The Credit Crunch: Rating Agencies in Crisis

Book: Structured Finance and Collateralized Debt Obligations | Author: Janet M. Tavakoli | Publisher: John Wiley & Sons (2008) | ISBN: 978-0-470-44344-6

Chapter 17 is where the whole book comes home. Everything Tavakoli warned about in earlier chapters – the junk science ratings, the super senior opacity, the correlation delusion, the underpriced risk in monoline guarantees – all of it landed at once in 2007 and 2008.

Rating Agencies, Regulators, and Junk Science

Tavakoli was writing about rating agency problems years before they became front-page news. In February 2007, she submitted a letter to the SEC recommending that the NRSRO designation be revoked from Moody’s, S&P, and Fitch with respect to structured finance products. Her letter was the first one posted on the SEC’s website in that comment round.

Her critique is precise. Statistics is the mathematical study of probability. You have to understand the character of the data before applying statistical methods to it. Rating agencies were applying historical corporate default data to new mortgage products with completely different risk characteristics. They were accepting issuer-provided information without independent verification. They were rating deals backed by stated-income loans – where borrowers claimed income with zero verification – without asking for any sampling of whether those income claims were accurate.

Tavakoli calls this combination “financial astrology, financial alchemy, and financial phrenology.” Using stale data is astrology. Setting subordination requirements without sound foundations is alchemy. Evaluating CDO managers without proper background checks is phrenology.

When challenged, rating agencies claimed journalist-like privileges in legal proceedings to avoid disclosing their analysis. Their defense was essentially: we only offer opinions. Tavakoli’s response: opinions based on junk science are still junk.

Savvy Investors Ignored Ratings

John Calamos Sr., whose funds have won 30 Lipper awards, told the University of Chicago Finance Roundtable in 2007 that his funds initially used Moody’s and S&P benchmarks. They “got smoked a couple of quarters.” After that they built their own credit models and stopped using rating agency benchmarks entirely.

His experience was common among the best investors. The hedge funds that shorted the ABX indexes in early 2006 were using their own analysis, not S&P’s models. They knew the underlying loans were bad. The ratings said otherwise.

Misfortune’s Formula

Tavakoli gives this name to the compounding effect of securitization layering. Flawed mortgage loans got securitized into RMBSs with flawed ratings. Those flawed tranches got used as collateral in CDOs, also with flawed ratings. Those CDO tranches got used as reference assets in synthetic CDO-squareds. Each layer amplified the errors in the underlying data.

The hybrid CDOs backed by subprime RMBSs showed substantial principal risk from the moment they closed. Tavakoli argues the rating agencies should have known this from the start – they had all the information they needed. They chose not to look carefully because looking carefully would have slowed down the fee-generating machine.

The ABCP Crisis and MLEC

By August 2007, the asset-backed commercial paper market began showing signs of stress. Investors wouldn’t roll their paper in SIVs. Treasury Secretary Hank Paulson announced a potential bailout vehicle called the Master Liquidity Enhancement Conduit (MLEC). The premature announcement froze the market while everyone waited to see what it would contain.

When it became clear that MLEC could not guarantee it would be free of contaminated CDO product, even well-run SIVs like Sigma and Theta were tainted by association. The MLEC idea was eventually abandoned. Citigroup, Bank of America, and JPMorgan Chase quietly shelved it. Banks began bringing SIV assets back on balance sheet.

Constellation CDOs: Falling Stars

By October 2007, CDOs with astronomical names – Carina, Sagittarius, and others – began hitting events of default. These constellation CDOs had been driven largely by one hedge fund doing the classic long equity/short mezzanine trade, with senior tranches wrapped by monoline insurers.

Adams Square Funding I is the clearest example. It closed December 15, 2006. In March 2007, the rating agencies defended their subprime CDO ratings publicly, saying investment-grade tranches would “mostly stay that way.” By December 2007 – less than 12 months after closing – Adams Square Funding I unwound. Investors got back less than 25 cents on the dollar. More than a 75 percent loss.

By the end of December 2007, 33 CDOs had issued event of default notices. The list was still growing.

New Flawed Models Replace Old Flawed Models

Financial institutions began building new models to assess their CDO exposures. Tavakoli observes that most of the new models were as worthless as the old ones because they didn’t capture deal-specific risks, didn’t account for moral hazard in portfolio management, and still relied on assumptions about recovery rates that were far too optimistic.

The prospectus for Adams Square Funding I actually stated in plain language: “Reliable sources of statistical information do not exist with respect to the default rates for all of the type of securities represented by the Collateral Debt Securities.” The deal was rated by both Moody’s and S&P. Both downgraded it several notches within a year.

Monoline Meltdown

By end of 2007, five of seven AAA-rated monoline insurers did not merit their ratings.

Tavakoli walked through each one using her own loss assumptions, which were significantly more pessimistic than the rating agencies':

• ACA: Capital cushion of $650 to $700 million against an estimated $3.5 billion needed just to reclaim single-A. The Maryland Insurance Commissioner signed a consent order. Inevitable collapse.
• FGIC: $300 to $350 million cushion against an estimated $4 billion needed for triple-A. Insured $315 billion in bonds.
• MBIA: $1.75 billion cushion against an estimated $3.5 billion needed (even after a proposed $1 billion infusion from Warburg Pincus). MBIA insured more than $2 trillion in securities.
• Ambac: $1.55 billion cushion against an estimated $2.1 billion additional needed.
• XL/SCA: Similar situation.

Fitch gave Ambac, MBIA, and FGIC more than a month to raise $1 billion in capital – Tavakoli considered this entirely insufficient to merit even a retained single-A rating, let alone triple-A.

By spring 2008, FGIC and XL had been downgraded below investment grade by at least one agency. They were in legal disputes with counterparty investment banks.

Overwhelming Losses and Poor Recoveries

Tavakoli’s subprime loss projections from early 2007: $270 billion to $340 billion for the approximately $1.5 trillion subprime market. Including all risky products, Alt-A and option ARMs included, her loss estimate ran to $450 billion to $560 billion.

In August 2007, she appeared on CNBC’s Squawk Box and stated these numbers publicly. At the time, Fed Chairman Bernanke said subprime losses would be $50 to $100 billion. Citigroup’s estimate was $100 billion. CSFB said $50 billion.

She told viewers the Fed had to stop letting Wall Street do its homework.

A major mortgage servicer told her that her recovery rate assumption of 30 percent was still too optimistic. The servicer was selling delinquent loans for three to six cents on the dollar. They were re-aging mortgages and skipping delinquency reporting to avoid triggering foreclosures, because actual foreclosures produced nearly zero recovery.

By January 2008, Wall Street had already reported more than $130 billion in losses. Most ARM resets hadn’t happened yet. Losses at hedge funds, insurance companies, and pension funds were not included in that figure.

Countrywide’s Bailout and Moral Hazard

On August 16, 2007, Countrywide drew down $11.5 billion from a 40-bank syndicate. The stock market dropped more than 340 points. Later that day, information about Fed action was apparently leaked; the market recovered to close down only 15 points.

The next day, the Fed cut the discount rate 50 basis points and extended borrowing terms to 30 days, accepting triple-A asset-backed collateral. The Fed effectively became a buyer of 30-day extendable Countrywide commercial paper. Countrywide, the largest U.S. mortgage lender, was deemed too big to fail.

In January 2008, Bank of America announced a $4 billion all-stock acquisition of Countrywide. They planned to drop the Countrywide name after closing.

Tavakoli’s conclusion on moral hazard: when a company that made catastrophically bad loans at massive scale gets bailed out by the central bank and then acquired at a price that still transfers its losses to a buyer, the lesson the market learns is that scale protects you from consequences.

Previous: Credit Funds and Hedge Funds in Structured Credit

Next: The Future of Structured Finance: Regulatory Failure and What Comes Next