Sunday, 30 March 2003 Try to load this version of the website..... blogger version Sunday, 23 March 2003 As if nothing had happended, I'm back.  Here's what's on my mind right now. What is the optimal number of risk ratings that should be used to evaluate corporate default risk?   Clearly there is a trade off between granularity and validation.  A very granular rating system is desirable because it allows companies with small differences in risk to be placed in separate rating categories.  Pricing and economic capital allocation can then be calibrated to an appropriate level for each rating.  Default risk models with provide default risk estimates (in basis points), such as KMV's Credit Monitor, are clearly geared to satifying the need for granularity. But granularity comes at the cost of validation.  To my knowledge, there is no way to certain that each of the default risk measures coming from models, like Credit Monitor, are accurate.  Validation work has largely focussed on evaluating a credit model's ability to properly rank-order companies on the basis of their default risk (power curves, or CAP curves), and to a lesser extent evaluating the total number of defaults predicted by a model against the actual number observed.  (For example, see Kurbat and Korablev)  But, this doesn't really tell us whether specific default risk probabilities are accurate. To manage this uncertainty, I think the credit risk analytical frameworks at most institutions, certainly most banks, involve sorting companies into a discrete number of ratings, and then using a single probability estimate for every company in the rating.  Yet, there is no consensus on the number of ratings that should be used. Moody's and S+P both use rating scales which have 16 categories which apply to companies that normally have access to the debt markets (AAA plus 3 subdivisions for each of AA, A, BBB, BB and B), and another 6 or 7 grades in the CCC and worse range.  That makes a total of about 23 non-default grades.  But this scale was developed years ago, when the desire for abosolute (rather than relative) accuracy was modest. If you look at the default rates for each of the rating agencies grades, it is hard to conclude that they are actually able to discriminate among 16 classess of default risk.  If we define a grade having successfully discriminated a level of default risk if the default rate for that grade is higher than all better grades, then there is precious little evidence that default rates of companies in the AA and AAA range, and to a lesser extent, in the  A range are actually different.  If fact, I think the agency ratings historical experience shows that they are able to resolve no more than 12 discrete classes of default risk. If this is true, then it seems logical that individual institutions, who have a smaller universe of companies in their portfolios, will never be able to demonstrate that they can resolve more than 10 or 12 grades of default risk. Monday, 19 January 2001 Better late than never, here are links to the 2000 default reports from the rating agencies: Standard and Poor's Moody's Read them with caution. Both agencies are inconsistent in their determination of when a restructured credit edges into default. Neither includes Conseco in their list of defaults. Eric Falkenstein Being a good risk manager was like being a good LAN administrator - hardly worth the effort given the lack of appreciation. It's only February, but Eric just might have given us the quote of the year. I'm definitely tempted to frame it an put in on the wall of my office.   Thursday, 25 January 2001 Erisk.com "Everybody is coming to the conclusion that commercial lending is not an attractive business, if they measure risk properly," says Stuart Brannan, Montreal-based vice president of portfolio management with Bank of Montreal (Registration required) Hey, that's me. FT.com Dresdner Bank, Germany's third-biggest bank, is to shut its commercial lending activities outside of Europe in a bid to cut costs and increase the profitability of core businesses. Erisk.com In this article, Dr Bob Mark and Dr.Michel Crouhy of CIBC discuss the implications for banks of the internal ratings based approach described in the latest proposals released on January 16. The "Shipley" Committee In addition to calling for more frequent public disclosure, the working group said financial information should be disclosed based on a firm's internal methodologies and exposure categories. It said quantitative information on a firm's risk exposure should be balanced with qualtitative information describing its risk management process. Thursday, 18 January 2001 Economist.com In the end, though, the fate of the new Basle rules will depend on how they are implemented. American and British regulators are likely to be vigorous enforcers, while Japan and Germany, as well as plenty of less-developed countries, will lag behind. Why does the UK press insist on referring to "Basle"? The spelling on the documentation is "Basel". Just wondering.... I am still just cracking the surface of the new Accord but the following thoughts come to mind..... ....Basel insists that obligors be assigned to risk grades and that the same default probability be used for every obligor in the grade. This is out of step with leading edge practice which tries to estimate a unique default probability for each obligor. ....the proposed standard definition of default is very comprehensive. This is good. There is a slight inconsistency in that any agreed postponement of principal, interest or fees is considered a default, but if no agreement is given and a borrower just misses a payment there is a 90-day grace period. The same 90-days should be applied in both scenarios. ....there is no recognition that collateral in the form of accounts receivable, inventory or prooperty, plant and equipment reduce credit risk. This is just wrong. ....the standards for a bank being considered eligible to use the "advanced IRB" approach are likely to be very high. The benefits might not be sufficient to justify the effort involved.   Wednesday, 17 January 2001 FT.com Basle Two looks to be a judicious advance on its predecessor - provided the regulators can master its complexity, remain vigilant and remember that more sophisticated vehicles do not on their own guarantee a safer ride. American Banker An element likely to raise industry hackles is the application of the accord to banks' consolidated holding companies. Fed, FDIC, OCC While the 1988 Capital Accord was applied to all banks in the United States, it has not been determined how broadly the new approach will be applied, particularly given the many complex elements that may not be needed for smaller, less complex institutions. FSA The proposed new framework goes far towards delivering the key objectives of a safer and more efficient financial system, by aligning regulatory rules with good practice in well managed banks and by providing incentives for all banks to raise standards. OSFI Safety and soundness in today's dynamic and complex financial system is best achieved by the combination of effective internal controls and bank management, market discipline and supervision," he added. "The end result should be a safer, sounder and more efficient banking system.   Monday, 15 January 2001 The new, hopefully improved,, Basel Committee capital framework is due to be released tomorrow but I won't be reading it right away. I'm going skiing with my son and Basel will have to wait until Wednesday. If anyone manages to get to it before I do, please email your thoughts to stuart@brannan.org. If I get enough I will set up a special page of reactions to Basel. You should be able to download it from www.bis.org/wnew on 16 January at noon New York Time.   Wednesday, 10 January 2001 FitchIBCA Of late, the credit derivatives markets have been roiled by the commonplace use of restructuring provision as a credit event. The controversy arose following a restructuring of Conseco, Inc.'s bank loans. Seizing upon this event, credit protection buyers specifically invoked the restructuring provision under various credit derivatives and demanded payment. The facts surrounding this event have raised questions about the future of restructuring and whether it will continue as a standard credit event.   Tuesday, 2 January 2001 Dianne Vazza (Standard and Poors) With more defaults on the way, speculative grade spreads will widen further before they firm in the first half of the year. Marie Cavanaugh (Standard and Poors) "We expect sovereign rating stability and default probability to converge with Standard & Poor's corporate ratios over time as thenumber of sovereign observations increases; both groups use the same rating definitions," said Marie Cavanaugh, director and head of criteria for Standard & Poor's sovereign ratings group. "However, what these data refute is the suggestion that sovereign ratings have been less stable than corporate ratings and thus have been a destabilizing force in global capital movements." Maybe its just me but this looks like gobbledygook of the first order. Maybe sovereign default rates will converge with corporate defaults, and sovereign rating transitions might converge with corporate transistions, but sovereign rating stability will never converge with "corporate ratios". And no matter how stable sovereign ratings were in 2000, how then can this "refute the suggestion" that they were less stable in the past? FT.com The [Financial Services Authority] says requiring banks to disclose the amount of capital it requires them to hold would mislead consumers and investors. It could undermine the relationship between banks and their regulator and shake public confidence in the financial system. The FSA is absolutely right... if the amount of required capital isn't calculated in a reasonable manner. Given the illogical capital rules currently in place, the FSA should be actively prohibiting banks from disclosing the amount of regulatory capital they hold, and their Cooke ratios. These convey absolutely no meaningful information about a bank's risk profile. On the other hand, if the capital calculation is based on sound principles, then the FSA should be actively promoting disclosure. Let's hope that 1) the Basel Committee comes up with a new capital regime that makes sense and 2) if they do, that the FSA actively promotes more disclosure, not less. Alan Mcnee (Erisk.com) It now looks as if Basle will make two key changes to capital regulation. The first is that regulator-approved banks will be able to use their internal ratings of customers to calculate the regulatory capital they must set aside to cover credit risk. To be a bit more accurate, banks need the ability to use their internally derived credit risk measures to determine regulatory capital. Some banks convert these credit risk measures into "grades" or "ratings", but others do not. Once again, lets hope that Basel comes out with a set of proposals that permit banks to use their own measures of risk to determine regulatory capital, and also gives local regulators some guidance on how to audit and validate the veracity of each banks' measurements. Credit Magazine The credit derivatives market is one of the success stories of the last five years. .... But it is now facing its first real crisis. In the last six weeks, market consensus and regular practice regarding the place of loan restructuring in credit default swaps has disintegrated completely. The difference between the discretion available to bank lenders and bondholders was noted at this site last August. Credit Magazine is a lot more expensive and they just picked up on it this month.   Tuesday, December 19 2000 nytimes.com One day after the government declared two credit unions insolvent, a Japanese newspaper released a survey showing that the loan portfolios of Japanese banks were deteriorating after several large corporate bankruptcies. Here we go again.   Monday, December 18 2000 This site has been getting lots of hits from ANZ Bank today. Greetings to them from the Great White North. FT.com "Corporate bond lenders . . . have been negligent in understanding what a new-age economy does to corporate bonds themselves. It produces an environment of no upside and substantial downside." Someone should have reminded these lenders when the salesmen came to call, that it pays to beware of Greeks bearing gifts. Dismal.com With some formidable energy utilities being downgraded due to a supply and demand debacle in California, regulators would be wise to check this out: While continued regulation might have kept prices artificially low this summer, the price hikes are the result of inadequate capacity, not deregulation. The new system has only highlighted the shortage. Regulators must resist all temptation to interfere with the market. That may prove painful in the short term, but a framework in which prices that truly reflect market conditions provide the right incentives, and in which consumers are free to choose their supplier among vigorous competitors, delivers the best results in the long term.   Thursday, December 14 2000 Economist.com Profit warnings have become a daily ritual. One by one, companies that had seemed effortlessly able to meet analysts' expectations have delivered bad news, and most have been punished by investors for their trouble. As suppliers complain that bills are not being paid on time by well-known companies that are using every trick in the book to eke out their profits, can it be long before even Cisco issues a profit warning-and how will investors react to their current poster-child proving fallible? (Subscription required.)   Wednesday, December 13 2000 Moodys.com Moody's estimates that between 385 and 450 companies out of the over 9,000 the index tracks will default in the next twelve months, more than triple the 114 that have defaulted since December 1999. And also from, Moody's: As default rates continue to climb and overall levels of corporate credit worth declines, companies with outstanding debt rated below-investment-grade could find themselves unable to tap the market to refinance their outstanding debt when that debt come due,   Monday December 11 2000 I was at the ICBI Risk Management conference in Geneva last week, and came back trying to make sense of lots of challenging ideas. One of the most interesting presenations was by Charles Monet of JP Morgan who revealed some of the thinking behind Morgan's capital allocation techniques. Some of the more interesting points he made were that: Morgan adds an explicit charge for large exposures in their portfolio, over and above the capital cost allocated by their portfolio model (CreditMetrics). I have been troubled for some time by the fact that KMV's Portfolio Manager doesn't allocate sufficient capital to large exposures to produce unambiguously unattractive returns. Seems like Morgan faced the same problem and just boosted the capital to discourage large exposures.   They don't think that RAROC delivers the "correct" incentives to the managers of their many businesses. Instead they calculate the appropriate cost of capital for each business, and then deduct this from each business's P+L. It is management's job to maximize the residual amount (the "economic value added") and increase it consistently over time. I don't understand how this can be applied to individual credit risk assets.   They assume 100% correlation for credits within certain emerging markets. This would certainly dampen the effect that we have observed of international assets being allocated unintuitively low levels of economic capital.   On the other hand, they assume that all committed credits will be 80% drawn at the time of defaut. This broad assumption of a consistently high usage given default doesn't sit well with me. I think that well structured short term facilities would have much lower drawdown rates than 80%. In another presentation by Stephen Kealhofer of KMV, he provided insights into two fundamental issues: First, if you look at bond spreads over long periods of time, and compare these to KMV's EDFs, then the market price for default risk appears to be very stable (with the possible exception of short periods of extreme stress). Compare this to the typical analysis of bond spreads by agency rating, which have been very unstable over the past few years. I find the fact that KMV can extract a consistent relationship between market remunaration and risk, using EDF's rather than agency ratings to be another brick that supports greater reliance on EDFs.   Second, Stephen explained that the reason that Merton-type default models tend to predict fewer defaults than have been observed historically is because the managers of a firm respond to negative developments within the firm by increasing, rather than decreasing, the firm's liabilitiies. In effect, Stephen proposed that, rather than curtailing capital spending, or reducing debt, management often attempts to supplement internal cashflow with external sources of cash. This means that when a firms level of risk increases, management often contributes to make it even more risky. It is only when a management sees that this risky strategy isn't working that they being to reduce liabilities, and in some cases its too late. Typical Merton-type models don't incorporate this perverse management behaviour and therefore underestimate the likelihood of default.   FT.com "It's beginning to feel a lot like 1990," said Harvey Miller, senior partner at Weil Gotshal & Manges in New York. He says that because of tightening credit and an economy running at "stall speed", companies are finding it difficult to refinance debt. economist.com If America were to experience a recession as bad as that of the early 1990s, J.P. Morgan reckons that it could wipe out over a year's after-tax profits for Crédit Lyonnais, and more than four-fifths of Deutsche Bank's, albeit over several years. Subscription required. Its worth remembering that The Economist also reported that Barclays had a phenomenal level of exposures to European telecom borowers. According to people I talked to at Barclays, they were way off the mark. Thursday 30 November 2000 Armstrong Holdings was downgraded to "D" by S+P today. Moody's still has them at "Ca", even though they failed to repay about US$50MM in commercial paper that matured last week. Armstong's bank lenders dodged one bullet this time around because the commerical paper "backup" facility matured in October so Armstrong wasn't able to pay out the CP holder by drawing on a bank facility. Apparently, the company and its banks were on the verge of signing up for an extension to the previous backup facility when Owens Cornng filed for bankruptcy, and the lenders got cold feet because Armstonrg hsa significant asbestos liabilities. Here is KMV's EDF history for Armstrong. Since S+P had this rated A- on January 1 this will show up as an investment grade default in their data base. If Moody's defers the recognifion of teh default for a few more weeks, they will be able to call this a Ca default (which has very differnt implications for the consistency of their ratings) I don't think that Moody's will be able to ignore this until the new year, but it does raise the interesing question of whether the rating agencies might be tempted to defer classifying a default at the end of the year because it would make their track record appear more reliable.   Wednesday, 29 November 2000 Look for my smiling face in a future issue of Credit Magazine. Charles Smithson conducted a survey of portfolio management practices at a number of banks and I was called upon to comment on the survey. When it is published, I will put a link here. John Plender (FT.com) The verdict must be that while there is not yet a credit crunch, there is a worrying combination of contracting credit, declining asset quality and weakening equities.   Tuesday, 28 November 2000 Greg M. Gupton, Daniel Gates, Lea V. Carty (Moody's Investors Service) This study updates Moody's previous Loss Given Default (LGD)research for bank loans by more than doubling the number of observations from the 58 in our November 1996 study to 121 defaults (representing 181 loans).Here,we look to secondary market price quotes of bank loans one month after the time of default -allowing markets to process the default news and revalue the debt. Reading this study just makes me mad. First of all, its not really a "loss given default" study in the normal sense. It's a "market price given default" study. These may or may not be the same thing; it depends on a lot of things like how accurately the market can predict the amount and timing of post-default cash payments, and also what is the appropriate discount rate to apply to these payments. The study goes to great lengths to point out that Moody's ratings aren't mere predictors of default, but incorporate additional factors such as recovery. Moody's analysts are apparently able to determine whether to "notch up" or "notch down" a particlar type of debt because of differences in the recovery potential. It would seem logical to try to answer the question: what is the difference in recovery rates between two rated obligations (by the same issuer) which are one notch apart, or two. Unfortunately Gupton, Gates and Carty don't even try. Instead they compare the recovery rates for obligors which shared a rating category at the time of default. Does this mean that notching is only meaningful for the last rating assigned prior to default? I am becoming more convinced by the day that Moody's notching for seniority/subordination is a largely subjective and unscientific process. Finally, the study makes the point that obligor industry classification isn't correlated with LGD, which I think is mostly true. And the authors seems to think that there is something meaningful in the observation that recoveries tend to be lower when a borrower has multple lenders. But one aspect of bank loans which I believe should be highly correlated with recovery, the existence of borrowing base or margin arrangement, isn't even mentioned. End of rant.   Wednesday, 22 November 2000 Peter A. Wuffli and David A. Hunt (The McKinsey Quarterly) ...a McKinsey study of credit risk management has found, both the underlying nature of the risk and its importance to a bank's profitability equation have fundamentally changed. So much so, in fact, that financial institutions need to develop a whole new technical and organizational approach to managing credit, which will radically alter the culture of traditional universal banks. Wuffli and Hunt have written a great summary of the actions that banks need to take to get on top of their large corporate credit risks. The amazing thing is that the article was written in 1993. Much of what they describe is now deeply embedded in the corporate culture at large banks. It's easy to focus on the work that still needs to be done, and castigate ourselves for not yet having reached the end-game. Sometimes we need to helicopter up a few thousand feet and see just how far we have come.   Monday, 20 November 2000 ERisk.com ING's investment-banking and corporate-finance operations are the least profitable part of the group's overall banking businesses, as measured by a risk-adjusted return on capital. In the first nine months of the year, the risk-adjusted return on capital for investment banking and corporate finance was 5%, less than the 22% posted by ING Europe, which includes the retail and commercial-banking operations, and the 44% at ING Asset Management. ING began calculating risk-adjusted return on capital only this year. Based on these results and one year's experience with RAROC, ING decided to drastically curtail its investment banking operations and put the US business up for sale. Decisive folks, those Dutch bankers.   Monday, 13 November 2000 Economist.com Carol Levenson of Gimme Credit, a research service, says she has never seen such a large divergence between the view of riskiness that is outlined by bond-rating agencies and that suggested by the market price of corporate debt. It would be best to hope that the experts in the agencies are right. (Subscription Required) I wouldn't bet on it.   Sunday, 12 November 2000 Mark Bruno (U.S. Banker) Fretting about a huge increase in problem loans, some regulators are blaming banks for caving into Wall Street's insistant demands for rapid-fire profit growth. Another article that points fingers at bank managers and equity analysts for leading banks to the brink of another asset quality debacle. BUT ONCE AGAIN, not a word about the lunacy of regulators who don't permit banks to reduce their capital if they pursue a low risk lending strategy.   Wednesday, 8 November 2000 Daewoo Motor filed for bankruptcy this week and Hyundai Engineering and Construction has defaulted. KMV's median EDF levels for Korean firms have been rising since mid-1999. The curious thing is that, even though default risk appears to be rising, many Korean banks have recently been upgraded by the rating agencies (Moody's, I haven't checked the others) How can this be? Surely it is the Korean banks that will bear the brunt of losses from their exposures to Daewoo, Hyundai and others yet to come.   Tuesday, 7 November 2000 Alan McNee (eRisks.com) Perhaps it's time for banks to start working on a standard format for reporting risk: it may not be long before pleading confidentiality ceases to be an option. Here's a suggestion for a place to start. It comes from discussions with John Mingo when we were working on preparing the RMA response to the Basel Committee's new capital rules. Every reasonably sophisticated bank should be able to describe their process for determining an EDF for borrowers in their portfolio, and also the process for assigning an LGD to indivudual loans. In fact the new regulatory rules are likely to expect this. Banks that are able to do this could then provide a set of matrices showing their exposure by EDF and LGD. A simple matrix like this one could be used to show the distribution of commitment, outstandings and capital. If a bank had a portfolio of loans totaling $100, it might look something like this:  EDF/LGD  <10%  10-30%  31-40%  41-60%  61-80%  >80%  <.5              .5 - .10           5   .11 - .20         5     .21 - .40      5 10       .41 - .80    5  10 10      .81 - 1.60    10 10  5      1.61 - 3.20  5  10  5        3.21 - 6.40  5            6.40 - 12.80              >12.80             This simple representation of the portfolio should show a tendency for loans to appear on the bottom-left to top-right diagonal as loans to the riskier borrowers (bottom) should be offset by sound structure resulting in a lower LGD (left). Loans to high quality borrowers would have low EDF's (top) and the bank may be prepared to provide minimal structure or even take a subordinate position (right). If every bank presented its portfolio information in this manner it would provide a good indication of the types of credit risks they are prepared to assume, and any significant differences between the information provided by banks would provide a reasonable basis for further inquiry by analysts, shareholders or regulators.   Monday 6 November 2000 Alexandra Berthault, David T. Hamilton, Lea V. Carty (Moody's Investor Services) Since 1972, when Moody's began rating CP programs, 45 issuers have defaulted on roughly $4.3 billion of rated and unrated CP. Historically, the U.S. market, the world's largest, has contributed the lion's share of defaults (31.1% since 1982). However, since 1992, only one rated USCP issuer, Mercury Finance Company defaulted. From the mid-1990s on, the bulk of CP defaults have been non-US issuers in nascent markets throughout Europe, Asia, and Latin America. 'Course they conveniently "forgot" about Conseco, which continues to operate due only to the forbearance of its lenders but which neither S+P nor Moody's have decided to call a default.   Wednesday, November 1, 2000 Economist.com There are now signs -- tentative for the moment, but worrying enough already -- that big banks are messing up again. Three problems stand out. The first is that problem loans in America are increasing, even though the economy is still bowling along at a fair rate. The second is that recent turmoil in the capital markets may well have caused difficulties for some of the commercial banks that have been rapidly expanding their investment-banking businesses in recent years. And one reason for this is the third problem: the banks' huge lending exposures to telecoms firms. (Subscription required) And buried in the text is this rather surprising comment: Britain's Barclays offers a good example of the depth of some banks' acute; involvement. At its peak, Barclays had telecoms exposures of some $20 billion. Most of these were (and still are) to just three companies: Vodafone, British Telecom and Orange. Its exposures have now been reduced somewhat, thanks partly to regulatory pressure. I find this very surprising because I believe that Barclays has been at the forefront of sound credit portfolio management over the past few years. If they are allowing concentrations of that magnitude to build up, then clearly I need to reconsider.			Home FAQ Feedback Archive July - Oct 2000