Credit risk ratings are becoming a risky business.

Yesterday’s filing of the complaint against S&P (MHP) centers, in essence, on the allegation of false advertising.   Stepping in to this issue for a moment, one of the key defenses offered by the raters is that their ratings are protected under the First Amendment rights to express an opinion (their “speech”).  But as law professor Eugene Volokh opines in his letter to the House Committee (May 2009) it is within the framework of commercial advertising that “speech aimed at proposing a commercial transaction – is much less constitutionally protected than other kinds of speech.”

In this case, the AG is not really focusing wholly on whether the ratings were wrong, as much as it’s saying that S&P advertised that it was following a certain code in ensuring the appropriate levels of independence and integrity were being brought to the ratings process.

A former SEC enforcement official, Pat Huddleston, once explained that “[when] I say the [financial] industry is dirty, I don’t mean to imply everyone in the industry is dirty,” … “[only] that the industry typically promises something it has no intention of delivering, which is a client-first way of operating.” This is essentially what the complaint argues: that S&P “misrepresented its objectivity” while offering a service that was “materially different from what it purported to provide to the marketplace.”

This goes back to, really, the key reform measure Mark proposed before the Senate in 2009 – that rating agencies would do well to separate themselves from commercial interests, by building a formidable barrier around the ratings process.

First, put a “fire wall” around ratings analysis. The agencies have already separated their rating and non-rating businesses. This is fine but not enough. The agencies must also separate the rating business from rating analysis. Investors need to believe that rating analysis generates a pure opinion about credit quality, not one even potentially influenced by business goals (like building market share). Even if business goals have never corrupted a single rating, the potential for corruption demands a complete separation of rating analysis from bottom-line analysis. Investors should see that rating analysis is virtually barricaded into an “ivory tower,” and kept safe from interference by any agenda other than getting the answer right. The best reform proposal must exclude business managers from involvement in any aspect of rating analysis and, critically also, from any role in decisions about analyst pay, performance and promotions.

Two other elements jump out immediately from the complaint:  First, the complaint specifically argues that the rating agency “misrepresented the factors it considered when evaluating structured finance securities.”  Next, the complaint tries to tie S&P’s actions to its publicly-advertised code of conduct, arguing that its actions were inconsistent with the advertised code.

In respect of actions being inconsistent with the code, certain of these arguments are commonplace, such as the contention that the rating agencies did not allocate adequate personnel, in opposition to what’s advertised in the code.  This of course becomes a contentious issue – you can see S&P coming back with copious evidence of situations in which they did “allocate adequate personnel and financial resources.”  But the complaint hones in on the factors considered in producing a rating, and it focuses on two parts of the code:

Section 2.1 of S&P’s Code states: “[S&P] shall not forbear or refrain from taking a Rating Action, if appropriate, based on the potential effect (economic, political, or otherwise) of the Rating Action on [S&P], an issuer, an investor, or other market participant.”

and…

Section 2.1 of S&P’s Code states: “The determination of a rating by a rating committee shall be based only on factors known to the rating committee that are believed by it to be relevant to the credit analysis.”

This brings back to mind, disturbingly, a recent New York Times article (Ratings Firms Misread Signs of Greek Woes) which focuses on the deliberations within Moody’s (MCO) and their concerns about the deeper repercussions of downgrading Greece – rather than the specifics of credit analysis:

“The timing and size of subsequent downgrades depended on which position would dominate in rating committees — those that thought the situation had gotten out of control, and that sharp downgrades were necessary, versus those that thought that not helping Greece or assisting it in a way that would damage confidence would be suicidal for a financially interconnected area such as the euro zone,” Mr. Cailleteau wrote in an e-mail.”

The question then, is whether rating committees were focused on credit analysis, or whether other concerns were at play, aside even from typical business interests.  The concerns for rating agencies, from a legal perspective, can become quite real when the debate centers not on ratings accuracy, but on whether the rating accurately reflected their then-current publicly available methodology.  There may be substantial risks, therefore, in delaying a downgrade of a systemically important sovereignty or institution (such as a too-big-to-fail bank or a key insurance company) if such downgrade is appropriate per the financial condition of the company or sovereignty, or in providing favorable treatment to certain companies or sovereignties based on the relative level of interconnectedness. 

The allegations of misrepresenting factors considered in their analysis opens another can of worms for rating agencies, as they’ll subsequently be increasingly focused on disclosing the sources of the information relied upon. There’s substantial concern that, to the extent they’re relying on the issuing entity (in cases in which the issuing entity is itself the paying customer) that such reliance becomes a disclosure issue to the extent the investor may otherwise have assumed the rating agency was independently verifying such information. This was a frequent problem in the world of structured finance CDOs such as those described in the AG’s complaint.

Last but not least, the complaint focuses on the effectiveness of ratings surveillance.  This is a topic of importance to us, as we feel that proper surveillance, alone, may have substantially diminished the magnitude of the crisis.  At the very least, certain securitizations that ultimately failed may not have been executed had underlying ratings been appropriately monitored, and several resecuritizations may have become impossible, limiting the the proliferation of so-called toxic assets.  See for example: Barriers to Adequate Ratings Surveillance

That’s all for now.  There’s a lot more to this complaint, so we suggest you check it out here.

First up, the Philadelphia Inquirer ran a story that suggests Collingswood NJ Mayor Jim Maley had failed to persuade Moody’s to adjust the borough’s rating.  Apparently, Moody’s had previously acknowledged that in its prior rating action, its analysts had miscalculated the borough’s liability on an $8.5 million loan for a struggling redevelopment project.  Reminiscent of S&P downgrading the rating of the United States after it was externally uncovered that they had made a $2 trillion miscalculation, Mayor Maley reportedly called the decision “an outrage” and explained that  “[it's] a shell game,” …. “It’s clear they’re not changing this rating because if they do it too quickly, they were wrong.”

Elisabeth Sexton has been covering, in Australian newspapers, the Federal Court proceedings concerning 13 Australian councils’ suit in conjunction with their purchase of CPDO notes.  (They’re suing Local Government Financial Services Pty Ltd, the investment bank that created them, ABN AMRO — now part of Royal Bank of Scotland – and Standard & Poor’s.)  Sexton’s coverage details concerns about the rating agency’s reliance on its own models. For example S&P’s analyst on the deal expresses surprise to learn that their CPDO ratings had nothing to do with the outcome of their model.  A separate S&P analyst was cross-examined about the possibility he gave in to the banker’s demands: an email he wrote to colleagues in May 2006 about a modeling assumption for volatility of credit spreads said that: ”We think it’s acceptable to lower the vol to 30 if it’s absolutely necessary for this trade.”  While the rating is well known to be the result of a ratings committee, rather than a model, this goes back to the whole question of ratings transparency, and what it means (see Rating Agencies: Transparent or Not).

Next, Kroll Bond Ratings produced a very interesting report on muni bonds this week. Among other things, this snippet caught our eye, as it speaks to a tangible discrepancy between how each rating agency defines default:

“For example, an S&P study (2000) documents 917 bond issues that went into default during the 1990s, 137 of which were rated and 780 of which held no rating. Similarly, Moody’s (2010) claims that just 54 (Moody’s) rated issuers defaulted on municipal bonds between 1970 and 2009. [1]  By contrast, various data sources indicate that 2,842 issuers defaulted over the same period.

[1] Moody’s study ignores technical defaults and defaults by issues that were wrapped by a financial guarantor.”

We feel that rating agency performance should be more comparable, and the definitions should be standardized, so that one can measure relative ratings performance in a way that comapres apples to apples.  See our piece on Why It’s a Problem That Only Rating Agencies Evaluate Their Performance

Last week, the Reason Foundation submitted a comment letter to the SEC that suggests an interesting angle to promoting healthy ratings competition.

First things first though.  We think it’s unhealthy to simply promote ratings competition.  That tends to lead to a decrease in standards as ever more raters fight for the same pie of new deals.  But if they can be encouraged to compete on accuracy, the reputational model appropriately dismissed by JP Hunt comes back into play. 

We think the Reason Foundation would agree with us on that.  They argue that a random (or arbitrary) assigning of credit rating business, by a government-driven public or private utility or self-regulatory organization to NRSROs could “remove incentives to compete in this manner, and could thus result in worse rating methodologies. If a firm obtains no competitive benefit from publishing a more analytically robust methodology, it may stop making the investment in research to improve its methodologies.”  (Our aside – business could or should better be allocated, proportionally, in line with historical performance.)

Here’s their coup de grâce.  They put forth as one alternative a 3rd party “challenge” process, that challenges what are thought to be incorrectly assigned ratings. (Us again – we might encourage this for stale ratings too, perhaps defined as having been inaccurate for a period longer than 6 months.)

“Under such a system, an independent credit assessment firm could submit a statement to the Federal Reserve challenging the rating assigned by an NRSRO. If the Federal Reserve determined that the challenge had merit, it could ask the NRSRO to respond and potentially hold a public hearing to determine the validity of the challenge. If the Federal Reserve determined that the rating was flawed, the affected securities would attract a maximum capital charge until the rating agency issued a rating supported by the challenger or another rating agency selected by the issuer published a rating.”

If the challenging party had a tennis-like restricition on the number of challenges it may make, the challenge process (or the threat of being challenged) would encourage raters to be better focused on maintaining up-to-date, consistent, reliable and defensible ratings. 

Perhaps if they were required to turn over ill-deserved or ill-gotten revenues for inaccurate ratings for the entire time they were inaccurate, we would see a desirable positive feedback loop: they would have a real financial incentive to remain accurate.

We delve into various rating agency conflicts that may hinder or impair the provision of objective credit ratings; and we offer ways to minimize or circumvent these “distractions.”

General duties

7.(1)  A credit rating agency must – …

(j) ensure that it at all times has the necessary knowledge and experience to issue credit ratings and perform its credit rating services;

Methodologies, models and key rating assumptions

9.  A credit rating agency must -

(a) adopt, implement and enforce adequate measures to ensure that the credit ratings it issues are based on a thorough analysis of all the information that is available to it and that is relevant to its analysis according to its rating methodologies;

(b) use rating methodologies that are rigorous, systematic, continuous and subject to validation based on historical experience, including backtesting;

(c) regularly review its methodologies, models and key rating assumptions such as mathematical or correlation assumptions, any significant changes or modifications to them and the appropriateness of those methodologies, models and key rating assumptions where they are used or intended to be used for the assessment of new financial instruments; and

(d) establish internal arrangements to monitor the impact of changes in macroeconomic or financial market conditions on credit ratings.

Credit ratings

10.(1)  A credit rating agency must –

(b) when publishing a credit rating –

(i) state clearly and prominently any attributes and limitations of the credit rating; and

(ii) provide an explanation of the key elements underlying the credit rating that an investor, potential investor or a member of the public, as the case may be, are able to understand how a rating was arrived; and

(c) monitor credit ratings and regularly review its credit ratings.

10.(4)  A credit rating agency must refrain from issuing a credit rating or withdraw an existing rating if the lack of reliable data, the complexity of a financial instrument or the quality of information available cannot result in a credible credit rating.

Disclosures

13.(2)  A credit rating agency must, every six months, disclose to the public and its subscribers data about the historical default rates of its rating categories.

It may be generally agreeable that several mistakes were made that (1) enabled credit raters to become too powerful, (2) allowed for their performance to continue unmeasured,  unmeasurable and unaudited, and (3) hindered our ability for holding them accountable.

The resulting lowering of ratings standards is not unexpected – it represents our failure as much as theirs, as we’ve classically conditioned them to push the boundary in seeking extended economic rents.  Let’s be honest, they’re for-profit companies providing opinions for which they aren’t being held accountable, and for which the negative-feedback-looping reputation mechanism is rendered inapplicable (they’re essentially a monopoly).  Do we expect them to say “no” to receiving free cash?

But while it may be the easiest solution to simply remove them, I think we can all agree that the average investor will be no better in measuring risk in their absence.  Yes we may agree their ratings hold little (or no) predictive content in certain areas of structured finance.  Perhaps we would argue they shouldn’t have rated some asset classes at all.  But does that mean that their expertise in rating corporate debt is superfluous, and should be foregone in its entirety?  Do their corporate ratings provide no economies of scale in respect of the analysis of corporations?

It’s certainly popular to say “we don’t care about their opinions” – but certain investors do.  Some investors are ill-equipped to perform the analysis required on a name-by-name basis.  Moreover, it’s not necessary that all rating agencies be good at evaluating all types of debt.  Perhaps some should focus on specific niche areas, and investors should learn to evaluate their relative levels of expertise in different areas.  Being able to independently measure their performance helps inform investors’ determination as to which raters they care to rely on in each specific scenario.

The short-term solution is simple. Rather than berating them as a public evil, let us empower rating agencies to be a public good.  It’s as fundamental as creating an incentive for them to be accurate, or a penalty for consistently being inaccruate.  Penalties can be effective: one only has to suspend one license held by one rating agency for a short period of time (a “sit out period”).  The other rating agencies will quickly ensure they’re dedicating the resources necessary to providing accurate ratings in that area.  (Alternative forms of penalties include ratings fee clawbacks and other forms of legal accountability.)

The Reuters article points out that “[fund] firms contacted by Reuters said rating agency research tended to be backward-looking and superficial…”  Perhaps we have only to incentive the rating agencies to be accurate, to dig deeper than endlessly assuming historical mean-reversion.

The standardisation of symbols ought to be a good thing — it’s something we’ve long been lobbying for (see  for example our Feb. 2010 piece on “Economies of (Ratings) Scales“) as it provides an advantageous medium for investor understanding. 

More important than the standardization of the symbols, however, is their definition, and it is in this aspect that the SEBI announcement is most wanting.

While it is rewarding to have ratings mean the same thing across companies and asset classes, it is more important that they have meaning – a clearly articulated, mathematically-defined, real meaning.  For ratings to be useful and predictive, or for ratings performance to be measurable, it is essential that ratings be linked to a quantitative scale.

For example, SEBI will now require credit rating agencies (CRAs) to define AA ratings to mean a “high degree of safety regarding timely servicing of financial obligations”; A ratings have an “adequate degree of safety regarding timely servicing of financial obligations.”

Similarly, B ratings have a “high risk” of default, whereas C-rated instruments have a “very high” risk of default.

The problem is that the terms “high” and “adequate” are subjective, giving the rating agencies a free pass to the extent they wish to purposefully inflate their ratings.  CRAs’ litigation risk is severely minimized: it is incredibly difficult (near impossible) to look through a ratings model and argue that the rating had only “adequate” safety, not “high” safety.  It’s equally difficult to separate, mathematically, the terms “high risk” and “very high risk.”

SEBI’s unwillingness to ensure the adoption of a quantitative scale means investors cannot reproduce ratings, and cannot encourage CRAs to adjust their ratings when they’re faulty.  In other words, in protecting the CRAs, SEBI has removed one of the few positive feedback loops from the ratings process, leaving investors all the more vulnerable to the potential for wide-scale ratings massaging.

Oddly, SEBI mentions in their release that their announcement is the product of their “consultation with the CRAs and considering the international practices,” but it makes no mention of having consulted investors.  Unfortunately, this release serves the CRAs very well — the investors, not so well.

Our general approach has been to recommend the removal, to the extent possible, of all external business incentives from the delivery of the ratings themselves.  Given many of the larger credit rating agencies are (rather unfortunately) divisions of publicly-traded corporations, this has been quite a difficult idea to push. 

Corporations like to sell analytics that rely on their having rated the underlying securities. They like to give their employees stock bonuses, the value of which depends on the number and frequencies of ratings being provided.  In these ways and others, rating agency analysts can become financially incentivized to rate any and all products.  We seek to reverse this misalignment of interest, and to encourage rating agencies to care deeply for the provision of accurate ratings, even if it means they have to walk away from rating certain products if their methodology doesn’t accomodate them, or if they cannot achieve the desired rating. 

Our earlier piece “Barriers to Adequate Ratings Surveillance” explored the combative interests rating agencies face when ratings and monitoring deals.  Rather ominously, we explained that “the rating agencies do not only lack the incentive to be accurate when monitoring deals, but are in fact potentially incentivized to be inaccurate.” 

We examined, in particular, the potential for analytics sales to have a hazardous affect on the quality of ratings, and so we naturally find it quite fulfilling to read the following excerpt from the SEC’s proposal vis-a-vis the credit rating agencies (emphasis added):

The proposed new absolute prohibition would be designed to address situations in which, for example, individuals within the NRSRO responsible for selling its products and services could seek to influence a specific credit rating to favor an existing or prospective client or the development of a credit rating methodology to favor a class of existing or prospective clients. With regard to methodologies, the Commission notes that its staff found as part of the examination of the activities of the three largest NRSROs in rating residential mortgage-backed securities (“RMBS”) and collateralized debt obligations (“CDOs”) linked to subprime mortgages that it appeared “employees responsible for obtaining ratings business would notify other employees, including those responsible for criteria development, about business concerns they had related to the criteria.” The absolute prohibition in proposed paragraph (c)(8) of Rule 17g5 would be designed to insulate individuals within the NRSRO responsible for the analytic function from such sales and marketing concerns and pressures.

This is important: the drive to increase ratings competition is associated with the economic principle that increased competition lends itself to a greater value proposition. We wanted to warn the EC that, in the case of ratings competition, the relationship between increased competition and increased quality may not be what they expect.

Back in November 2010, Credit-Rating — a credit rating agency in the Ukraine — complained (perhaps not unreasonably) of the very pressures exerted on their ratings by the proliferation of newly authorized rating agencies:

“During the past 6 months the three credit rating agencies have assigned credit ratings to certain companies and banks, which earlier bore ratings from Credit-Rating. In all these cases a drastic overrating (by up to 6 notches) produced by the new participants of the market has taken place.”

Indeed Credit-Rating’s protestations seem to have left them in no better a position.  Today’s press release explains their decision to litigate against the State Commission for Securities and Stock Market of Ukraine for withdrawing their state-authorized status, “as this decision has been adopted with no proper reasons, flagrantly violating required procedures and in the interests of third parties.”

Credit-Rating also seems to think the newly authorized rating agencies had a hand in this:

“Credit-Rating will seek not only for nullifying the Commission’s decisions in a court, but also to determine the extent of involvement of IBI-rating, Rurik, and Expert-Rating in these actions.”

At a time when disclosure and transparency are key directives of the new regulatory regime (under the Dodd-Frank Act), the American Banker has produced a timely article, critical of the state of affairs.  With their permission, we’re providing for you certain key excerpts from their Feb. 28 article (click here for full article): (emphasis added)

Despite widespread allegations of botched loan administration and conflicts of interest, a review of the three main ratings agencies’ quality evaluations of servicers (including banks) show that virtually every servicer with a published rating is ranked as “proficient,” “average” or better.

Fitch, which rates servicers on a scale of 1-5, has no servicers rated below a 3 — a level it designates “proficient.”

Standard & Poor’s rates servicers according to how they stack up against “average” performance. Out of the more than 30 servicers on which it offers a public rating, all are average or better. Subpar scores exist under S&P’s system, a spokesman said, but are usually kept confidential by the servicer client paying for them until they can improve their performance.

Moody’s also rates servicer performance relative to “average,” with 1 being “strong” and 5 being “weak.” Out of the 25 companies it publicly evaluates, only the bankrupt Lehman Brothers’ Aurora Loan Services ranks as a 4, or “below average.”

…

Moody’s vice president Bill Fricke said the generally average or better ratings of its clients reflect that only larger and more established servicers tend to commission ratings.

The unusual distribution of ratings came to American Banker‘s attention during research for a recent article on Carrington Mortgage Services. The Carrington Capital Management unit has pursued certain unorthodox strategies, such as holding on to foreclosed homes for long periods of time, that some investors allege benefit its holdings of bottom-ranked securities in deals it services. Carrington has argued in court documents that the interests of its holdings “may be adverse” to other investors’, and the company is being sued by the state of Ohio for unworkable modifications and “incompetent, inadequate and inefficient customer service.” Carrington has offered an extensive defense of its practices, which it asserts are in the interests of the trusts it serves as a whole.

Two ratings agencies evaluate Carrington, S&P and Fitch, though S&P has not made its evaluation public.

…

In its public rating of Carrington, however, Fitch cited neither investors’ allegations nor Carrington’s rebuttal. In December, Fitch issued a “proficient” rating for Carrington and praised the company’s “experienced management team and capable default-handling practices.”

…

“Carrington’s ’3′ ranking denotes the lowest acceptable stand-alone servicer ratings in Fitch’s scale and is in part due to its unorthodox loss mitigation and liquidation strategies,” said Fitch managing director Diane Pendley in the e-mail.

Asked to reconcile that statement with the complimentary language in its public servicer quality rating materials, Fitch offered no additional comment.

“I don’t know how they could have possibly missed this,” said Amherst Securities analyst Laurie Goodman, who argues that Carrington’s actions have harmed investors. “This is something that the market has been aware of for years. You look at Carrington’s capitalization mod rate, and it just flashes, warning, warning, warning.”

Golden Knight II CLO, Ltd., issued in March 2007, is a collateralized loan obligation backed primarily by a portfolio of senior secured loans.

In July 2009, when Moody’s downgraded to single C the class E notes issued by Golden Knight II, it sparked my interest: a Moody’s single C rating maps to an idealized expected loss of 100% (to see how this works, visit page 33 of ”Moody’s Approach to Rating Collateralized Loan Obligations,” Aug. 12 2009). 

At a time when loan prices were on the way up, could Moody’s be absolutely certain that this actively managed loan portfolio would not make a comeback - that no more moneys would ever be due to this tranche?  Or was this a ratings error waiting to happen? (A “Type II” ratings error is the rating too low of a security that ultimately pays off.)  Single C is Moody’s lowest rating – it cannot go any lower.

Last Friday, while many of us were readying ourselves for the long President’s Day weekend, I was there to witness the astonishing.  An upgrade: Golden Knight CLO II tranche E upgraded to Caa2.

Why?

Moody’s noted in its press release that (now) “all related overcollateralization tests are currently in compliance” and that  the “Class E Notes are no longer deferring interest and that all previously deferred interest has been paid in full.”

Marvelous!

Oddly enough, though, Moody’s reminds us that one ought not to have too much confidence in the accuracy of these ratings: 

“this transaction is subject to a high level of macroeconomic uncertainty, as evidenced by 1) uncertainties of credit conditions in the general economy and 2) the large concentration of speculative-grade debt maturing between 2012 and 2014 which may create challenges for issuers to refinance. CDO notes’ performance may also be impacted by 1) the manager’s investment strategy and behavior and 2) divergence in legal interpretation of CDO documentation by different transactional parties due to embedded ambiguities.”

My only question then is how, with all this confusion and variability, could one justify the original downgrade to single C?

THIS POST IS PROVIDED BY A CONTRIBUTOR WHO WISHES TO REMAIN ANONYMOUS

If a methodology is well defined, users can spot when the rating agencies have slipped up and encourage them to correct it.  This provides a positive, reinforcing feedback loop: it limits the potential for damage (i.e., large downgrades when errors are noticed suddenly).  Of course, to the extent investors can show that the ratings used do not adequately reflect the rating agencies’ then-current methodologies, the rating agencies may worry they’re increasingly exposed to litigation risks.

The problem is that ratings errors — actual system failures – are not infrequent (see here and here for examples).  Similarly, their methodologies change quite frequently, which is why we think well articulated criteria are important.

When Moody’s in April 2009 downgraded the ratings of Ambac’s insurance subsidiaries (Ambac Assurance Corp. and Ambac Assurance UK Limitied) to Ba3 (junk) from Baa1, Ambac responded by saying: 

“While Ambac believes that Moody’s is entitled to its opinion of Ambac’s financial strength, it notes that this is the tenth such opinion change since January 2008.”

We wrote a piece back in May about the lack of transparency in rating agency methodologies.  The piece provided both sides of the argument: why the rating agencies felt they were being transparent, and why market participants felt otherwise.

Our recent experience is that despite the encouragement of the Dodd-Frank Act towards enhancing ratings transparency, the large rating agencies — with the possible exception of Moody’s — are in many cases being increasingly less transparent about their ratings methodologies. 

This is a particular concern given the increasing interest the rating agencies are themselves seeing in non-traditional investments (see here) and that Goldman Sachs’ president Gary Cohn is seeing in pushing risky activities into the less regulated, opaque shadow banking system (see here).

One has to wonder if it’s purely lip-service when S&P president Deven Sharma writes for the Financial Times that “capital markets thrive on certainty” while at the same time their ratings methodologies change rapidly. 

Yesterday SCI printed a criticism of S&P’s newly proposed bond insurance ratings criteria.  The author, managing principal of Grenadier Capital, notes that the criteria are “confusing,” that there is a lack of disclosure surrounding the circularity inherent in the process, that the model inputs are ill-defined, that “S&P gives itself wiggle-room to blame the management of the bond insurers, as opposed to its own failings in the actual ratings process.” 

He questions whether the convoluted components and stresses proposed in S&P’s 39-page manual indicate that S&P lacks confidence in its own ability to rate the insurers’ guaranteed transactions.  He wonders whether investors will ever again gain comfort in a bond insurer’s ratings if the proposed criteria are adopted.

We wonder, too.

In the corporate world, a low debt rating often hinders an issuer’s ability to issue debt and typically makes any debt issued more expensive to the issuer.  In the worst of cases, such as the case of an insurer like AIG or MBIA, downgrades can affect their ability to continue operations, as their ability to insure counterparty exposure hinges on their maintaining a high (often superior) rating, and certain rating triggers — such as a downgrade to below investment grade — might allow counterparties the right to terminate existing contracts.  

More generally, holding a lower-rated security requires maintaining higher regulatory capital reserves, or collateral margin, than for higher-rated securities.  In other words, low ratings can hinder growth and expansion potential.  As such, Type II errors are harmful to investors (who are holding more collateral than is warranted) and to debt issuers (who are required to pay a higher-than-warranted interest rate on their debt and/or may find debt issuance to become prohibitively expensive). 

There’s also a particular susceptibility to this error in a market in which investors are worried about highly rated securities defaulting.  To protect against that, there may be a tendency to lower ratings. Hence today’s focus on the Type II error.

One solution, or mitigant, we provide for in our submission is to require rating agencies to describe not only the reason for their rating action (“the upgrade was due to the amortization of the notes”) but to justify the action, from a data perspective, and explain in what ways the security performed differently to the rater’s expectations.  We call, thus, for disclosures that are meaningful, informative, justifiable and verifiable disclosures.

For example, Tropic  2003-1 class A1L (CUSIP 89707QAA9) was rated Aaa/AAA/AAA by Moody’s, Fitch and S&P in 2003.

-          Fitch has to this day never changed the rating of the bond.

-          S&P in August 2010 and again last month downgraded this bond from AAA to AA to A+, at which rating it currently resides.

-          Moody’s last year downgraded the bond from Aaa to A3 (March ‘09) to Baa3 (October ’09) and yesterday upgraded it to Aa3.

All three raters have a markedly different approach and so eyebrows ought to be raised and questions asked: what happened in October and November of this year that would encourage S&P to downgrade the bond while encouraging Moody’s to upgrade it significantly (6 rating subcategories)?; and what level of confidence does each rating agency have in their now-current rating given the volatility of their recent actions?

If, for example, one rating agency were forced to describe in detail that “the bond paid down slower/faster than expected, at rate x% as opposed to y%,” investors would be empowered to consider the effect that a similar amortization might have on their other bonds, and they will be able to pressure the other rating agency to act in a similar fashion. 

In sum, detailed justification keeps a rating agency honest, and promotes investor due diligence.

(For further reference, consider the senior notes of Allmerica CBO I, Limited/Corp (CUSIP 01975EAA6) which was paid in full earlier this year, despite being rated triple C by two different rating agencies at the time of full pay-down.)

Fitch Ratings’ Grossman and Hansen pulled them up in a comment letter which explains that CDS spreads provide an imperfect measure when contemplating default probability:

“when using CDS spreads as default risk indicators, it is important to note that spreads can be driven by a number of factors not directly related to an entity’s fundamental creditworthiness, such as the leverage inherent in CDS trading, liquidity conditions, counterparty risk and the risk aversion of market participants.”

More important, in our opinion, is the examination of the superior measure and the environment in which it remains the most reliable measure.  Certainly, each measure has its pros and cons, as Fitch points out.  Nor will any measure be immune to the parties estimating it: within each measure, for example, each CDS trader has his profitable bets and his losers, and each rating agency has its successes and its failures.  (If in general, traders of a certain company’s CDS tend to be poorly informed, a better-informed rating agency analyst might tend to be more accurate, even if it happened to be determined that CDS spreads typically outperform ratings from an accuracy perspective.)

In other words, both the rating agency and the market could be wrong: it therefore serves no purpose to consider a rating to be an informed opinion simply because it tends to agree with the market.  In the worst case, a rating agency might actually use CDS spreads as a tool in coming up with the rating, in which case the ratings provided will be highly (if not perfectly) correlated to the CDS spreads.  In such a scenario, the FT would doubtless find the rating to be entirely accurate.  But what would be the usefulness of the rating?

Rather, we should hold the ratings and the market to a different accuracy standard, testing whether their opinions provided useful and meaningful predictive content, over time, of whatever they were estimating — in this case probability of default.  Is there a strong correlation between CDS spreads and eventual default (over a certain tested time horizon); and is there a strong correlation between the default probability rating provided and the eventuality of default?

The following is an example of something that opens our eyes, and would be humorous if only it weren’t so serious, or so regular an occurrence.

On March 4, 2010 S&P downgraded Structured Asset Mortgage Investments Trust 2003-CL1 (CUSIP 86358HSY6) from AAA directly to D (D=defaulted).

A month later, Moody’s placed the same bond on watch negative at its then-current Moody’s ratings of Aa2. 

This disconnect was resolved this morning, when S&P corrected its rating from D back to AAA.

The reason:

On March 4, 2010, we lowered our rating on class I-PO, a principal-only (PO) certificate, to ‘D (sf)’ based on trustee remittance reports that indicated a cumulative principal write-down. Our current ‘AAA (sf)’ rating on class I-PO reflects our assessment of information in corrected remittance reports dated April 2009 through July 2010, as well as our analysis of the credit support available to this class to cover the projected loss amounts as of the August 2010 remittance report.

Considering ratings data made available on their websites, we have noted a striking increase in ratings surveillance activity among all three of the NRSROs we examined.  If one goal of reform is to encourage NRSROs to be more efficient in their upgrades and downgrades, then clear progress has been made.

For the purposes of this analysis, if a security was downgraded twice during a given year, it was considered to have been acted on for that year (i.e., one “action” is considered to have taken place).  Our analysis of approximately 750,000 securities, displayed a noteworthy increase in monitoring activities in the most recent year, as compared against the prior eight years.

                           % Actions Per Annum 

(Separately, we sought to compare the frequency of these NRSROs’ ratings transitions to those of a smaller non-NRSRO rating agency.  This smaller rating agency rated roughly 3,000 securities in the most recent 12 month-period, and acted on almost 2,500 of them at least once.  Its percentage of securities acted upon per annum: 80.8%.) 

We are seeing other anecdotal evidence of heightened attention being given to post-purchase ratings surveillance:

(1)    Fitch Ratings recently responded to a National Association of Insurance Commissioners (NAIC)-driven request in advertising that “99% of its U.S. Structured Finance ratings have been formally reviewed during the past 12 months as of August 2010.”

(2)    Earlier this week S&P ceased monitoring its ratings on a security called Invicta Credit LLC/Invicta Capital LLC, a credit derivative product company (CDPC).  The reason: “We … withdrew our ratings on Invicta because, in our view, we do not have sufficient information to maintain our ratings …” This may be instrumental if it demonstrates a level of care in the surveillance process.  Given S&P is almost certainly compensated for monitoring its outstanding ratings, their withdrawal shows a willingness to subordinate short-term revenues to the longer-term benefits of ratings quality. 

Whether these new focuses are driven by the legislative or judicial branch ultimately matters not.  Let us only hope that these are preliminary indications of long-term trends that (i) raters are encouraged to be more proactive about monitoring their ratings and/or (ii) raters are incentivized to walk away from rating a security that cannot be adequately estimated by a then-current ratings methodology.  Will ratings agencies start to withdraw ratings on all securities whose assessments depend on underlying data the raters know to be sparse, flawed, incomplete, or fraudulent?  Time, alone, will tell.