McGraw-Hill did this fully aware that, in handing over some twenty million pages of e-mails to government prosecutors, it presented them with some pretty damaging stuff, including a number of messages in which S. & P. employees criticize the firm’s rating process and seemingly acknowledge that it issued generous ratings to please its customers—the Wall Street banks that issued subprime bonds—and avoid losing market share to rivals Moody’s and Fitch. In one instant message, turned over by the company and included in the government’s complaint, an S. & P. analyst says the firm would have given a good rating to a deal put together by cows.
If the case does go before a judge—the two sides could still settle before trial—S. & P. will have the opportunity to place the offending communications in context, and to counterbalance them with more exculpatory materials. But the firm also faces the risk that some of its employees, and former employees, will say incriminating things on the witness stand, and that the prosecutors could be holding even more embarrassing exhibits in reserve. The ones contained in the one-hundred-and-nineteen-page complaint are bad enough, although they don’t necessarily ensure a government victory in the case.
The case revolves around the activities of two teams at S. & P., one that rated residential-mortgage-backed securities (R.M.B.S.), and another that rated collateralized debt obligations (C.D.O.’s). The underlying collateral for both of these types of securities was largely the same: subprime mortgages taken out by borrowers who didn’t have a high enough credit rating to get a normal mortgage. The Wall Street firms that put together the securities paid S. & P. hefty fees to rate them: one hundred and fifty thousand dollars for a R.M.B.S., five hundred thousand dollars for a regular C.D.O. based on actual mortgages, and seven hundred and fifty thousand dollars for a fancy (“synthetic”) C.D.O. based on derivatives tied to the mortgages. It was a big business. Between 2005 and 2007, S. & P.’s C.D.O. division alone generated four hundred and sixty-one million dollars in revenue.
S. & P. claimed all of its ratings were independent and made in good faith, based on sound reasoning. The government claims otherwise. In particular, it makes two charges. First, between 2004 and 2007, S. & P. deliberately “limited, adjusted, and delayed” changes to its statistical models and ratings criteria for subprime securities that would have resulted in the firm issuing lower ratings. Second, between March and October of 2007, when it was perfectly clear that the housing bubble had burst, the firm “knowingly disregarded the true extent of the credit risks associated with” securities it rated.
Most of the headline-grabbing e-mails and messages come from that period in 2007, when S. & P.’s employees were tussling with data showing an “unprecedented deterioration” in the subprime-mortgage market, with loan delinquencies rising sharply. The numbers were so bad, the complaint says, that some S. & P. “analysts initially thought the data contained typographical errors.” But rather than immediately issuing a mass downgrade of lower quality mortgage bonds, which would have had a devastating impact on the C.D.O. ratings as well, S. & P. reacted slowly and cautiously, placing just a few products on a credit watch for a possible downgrade.
Internally, some of the firm’s employees were rather more explicit in their views. In March of 2007, a person identified as “Analyst D,” who had studied a series of R.M.B.S. issued in 2006, wrote an e-mail to several colleagues with the subject line “Burning Down the House—Talking Heads.” This is what it said:
Watch out
Housing market went softer
Cooling down
Strong market is now much weaker
Subprime is boi-ling o-ver
Bringing down the house
Hold tight
CDO biz—has a bother
Hold tight
Leveraged CDOs they were after
Going—all the way down, with
Subprime mortgages
Own it
Hey you need a downgrade now
Free-mont
Huge delinquencies hit it now
Two-thousand-and-six-vintage
Bringing down the house
A few minutes after sending this message, Analyst D followed up with another one:
For obvious professional reasons, please do not forward this song. If you are interested, I can sing it in your cube.
That same month, March, 2007, S. & P. gave investment-grade ratings to sixty-one C.D.O.’s valued at fifty-one billion dollars. In April, it rated another forty-seven C.D.O.’s valued at twenty-four billion dollars. With big investment banks like Bear Stearns and Merrill Lynch desperately trying to unload all the junk mortgages they had taken onto their balance sheets, the ratings agency was doing record business.
On April 5, 2007 two C.D.O. analysts at S. & P. had an instant-message exchange about one of the subprime issues the firm had given a sound rating, and the model used to justify that action.
Analyst 1: btw that deal is ridiculous
Analyst 2: I know right…model def does not capture half of the…risk
Analyst 1: We should not be rating it.
Analyst 2: we rate every deal…it could be structured by cows and we would rate it.
In July, 2007, following a lot of criticism in the media, S. & P. finally announced a mass downgrade. Even then, though, it continued to give investment-grade ratings to some new C.D.O.’s. On July 5th, an analyst, who was a recent hire with the firm’s structured-finance group, wrote to an investment-banking client:
The fact is, there was a lot of internal pressure in S&P to downgrade lots of deals earlier on before this thing started blowing up. But the leadership was concerned about p*ssing off too many clients and jumping the gun ahead of Fitch and Moody’s.
The other batch of e-mails and documents concern the intellectual criteria that S. & P. employed in its ratings decisions. During 2004 and 2005, there was an internal debate about whether to update the principal model the company used to analyze R.M.B.S., a mathematical gizmo called the “Loan Evaluation & Estimate of Loss System” or “LEVELS.” Armed with new data, some of S. & P.’s quantitative analysts had produced a new version of the LEVELS model that would have required Wall Street issuers to beef up the “loss-coverage enhancements” in their R.M.B.S. Such a change would have made the mortgage bonds safer, but it also would have made issuing them less profitable for the banks. Some S. & P. analysts who worked on the actual ratings—as opposed to the methodology used to create them—were against to the change; they thought that S. & P.’s methods were already too conservative. In May, 2004, an analyst sent an e-mail to her bosses:
We just lost a huge Mizuho RMBS deal to Moody’s due to a huge difference in the required credit support level. What we found from the arranger was that our support level was at least 10% higher than Moody’s…this is so significant it could have an impact on future deals. There’s no way we can get back on this one but we need to address this now in preparation for future deals.
According to the government, the new, tougher, version of LEVELS “was never released by S. & P.” Instead, the ratings agency updated its older model in a way that didn’t increase the level of loss coverage that Wall Street issuers were required to include in their subprime deals. The decision not to roll out the new model didn’t please everybody inside the firm. In March, 2005, a senior analyst wrote:
Version 6.0 could’ve been released months ago and resources assigned elsewhere if we didn’t have to massage the sub-prime and Alt-A numbers to preserve market share.
Inside the C.D.O. department, there were similar issues related to a proposed upgrade of its statistical model, the “C.D.O. Evaluator.” Within the team, there were concerns that a new version of the model, known as “E3,” would generate lower ratings for subprime C.D.O.’s, which would obviously displease the banks that issued them. Some S. & P. executives and analysts considered the possibility of discussing the proposed changes to the model with the Wall Street firms. In a February, 2005, e-mail, Andrea Bryan, who headed the ratings team for synthetic C.D.O.’s, wrote:
So we may have to put this beta model in the hands of a few trusted souls and let them help us understand their tolerance level.
A few months later, an S. & P. analyst also suggested that “the only way I can see to move this forward” was to run the proposed changes past some Wall Street bankers, but, he acknowledged, “this looks too much to me as if we are publicly back into a set of levels driven by our clients.” A London-based senior analyst joined the debate:
Remember the dream of being able to defend the model with sound empirical research? The sort of work a true quant…should be doing perhaps. If we are just going to make it up in order to rate deals, then quants are of precious little value. I still believe that people want the model to be consistent with history, and that the impact of the model will not destroy the business. If I’m wrong, then so be it.
Throughout the summer of 2005, S. & P. beta-tested the new version of the model. Bear Stearns, which was a big player in the subprime C.D.O. market, including the then-fast-growing synthetic C.D.O. market, gave it a particularly negative reaction, prompting S. & P. to slow down its rollout. On July 20, 2005, Patrice Jordan, the head of the global C.D.O. group, reported this development to his boss, Joanne Rose, who was in charge of rating all structured products.
Due to the not insignificant impact on lowly rated (BBB and down) synthetic reference pools…we have toned down and slowed down our roll out of E3 to the market, pending further measures to deal with such negative results.
These further measures included the development of a less demanding alternative to E3 called “E3Low,” which, according to the government, wasn’t based on any historical research or analytical data. “Rather,” the complaint claims, “the rationale behind this weaker model was to preserve S&P’s market share.” In December, 2005, S. & P. finally introduced this hybrid model, instructing its analysts as follows:
If the transaction passes E3.0, GREAT!! The deal is modeled, rated, and surveilled with E3.0…If the transaction failed E3.0, then use E3Low.
From S. & P.’s perspective, obviously, none of this reads well. But bad publicity doesn’t necessarily equate to a defeat in court. To prove its case, the government will have to persuade a jury that the firm deliberately misled the purchasers of sub-prime mortgages, and that it wasn’t merely caught up, along with its rivals, in the euphoria of a bubble. Rather than castigating the firm as a whole, prosecutors will have to show that some specific ratings were wrong, and that the analysts who issued them knew it, or should have. S. & P., for its part, will surely argue that those ratings were no different than the ones issued by Moody’s and Fitch, which haven’t been charged with any wrongdoing.
As with most financial-fraud cases, the outcome is tough to predict. (I, for one, wouldn’t dismiss S. & P.’s chances of beating the rap.) For today, though, let’s applaud the Justice Department. Ever since the financial crisis of 2007–08, politicians and commentators have been calling on the authorities to hold the ratings agencies accountable for their shameful behavior during the boom. Five years on, the government has finally acted, making creative use of some anti-fraud legislation that Congress passed in wake of the Savings & Loan crisis. That’s just the sort of initiative the D.O.J.’s critics have been calling for. To be sure, in bringing the case Justice is taking a risk. So be it.
Original Article
Source: newyorker.com
Author: John Cassidy
No comments:
Post a Comment