpromote precision and consistency in grading byseparately recording a rater’s judgments about PDand EL rather than mixing them together.

A few banks said they had plans to shift to asystem in which the borrower grade reflects PD butthe facility grade explicitly measures LIED. The raterwould assign a facility to one of several LIED cate-gories on the basis of the likely recovery rates asso-ciated with various types of collateral, guarantees, orother considerations associated with the facility’sstructure. EL for a facility would be calculated bymultiplying the borrower’s PD by the facility’sLIED.8

Rating Scales at Moody’s and S&P

At the agencies, as at many banks, the loss concepts(PD, LIED, and EL) embedded in the ratings aresomewhat ambiguous. Moody’s states that ‘‘ratingsare intended to serve as indicators or forecasts of thepotential forcredit lossbecause of failure to pay, adelay in payment, or partial payment.’’ Standard &Poor’s states that its ratings are an ‘‘opinion of thegeneral creditworthiness of an obligor, or . . . of anobligor with respect to a particular . . . obligation . . .

8. Systems recording LIED rather than EL as the second grade canpromote precision and consistency in grading. PD-EL systems typi-cally impose limits on the degree to which differences in loan struc-ture permit an EL grade to be moved up or down relative to the PDgrade. Such limits can be helpful in restraining raters’ optimism but,in the case of loans with a genuinely very low expected LIED, suchlimits can materially limit the accuracy of risk measurement. Anotherbenefit of LIED ratings is the fact that raters’ LIED judgments can beevaluated over time by comparing them to loss experience.

1. Example of a two-dimensional risk rating systemusing average LIED values

Grade

Borrowerscale:

borrower’sprobabilityof default

(PD)(percent)

(1)

Assumedaverageloss onloans inthe eventof default(LIED)

(percent)(2)

Facilityscale:

expectedloss(EL)

on loans(percent)(1 × 2)

1—Virtually no risk . . 0 02—Low risk . . . . . . . . . .1 .033—Moderate risk. . . . . .3 .094—Average risk. . . . . . 1.0 .305—Acceptable risk . . . 3.0 30 .906—Borderline risk . . . 6.0 1.807—OAEM1 . . . . . . . . . . 20.0 6.008—Substandard. . . . . . 60.0 18.009—Doubtful . . . . . . . . . 100 30.00

1. Other Assets Especially Mentioned.

2. Moody’s and Standard & Poor’s bond rating scales and average one-year default rates

Category

Moody’s Standard & Poor’s

GradeAverage default rate (PD)

per year, 1970–95(percent)

GradeAverage default rate (PD)

per year, 1981–94(percent)

Investment grade. . . . . . . . . . . . . . . . . . . . . . . . . . . . Aaa .00 AAA .00Aa, Aa1, Aa2, Aa3 .03 AA+, AA, AA− .00A, A1, A2, A3 .01 A+, A, A− .07Baa, Baa1, Baa2, Baa3 .13 BBB+, BBB, BBB− .25

Below investment grade (‘‘junk’’). . . . . . . . . . . . Ba,Ba1, Ba2, Ba3 1.42 BB+, BB, BB− 1.17B, B1, B2, B3 7.62 B+, B, B− 5.39Caa, Ca, C n.a. CCC, CC, C 19.96

Default . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D . . . D . . .

Note. Grades are listed from less risky to more risky, from top to bottom andfrom left to right.

n.a. Not available.. . . Not applicable.

Source. Moody’s Investors Service Special Report,Corporate BondDefaults and Default Rates 1938–1995(January 1996). Standard & Poor’sCreditweek Special Report,Corporate Defaults Level Off in 1994(May 1,1995).


based on relevant risk factors.’’ On balance, a closereading of Moody’s and Standard & Poor’s detaileddescriptions of rating criteria and procedures sug-gests that the two agencies’ ratings incorporate ele-ments of PD and LIED but are not precisely ELmeasures.9

Risk tends to increase nonlinearly on both bankand agency scales. For example, on the agency scales,default rates are low for the least risky grades but riserapidly as the grade worsens (table 2).

Administrative Grades

All the banks we interviewed maintain some sort ofinternal ‘‘watch’’ list as well as a means of identify-ing assets that fall into the ‘‘regulatory problemasset’’ categories (table 3). Although watch and regu-latory problem-asset designations typically identifyhigh-risk credits, they have administrative meaningsthat are conceptually separate from risk per se. Spe-cial monitoring activity is usually undertaken forwatch and problem assets, such as formal quarterlyreviews of status and special reports that help seniorbank management monitor and react to importantdevelopments in the portfolio. However, banks maywish to trigger special monitoring for credits that arenot high-risk and thus may wish to separate adminis-trative indicators from risk measures (an examplewould be a low-risk loan for which an event thatmight influence risk is expected, such as a change inownership of the borrower).

Among the fifty largest banks, all but two havegrades corresponding to the regulatory problem-assetcategories Other Assets Especially Mentioned(OAEM), Substandard, Doubtful, and Loss (someomit the Loss category).10 All other assets are collec-tively labeled ‘‘Pass’’ by regulators. The bank super-visory agencies do not specifically require that banksmaintain regulatory categories on an internal scalebut do require that recordkeeping be sufficient toensure that loans in the regulatory categories can bequickly and clearly identified. The two banks that useprocedures not involving internal grades appear to doso because the regulatory asset categories are notconsistent with the conceptual basis of their own

grades.11 Moreover, banks and regulators may some-times disagree about the riskiness of individual assetsthat fall into the various regulatory grades.12

Watch credits are those that need special monitor-ing but do not fall in the regulatory problem-assetgrades. Only about half the banks we interviewedinclude a watch grade on their internal rating scales.Others add a watch flag to individual grades, such as3W versus 3, or simply maintain a watch list sep-rately, perhaps by adding an identifying field to theircomputer systems.

9. Moody’s Investors Service,Global Credit Analysis(IFR Pub-lishing, 1991), p. 73 (emphasis in the original); Standard & Poor’s,Corporate Ratings Criteria(1998), p. 3. Other rating agencies playimportant roles in the marketplace. We omit details of their scales andpractices only for brevity.

10. A few break Substandard into two categories, one for perform-ing loans and the other for nonperforming loans.

11. Although the definitions are standardized across banks, ourdiscussions and inspection of internal documents imply that banksvary in their internal definition and use of OAEM. Among the regu-latory categories, OAEM in particular can have an administrativedimension as well as a risk dimension. Most loans identified asOAEM pose a higher-than-usual degree of risk, but some loans maybe placed in this category for lack of adequate documentation in theloan file, which may occur even for loans not posing higher-than-usualrisk. In such cases, once the administrative problem is resolved, theloan can be upgraded.

12. Examiners review problem loans and evaluate whether theyhave been assigned to the proper regulatory problem-asset grades andalso review a sample of Pass credits. Examiners heretofore havegenerally not attempted to validate or evaluate internal ratings of Passcredits.

3. Regulatory problem asset categories

Category Regulatory definition

Recommendedspecificreserve

(percent)

Special Mention(OAEM)1 . . Has potential weaknesses that

deserve management’s closeattention.

If left uncorrected, these potentialweaknesses may,at some futuredate, result in the deterioration ofthe repayment prospects for thecredit.

Norecommendation

Substandard. . . . . Inadequately protected by currentworth/paying capacity of obligor orcollateral. Well-defined weaknessesjeopardize liquidation of the debt.

Distinct possibility that bank willsustain some loss if deficiencies arenot corrected.

15

Doubtful . . . . . . . . All weaknesses inherent insubstandard, AND collection/liquidation in full, on basis ofcurrently existing conditions, ishighly questionable or improbable.

Specific pending factors maystrengthen credit; treatment as lossdeferred until exact status can bedetermined.

50

Loss . . . . . . . . . . . . Uncollectible and of such littlevalue that continuance as bankableasset is not warranted.

Credit may have recovery orsalvage value, but notpractical/desirable to defer writingit off even though partial recoverymay be effected in future.

100

Note. Assets that do not fall into one of these categories are termed Pass bythe federal banking regulators.

1. Other Assets Especially Mentioned.


Number of Grades on the Scale

The number of grades on internal scales varies con-siderably across banks. In addition, even where thenumber of grades is identical on two different banks’scales, the risk associated with the same grades (forexample, two loans graded 4) is almost always differ-ent. Among the fifty largest banks, the number ofPass grades varies from two to the low twenties. Themedian is five Pass grades, including a watch gradeif any (chart 1). Among the ten largest banks, themedian number of Pass grades is six and the mini-mum is four. As noted, the vast majority of largebanks also include three or four regulatory problem-asset grades on their internal scales.

Internal rating systems with larger numbers ofgrades are more costly to operate because of the extrawork required to distinguish finer degrees of risk.Banks making heavy use of ratings in analyticalactivities are most likely to choose to bear these costsbecause fine distinctions are especially valuable insuch activities (however, at least a moderate numberof Pass grades is useful even for internal reportingpurposes). Banks that increase their analytical use ofratings may persist for a while with a relatively smallnumber of Pass grades because the costs of changingrating systems can be large. Nonetheless, those banksthat have recently redesigned their rating systemshave all increased the number of grades.13

The proportion of grades used to distinguish amongrelatively low risk credits versus the proportion used

to distinguish among the riskier Pass credits tends todiffer with the business mix of the bank. Amongbanks we interviewed, those that do a significantshare of their commercial business in the large corpo-rate loan market tend to have more grades reflectinginvestment-grade risks. The allocation of gradesbetween the investment-grade and below-investment-grade categories tends to be more even at banksdoing mostly middle-market business.14 The differ-ences are not large: The median middle-market bankhas three internal grades corresponding to agencygrades of BBB−/Baa3 or better and three riskiergrades, whereas the median bank with a substantiallarge-corporate business has four investment gradesand two junk grades. Such a difference in ratingsystem focus is sensible in that an ability to make finedistinctions among low-risk borrowers is quite impor-tant in the highly competitive large-corporate lendingmarket. In the middle market, fewer borrowers areperceived as posing AAA, AA, or even A levels ofrisk, so such distinctions are less crucial.

However, a glance at table 2 reveals that a gooddistinction among risk levels in the below-investment-grade range is important for all banks.For example, the range of default rates spanned bythe agency grades BB+/Ba1 through B−/B3 is ordersof magnitude larger than the risk range for, say,A+/A1 through BBB−/Baa3, and yet the median largebank we interviewed uses only two or three grades tospan the below-investment-grade range, one of themperhaps being a watch grade. More granularity—finerdistinctions of risk, especially among riskier assets—can enhance a bank’s ability to analyze its portfoliorisk posture and to construct accurate models of theprofitability of its broader business relationships withborrowers.

Systems with many Pass categories are less usefulwhen loans or other exposures tend to be concen-trated in one or two grades. Among large banks,sixteen institutions, or 36 percent, assign half or moreof their rated loans to a single risk grade (chart 2).Such systems appear to contribute little to the under-standing and monitoring of risk posture.15

13. The average number of grades on internal scales appears tohave increased somewhat during the past decade. See Gregory F.Udell, Designing the Optimal Loan Review Policy: An Analysis ofLoan Review in Midwestern Banks(Prochnow Reports, Madison,Wis., 1987), p. 18.

14. The term ‘‘large corporate’’ includes nonfinancial firms withlarge annual sales volumes as well as large financial institutions,national governments, and large nonprofit institutions. Certainly theFortune 500 firms fall into this category. Middle-market borrowers aresmaller, but the precise boundary between large and middle-marketand between middle-market and small business borrowers varies bybank.

15. Such failure to distinguish degrees of risk was recently cited inFederal Reserve examination guidance as a potentially significantshortcoming in a large institution’s credit risk management process.See Supervision and Regulation Letter SR 98-18, ‘‘Lending Standardsfor Commercial Loans’’ (June 23, 1998). For additional information

1. Fifty largest U.S. banks, distributed by numberof Pass grades

1 to 3 4 5 6 7 8 or more

Number of grades

3

6

9

12

Number of banks

Note. Shown are the forty-six banks for which this measure was available.


The majority of the banks that we interviewed(and, based on discussions with supervisory staff,other banks as well) expressed at least some desire toincrease the number of grades on their scales and toreduce the extent to which credits are concentrated inone or two grades. Two kinds of plans were voiced:Addition of a +/− modifier to all existing grades, anda split of existing riskier grades into a larger numberof newly defined grades, leaving the low-risk gradesunchanged.16 The +/− modifier approach is favoredby many because grade definitions are modifiedrather than completely reorganized. For example, thebasic meaning of a 5 stays the same, but it becomespossible to distinguish between a strong and a weak 5with grades of 5+ and 5−. This approach limits thedisruption of staff understanding of each grade’smeaning (as noted below, such understanding islargely cultural rather than being formally written).

THE OPERATINGDESIGN OFRATING SYSTEMS

In essentially all cases, the human judgment exer-cised by experienced bank staff is central to the

assignment of a rating. Banks thus design the opera-tional flow of the rating process in ways that areaimed at promoting the accuracy and consistency ofratings while not unduly restricting the exercise ofjudgment. Balance between these opposing impera-tives appears to be struck at each institution on thebasis of cost considerations, the nature of the bank’scommercial business lines, the bank’s uses of ratings,and the role of the rating system in maintaining thebank’s credit culture.

Key operating design issues in striking the balanceinclude the organizational division of responsibilityfor grading (line staff or credit staff), the nature ofreviews of ratings to detect errors, the organizationallocation of ultimate authority over grade assign-ments, the role of external ratings and statisticalmodels in the rating process, and the formality of theprocess and specificity of formal rating definitions.

What Exposures Are Rated?

At most banks, ratings are produced for all commer-cial or institutional loans (that is, not consumerloans), and in some cases for large loans to house-holds or individuals for which underwriting proce-dures are similar to those for commercial loans. Ratedassets thus include commercial and industrial loansand other facilities, commercial lease financings,commercial real estate loans, loans to foreign com-mercial and sovereign entities, loans and other facili-ties to financial institutions, and sometimes loansmade by ‘‘private banking’’ units. In general, ratingsare applied to those types of loans for which under-writing requires large elements of subjective analysis.

Overview of the Rating Process in Relation toCredit Approval and Review

Ratings are typically assigned (or reaffirmed) at thetime of each underwriting or credit approval action.The analysis supporting the ratings is inseparablefrom the analysis supporting the underwriting orcredit approval decision. In addition, the rating andunderwriting processes, while logically separate, areintertwined. The rating assignment influences theapproval process in that underwriting limits andapproval requirements depend on the grade, whileapprovers of a credit are expected to review andconfirm the grade. For example, an individual staffmember typically proposes a risk grade as part of thepre-approval process for a new credit. The proposedgrade is then approved or modified at the same time

about current bank lending practices, see William F. Treacy, ‘‘TheSignificance of Recent Changes In Underwriting Standards: Evidencefrom the Loan Quality Assessment Project,’’Federal Reserve SystemSupervisory Staff Report(June 1998); and U.S. Comptroller of theCurrency,1998 Survey of Credit Underwriting Practices(NationalCredit Committee, 1998).

16. At the time of the interviews, however, the majority of thebanks voicing plans to increase the number of their grades had noactive effort in progress. Many of those institutions actively moving toincrease the number of their Pass grades do not now have concentra-tions in a single category.

2. Fifty largest U.S. banks, distributed by percentageof outstandings placed in the grade with themost outstandings

Lessthan 20

20–29 30–39 40–49 50–59 60–69 70–79 80 ormore

Percentage in single grade

3

6

9

12

Number of banks

Note. Shown are the forty-five banks for which this measure was relevant.


that the transaction itself receives approval and mustmeet the requirements embedded in the bank’s creditpolicies. In nearly all cases, approval requires assentby individuals with requisite ‘‘signature authority’’rather than by a committee. The number and level ofsignatures needed for approval typically depend onthe size and (proposed) risk rating of the transaction:In general, less risky loans require fewer and perhapslower-level signatures. In addition, signature require-ments may vary according to the line of businessinvolved and the type of credit being approved.17

After approval, the individual that assigned theinitial grade is generally responsible for monitoringthe loan and for changing the grade promptly as thecondition of the borrower changes. Exposures fallinginto the regulatory grades are an exception at someinstitutions, where monitoring and grading of suchloans becomes the responsibility of a separate unit,such as a workout or loan review unit.

Who Assigns and Monitors Ratings, and Why?

Ratings are initially assigned either by relationshipmanagers or the credit staff. Relationship managers(RMs) are lending officers (line staff) responsible forthe marketing of banking services. They report tolines of business that reflect the strategic orientationof the bank.18 All institutions evaluate the perfor-mance of RMs—and thus set their compensation—onthe basis of the profitability of the relationships inquestion, although the methods of assessing profit-ability and determining compensation vary. Evenwhen profitability measures are not risk-sensitive,ratings assigned by an RM can affect his or hercompensation.19 Thus, in the absence of sufficientcontrols, RMs may have incentives to assign ratingsin a manner inconsistent with the bank’s interests.

The credit staff is responsible for approving loansand the ratings assigned, especially in the case oflarger loans; for monitoring portfolio credit qualityand sometimes for regular review of individual expo-sures; and sometimes for directly assigning theratings of individual exposures. The credit staff is

genuinely independent of sales and marketing func-tions when the two have separate reporting structures(that is, ‘‘chains of command’’) and when the perfor-mance assessment of the credit staff is linked to thequality of the bank’s credit exposure rather than toloan volume or business line or customer profitabil-ity. Some banks apportion the credit staff acrossspecific line-of-business groups. Such arrangementsallow for closer working relationships but in somecases lead to linkage of the credit staff’s compensa-tion or performance assessment with profitability ofthe business line; in such cases, incentive conflictslike those experienced by RMs can arise. At otherbanks, RMs and independent credit staff produceratings as partners and are held jointly accountable.Whether such partnerships are effective in restrainingincentive conflicts is not clear.

The primary responsibility for rating assignmentsvaries widely among the banks we interviewed. RMshave the primary responsibility at about 40 percent ofthe banks, although in such cases the credit staff mayreview proposed ratings as part of the loan approvalprocess, especially for larger exposures.20 At 15 per-cent of interviewed banks the credit staff assigns allinitial ratings, whereas the credit staff and RMs ratein partnership at another 20 percent or so. About30 percent of interviewed banks divide the responsi-bility: The credit staff has sole responsibility forrating large exposures, and RMs alone or in partner-ship with the credit staff rate middle-market loans. Inprinciple, both the credit staff and RMs use the samerating definitions and basic criteria, but the differentnatures of the two types of credit may lead to somedivergence of practice.

A bank’s business mix appears to be a primarydeterminant of whether RMs or the credit staff areprimarily responsible for ratings. Those banks weinterviewed that lend mainly in the middle marketusually give RMs primary responsibility for ratings.Such banks emphasized informational efficiency,cost, and accountability as key reasons for theirchoice of organizational structure. Especially in thecase of loans to medium-size and smaller firms, theRM was said to be in the best position to appraise thecondition of the borrower on an ongoing basis andthus to ensure that ratings are updated in a timelymanner. Requiring that the credit staff be equally wellinformed adds costs and may introduce lags into theprocess by which ratings of such smaller credits areupdated.

17. If those asked to provide signatures believe that a loan shouldbe assigned a riskier internal rating than initially, additional signaturesmay be required in accordance with policy requirements. Thus, dis-agreement over the rating can alter the approval requirements for theloan in question.

18. Lines of business may be defined by the size of the businesscustomer (such as large corporate), by the customer’s primary indus-try (such as health care), or by the type of product being provided(such as commercial real estate loans).

19. For example, because loan policies often include size limitsthat depend on ratings, approval of a large loan proposed by an RMmay be much more likely if it is assigned a relatively low risk rating.

20. At most banks, RMs have signature authority for relativelysmall loans, and the credit staff might review the ratings of only afraction of small loans at origination.


The institutions at which an independent creditstaff assigns ratings tend to have a substantial pres-ence in the large corporate market. Placing the ratingprocess primarily in the hands of the credit staffoffers greater assurance that grading will be purely onthe basis of risk, without coloration by possible rami-fications for customer or business line profitability. Inaddition, because the credit staff is small relative tothe number of RMs and is focused entirely on riskassessment, it is in a better position to achieve consis-tency in its ratings (that is, to assign similar grades tosimilarly risky loans, regardless of their other charac-teristics). Moreover, the costs of having the creditstaff perform all analysis are small relative to therevenues generated by large corporate loan transac-tions. In contrast, such costs can be large relative tothe transaction revenues for middle-market loans.

Our impression is that middle-market lending rep-resents a much larger share of the business of bankswe did not interview. If the pattern described aboveholds, the proportion of all large banks using RM-centered rating processes is probably higher thanamong our interviewees. Unfortunately, policy docu-ments for those we did not interview generally do notreveal details of this aspect of the process.

Almost all the banks we interviewed are at leastexperimenting with consumer-loan-style credit scor-ing models for small commercial loans. For expo-sures smaller than some cutoff value, such models areeither a tool in the rating process or are the sole basisfor the rating. If, however, models are the sole basis,performing loans are usually assigned to a singlegrade on the internal rating scale rather than makinggrade assignments sensitive to the score value.

How Do They Arrive at Ratings?

Both assigners and reviewers of ratings follow thesame basic thought process in arriving at a rating fora given exposure. The rater considers both the riskposed by the borrower and aspects of the facility’sstructure. In appraising the borrower, the rater gathersinformation about its quantitative and qualitativecharacteristics, compares them with the standards foreach grade, and then weights them in choosing aborrower grade. The comparative process often isas much one of looking across borrowers as one oflooking across characteristics of different grades:That is, the rater may look for already-rated loanswith characteristics close to those of the loan beingrated and then set the rating to the grade alreadyassigned to such borrowers.

Models and Judgment

Although in principle the analysis of risk factorsmay be done by a mechanical model, in practice therating process at almost all banks relies heavily onjudgment. We suspect most banks are hesitant tomake models the centerpiece of their rating systemsfor three reasons: (1) Different models would berequired for each asset class and perhaps for differ-ent geographic regions; (2) data to support estimationof such models is currently rarely available; and(3) the reliability of such models would becomeapparent only over time, exposing the bank topossibly substantial risks in the interim. Those fewbanks moving toward heavy reliance on modelsappear to feel that models produce more consistentratings and that, in the long run, operating costs willbe reduced in that less labor will be required toproduce ratings.

As part of their judgmental evaluation, most of thebanks we interviewed either use statistical models ofborrower default probability as an input (about three-fourths do so) or take into consideration any availableagency rating of the borrower (at least half, andprobably more, do so). Such use of external points ofcomparison is common for large corporate borrowersbecause they are most likely to be externally ratedand because statistical default probability models aremore readily available for such borrowers. In addi-tion, as described further below, many banks useexternal ratings or models in calibrating their ratingsystems and in identifying likely mistakes in gradeassignments.

Factors Considered

Bank personnel base their decisions to assign a par-ticular rating on the criteria that define each grade,which are articulated as standards for a numberof specific risk factors. For example, a criterionfor assignment of a grade ‘‘3’’ might be that theborrower’s leverage ratio must be smaller than somevalue. Risk factors include the borrower’s financialcondition, size, industry, and position within theindustry; the reliability of the borrower’s financialstatements and the quality of its management; ele-ments of transaction structure (for example, collat-eral); and miscellaneous other factors. The risk fac-tors are generally the same as those considered indeciding whether to extend a loan and are similar tothe factors considered by rating agencies. Banks varysomewhat in the particular factors they consider andin the weight they give each factor. What follows is a


description of the factors considered by a typicalbank among those we interviewed.21

Financial statement analysis is central to apprais-ing the likely adequacy of future cash flow and thusthe ability of the borrower to service its debt. Thefocus of analysis is on the borrower’s debt servicecapacity, taking account of its free cash flow, theliquidity of its balance sheet, and the firm’s access tosources of finance other than the bank. Historical(and to a lesser extent, projected) earnings, operatingcash flow, interest coverage, and leverage are typi-cally analyzed, with exact definitions of financialratios used in the analysis varying across banks and,in some cases, across borrowers or loan types. Theanalysis yields an assessment of the differencebetween current or projected performance and liquid-ity on the one hand and projected debt service obliga-tions on the other. The larger the cushion, in general,the more favorable the rating.

As a context for financial statement analysis, thecharacteristics of the borrower’s industry are oftenconsidered (such as cyclicality, general volatility, andtrends in cash flow and profitability). Indeed, thefinancial analysis often includes a formal comparisonof the borrower’s financial ratios to prevailing indus-try norms.22 Firms in declining industries are consid-ered more risky, as are those in highly competitiveindustries, whereas firms with diversified lines ofbusiness are viewed as less risky. A related factor, theborrower’s position in its industry, is also an impor-tant factor in determining ratings. Those borrowerswith substantial market power or that are perceived tobe ‘‘market leaders’’ in other respects are consideredless risky because they are thought to be less vulner-able to competitive pressure.

One of the most important reasons that rating isusually a judgmental process is that the details offinancial statement analysis vary with the borrower’sother characteristics. In contrast, statistical models ofdefault probability tend to analyze fixed sets of finan-cial ratios and to apply fixed weights to each ratio inarriving at a default probability, perhaps with somevariation in weights by industry. Subjective factorsplay at most a minimal role. This relative inflexibility

of models leads most banks to regard their resultsonly as generally suggestive of an appropriate rating.When internal ratings are produced primarily bymodels, several models may be needed for differentborrowers or loan types and continual tuning of themodels is likely to be required.

Raters also appraise the quality of financial infor-mation provided by the borrower. For example, rat-ers have much more confidence in financial state-ments that are audited by a major accounting firmthan in those that are compiled or unconsolidated orthat are audited but accompanied by important quali-fications. When statement quality is poor or uncer-tain, financial analysis may produce a distorted viewof the borrower’s condition, adding substantially torisk.

A primary difference between banks and publicrating agencies is whether the financial analysis iskeyed to a downside (or ‘‘stress’’) scenario or to a‘‘base’’ (or ‘‘most likely’’) case. As noted previously,banks assign ratings on the basis of the borrower’scurrent condition and most likely outlook, whereasthe rating agencies assign grades on the basis of adownside scenario.

In another departure from practice at the ratingagencies, most banks formally consider both firmsize (sales revenue or total assets) and the book ormarket dollar value of a firm’s equity in assigningratings. Interviewees noted that small firms—including many that would be considered middlemarket—usually have limited access to externalfinance and often have few or no assets that can besold in an emergency without disrupting operations.In contrast, larger firms were characterized as havingmore ready access to alternative financing, more sale-able assets, and a more firmly established marketpresence. For these reasons, many banks require thatsmall borrowers be assigned relatively risky gradeseven if their financial characteristics might suggest amore favorable rating.

Almost all internal rating systems cite the borrow-er’s management as an important consideration inassigning the risk grade. Such assessments are nec-essarily subjective and may reveal weaknesses in anumber of areas related to competence, experience,integrity, or succession plans. Vulnerability of man-agement to the retirement or departure of key indi-viduals is usually considered. Some institutions (simi-lar to the rating agencies) appear to give considerableweight to the rater’s appraisal of management’s abil-ity and willingness to manage the firm to achieve ahigh level of financial performance throughout thebusiness cycle and to its attitude toward protectingthe interests of lenders.

21. We reviewed thewritten criteria for those banks among thefifty largest that we did not interview. Our experience with inter-viewed banks indicates that conclusions should be drawn with carefrom written documents alone. However, the description of risk fac-tors herein is probably representative of the factors used by almost alllarge banks.

22. Staff at the banks interviewed appeared to be well aware of thepotential pitfalls of such comparisons. For example, a borrower with afive-year history of stable cash flow might still be considered ratherrisky if the particular five-year period contained no recession and theborrower’s industry is highly cyclical.


The borrower’s country of domicile or operationsis an important determinant of the rating in somecases. Especially when transfer risk or political risk issubstantial, general practice seems to be that a bor-rower’s grade may be no less risky than the gradeassigned to the borrower’s country by a special unitin the bank. Such country grades can be significantlyaffected by the country risk grade assigned by regula-tors as part of an annual cycle.

Ratings may also be influenced by exposure toevent risks, such as litigation, environmental liability,or changes in law or national policy.

A handful of considerations reflecting the structureof the transaction being rated also enter into consider-ation because they can affect LIED. Adequate collat-eral can in many cases improve the rating, particu-larly if that collateral is in the form of cash or easilymarketed assets such as U.S. Treasury securities.23

Guarantees can generally enhance the rating as well,but not beyond the rating that would be assigned tothe guarantor if it were the borrower. The term tomaturity of the loan is a factor in grade assignmentsat only a few large banks. Similarly, few banks adjustthe risk grade on the basis of other elements of theloan structure, such as financial covenants.

Written and Cultural Definitions

Large banks’ written definitions of ratings specifyrisk factors to be used in assigning ratings, but usu-ally the discussion is brief and broadly worded, andgives virtually no guidance regarding the weight toplace on each factor.24 According to interviewees,such brevity arises partly because some factors arequalitative but also because the specifics of quantita-tive factors and the weights on factors can differ agreat deal across assets. Some noted that the number

of permutations is so great that attempting to writethem down would be counterproductive. Instead,raters learn to exercise judgment in selecting andweighting factors through training, mentoring, andespecially by experience. The actual meanings ofwritten rating definitions and the specifics of assign-ing ratings take the form of common,unwrittenknowledge embedded in the bank’s credit culture.

Formality of Procedure

Most banks require some sort of written justificationof the grade as part of the loan approval package, buta few employ forms or grids on which the rateridentifies the relevant factors. Such forms or gridsmay also suggest a structure for the rating analy-sis and serve to remind the rater to consider a broadset of risk factors and to weight them appropri-ately. The stated motivation for such formalism isbetter consistency across asset types and geographicregions.

Reviews, Reviewers, and the ‘‘Keepers of theFlame’’

Reviews of ratings are threefold: Monitoring by thosewho assign the initial rating of a transaction, regu-larly scheduled reviews of ratings for groups of expo-sures, and occasional reviews of a business unit’srating assignments by a loan review unit. Monitoringmay not be continuous, but it is intended to keep therater well enough informed to recommend changes tothe internal risk grade in a timely fashion as needed.All institutions interviewed emphasized that failureto recommend changes to risk grades in a timelyfashion when warranted is viewed as a significantperformance failure for the relationship manager, thecredit staff, or both, and can be grounds for internallyimposed penalties.25

Most institutions also conduct annual or quarterlyreviews of each exposure, which may be in additionto those that are part of the credit approval process atthe time facilities are renewed. The form of regularreviews ranges from a periodic signoff by the rela-tionship manager working alone to a committeereview involving both line and credit staff. Bankswith substantial large-corporate portfolios tend toreview all exposures in a given industry at the sametime, with reviews either by the credit specialistfor that industry or by a committee. Such industry

23. Different rules are often used in grading certain classes oftransactions, especially asset-based lending. At best, asset-based bor-rowers would be only marginally acceptable risks for banks in theabsence of the detailed field audits of collateral that asset-basedlenders demand. With such close monitoring, which typically includessome degree of bank dominion over accounts receivable and inven-tory, the expected loss associated with a default is dramaticallyreduced, and a more favorable rating can be assigned.

24. Written definitions are intended to address a broad range ofcredit classes and borrower types. At a few banks, a supplementarygrid of nonbinding quantitative standards or financial ratios is pro-vided (for example, for leverage or debt service coverage), but guid-ance is generally sketchy as to how such ratios should be weightedagainst each other or against more qualitative considerations. Inter-viewees indicated that even when reference grids are provided, theratios and standards are generally not binding. Similarly, some banksprovide supplemental descriptions of risk factors to be considered forparticular business lines or loan types, but such supplements oftenclosely resemble the core risk rating definitions.

25. Updates to the risk grade usually require approvals similar tothose required to initiate or renew a transaction.


reviews were said to be especially helpful in reveal-ing inconsistently rated credits.

Ratings are also checked by banks’ independentloan review units, which usually have the finalauthority to set grades. Such departments examineeach business unit’s underwriting practices, and itsadherence to administrative and credit policies, on aone- to three-year cycle. Not unlike bank examiners,the loan review staff typically inspects only a sampleof loans in each line of business. Although the sam-pling procedures used by different institutions varysomewhat, most institutions weight samples towardloans perceived to be riskier (such as those in high-risk loan grades), with the primary focus on regu-latory problem-asset categories. In general, however,an attempt is made to review some loans made byeach lender in the unit being inspected.26

At a few banks, the loan review unit inspectsinternal ratings assigned to Pass loans only to confirmthat such loans need not be placed in the watch orregulatory grades. Thus, as a practical matter, theloan review unit at these banks has little role inmaintaining the accuracy of assignments within thePass grades. In this regard, the loan review staff atthese banks follows the same pattern as bank examin-ers. These banks tend to make relatively little use ofPass grade information in managing the bank.

Because operational rating definitions and proce-dures are embedded in bank culture rather than writ-ten down in detail, the loan review function at mostinstitutions is critical to maintaining the disciplineand consistency of the overall rating process. Theloan review unit, as the principal entity looking atratings across business lines and asset types, oftenbears much of the burden of detecting discrepanciesin the operational meaning of ratings across lines.

Because the loan review unit at most institutionshas the final say about ratings, it can exert a majorinfluence on the culturally understood definition ofgrades.27 Typically, when the loan review staff findsgrading errors, it not only makes corrections butworks with the relevant staff to find the reasons for

the errors. Misunderstandings are thus corrected asthey become evident.28

Loan review units generally do not require that allratings produced by the line or credit staff be iden-tical to the ratings that loan review judges to becorrect. At almost all banks we interviewed, loanreview units treat only two-grade discrepancies forindividual loans as warranting discussion. With atypical large bank having four to six Pass categories,such a policy permits large discrepancies for indi-vidual exposures, potentially spanning two or morewhole letter grades on the Standard & Poor’s scale.However, most institutions interviewed indicated thata pattern of one-grade disagreements within a givenbusiness unit—for example, a regional office of agiven line of business—can lead to a quick anddecisive response.

All interviewees emphasized that the number ofcases in which the loan review staff changes ratings isusually relatively small, ranging from essentiallynone to roughly 10 percent of the loans reviewed,except in the wake of large cultural disruptions suchas mergers or major changes in the rating system. Alow percentage of discrepancies does not imply thatthe loan review function is unimportant but ratherthat, in well-functioning systems, the cultural mean-ing of ratings tends to remain stable and widelyunderstood. One element of a well-functioning sys-tem is the rater’s expectation that the loan reviewstaff will be conducting inspections.

The interviews also indicated that differences ofopinion tend to become more common when thenumber of ratings on the scale is greater, creatingmore situations in which ‘‘reasonable people candisagree.’’ More direct linkage between the risk gradeassigned and the incentive compensation of relation-ship managers also tends to produce more disagree-ments. In both cases, resolution of disagreementsmay consume more resources.

Loan review units usually have a role apart frominspections in maintaining rating system integrity.For example, when a relationship manager and thecredit staff are unable to agree on a rating for a newloan, they will consult with the loan review unit onhow to resolve the dispute. In its consultative role,the loan review staff guides the interpretations ofrating definitions and standards and, in novel situa-tions, establishes and refines the definitions.

26. For an analysis of the broader role of loan review units, seeUdell, Designing the Optimal Loan Review Policy;and Gregory F.Udell, ‘‘Loan Quality, Commercial Loan Review, and Loan OfficerContracting,’’ Journal of Banking and Finance,vol. 3 (July 1989),pp. 367–82.

27. Interviews and discussions with supervisory staff suggest, how-ever, that the notion of ‘‘final say’’ is murkier than suggested bywritten policy and stated practice. Important informal elements ofrating processes, such as negotiation among various organizationalunits, may lead to a consensus rating or understanding. Such negotia-tion would not compromise the integrity of the rating system so longas loan review retains its independence and objectivity. Such informalunderstandings might make it more difficult, however, for an outsiderto understand (much less validate) the ratings being assigned.

28. The loan review staff generally uses the same definitions of riskgrades, at the same level of detail, as relationship managers and theindependent credit staff. At a few banks, however, loan review alsorelies on older policy documents that are far more detailed thancurrent policies. Thus, the older, more specific policies remain essen-tially in effect.


Because of its central role in maintaining the integ-rity of the rating system, the loan review unit musthave both substantial independence and staff mem-bers who are well versed in the bank’s credit cultureand the meaning of ratings. All loan review units atbanks we interviewed report to the chief auditor orchief credit officer of the bank, and many periodicallybrief the board (or a committee thereof) on the resultsof their reviews.

Loan review units may be less critical to the integ-rity of rating systems at banks that are primarily inthe business of making large corporate loans and atwhich all exposures are rated by a relatively small,highly independent credit staff. Although few bankscurrently fit this description, they provide an interest-ing contrast. Such banks’ credit units tend to conductthe annual industry-focused reviews mentioned previ-ously and thus are likely to detect rating discrepan-cies. Having such reviews conducted by broadlybased committees rather than only by industry spe-cialists tends to restrain any drift in the meaning ofratings as applied to different industries. In suchcircumstances, the small credit staff is in a goodposition to function as the ‘‘keeper of the flame’’ withregard to the credit culture because it essentiallycarries out the key rating oversight functions of tradi-tional loan review units.

Rating Systems and Credit Culture

‘‘Credit culture’’ refers to an implicit understandingamong bank personnel that certain standards ofunderwriting and loan management must be main-tained, even in the face of constant pressures toincrease revenues and bring in new business. Mainte-nance of a credit culture can be difficult, especially atvery large banks serving many customers over a widearea. Of necessity, substantial authority must be dele-gated to mid-level and junior personnel, and unduerelaxation of standards may not appear in the form ofloan losses for some time.

At some of the banks we interviewed, senior man-agers indicated that the internal rating system is atleast partly designed to promote and maintain theoverall credit culture. At such banks, relationshipmanagers are held accountable for credit qualitypartly by having them rate all credits, including largeexposures that might be more efficiently rated by thecredit staff. Strong review processes aim to identifyand discipline relationship managers who produceinaccurate ratings. Such a setup provides strongincentives for the individual most responsible fornegotiating with the borrower to assess risk properly

and to think hard about credit issues at each stage ofa credit relationship rather than relying entirely onthe credit staff. An emphasis on culture as a criticalconsideration in designing the rating system wasmost common among institutions that had sufferedserious problems with asset quality in the past ten orfifteen years.

Tensions can arise when rating systems both main-tain culture and support sophisticated modeling andanalysis. As noted, the latter applications introducepressures for architectures involving fine distinctionsof risk, and the frequency of legitimate disagreementsabout ratings is likely to be higher when systemshave a large number of Pass grades. If not properlyhandled by senior management and the loan reviewunit, a rating system redesign that increases the num-ber of grades may make cultural norms fuzzier andthe rating system less useful in maintaining the creditculture.

Mergers and Expense Pressure

Some of our interviews involved banks that hadrecently been involved in mergers, and the dis-cussions clearly indicated that mergers can causeupheaval in credit processes and systems, credit cul-ture, and traditional sources of rating discipline. Aftera rating system architecture is chosen for the com-bined institution, mechanical issues of converting thepredecessor banks’ ratings to the new scale can bechallenging, especially when the predecessors’ rat-ings of the same borrower suggest differing assess-ments of that borrower’s risk. Cultural disruptionsarising from the merger are usually even more prob-lematic than the mechanical issues because, as noted,the operational definitions of ratings are a matter ofculture. Even if the architecture of one of the prede-cessors is used as-is, the staff of the other bank mustabsorb and adjust to the new culture.

Merging institutions face a difficult choice betweenmoving very quicky to convert the ratings of allassets to the new system, in which case stresses arehigh, and converting the ratings over time, whichreduces the intensity of stress but also can reduce thereliability of internal rating information during thelonger transition. In one version of the slower transi-tion, which is especially common when a large bankacquires a much smaller bank, all of the acquiredbank’s performing loans are assigned to the riskiestnonwatch Pass grade. Each loan is then reassigned asappropriate at the time of its next review. Althoughsuch a practice may be viewed as conservative, itmasks the true risk posture of the bank during the


transition period. Regardless of the speed of transi-tion, loan review units are under substantial pressureduring and immediately after the transition.

Expense control has also been a focus of the bank-ing industry in recent years. The emphasis on econ-omy naturally puts pressure on the resources devotedto operating and maintaining the rating system, andespecially to reviews. Although reviews can be cur-tailed or eliminated in the short run without apparentdamage to rating system integrity, inadequate reviewactivity may lead to biased and inconsistent ratingsover the longer term. Another possible expense-reduction strategy is to rely more heavily on statisti-cal models in assigning ratings, reducing the degreeof judgment and, thus, the amount of labor requiredto produce each rating. The long-run success of such

a strategy depends on the adequacy of the models,including their ability to incorporate subjective fac-tors and their robustness over the business cycle. Ourimpression is that, at present, such adequacy isuncertain.

Summary Observations on Operating Design

The rating process has many interlinked elements,as illustrated in diagram 1. At almost all large banks,internal rating systems rely importantly on the judg-ment of staff operating with relatively little writtenguidance. The operational definition of each gradeis largely an element of credit culture that is deter-mined and communicated by informal means.

Diagram 1

Risk Rating Processes

Factorsconsideredin rating

Financial analysis

Industry analysis

Quality offinancial data

External ratings

Analyticaltools/models

Firm size/value

Management

Terms offacility/LIED

Otherconsiderations

Ratings criteria

Written/formalelements

Subjective/informalelements(cultural)

Rater’s ownexperience andjudgment

Preliminaryrating

proposedfor loanapprovalprocess

Relationshipmanagerand/or

credit staff

Approvalprocess

(perpolicy)assigns

finalrating

Riskrating

Quantitativeloss

characteristics

Portfolio monitoring

Loan loss reserve analysis

Loan/business line pricingand profitability analysis

Internal capital allocationand return on capitalanalysis

Generalcreditquality

characteristics

Assessing attractivenessof customer relationship

Evaluation of ratereffectiveness

Administrative and moni-toring requirements

Frequency of loan review

Line/credit review Watch processes Loan review

Ongoing review by initialrater

Periodic review of eachcustomer relationship

Aimed at reviewingprofitability/desirability as well ascondition

Generally conducted bysame authorities thatapprove loans

Quarterly process focused on loans that exhibit current or prospective problems only

Aimed at identifyingbest path to improve or exit credit at lowest cost

Conducted by sameauthorities to improveloans, although others may participate as well (e.g., workout group)

Review of adequacy ofunderwriting andmonitoring fromrandom sample

Sample weighted towardhigher-risk loans

Loan review judgment is‘‘final say’’

Negative consequences for initial rater if con-sistent disagreements

Assignment of ratings Use of ratings

Review processes


Review activities, especially those conducted by loanreview units, are crucial for maintaining the culturein that the feedback they give is critical to commonunderstanding and discipline. The credit culture canbe disturbed or unbalanced by changes in the incen-tives faced by the staff; such changes typically arisewhenever the rating system is required to supportadditional functions or uses. The systems of banks atwhich all ratings are assigned by credit staff arerelatively immune to such shocks, but the importantrole of middle-market loans in most banks’ portfoliosoften makes rating assignment by relationship man-agers cost-effective. In the latter case, the ratingsystem’s resilience to shocks depends to a consider-able extent on the loan review unit’s ability to detectand correct problems in a timely manner. Strongsupport of loan review by senior management andboards of directors appears to be quite important.

Points of external comparison, such as agencyratings or results of statistical models of borrowerdefault probability, can be helpful in maintaining theintegrity of internal ratings. A few banks are movingtoward models as the primary basis for internalratings. Such an operating design largely removes theproblems of culture maintenance and conflictingincentives that make management of judgmentalrating systems challenging. However, the ability ofmodels to produce sufficiently accurate ratings forthe broad range of assets on the typical large bank’sbalance sheet remains in question.

BANK SYSTEMSRELATIVE TORATING AGENCYSYSTEMS

Credit risk ratings have played an important role incapital markets for most of the twentieth century.Ratings of publicly issued bonds were first producedduring the early 1900s by predecessors of the currentrating agencies Moody’s and Standard & Poor’s. Inthe decades after 1920, other agencies, both domesticand foreign, were formed and commenced publica-tion of ratings. Today a variety of instruments arerated, such as commercial paper, bank certificates ofdeposit, commercial loans, and hybrid instruments.

Agency and bank rating systems differ substan-tially, mainly because rating agencies themselvesmake no investments and thus are not a party totransactions between borrowers and lenders. Theirrevenue comes from the sale of publications and fromfees paid by issuers of debt. Such fees can be substan-tial: S&P’s fee for rating a public corporate debt issueranges from $25,000 to more than $125,000, with theusual fee being 0.0325 percent of the face amountof the issue. Fees are a reflection of the substantial

resources the agencies typically devote to producingeach rating, especially the initial rating.

At banks, the costs of producing ratings must becovered by revenues on credit products. Thus,although a bank might expend resources at a ratesimilar to that of the rating agencies when underwrit-ing and rating very large loans, the expenditure of somuch labor for middle-market loans would make thebusiness unprofitable.

Agency ratings are used by a large number andvariety of parties for many different purposes. Toensure wide usage (and thus their ability to collectfees), the agencies strive to be deliberate, accurate,and evenhanded. They also produce relatively finedistinctions of risk on rating scales having architec-tures and meanings that are stable over time. Accu-racy and evenhandedness are crucial to the ratingagency business—for example, an agency suspectedof producing the most favorable ratings for those thatpay the highest fees would soon be out of business:Investors would cease paying attention to its ratings,and issuers would thus have no incentive to pay.

Similarly, changing the rating scale can confusethe public and at least temporarily degrade the valueof an agency’s product. The agencies also have incen-tives to be relatively open about their process and toproduce written explanations of each rating assign-ment or change. Clarity helps investors use the rat-ings and helps assure issuers that the process is asobjective as possible.

At banks, ratings are kept private, and the costsand benefits of rating systems are internal; hence,pressures for accuracy, consistency, and fine distinc-tions of risk are mainly a function of the ways inwhich ratings are used in managing the portfolio.Moreover, the rating system can be tailored to fit therequirements of the bank’s primary lines of businessand can be restructured whenever the internal bene-fits of doing so exceed the costs.

Agencies and banks both consider similar risk fac-tors, and both rely heavily on judgment and culturalelements rather than on detailed and mechanical guid-ance and procedures. However, the agencies publishsupplementary descriptions of rating criteria that aremuch more detailed than banks’ internal guidance,partly because agency ratings must be understood byoutsiders. In addition, the agencies track the financialcharacteristics of borrowers receiving their ratingsand publish both default histories for each grade andfinancial profiles of the ‘‘typical’’ borrower in eachgrade, thus providing additional referents to outsidersseeking to understand the meaning of their ratings.

Agencies have nothing comparable to a bank’sloan review unit. The rating culture at agencies is


maintained instead by a combination of market dis-cipline and a committee system. Market disciplinearises because the agencies stand between investorsand issuers, with the former typically preferring con-servative ratings and the latter preferring optimism.Thus, the agencies quickly hear from investors orissuers about any perceived tendency toward exces-sive optimism or pessimism. Although a singleagency analyst is primarily responsible for proposinga rating, committees make the final determinations.The membership of a committee changes from onerating action to the next so that agency staff mem-bers participate in many rating decisions and a cul-tural understanding of the meaning of each grade ismaintained.

BANKS’ ATTEMPTS TOMEASURELOSSCHARACTERISTICS BYGRADE

Consistent and accurate rating assignments and reli-able quantitative estimates of the risk associated witheach internal grade are useful in a bank’s efforts toanalyze risk posture, establish its appetite for risk,and evaluate the effectiveness of its risk rating cri-teria. At most banks, however, the primary demandsfor quantitative information about PD, LIED, and ELhave come from those involved in the loan lossreserve process and from credit modeling groups(those building and implementing quantitative mod-els of portfolio risk, capital allocation, profitability,and pricing). Internal ratings are key inputs into suchprocesses. Empirical analysis of loss characteristicsby grade appears to be an area where industry prac-tice is developing rapidly.

Problems in Evaluating the Accuracy andConsistency of Ratings

If internal ratings are to be accurate and consistent interms of the system’s loss concepts (that is, PD,LIED, or EL), different assets posing a similar levelof risk should receive the same grade. Such quantitiesare not observable ex ante, however, and thus ratingsystems rely on criteria that are thought to predictloss. Accuracy and consistency require that ratingcriteria be adjusted as necessary to ensure that expo-sures posing similar risk are grouped together (dia-gram 2 illustrates what is involved in the adjustmentprocess).

As a practical matter, alignment of the ex anterating criteria to achieve accuracy and consistency inthe economic meaning of each rating—that is, quan-

titative loss characteristics—is a difficult task. Twoproblems arise: How to ensure that criteria are cali-brated so that different assets of the same generaltype in the same grade have the same loss character-istics, and how to address diversity among assettypes. Within a narrowly defined asset class, such asloans to large commercial firms in the same industry,comparisons across firms are relatively manageable,so the main problem is defining the boundaries ofrating categories and inferring the default or lossrates for each category. That by itself is not easy, butthe problem becomes much more difficult when verydifferent types of assets must be compared. For exam-ple, how would a loan to a well-established com-mercial real estate developer, featuring a 70 percentloan-to-value ratio, compare with a term loan to afirm in a relatively stable manufacturing industrywith a current debt to equity ratio of 1:1 and aninterest coverage ratio of 3?

Because the rating criteria differ so greatly fordifferent asset classes, some information about therelationship of borrower and asset characteristics tohistorical loss experience would appear to be nec-essary. Especially with loss experience data coveringa fairly long period of time, say a couple of creditcycles, it would be possible to make at least roughinferences about relative risks across asset classes.

Unfortunately, to the best of our knowledge, few ifany banks have available the necessary data, espe-cially for a variety of asset classes. At a minimum,information on the performance of individual loansand their rating histories is required. Because ratingcriteria have changed over time at most large institu-tions, information about borrower and loan character-istics is also required, so that the risk implications ofdifferent rating criteria can be assessed.

Historically, banks have retained performance databy loan type (for example, data provided on CallReports) or by line of business in the aggregate, butnot by risk grade. Because of mergers, even at banksthat have tracked performance by grade, data may notcover the whole of the current institution but ratheronly one predecessor institution. Mergers often causeupheaval not only in rating processes but also in datasystems and, in particular, contribute to the loss orobsolescence of historical data.

Although data collection is costly, many largebanks have recognized its importance and have begunprojects to build databases of loan characteristics andloss experience. However, the costs of extractingfrom archival files historical data on the performanceof individual loans appear to be prohibitively high.Thus, those banks that are collecting data indicatedthat they are several years away from having data


sufficient to support empirical analyses of their ownportfolios that are comparable to the studies beingdone for publicly issued bonds.29

In the absence of data, our impression is that thetraditional means of tuning both rating criteria andunderwriting standards relies heavily on the judg-ment and experience of the senior credit staff withlong tenure at their institution. Over a period encom-passing multiple credit cycles, these staff membersaccumulate an individual and collective memory of

the credit problems experienced by the institution andof the implications for risk of various borrower andloan characteristics. Such experience is likely suffi-cient to support meaningful tuning of rating systemsthat have small numbers of Pass grades (each cover-ing a broad band of risk) and that are used to ratetraditional banking assets. The precision with whichsystems involving a large number of Pass grades canbe tuned by experience alone is not clear.

Mapping to Agency Grades as a PartialSolution

Because little information is available internally,many banks have estimated the quantitative loss char-acteristics of their ratings by using the extensive dataavailable on the loss performance of publicly issuedbonds. As noted, rating agencies and others fre-quently publish studies covering many years of bonddefault and loss experience by grade, and publiclyavailable databases of bond issuer characteristicsmake it possible to relate loss experience to potential

29. The situation is somewhat better with respect to loss in theevent of default (LIED) in that historical studies require informationonly on the bad assets. Often their number is small enough thatgathering data from paper files is feasible, and thus many banks arebeginning to accumulate LIED information from their own portfolioexperience. A few publicly available studies have also appeared.Estimating PD and EL requires much more data in that information onboth performing and nonperforming assets are required. Studies withLIED statistics include Lea V. Carty and Dana Lieberman,SpecialReport: Defaulted Bank Loan Recoveries(Moody’s Investors Service,1996); Elliot Asarnow and David Edwards, ‘‘Measuring Loss onDefaulted Bank Loans: A 24-Year Study,’’Journal of CommercialLending,vol. 77 (March 1995), pp. 11–23; and Society of Actuaries,1986–92 Credit Risk Loss Experience Study: Private Placement Bonds(Society of Actuaries, Schaumberg, Ill., 1996).

Diagram 2

Tuning the Rating Criteria

Factors, definitions,and weighting

Written/formalelements

Subjective/informalelements

Rater’s experienceand judgment

Asset type

Quantitative losscharacteristics

Probability ofdefault (PD)

Loss in event ofdefault (LIED)

Expected loss (EL)

Distribution of lossexperience

Criteria chosento obtain desiredcharacteristics

When available,loss experience analyzed

to evaluate ratingeffectiveness

Promote accuracyand consistency

Controls and validation processes

Incentives and training

Documentation and approval requirements

Validation processes, especially loan review


rating criteria. Indeed, S&P occasionally publishestables of indicative or average financial ratio valuesby grade (while noting that many other factors enterinto its rating decisions).

To use data on bond loss experience, a bank mustdevelop or assume some correspondence betweenagency ratings and its own internal grades. Inter-

views suggest that the basis of such mappings isthreefold: (1) The internal grades assigned to borrow-ers who have also issued publicly rated bonds;(2) analysis of the ‘‘typical’’ financial characteristicsof bank borrowers in each internal grade vis-a`-vis thecharacteristics of the firms with bonds in each agencygrade; and (3) subjective analysis.

Mappings and the Problem of Different Architectures

Both banks and rating agencies assign ratings based oncriteria that are predictive of a borrower’s probability ofdefault (PD) or a loan’s expected loss (EL). However,because no mechanical formula exists that converts criteriainto values of PD or EL for each grade, such values must beobtained from historical loss experience. As noted, banksrarely have databases of such experience, but the majorrating agencies do. A mapping of internal grades to agencygrades permits a bank to use statistics from the agencies’bond default studies to assign values of PD to each of itsinternal grades.

For simplicity, we focus here only on PD. Four problemscan cause a mapping to lead to a materially inaccurateestimate of PD for internal grades:

(1) A bank’s rating system may place loans with widelyvarying levels of PD into the same grade and similar levelsof PD into different grades. In this case, grades bear littlerelation to PD values and thus mapping will not providegood estimates of PD.

(2) Default rates on publicly issued bonds may differsystematically from loan default rates.

(3) The mapping exercise may simply associate thewrong agency grades with internal grades.

(4) The implications of differences between banks’point-in-time and agencies’ through-the-cycle rating phi-losophies may not be taken into account.

Even when the first three problems do not apply, thefourth, which is a characteristic of the most common map-ping approach, can produce materially biased estimates ofPD for internal grades. Such bias can confuse attempts totune rating criteria and can seriously distort internal analy-sis of business line profitability, loan loss reserves, andcapital allocation.

Bias arises in the most common approach to mappingbecause bank internal ratings change as the borrower’scondition changes, whereas the PD associated with eachinternal grade is stable. In contrast, agency ratings tend tostay the same, while default probabilities for each ratingvary with the economic cycle. Thus, mapping exercisesshould take into account the current point of the economiccycle and should draw default rates from the agencies’historical studies for similar points in prior cycles.

The fourth problem is illustrated here with an example:Suppose that a hypothetical large bank, BigBank, has an

I. BigBank’s Pass rating scale

Grade

True PD for rating system,but precise values not

known to bank(percent)

1—Virtually no risk . . . . . . . . . . . . . . . . 02—Low risk . . . . . . . . . . . . . . . . . . . . . . . .103—Moderate risk. . . . . . . . . . . . . . . . . . .254—Average risk . . . . . . . . . . . . . . . . . . . 1.005—Acceptable risk. . . . . . . . . . . . . . . . . 2.006—Borderline risk . . . . . . . . . . . . . . . . . 5.00

internal rating system with six Pass grades, and suppose ithas two hypothetical borrowers, OK Corp. and Less-OKCorp. To focus on the point-in-time vs. through-the-cycleissue, suppose we know that BigBank’s rating criteria andrating system will always group borrowers with similarvalues of PD into the same grade, that the ‘‘true’’ PD foreach grade is as shown in table I, and that BigBank doesnot know the values of PD associated with its grades.Similarly, as shown in the top section of table II, the truePD for OK Corp. is 1 percent in upturns and 2 percent indownturns, whereas Less-OK Corp.’s true PD is 3 percentduring upturns and 6 percent in downturns. However,because neither BigBank nor the rating agencies knowthese true PD values, they rate on the basis of observableborrower characteristics.

Having no data on its historical loss experience, Big-Bank maps its internal grades to agency grades simply byidentifying the agency ratings assigned to those borrowerswith such ratings in each internal grade. BigBank then usesthe correspondinglong-term historical average one-yeardefault rate identified in agency default studies as anestimate of theexpected one-year default ratefor all loansin each internal grade.

II. Borrowers used for mapping, and their characteristics

CharacteristicBorrower

OK Corp. Less-OK Corp.

PD in upturns. . . . . . . . . . . . . . . . . . 1percent 3 percentPD in downturns. . . . . . . . . . . . . . . 2percent 6 percent

BigBank rating in upturns. . . . . . 4–Average risk 5–Acceptable riskBigBank rating in downturns . . . 5–Acceptable risk 6–Borderline risk

Agency ratings (stable throughcycle) . . . . . . . . . . . . . . . . . . . . BB or Ba B+ or B1


When mapping is done by comparing the inter-nally assigned grades of publicly rated borrowerswith ratings assigned by agencies, the danger ofcircularity arises. In most cases, agency grades area rating criterion, and even when agency grades arenot written into rating definitions, assigners of inter-

nal ratings always know the agency grade for a givenborrower and have an idea of the borrower’s likelyposition on the internal scale. Obviously, if theagency rating is the sole criterion used in assigninginternal grades to agency-rated borrowers, rated andunrated borrowers within a given internal grade might

Mappings and the Problem of Different Architectures—Continued

Because it rates on a point-in-time basis, BigBank doesnot allow the PD values for each grade to vary through theeconomic cycle; loans whose one-year PDs increase incyclical downturns are downgraded to a riskier internalgrade. As shown in the middle section of table II, BigBankassigns ratings that are appropriate for varying risk: It ratesOK Corp. a 4 inupturns and a 5 indownturns, and it ratesLess-OK Corp. one grade worse—a 5 in upturns and a 6 indownturns. The rating agencies are similarly accurate intheir assessment of risk (bottom section of table II), butbecause they rate through the cycle (that is, according to theborrower’s condition when under stress), they rate OKCorp. as BB/Ba and Less-OK Corp. as B/B in both upturnsand downturns.

Suppose that BigBank conducts its mapping exerciseduring an upturn. As shown in the top section of table III, itwill assume that its grade 5 is equivalent to the agencies’B grades because Less-OK Corp. is in relatively good shapeduring upturns and achieves a point-in-time internal ratingof 5 even though its through-the-cycle agency grade is B.BigBank should infer the PD for grade 5 from the averagedefault frequency of B-rated public bondsonly in upturns,which is the good-year average of 4 percent (table III); butif it follows common practice it will use theoverall averagedefault frequency of B-rated bonds, which is 5.5 percent.

Next, suppose BigBank conducts its mapping exerciseduring a downturn. As shown in the bottom section of tableIII, it will assume that its grade 5 is equivalent to BB/Babecause OK Corp. will be rated 5 (Less-OK Corp. is down-graded to 6 during downturns). BigBank should infer thePD for grade 5 from thebad-yearaverage PD of BB/Barated bonds (2 percent), but instead it uses theoverallaverage of 1 percent.

III. BigBank mapping and PD estimation exercise basedon borrower ratings

Period ofmapping

Internalgrade

Equivalentagencygrade

Average one-year PDfor bonds

Overall Good year Bad year

Upturn . . . . . . 4 BB/Ba 1.00 .75 2.005 B/B 5.50 4.00 6.506 . . . . . . . . . . . .

Downturn . . . 4 . . . . . . . . . . . .5 BB/Ba 1.00 .75 2.006 B/B 5.50 4.00 6.50

. . . Not applicable.

In this example, BigBank’s and the agencies’ ratingsystems both do an excellent job of assigning ratings thatare consistent with the borrower’s true PD, but mappingwithout regard to the difference between point-in-time vs.through-the-cycle rating causes BigBank to badly mis-estimate the PD.Using the most common mapping prac-tices, BigBank might estimate the PD of its grade 5 at1 percent to 5.5 percent, whereas the true PD of grade 5 is2 percent.If the mapping is done simplistically, as in thisexample, and during an upturn, BigBank likely overesti-mates the PD, whereas during a downturn it likely underes-timates the true value. If BigBank had used average defaultfrequencies from the agencies’ studies that were appro-priate to the point in the cycle at which the mapping wasdone, it might still have obtained inaccurate estimates, butthey would have been closer to the truth. BigBank mightstill have been somewhat uncertain about whether to con-sider category 5 as equivalent to BB/Ba or B, but anysuch equivalence can never be exact because BigBank’sscale and the agency scales have different conceptualfoundations.

We consider the numbers in the example to be fairlyrealistic and thus the mis-mapping problem at most banksto be potentially serious. The problem of mis-estimatedPDs is much more important at the higher-risk end ofrating scales. Precision is especially important at that endbecause differences in reserve and capital allocations canbe large, whereas dollar differences in allocations acrossdifferent classes of low-risk assets are typically small. Inaddition, default studies and other analyses tend to showthat variations in one-year default rates on investment-grade assets tend to be driven by idiosyncratic factorsrather than the credit cycle.

Mapping processes are further complicated if, over time,a borrower’s agency rating is allowed to be the dominantcriterion in assigning an internal grade. In general, such apractice would tend to reduce the likelihood that a loanwould be appropriately downgraded during a recession—the borrower’s agency rating would not change unless itsperformance or prospects deteriorated more than antici-pated in the agency’s through-the-cycle risk analysis. Thisprocedure could effectively turn BigBank’s ratings intothrough-the-cycle rather than point-in-time, putting lossestimates potentially out of line with management analysesthat assume point-in-time grading.


differ substantially in risk. In such circumstances themapping is circular because borrowers are assignedto internal grades based on the agency rating, and theagency rating corresponding to each internal gradeis inferred only from such rating assignments. Thebanks we interviewed maintain that agency ratingsare used only as a starting point in their rating pro-cesses, not as the sole criterion.30

Mapping and the Problems Caused byInconsistent Architectures

Because major agencies rate borrowers with theexpectation that the rating will be stable throughnormal economic and industry cycles, only thoseborrowers that perform much worse than expectedduring a cyclical downturn will be downgraded (will‘‘migrate’’ to riskier grades). In contrast, rating sys-tems that focus on the borrower’s current condition(virtually all bank systems) are likely to feature muchmore migration as cycles progress but, in principle,should exhibit somewhat less cyclical variation indefault rates for each individual grade.

Though apparently subtle, this difference in archi-tectures has important implications for mapping exer-cises and the inference of PD values for internalgrades. Both the point in the economic cycle at whichthe mapping exercise is done and the exact nature ofthe PD statistics drawn from the agencies’ studies oflong-term default history can have a dramatic effecton the mapping (see box ‘‘Mappings and the Problemof Different Architectures’’). Values of PD attributedto internal grades can differ by several percentagepoints depending on how the mapping is done. PDsare most likely to be badly estimated for the higher-risk Pass grades, but precision is also especiallyimportant for such grades in that allocated reservesand capital are most sensitive to assumptions aboutriskier assets.

Obtaining reasonably accurate mappings is mainlya matter of paying attention to the stage of the cycleat which the mapping is being done and of usinghistorical average PD values from either good-experience or bad-experience years as appropriate.However, interviews left us with the impression thatfew banks carefully consider cyclical issues whenmapping their internal grades to agency grades.

AN AGGREGATEBANK RISK PROFILE

Mapping between internal and agency grades facili-tates a bank’s quantitative loss analysis and the inte-gration of publicly available information into ratingdecisions. Such mappings also make possible an esti-mate of the risk profile of the internally rated portionof bank loan portfolios on a standardized scale. Infor-mation about the risk profile of bank credit helps putmany rating system issues in perspective.

As part of the analysis leading to this article, wereviewed internal reports showing distributions ofrated assets across internal grades for the fifty largestconsolidated domestic bank holding companies. Inaddition, we obtained mappings of internal grades toagency equivalents from twenty-six of them. Themappings allow us to allocate internally rated bal-ances to grades on a rating agency scale. To ourknowledge, this is the first time that such a character-ization of the overall risk profile of a large portion ofthe banking industry’s commercial loan portfolio hasbeen possible.

The twenty-six banks accounted for more than75 percent of aggregate banking industry assets atyear-end 1997. Rated loans outstanding at individuallarge banks usually represent 50 percent to 60 percentof their total loans.31

In general, we cannot judge whether the mappingsprovided by banks are correct. Inaccuracy can arisefrom errors or inconsistency in assigning the internalratings themselves, problems of cyclicality or cir-cularity in the mapping process, inconsistenciesbetween large corporate and middle market lines ofbusiness, or other difficulties. In addition, mappingsat some institutions are more precise in form than atother institutions in that they distinguish amongmodified agency grades, such as BB and BB+. Still,such mappings are an element of banks’ day-to-dayoperating procedures and analysis, which suggeststhat the twenty-six banks have endeavored to makethem as accurate as possible within the constraints oftheir rating systems. It thus appears that aggregationand comparison of these mapped balances representsa reasonable—albeit crude and broad—first approxi-mation of the actual risks in banks’ portfolios.

Chart 3 displays the aggregate weighted-averagedistribution of internally rated outstanding loans atyear-end 1997 for the twenty-six consolidated bankholding companies. About half of aggregate ratedloans pose below-investment-grade risks (were ratedthe equivalent of BB+/Ba1 or riskier), and about65 percent of outstandings were concentrated around30. Even when circularity is avoided, heavy use of bond experi-

ence data in defining criteria for each grade might lead to exclusion ofcriteria needed to capture the risk of unrated borrowers, such asmiddle-market firms. 31. Total loans includes consumer loans, which are rarely rated.


the boundary between investment and below-investment grades (rated BBB or BB).

Banks’ loan loss experience during 1997 is consis-tent with the credit quality distribution shown inchart 3. Using the 1997 default frequencies for eachgrade drawn from S&P’s latest annual study and anassumption that the average LIED for loans is about30 percent, an aggregate portfolio with the qualitydistribution for the twenty-six banks would beexpected to have an annual credit loss rate of roughly0.20 percent. Although this rate is roughly equal tothe actual loan loss experience of the banking indus-try’s aggregate commercial loan portfolio during1997 (0.21 percent), this simple exercise shouldnot be taken as proof that the distribution in chart 3is representative; nonetheless, the results aresupportive.32

Chart 4 displays the percentages of internally ratedassets that are below investment grade as of year-end 1997 for twenty-six banks in three peer group-ings: major loan syndication agents; smaller banks(less than $25 billion in total assets at year-end1997); and the remainder of the twenty-six, labeled‘‘regionals’’ (many other peer groupings are possible,of course). The three peer groups display systematicdifferences in risk posture. On average, the majoragents have 45 percent of rated assets in categoriescorresponding to BB and riskier, compared with

about 60 percent for regionals and 75 percent forsmaller banks.33

USES OFINTERNALRISK GRADES

Banks use internal ratings in two broad categories ofactivity: analysis and reporting, and administration.Analytic uses include reporting of risk postures tosenior management and the board of directors; loanloss reserving; and economic capital allocation, prof-itability measurement, product pricing, and (indi-rectly) employee compensation. Administrative usesinclude loan monitoring, regulatory compliance, andcredit culture maintenance. In addition, external enti-ties such as investors or regulators may become moresignificant users of internal ratings information. Dif-ferent uses place different stresses on the rating sys-tem and may have different implications for the inter-nal controls needed to maintain the system’s integrity(diagram 1 shows such uses).

Portfolio Reporting

Virtually all large banks report total asset balancesin each of the regulatory problem-asset grades to

32. Actual loss experience is measured as the average annualizednet charge-off rate for bank loans in the commercial and industrial,commercial mortgage, and agricultural loan categories as reportedon the quarterly Report of Condition—or Call Report—filed by allbanks.

33. That the fraction of loans posing below-investment-grade risksis much larger at some institutions than at others does not imply apriori that such institutions are operating in an unsafe or unsoundfashion. In general, provided a bank is aware of its risk posture, hasadequate processes to manage risk, is pricing loans to reflect the risk,and has reserves and capital that are adequate to the risks, a portfoliowith a large fraction of below-investment-grade exposures can besafe, sound, and profitable.

3. Percentage of aggregate internally rated outstandingsplaced in each agency rating category at banksmapping to agency scale, year-end 1997

10

20

30

40

Agency rating category

Percent

AAA/aaa AA/aa A BBB/Baa BB/Ba B Cs D

Note. The banks are twenty-six of the fifty largest.

4. Percentage of aggregate internally rated outstandingsbelow investment grade at banks mapping to agencyscale, by bank group, year-end 1997

Major agents Regionals Smaller banks1

25

50

75

Percent

Note. The banks are twenty-six of the fifty largest.1. Less than $25 billion in total assets. Regionals are those that are not major

syndication agents or smaller banks.


senior management and the board of directors. About80 percent also internally report balances in each oftheir Pass grades. In the latter case, such reportsappear to be used either by management or the creditstaff as a means of detecting changes in portfolio mixand are only infrequently shown to boards of direc-tors.34 Balances in the regulatory grades give a senseof the share of bank assets that are troubled, whereasa profile of balances in Pass grades can provide aforward-looking sense of trends in the bank’s riskposture so long as Pass grade assignments meaning-fully distinguish risks; internal reports are much lessinformative when a large share of rated assets fallsinto only one or two Pass grades.

Reserving

Although many accounting and regulatory policiesinfluence the setting of loan loss reserves and provi-sions, balances in the regulatory grades are integralto reserve analysis at all banks. Supervisors requirea specific reserve of at least 50 percent of Doubtfulloans plus 20 percent of Substandard loans; banks setthe amount of additional reserves for OAEM andPass loans according to their judgment, subject toevaluation by examiners.35 Many banks develop re-serve factors specific to each Pass category. Accord-ing to accounting and regulatory standards, loan lossreserves are to cover losses already ‘‘embedded in theportfolio,’’ and the generally accepted interpretationis that reserves for Pass loans should cover expectedlosses over a period of one year. Thus, if an institu-tion can identify a reasonable estimate of expectedloss for each Pass grade, a reserve analysis sensitiveto balances in the different Pass grades provides agood estimate of embedded losses.

A significant number of the banks we intervieweddo not differentiate among the Pass grades in per-forming reserve analysis. In such cases, a singleexpected-loss (EL) weight is applied to balances inall Pass grades. Such a simplification is least costly interms of accuracy of the reserve analysis when loan

balances are concentrated in a single category orwhen the composition of the Pass portfolio by riskgrade is very stable.

Profitability Analysis, Pricing Guidelines,and Compensation

All banks we interviewed conduct internal profitabil-ity analyses (of different business lines, for example).Some banks do not use internal ratings at all in suchanalyses, whereas others include a rating-sensitiveexpected-loss cost but no rating-sensitive capital cost.The most sophisticated analyses involve bothexpected-loss costs and costs of allocated capital thatvary by internal rating. The higher such costs, thelower the measured profitability of a business unit orindividual transaction. The use of rating-sensitiveprofitability analysis thus has significant implica-tions for the design and operation of internal ratingsystems.

To implement rating-sensitive profitability analy-sis, the bank must estimate expected losses for assetsin each grade as well as the amount of economiccapital to allocate (if it allocates capital). Economiccapital for the bank as a whole is that needed tomaintain the bank’s solvency in the face of unexpect-edly large losses. The process of estimating the addi-tional economic capital needed as a result of bookingany given loan is complex, but as a practical matter,the loan’s internal rating is a primary (if not the sole)day-to-day determinant of the capital allocationsimposed by risk-sensitive profitability models.36

The measured profitability of business units isan important factor in management decisions aboutwhich units should grow or shrink. When risk-sensitive profitability is appraised at the level of theindividual loan or relationship, unprofitable loans arenot made and unprofitable relationships are even-tually dropped. At a growing number of banks,employee compensation is formally tied to profitabil-ity measured by such systems.

34. At some banks, portfolio composition is reported as a weighted-average risk grade. Such averages weight the balances by the grade’snumeric designator. For example, assets in grade 4 are treated as beingtwice as risky as assets in grade 2. This can produce misleadingaverages because risk—whether PD or EL—tends to increase morethan linearly with grade (table 2). At those banks we interviewed thatused this measure, the staff seemed to understand that it does notreflect portfolio risk—it can indicate only whether the mix haschanged.

35. Federal Financial Institutions Examination Council,Inter-agency Policy Statement on the Allowance for Loan and Lease Losses(December 1993).

36. Mechanically, one can think of economic capital for the creditrisk of a whole portfolio as that amount necessary to cover (forexample) 99.9 percent of the possible portfolio loss rates. Capitalrequired to support a given loan can be viewed as that increment tototal bank capital that will keep the bank insolvency probabilityconstant if the given loan is added to the portfolio. Conceptually, totalcapital can be split into expected and unexpected loss portions. In anaccounting sense, the loan loss reserve might be viewed as coveringthe expected loss and equity as covering the unexpected loss. Formore details, see ‘‘Credit Risk Models and Major U.S. BankingInstitutions: Current State of the Art and Implications for Assessmentsof Capital Adequacy,’’Federal Reserve System Task Force on InternalCredit Risk Models(Board of Governors of the Federal ReserveSystem, May 1998).


Interviews indicated clearly that the introductionof risk-sensitive profitability analysis puts signifi-cant new pressures on the risk grading system. Pres-sure to rate loans favorably arises because expectedlosses and capital allocations are lower for lower-risk loans. Some institutions found that many loanswere upgraded shortly after the introduction ofprofitability analysis, although the overall degreeof the shift was small. One institution specificallymentioned an upward bias of about one-halfgrade relative to previous rating practice. Manynoted that the number of disagreements in whichrelationship managers pressed for more favorableratings increased once such systems were put intoplace.

In addition to pressure for more favorable ratings,rating-sensitive profitability analysis also createspressure to increase the number of rating categories.This pressure, which comes both from the businessline staff and the profitability analysis unit itself,arises because some of the loans in any given gradeare less risky than other loans in that grade and thusshould bear smaller credit costs. Creation of moregrades allows for better recognition of such riskdifferences. Institutions reported that the pressure toincrease the number of grades has become morepronounced in recent years as competitive forceshave compressed loan spreads; in this setting, reduc-ing expected loss factors by a few basis points, orslightly reducing the amount of capital allocated tothe loan, may be the difference between a transactionthat meets internal profitability ‘‘hurdles’’ and onethat does not.

These stresses place increased pressure on the loanreview unit to maintain discipline and enforce con-sistency, stability, and accuracy. Controlling ratingbiases is always a challenge. As the number of gradeson the scale increases and the distinctions of riskbecome finer, disagreements about ratings naturallyarise more frequently, and the control of biasesbecomes even more difficult. The difficulty seemslikely to be greatest just after the number of grades isincreased because the loan review staff must enforce(and if necessary, develop) new cultural definitionsfor the grades. The latter task is somewhat easier atbanks that use external referents in assigning orreviewing ratings, such as default probability modelsand agency ratings of borrowers; such referents giveloan reviewers objective benchmarks to use in identi-fying problems and communicating with staff. Rede-signs of the rating scale that split existing gradesinto smaller compartments are also easier to imple-ment because the existing cultural definitions can berefined rather than replaced.

Risk-sensitive profitability analysis also increasesthe demand for internal data on loss experience andfor mappings to external referents because the analy-sis demands relatively precise quantification of therisk characteristics of each grade. However, suchanalysis can also make existing data and mappingsless useful, at least in the short run, because ratingpressures or changes in architecture may, to someextent, change the effective meaning of grades.

Using Ratings to Trigger AdministrativeActions

As noted, many banks include an internal watchgrade on their scales in addition to the regulatoryproblem-asset grades (formally, the watch gradewould be counted among the Pass grades). Reassign-ment of a loan to watch or regulatory grades typicallytriggers a process of quarterly (or even monthly)reporting and formal reviews of the loan. At institu-tions where the main use of ratings is for monitoringand regulatory reporting, RMs’ incentives are oftenthe opposite of those introduced by rating-sensitiveprofitability analysis: Their main interest is to avoidgetting caught assigning ratings that arenot riskyenough. Getting caught can have negative careerimplications, and thus RMs have an incentive toassign credits to the riskiest Pass grade that is notwatch. For example, some banks are especially likelyto penalize RMs when a loan review reassigns a Passcredit from one of the less risky grades into a regula-tory grade. Penalties can be forthcoming even when aloan is reassigned from a less risky Pass grade intowatch, but are likely to be less severe. Thus, in theabsence of carefully designed controls, the presenceof administrative grades in a rating system can reducethe accuracy of non-administrative Pass grade assign-ments. This sort of bias is less likely at the largestbanks because the countervailing incentives of rating-sensitive profitability analysis are most likely to oper-ate there.

However, incentives associated with rating-sensitive profitability analysis can reduce the effec-tiveness of administrative management of problemloans. The staff may delay assigning credits to watchor regulatory grades because of the negative implica-tions for measured profitability. Thus, there is a cer-tain tension in the simultaneous use of rating systemsfor administrative purposes and for profitabilityanalysis. Such tension can be overcome with properoversight, the implementation of which representsanother burden on loan review functions.


Potential Uses of Internal Ratings by ExternalEntities

Internal ratings are a potential source of informationfor bank investors and regulators. For example, dis-closure of the profile of a bank’s loans across itsinternal rating categories might enhance the ability ofshareholders and analysts to assess bank risk.

Moreover, investors in securitizations of traditionalcommercial loans might benefit from informationabout the credit quality of the underlying assets.Some banks are reportedly considering using internalrating information in structuring such securitizations.For example, when loans in the securitized poolare paid off, the new loans replacing them may berequired to be drawn from a particular internal grade.Obviously, to evaluate the attractiveness of the pool,investors (or rating agencies) must be able to under-stand the loss characteristics of each internal gradeand must have confidence that such characteristicswill remain stable over time. Thus, external valida-tion of rating systems becomes necessary if internalratings are to be used in securitizations. Such valida-tion would appear to be quite difficult because eachbank’s rating scale is different, and the meaning ofratings is largely embedded in culture rather than inwriting. Moreover, most banks do not have sufficienthistorical data on loss experience by internal grade tosupport objective measurements.

Internal ratings might also be used in bank super-vision and regulation. As a banking supervisor, theFederal Reserve has long emphasized the impor-tance of strong risk management practices at banksand has stated its desire to orient its activities moretoward testing of risk management and control pro-cesses and somewhat away from testing of individualtransactions. This preference allows for less intru-sion into the operation of the bank and minimizesthe restrictive effect of supervision on bankinginnovation.

Information on a bank’s risk profile by internalgrade and shifts in that profile over time couldbecome a useful supervisory tool. Supervisors coulduse internal profile information as one considerationin evaluating the asset quality and credit risk manage-ment of large banks, probably on balance reducingthe overall burden of supervision. For those institu-tions that map their internal ratings to external refer-ence points, such as the S&P scale, supervisors coulduse the mapping to put large institutions roughly on acommon scale (in a fashion similar to that shown inchart 3). While bearing in mind that this technique isvery crude, analysis of risk profiles and of trends inprofiles could provide valuable insights into credit

conditions and standards in the industry as well as atindividual institutions. Continuing work by indi-vidual institutions to better understand the loss char-acteristics of loans in their own risk grades will beimportant to refining and interpreting such compari-sons over time.

Internal risk grades could also become an explicitelement in the evaluation of capital adequacy. Thecurrent risk-based capital regime (based on the 1988Basle Accord) provides for lower capital weights oncertain low-risk assets (for example, those that aregovernment-issued or guaranteed) but applies thesame capital requirement (that is, 8 percent) to essen-tially all loans to private borrowers regardless of theunderlying risk. Internal risk grades might becomeone consideration in scaling capital requirements onbusiness loans more closely to the loss characteristicsof a bank’s loan portfolio.

Greater supervisory reliance on internal credit riskratings would require that supervisors be confident ofthe rigor and integrity of internal rating systems.Heretofore, examiners have sought to validate assign-ments to internal grades only as they relate to theregulatory problem-asset grades. If supervisors are torely more heavily on Pass grade information, somedegree of validation and testing would have to beextended to those grades as well.

External use of internal ratings would introducenew stresses on internal rating systems. In somerespects, the stresses would parallel those associatedwith rating-sensitive profitability analysis. That is,incentives would arise to grade optimistically and toalter the rating system to produce more fine-graineddistinctions of risk. However, new incentive conflictswould arise between outsiders on the one side and thebank as a whole on the other. Such new conflictscould overwhelm the checks and balances currentlyprovided by internal review functions. Even in theabsence of such incentive conflicts, external usersmight demand a greater degree of accuracy or consis-tency in rating assignments than that required inter-nally. For both reasons, external reviews and vali-dation of the rating system might be necessary. Inaddition, banks and external parties should both beaware that the additional stress imposed by externaluses, if not properly controlled, could impair theeffectiveness of internal rating systems as a tool formanaging the bank’s credit risk.37

37. In the early 1990s, the National Association of InsuranceCommissioners (NAIC) introduced a system of risk-based capitalrequirements for insurance companies in which requirements varywith the ratings of assets. Although such ratings are assigned by theNAIC’s Securities Valuation Office (SVO), the SVO does take intoaccount any ratings of assets published by major rating agencies. In


CONCLUDING COMMENTS

A bank’s decisions about its internal rating systemcan have a material effect on its ability to managecredit risk. But development of internal rating systemarchitectures and operating designs that are appropri-ate to the uses made of the ratings is an especiallycomplex task. The central role of human judgment inthe rating process and the variety of possible uses forratings mean that internal incentives can influencerating decisions. Thus, careful design of controls andinternal review procedures is a crucial considerationin aligning form with function.

No single internal rating system is best for allbanks. Banks’ systems vary widely largely becauseof differences inbusiness mixand in theuses towhich ratings are put. Among variations in businessmix, the share of large-corporate loans in a bank’sportfolio has the largest implications for its internalrating system. Banks with a substantial large cor-porate market presence are likely to benefit from arating system that achieves fine distinctions amongrelatively low-risk credits, while other banks mayfind significantly less value in such distinctions. Inaddition, an independent credit staff is often solelyresponsible for rating large loans. Such an arrange-ment can greatly reduce potential incentive conflicts,but may involve per-loan costs that are too large to beeconomic for smaller loans, which are often rated byrelationship managers. Smaller loans also pose lessrisk to bank earnings and capital, and thus gradingerrors and biases may be more tolerable.

Among the various uses of internal ratings, profit-ability analysis and product pricing models have themost significant implications for the rating system.At banks where such analysis is in place, ratings canhave a material effect on the measured profitability oftransactions and relationships and can directly orindirectly influence the compensation of bank staff.Thus, careful attention to review and control pro-cedures that limit biases in ratings is important tothe accuracy and consistency of internal ratings.

Profitability analysis also introduces pressures forrating systems with more risk grades. Relationshipmanagers may press for such systems because of adesire to subdivide grades that cover broad rangesof risk, thereby allowing different expected loss andcapital charges for exposures at different ends of theranges. The groups that develop and maintain the

profitability analysis systems may also press for fine-grained distinctions in order to support better balanc-ing of risk and return. However, internal rating sys-tems with many grades may make review and controlof grading both more difficult and more expensivebecause reasonable people are more likely to differ intheir subjective judgments when differences betweengrades are small rather than large.

Our interviews indicate that certain practices canimprove the quality of any internal rating system andare especially helpful to rating systems that supportanalytical functions such as profitability analysis andportfolio management. First, a bank with appropriatedata describing its historical loss experience by inter-nal grade and by different risk factors is better able toassess the predictive power of its ratings criteria andto estimate values of parameters needed for its analy-ses (such as grade-specific values of PD or EL).Second, assigning or reviewing ratings with the aidof agency ratings, statistical models of default prob-ability, or other objective criteria helps limit themagnitude of rating biases. However, care must beused in mapping internal grades to external grades orother indicators to ensure that the desired results areachieved. Finally, internal ratings grounded in clearloss concepts are helpful in grade assignment andreview because rating criteria can be clearly linked todifferent aspects of risk. For example, a system thathas separate grades for default probability and loss inevent of default can incorporate different effects for awide variety of types of collateral. All three of thesepractices are likely to be helpful in refining the sub-jective judgments that are central to almost all ratingsystems.

By their nature, banks’ credit cultures typicallyadapt slowly to changes in conditions. The rapid paceof change in risk management practice and thetrend toward risk-sensitive profitability analysis hasrecently increased the stresses on credit culturesin general and internal rating systems in particular.Careful attention to the many considerations noted inthis article can help accelerate the process of adjust-ment and thus the easing of stresses.

The use of internal ratings by external entities suchas regulators and investors has the potential to intro-duce new stresses in which incentives conflicts thatpit banks’ interests against those of the external enti-ties compound existing internal tensions. Use ofinternal ratings by entities outside the bank wouldprobably require some external validation of theratings and the systems that generate them. In ourview, such validation is probably feasible, but carefuldevelopment of a new body of practice will berequired.

the wake of this and other developments in the insurance industry, therating agencies experienced substantial pressure from both issuers andinvestors (insurance companies) to assign favorable ratings to someassets, a new and difficult development for the agencies in that issuersand investors had traditionally applied opposing pressures.


Date post:	08-Feb-2018
Category:	Documents
Upload:	lydieu
View:	215 times
Download:	1 times

Credit Risk Rating at Large U.S. Banks - United States dollar · PDF fileCredit Risk Rating at...

Documents