By using 2000/2001 as a benchmark, Kanef wanted to mislead Congress and the public into believing that Moody's rated these deals "through the cycle," that the ratings were calibrated according to how the deals would fare during a cyclical downturn. In fact, Moody's, S&P and Fitch did the opposite. They rated through the bubble. In order for their ratings to work, the housing bubble had to continue its upward climb. If home prices experienced the gentlest of soft landings and fell even slightly, hundreds of billions of investment grade mortgage securities got wiped out.
Since time immemorial, real estate lending has been governed by two rules: (1) Location, location, location; and (2) Timing is everything. Simply put, after a loan is booked, the rate of home price appreciation, upward or downward, is the most important factor in determining whether a lender gets his money back. The impact is exponential. That is, when prices are rising at a brisk pace, the risk that a borrower will default is exponentially lower. And if a borrower does default in a rising market, then the rate of loss on any foreclosure sale is exponentially lower.
This is especially true in the subprime sector. Goldman's rule of thumb was that every 5% of annualized home price increase reduced subprime defaults by 6% by the 6th year. In other words, Goldman's data showed if home price appreciation ranged between 10% and 15%, the six-year cumulative default rate was about 5%; if price appreciation ranged 0% and 5%, the default rate was about 18%.
The risk of flat or falling home prices is much higher for private label mortgage securities than it is for bank lenders, who can diversify their market risk. An investor in a private label deal needs to worry buying loans at the peak of the cycle. A bank holds loans that were originated before the peak, and it can profit from new loans booked after the peak. And while private label deals are generally nonrecourse, GSE mortgage bonds have a guarantee from a bank, Fannie or Freddie, which holds loans originated throughout the cycle.
There's another rule about residential real estate. There's no such thing as short cycle. It goes through multi-year periods of feast or famine. If any real estate market experiences a feast of sharp annual price increases, a multi-year correction invariably follows. In 2005, the FDIC published a study titled, "U.S. Home Prices: Does Bust Always Follow Boom?" The answer was, "No," because the FDIC defined a bust as a 15% price decline over a five-year period. But in every boom market, there was a subsequent multi-year famine when prices showed nominal appreciation or some kind of decline. And even though a bust did not always follow a boom, the correlation was very high.
In other words, by 2005 there was a mountain of evidence indicating that future home price the loss rate on newly-booked subprime mortgages would be several times higher than that observed for 1999 and 2000-vintage subprime mortgage deals, which had an average six-year cumulative loss rate of 6%. During the six-year stretch following 2000, the 20-city Case Shiller composite appreciated by 80%. But when Moody's rated subprime deals in 2005 through 2007, the estimated loss was usually 6% or lower. For instance, the 20 deals referenced in the ABX 2006-2 composite had estimated losses ranging from 4.2% to 6.6%, with an average of 5.4%. In other words, Moody's estimated losses to match the loss rate seen during the bubble.
When home prices stopped rising in 2006, subprime delinquencies spiked upward and it quickly became obvious that the lower-rated tranches of subprime bonds would default. But if Moody's were going to issue downgrades, it needed a good explanation, something other than the truth, which was that they screwed up. So management began working on an elaborate strategy to rationalize what they had been doing all along. One of the foundations of this disinformation strategy was to benchmark loan performance based on data observed during the housing bubble, which is what Kanef, the former head of Moody's structured finance, attempted to do in his written testimony.
Ignoring Leverage
The rating agencies found other ways to repudiate the truism that home equity drives loan performance, most notably when they decided that simultaneous second liens didn't matter. From a credit or common sense perspective, this is like saying that the sun sets in the east or that water flows uphill. S&P's press release, dated August 23, 2000, is a real gem of sophistry and doublespeak:
"Standard & Poor's Residential Mortgage group has revised the foreclosure frequency multiples it uses when determining loss coverage amounts for first-lien loans originated with simultaneous second-lien loans (simultaneous seconds)...The new criteria specify that...A first-lien loan with a simultaneous second lien will have its risk grade determined by the LTV of the first-lien loan and not by the CLTV."
In plain English, this means that an 80% first lien loan extended to a borrower who buys his home with a 20% cash down payment is no less risky than an identical loan extended to a borrower who finances the remaining 20% with a simultaneous second lien loan. Does anyone with half a brain really think that the second borrower is no more likely to walk away in case he fell on hard times or if home prices fell? In California and Arizona, a homeowner could walk away and the lender would have no recourse beyond the underlying real estate collateral. Mind you, there has always been a mountain of evidence that borrowers, of any kind, who have an appetite for high leverage also have an appetite for risky behavior that leads to default.
In theory, the second lien lender gets nothing until the first lien lender is fully repaid in a foreclosure sale. But, as Lew Ranieri explained to the deaf ears of rating agency executives, a second lien lender can throw a monkey wrench into any good faith efforts to work out or restructure a problem loan. Which is exactly what's happening now.
How did S&P decide to gut this preexisting credit standard? Based on the selective disclosures of a few lenders during the time that mortgage delinquencies were at a 28-year low. It seems a safe bet that the hypothesis was not tested over the extended life span of a real estate cycle. One of the problems with second lien mortgages was that disclosure was inconsistent, so, in acknowledging that they didn't know what they didn't know, S&P assumed it wasn't a problem.
Consistency That Strains Credulity
Markopolos didn't believe that the Madoff Fund generated returns that were so remarkably consistent over time. The remarkably consistent credit ratings among mortgage securities also raise suspicions. Check out the capital structures and ratings of different Alt-A or subprime deals, and you'll notice that they're all very much alike, almost unbelievably so. With subprime deals, the top 80% is always rated triple-A, the next 10% is rated double-A, the next 6-7% is rated single-A and triple-B, with the remainder unrated. Here's an example of two subprime deals, of a similar vintage, with vastly different risk factors and substantially identical ratings. It sure looks as if the models and methodologies were rigged to come up with the desired results, no matter what the inputs. Moody's didn't use a financial model for rating subprime deals until the bubble had peaked in 2006. Prior to that point, it used a benchmarking system that resulted in standardized cookie cutter results.
MASTR Asset Backed Securities Trust 2005-NC2 was a $903 million deal underwritten by UBS in November 2005. Consider that:
Next Page 1 | 2 | 3 | 4 | 5 | 6 | 7
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).
