As the credit rating agencies (CRAs) begin to rebuild their brand, market experts are questioning whether the changes these firms are implementing are tackling the key problems that have caused rating inaccuracies.
The failure of the CRAs and of the ratings they assigned has been blamed for the difficulties plaguing the economy today. This failure has created a buyer's market that has no confidence in either the credibility of the CRAs or the reliability of the U.S. economy.
However, CRAs believe they emerged as easier targets for people to criticize, and rating analysts argue that no one could have had models in place there years ago that could have predicted such unprecedented volatility in certain segments of the securitization market.
"What we experienced in U.S. RMBS was at least three to four times worse than the scenario we predicted," said Diane Westerback, managing director of structured finance at Standard & Poor's. "We were certainly one of many players in the industry that did not predict this."
Westerback believes that S&P is in a much better position today. "We are running portfolios more frequently, more diligently and more deeply than before, and we can execute a much faster analysis on the whole universe of bonds."
S&P published refined methodologies and assumptions for rating U.S. RMBS transactions backed by prime, Alt-A and subprime mortgages. The criteria changes are a significant update to previous methodologies and assumptions for determining credit enhancement levels. "The core of our approach was to establish an 'AAA' credit enhancement level that, in our view, should be sufficient to enable securities rated at that category to withstand an extreme economic downturn without defaulting," said Francis Parisi, a managing director in the structured finance ratings group at S&P.
The updated criteria define an "archetypical pool" and its associated credit enhancement level at the 'AAA' rating category. The analysis under the revised criteria arrived at a 7.5% 'AAA' credit enhancement level for the archetypical prime pool.
The 7.5% 'AAA' credit enhancement level for the archetypical prime pool is a significant increase over the credit enhancement levels under S&P's previous criteria and is primarily derived from S&P's assessment of potential borrower default behavior and property value declines under conditions of extreme stress, such as during the Great Depression.
"This represents a major recalibration of our RMBS criteria and is intended to make U.S. RMBS ratings more comparable with ratings in other sectors, such as corporates, municipals, sovereigns and other areas of structured finance," Parisi said.
Parisi added that the changes are a step in the right direction toward restoring investor confidence.
"We have tightened our standards for triple-A ratings on pools of mortgage loans and, in general, these pools will need higher credit enhancement levels than in the past to achieve the same ratings," he said. "From the market's point of view, the general assumption is that prime RMBS will be the first to come back, and from our side we are seeing loan pools with very pristine loan characteristics and a higher standard of underwriting."
The agency has also recently published revised methodologies and assumptions it uses to rate global CDOs backed by corporate debt.
The most notable update in the criteria is the addition of qualitative and quantitative tests that supplement the default simulation model used in S&P's portfolio analysis.
The updated criteria also adjusted the asset default rates, correlation, recovery and other model parameters to produce asset portfolio default and loss results commensurate with the agency's rating definitions.
"We believe that adding quantitative and qualitative elements to our analysis - entirely apart from the Monte Carlo default simulations we run - will provide a more robust analysis than using only simulation models," said Henry Albulescu, global criteria officer, structured credit at S&P. "By achieving specific targeted portfolio default rates in the CDO Evaluator, we have made it easier and more transparent for investors to understand our ratings and analyses."
Another important update to the criteria is that a review of a corporate CDO transaction now also includes sensitivity to model parameters. This tests what the effects would be on a CDO tranche's ratings if the rating agency changes four key portfolio parameters: correlation, recovery, spreads and default bias. This aspect of the criteria is intended to assess whether the model results and transaction structure display high sensitivity to changes in input parameters.
Analysts said that the revision represents a significant recalibration of the agency's CDO criteria that will enhance the comparability of CDO ratings with ratings in other sectors.
S&P said it expected that the criteria update would result in downgrades to many rated corporate CDO tranches. Globally, the agency anticipated it will place on CreditWatch negative approximately 4,790 public rated corporate CDO tranches.
"Our preliminary estimates are that outstanding synthetic CDOs will likely experience an average downgrade of 4 notches," said Thomas Gillis, chief credit officer for S&P. He added that structured finance ratings super senior 'AAA' tranches of such transactions will probably be affected less, with an estimated average downgrade in the 2-3 notch range. Tranches rated 'AAA' will likely be affected more, with an estimated downgrade of 4-5 notches.
"Outstanding cash flow CDOs will likely experience an average downgrade of 3 notches," Gillis said. "The senior-most 'AAA' rated cash flow tranches that are above other junior 'AAA' rated tranches will likely be affected less, with an estimated average downgrade of 1-2 notches."
Westerback also pointed to the efforts the agency has made on developing new approaches that give investors better depth into transaction performance. She believes that that recovery ratings/analytics will go a long way to instill confidence in ratings again because they provide a level of transparency that helps investors better understand what they are holding.
"Recovery ratings are based on the same criteria used for our letter ratings, which include economic factors and deal level factors," Westerback said. However, with the recovery analytics we will have updates every month as new performance data comes in. The rating agency will be calculating the recovery ratings/analytics, for the most part, on loan level assumptions and flow these results through the current bond waterfall. Westerback said that they apply the same criteria used to rate the bond but focus on pulling out the principle recovery of the bond.
Fitch's Side of Things
Fitch Ratings already had in place a system for recovery ratings, but the current environment has given the agency more data to work into the system.
"When the problems started to appear in 2007, it became clear that the ratings had underestimated the impact of certain types of mortgages," said Huxley Somerville, group managing director and U.S. RMBS group head at Fitch. "We recognized that higher loan-to-value mortgages or no-documentation loans were more risky, but we underestimated the impact that those loans could have on the deal."
Updated criteria since 2006 didn't mean much until the triple-As were downgraded. The ratings don't tell how much loss will be achieved; they just tell you when you have the first dollar of loss. Recovery ratings look at any bond that is distressed in the base case and assume that those will suffer a loss in the future.
Recovery ratings also provide an indication of loss you can expect. "A lot of former triple-As were downgraded because we expected even just a dollar of loss, but in some cases you can expect recovery up to 90%," Somerville said. "We looked at the criteria used 22,000 recovery ratings that provide investors with a fuller picture and also introduced loss severity ratings that measure what the loss would be like if the bond defaulted. Triple-A, for example, is not expected to default, but for the mezzanine tranche it could be a different situation, and if the bond were to default, what cover or protection is there?"
However, Fitch has also undertaken a recalibration of its methodologies. Somerville said that in RMBS Fitch continues to use the same model but better attuned to the performance of those weak attributes. "We have been able to understand how those hazards work so that the new ratings are much stronger," he explained.
For example, second lien mortgages are given a higher default probability. "So basically we are rating transactions in the same way, but we have made the adjustments for the weaker criteria, including probability of default," Somerville said. "But we are talking about only two areas where it's been stressed - otherwise our ratings have held up well - and that is U.S. RMBS and CDO. To be fair, those two have performed miserably."
Somerville said that the agency is looking more at loss metrics, measures that are free to investors and provide them with a bond-by-bond, deal-by-deal analysis breakdown for prime, Alt-A and subprime, which provides transparency on bond level and deal level.
Putting Your Money Where Your Method Is
The rating reform measures proposed so far tackle only issues peripheral to the key problems that caused rating inaccuracies and the widespread dependencies on these ratings, according to PF2 Securities Evaluations in a recent special report titled First Steps Toward Real Rating Agency Reform: Knowing Where We Need to Go.
"Overall we do feel that changes have been made and improvements have been made that make rating methodologies more transparent, but the changes have fallen short of where they need to be," said Gene Phillips, a director at PF2. "Both Fitch and [Moody's Investors Service] have become more explanatory about the individual components of each rating - for example, the bifurcation of some expected loss ratings into its separate components of recovery rate and default probability. You used to get just one piece of the puzzle, but now we're getting more."
But Phillips believes that the reform measures that have been implemented thus far won't be enough to restore investor confidence.
"We still aren't getting where we need to be. On CDOs secured by RMBS, for example, rating agencies will continue to make assumptions based largely on their own ratings of RMBS securities, which remain poor estimates (they continue to be downgraded; they don't carry much historical significance or basis for recovery rate or correlation assumptions; and they don't provide a prepayment or amortization schedule, so the rating agencies cannot rely on their internal analysis to complete the analysis - a major failing point)," he said. "These were seen as a huge failure at the beginning of the crisis and are still a problem."
Phillips said that the rating agencies need to start at the most fundamental level with the basic product. and that is the mortgage itself. The assumption needs to be made on the performance of the mortgages and not on the RMBS; otherwise the rating isn't all the way through the CDO structure and will lose out on concepts such as payment timing, or loss timing, which are crucial to the analysis of the CDO.
However, a move away from the current model might be too hard to execute. "The correlation assumption on the RMBS was always fully described but the methodology was never fully explained, and where it used to be too lax, it is now too conservative," he said. "But the thing is that the rating agencies really have no right to make these assumptions because they are based on a very limited history of defaults and recoveries on RMBS. Essentially, it prevents the agencies from making an informed decision."
Phillips added that, while the CRAs have recalibrated models to provide the recovery rate and correlation on RMBS, it is done so without good supporting data. "The rating agencies basically don't understand the payouts on these RMBS," he said. "They are relying on amortization, which is not supported by internal analysis."
Regulating CRAs from the Outside
The Securities and Exchange Commission (SEC) last month set out proposals to make credit agencies more accountable for their credit opinions that would motivate credit rating agencies to get credit ratings right and would provide incentives beyond monetary compensation to be accurate with ratings going forward.
Phillips believes that creating a "one stop shop" where individually a rating agency could rate all sort of deals maybe wasn't the best approach. "In some cases the Wal-Mart approach is the best solution," he said. "If you can't do the business, turn it down. If you don't have the tools in place or the necessary levels of sophistication to rate a security, you really need to stop rating that security. Unless, of course, there's no liability for being wrong, in which case, well, why not rate everything 'AAA' and hope for the best?
"Even if you managed to convince yourself upfront that your methodology was sound, upon realizing its flaws you've really either got to adopt a new methodology or stop rating the asset class," he said. "That's not what we've seen; what we've seen is a continuation of rating the same asset class with new, differently flawed assumptions, but flawed potentially in the opposite direction, becoming too punitive or conservative where they were previously too lax.
Phillips argued that the market continues to have a major problem with a rating agency's stamping its opinion on something it has inadequately evaluated - something, he said, it knows it should not be rating.
"There needs to be an incentive for the rating agency to get it right," he added. Phillips cited Arturo Cifuentes, who was formerly managing director for the structured finance department at R.W. Pressprich & Co., who said, "I think we have moved past the day in which ratings were purely an opinion - the economy is carefully hinged on ratings, and if people are paying for ratings and we're relying on them for risk measurements and capital adequacy ratios, then they become more than simply opinions."
Phillips proposes a more radical approach and argues that investor confidence might be further propelled if the market witnessed one of the CRAs go under for providing inaccurate ratings. "Unfortunately, any lawsuit that may bring one under may in fact bring all of the "Big 3" under," he said. "But the concept remains that if investors notice CRAs going under for providing inaccurate ratings, the investors may have better reason to believe that the ratings they're getting from the surviving CRAs are reliable (since CRAs may disappear - a punishment - for being inaccurate)."
The SEC might require banks and other issuers to disclose preliminary ratings to prevent them from shopping around for better ratings, and it is also considering requiring credit agencies to reveal more information about past ratings so that investors could compare their relative performance.
The regulator is also expected to issue a general discussion paper that questions whether credit agencies should be regulated as "experts" under securities law, and thus subject to tougher standards of liability.
And the SEC isn't the only regulatory body looking to make rating agencies more accountable. Since April, Japan has also been investigating how credit rating firms determine risk levels for financial products.
The Australian Securities & Investments Commission (ASIC) has also recently said that it will require credit rating agencies to obtain an Australian Financial Services license starting January 2010. The ASIC is also considering, according to market reports, lifting a rule that prevents S&P, Moody's and Fitch from being held accountable for their ratings in product disclosure documents.
In Europe, Much of the Same View
The Bank for International Settlements (BIS) published an article called The future of securitization: how to align incentives? In the paper, the BIS admitted to its part in creating an overreliance on the ratings process under requirements set forth in Basel II and CRD where the risk-weighting depends strongly on ratings.
In the article, the BIS recommends that, first, regulation no longer be based on ratings and, second, the quality of ratings needs to be improved. "Of course, we do not see how banking regulation (Basel II, the CRD) could be rewritten so that it is not ratings-based in the short or medium term," Reto Bachman, co-head of European ABS research at Barclays Capital said. "If a non-ratings-based approach is feasible, it would seem to be a project for Basel III."
The BIS noted that rating agencies have responded to regulatory pressure but said that some of the measures "suggested" by the rating agencies are largely gimmicks, such as including statements in credit rating opinions to the effect that credit ratings reflect credit risk and not other risks, or giving ABS credit ratings different labels from those used for corporate credit ratings.
Better ideas include composite ratings, with the first component reflecting an expectation (such as the loss probability or the expected loss, as credit ratings have traditionally done) and a second component reflecting the uncertainty regarding this expectation (as suggested by this author long ago).
But the BIS agrees that the main result of the pressure on rating agencies has been harsher rating methodologies for ABS bonds, contributing to a flood of downgrades.
"The consequences have been twofold: with the methodologies in flux, it has become more difficult to interpret bond ratings and rating changes, as current ratings are no longer directly comparable with earlier ratings; and, because Basel II and the CRD are ratings-based, investing banks have been forced to hold more regulatory capital against their ABS investments in the middle of a recession," Bachman said. "Hence, the methodology changes have made Basel II even more pro-cyclical than it already was."
(c) 2009 Asset Securitization Report and SourceMedia, Inc. All Rights Reserved.