Tonko Gast, CIO, Dynamic Credit Partners LLC

Sept. 25 (Bloomberg) — In August 2004, Moodys Corp. unveiled a new credit-rating model that Wall Street banks used to sow the seeds of their own demise. The formula allowed securities firms to sell more top-rated, subprime mortgage-backed bonds than ever before.

A week later, Standard & Poors moved to revise its own methods. An S&P executive urged colleagues to adjust rating requirements for securities backed by commercial properties because of the «threat of losing deals.

The worlds two largest bond-analysis providers repeatedly eased their standards as they pursued profits from structured investment pools sold by their clients, according to company documents, e-mails and interviews with more than 50 Wall Street professionals. It amounted to a «market-share war where criteria were relaxed, says former S&P Managing Director Richard Gugliada.

«I knew it was wrong at the time, says Gugliada, 46, who retired from the McGraw-Hill Cos. subsidiary in 2006 and was interviewed in May near his home in Staten Island, New York. «It was either that or skip the business. That wasnt my mandate. My mandate was to find a way. Find the way.

Wall Street underwrote $3.2 trillion of loans to homebuyers with bad credit and undocumented incomes from 2002 to 2007. Investment banks packaged much of that debt into investment pools that won AAA ratings, the gold standard, from New York-based Moodys and S&P. Flawed grades on securities that later turned to junk now lie at the root of the worst financial crisis since the Great Depression, says economist Joseph Stiglitz.

`Would Have Stopped Flow

«Without these AAA ratings, that would have stopped the flow of money, says Stiglitz, 65, a professor at Columbia University in New York who won the Nobel Prize in 2001 for his analysis of markets with asymmetric information. S&P and Moodys «were trying to please clients, he said. «You not only grade a company but tell it how to get the grade it wants.

Presidential candidates John McCain and Barack Obama lay responsibility for the carnage with Wall Street itself. The Securities and Exchange Commission in July identified S&P and Moodys as accessories, finding they violated internal procedures and improperly managed the conflicts of interest inherent in providing credit ratings to the banks that paid them.

S&P and Moodys earned as much as three times more for grading the most complex of these products, such as the unregulated investment pools known as collateralized debt obligations, as they did from corporate bonds. As homeowners defaulted, the raters have downgraded more than three-quarters of the AAA-rated CDO bonds issued in the last two years.

`Cut Too Many Corners

Facing the threat of lawsuits and tighter regulation, Moodys and S&P now say they are adopting tougher requirements to more accurately evaluate and monitor debt.

«We have made significant progress in achieving these goals, Chris Atkins. S&Ps vice president of communications, wrote last week in a statement to Bloomberg. «Working with policy makers and market participants around the world, we will continue to take steps to meet and exceed the high standards for quality we have put in place. He wouldnt respond to specific questions for this story.

Moodys spokesman Anthony Mirenda declined to comment after Bloomberg submitted questions in writing and by phone.

«The rating agencies models were too flawed and cut too many corners, and the raters got pressured by the bankers, says Tonko Gast, the chief investment officer of the $5.1 billion New York hedge fund Dynamic Credit Partners LLC. He reverse-engineers the raters models as part of his investing strategy.

«Thats how the race to the bottom was kind of invisible for a lot of people, he says.

Favoring Diversity

Starting in 1996, Moodys used a framework known as the binomial expansion technique for rating CDOs, structured funds consisting of aircraft leases, franchise loans, high-yield bonds, hotel mortgages and mutual-fund fees. On the theory that diversification reduced risk, the BET formula rewarded balanced portfolios and punished concentrations of assets, using a proprietary measurement Moodys called the diversity score.

On Aug. 10, 2004, Moodys Managing Director Gary Witt introduced a new CDO rating method that dispensed with the diversity test and made other adjustments to the evaluation of structured-finance products.

«People were just starting to do deals that were all residential, says Witt, 49. He retired from Moodys this year and is now an assistant professor of statistics at Temple Universitys Fox School of Business in Philadelphia. The BET model «is not as well suited for the highly correlated portfolios that were becoming common in 2004, he says.

The emphasis on diversity turned into an obstacle after the 2001 recession, when some assets plummeted in value. Home mortgages, auto loans and credit-card receivables offered higher returns for CDO managers. As mortgage rates fell and the market boomed, investment firms argued the risks in housing were small.

`Moodys Obliged

«I know people lobbied Moodys to accommodate more concentrated residential mortgage risk in CDOs, and Moodys obliged, says Douglas Lucas. 52, the head of CDO research at UBS Securities LLC in New York. The former Moodys analyst says he invented the diversity score in the late 1980s.

A statistical tabulation appearing in the appendix of Witts paper represented the new formula as more rigorous in calculating risks than the BET. A side-by-side comparison showed that the new model projected losses that were 24 to 165 percent higher than forecast by the old, on a hypothetical investment pool.

Bankers «could put together a deal with greater concentrations in one area or another, says Jeremy Gluck. 52, a former Moodys managing director, who worked with Witt at the time.

Fewer Defaults Projected

In September 2005, Witt and colleagues published a follow-up analysis. Compared with the BET, the new model now projected that the likelihood of collateral defaults affecting CDO bonds rated at least Aa could be 73 percent lower at the extreme, in a range of possibilities.

The new comparison was based on a hypothetical investment pool in which 75 percent of the assets were residential mortgage-backed securities, including 30 percent that were subprime.

Moodys could produce a lower default rate by incorporating a decade of ratings stability for structured finance into its assumptions. The average five-year loss rate on U.S. structured finance products between 1993 and 2003 was 1.9 percent, compared with 6.3 percent for corporate bonds, the company had said in September 2004. A drawback was that raters didnt have data going back to the 1920s, as they did on corporate bonds.

In a press release on the report, Moodys said «structured-finance ratings are broadly comparable in quality to the ratings of corporate bonds.

`More Aaas

Philippe Jorion, 53, a finance professor at the University of California, Irvine, criticizes the Moodys decision to factor ratings stability into its evaluations.

«This uses the output of their model as input into their models, Jorion says. «This type of model is totally out of touch with the underlying economic reality.

Witt declined in an e-mail exchange to discuss the September 2005 findings or his earlier projections from August 2004.

«The effect that had on structures was to create more Aaas, says Thomas Priore. 39, chief executive officer of Institutional Credit Partners LLC in New York, which oversees $13 billion of fixed-income investments.

The Moodys share of the market for rating CDOs was falling before the change, to 76.8 percent in 2004 from 91.5 percent a year earlier, according to the industry publication Asset-Backed Alert in Hoboken, New Jersey. It climbed afterward, to 85.1 percent in 2005 and 96.8 percent in 2006. S&P had 97.5 percent that year, the publication said. Underwriters made obtaining a top grade from one or both raters a condition for the sale of the investment pools.

E*Trades CDO

The value of asset-backed CDOs tripled to $30 billion in the fourth quarter of 2004, according to the London-based monthly journal Creditflux. The yearly total increased 87 percent to $104.3 billion in 2005. Subprime mortgages came to account for about half the collateral on all asset-backed CDOs issued that year, according to a Moodys estimate.

In December 2005, New York-based E*Trade Financial Corp. raised $300 million to fund E*Trade ABS CDO IV Ltd. It followed the formula Witt and co-authors outlined in the September paper.

Moodys assigned Aaa grades to three-quarters of the CDOs rated bonds, which invested 73.5 percent of the funds assets in mortgages backed by loans to homeowners with bad credit and limited income documentation. As the subprime market deteriorated, the company in June 2008 lowered $137.3 million of the bonds initially rated Aaa to Baa2 and the rest to speculative.

Gasts Reverse Engineering

Investors began to recognize that Aaa ratings on asset-backed CDOs werent equivalent to top grades on corporate debt. Dynamic Credit Partners Gast, 35, a Dutch-born quantitative analyst, says he began to discern that Aaa ratings werent consistent even between CDOs. He says this demonstrates erosion in the rating companies standards.

«In 05 already what was clear, I think, was that there was a deteriorating underwriting trend, Gast says. «Because we had it all in-house, we were able to figure out: `Wow, you can tweak so many different parameters to come to the same result.

Two CDOs issued two years apart illustrate the point. Both invested in subprime and other mortgage securities, and received AAA ratings on their senior bonds from Moodys and S&P.

In December 2004, NIB Capital Bank NV of The Hague, the Netherlands, and UBS AG of Zurich jointly issued the $1 billion Belle Haven ABS CDO Ltd. The fund manager, an arm of NIB Capital, borrowed 45 times investors equity to buy real estate securities and other assets, according to the prospectus. That magnified potential gains, while also increasing possible losses.

Tale of Two CDOs

The least-protected bondholders were backed by collateral equal to 102.26 percent of their stake, according to the prospectus, providing a cushion against declines.

Two years later, in December 2006, a U.S. arm of the Zurich investment bank Credit Suisse Group AG issued the $1.5 billion McKinley Funding III Ltd. CDO. The manager, New York-based Vertical Capital LLC, borrowed 84 times investors equity, and junior investors were backed by a narrower cushion of assets equal to 100.98 percent of their stake, the prospectus shows.

While Belle Haven could put 20 percent of its assets in other CDOs, further magnifying the risks and returns, McKinley could place twice that percentage.

Both CDOs were downgraded as the subprime market deteriorated, with the earlier CDO holding up better than the later one.

Belle Havens most senior Aaa tranche today retains a Moodys investment-grade rating of A1- and an S&P grade of BBB-. By contrast, Moodys lowered the top tier of the McKinley CDO to junk status, Ca, on September 23. S&Ps rating is CCC-.

Because the funds are registered in the Cayman Islands and dont disclose holdings, it isnt known how much investors may have lost.

S&Ps Model Changes

Meanwhile, S&P tinkered with its methodology for grading CDOs that bought commercial mortgage securities backed by apartments, hotels, offices and stores, according to an Aug. 17, 2004, e-mail obtained by Bloomberg. Managing Director Gale Scott warned of the «threat of losing deals to Moodys unless the company relaxed its rating requirements.

«OK with me to revise criteria, replied Gugliada, then S&Ps top CDO-rating executive, the e-mail exchange shows.

In an interview, Gugliada confirmed the e-mails contents and said it led to one of S&Ps adjustments to accommodate clients. He says Scott did research supporting a relaxation of S&Ps assumptions about how closely correlated the default probabilities were for commercial real estate securities.

More Flexibility

The changes gave S&Ps clients more flexibility. The switch directly preceded «aggressive underwriting and lower credit support in the market for commercial mortgage-backed securities from 2005 to 2007, according to an S&P report that Scott co-wrote in May 2008. This led to growing delinquencies, defaults and losses, the report said.

Scott left in August as S&P cut staff. The company declined to make her available for comment before her departure, and subsequently she couldnt be reached.

Errors sometimes worked their way into the analysis. Kai Gilkes. 40, a former S&P quantitative analyst in London, says he discovered a flaw in the companys main CDO model, the CDO Evaluator, which he updated in late 2005.

In some cases, the S&P system overstated the quality of synthetic CDO Squared securities, Gilkes says. These complex investment pools are based on credit default swaps, a type of insurance against corporate default.

«On collateral rated AA or higher, the S&P model did not properly stress the default behavior of the underlying CDOs, resulting in assets with a lower default probability than their ratings suggested, Gilkes says.

`Line in Sand Shifts

He says he fixed the glitch during «a major revision that December and doesnt know whether any investment was inappropriately rated as a result of the error.

Still, Gilkes says he believed that competitive considerations, as communicated by management, intruded on S&Ps ratings decisions up until he left the London office in 2006.

«The discussion tends to proceed in this sort of way, he says. «`Look, I know youre not comfortable with such and such assumption, but apparently Moodys are even lower, and, if thats the only thing that is standing between rating this deal and not rating this deal, are we really hung up on that assumption? You dont have infinite data. Nothing is perfect. So the line in the sand shifts and shifts, and can shift quite a bit.

`Golden Goose

Gugliada says that when the subject came up of tightening S&Ps criteria, the co-director of CDO ratings, David Tesher. said: «Dont kill the golden goose.

S&P declined to make Tesher available for comment.

In the SECs July 8 report examining the role of the credit rating companies in the subprime crisis, the agency raised questions about the accuracy of grades on structured-finance products and «the integrity of the ratings process as a whole.

«Lets hope we are all wealthy and retired by the time this house of cards falters, one unidentified analyst told a colleague in a December 2006 e-mail, according to the SEC report. The e-mail was signed with a computerized wink and smile: «;o).

Moodys stock peaked at $74.84 on Feb. 8, 2007, a day after London-based HSBC Holdings Plc said it would set aside about $10.56 billion for losses on U.S. home loans. That statement was among the first signs of the subprime meltdown.

The reckoning swept Wall Street five months later. On July 10, Moodys cut its grades on $5.2 billion in subprime-backed CDOs. That same day, S&P said it was considering reductions on $12 billion of residential mortgage-backed securities.

More Aaas

Still, they continued stamping out AAA ratings.

Moodys announced Aaa grades on at least $12.7 billion of new CDOs in the last week of August 2007. Five of the investments were lowered by one or both companies within three months. The rest were cut within six months.

«The greed of Wall Street knows no bounds, says Stiglitz, the Nobel laureate. «They cheated on their models. But even without the cheating, their models were bad.

By last month, Moodys had downgraded 90 percent of all asset-backed CDO investments issued in 2006 and 2007, including 85 percent of the debt originally rated Aaa, according to Lucas at UBS Securities. S&P has reduced 84 percent of the CDO tranches it rated, including 76 percent of all AAAs.

«Credit in Latin means `to believe, says former Moodys analyst Sylvain Raynes. 50, now the head of his own New York bond-analysis firm, R&R Consulting. «Trust and credit is the same word. If you lose that confidence, you lose everything, because that confidence is the way Wall Street spells God.

(Failing Grades on Wall Street: Part 2 of 2.)

To contact the editor responsible for this story: Robert Simison at rsimison@bloomberg.net.