Inside ACE Securities’ HEL Trust, Series 2005-HE5
The nation is running out of magazine covers on which to announce the coming collapse of house prices. From which fact it could be inferred that Mr. Market is running out of sellers of the statistically cheap housing stocks. Is there even one surviving bull on Toll Brothers or Countrywide Financial or New Century Financial Corp. who doesn't know that the house-price bubble has burst?
Maybe not. But the news has strangely failed to register in the mortgage-backed securities market. For the buyers of CDOs, HEL trusts, RMBS and every other alphabetic variation on the words "mortgage debt," the year might as well be 2004, not 2006. As far as the bond bulls seem to know, house prices are still climbing, homeowners are still painlessly extracting cash from their bricks and granite countertops, and foreclosures are just a tiny cloud in an otherwise clear blue sky. The worse the news from the home front, the closer mortgage yields seem to hug the Treasury yield curve--and the more determined the bidding by Wall Street's asset-backed securities mills for First Franklin, Saxon Capital and the other mortgage originators lately put on the auction block. (The world returned to its desk after the Labor Day weekend to discover that Merrill Lynch had agreed to buy National City Corp.'s home-mortgage subsidiary for $1.3 billion.)
This paradox is the subject at hand. Our approach is at once bottom-up and top-down: a clinical examination of the mortgage security named in the headline as well as a review of the micro and macro forces that have contributed to its stunning overvaluation. Now the cat's out of the bag. "Overvalued," we, in fact, judge trillions of dollars of asset-backed securities and collateralized debt obligations to be, and we are bearish on them. Housing-related stocks may or may not be prospectively cheap; they at least look historically cheap. But housing-related debt is cheap by no standard of value. For institutional investors equipped to deal in credit default swaps, there's an opportunity to lay down a low-cost bearish bet.
The sheer volume of issuance of non-Fannie and non-Freddie residential mortgage-backed securities may surprise you. In the first six months of this year, $303 billion was minted vs. $490 billion in all of 2005. As recently as 2000, such issuance totaled a mere $58.5 billion. If you've guessed that there's money to be made in the creation and distribution of these mortgage conflations, you're well on your way to penetrating the mystery of why the Bloomberg/Bear Stearns Home Equity HELOC Index is trading at the tightest spread to the Treasury curve in the past 10 years (for ocular evidence, see page 2).
A Moody's managing director, John Kriz, helped to sort things out in a recent article in the American Banker. Why, he was asked, is the value of M&A activity in mortgage-origination businesses on its way to hitting a decade high? Why are Wall Street's best and brightest so keen to own the companies that lend against the no-longer gold-plated collateral of residential real estate?
"If you have a significant distribution platform," replied Kriz, "there are many things you can do to move those assets--through securitizations and outright resale, among other things. What you need is product to feed the machine." This machine is one of Wall Street's most treasured. It processes mortgages into asset-backed securities and ABS tranches into collateralized debt obligations and CDO tranches into CDOs squared (a CDO squared is, of course, a CDO of a CDO). It is a wondrous kind of machine that spits out fees for its owners at every step of the manufacturing process.
Last month, Reuters took note of the burgeoning sale of home equity loans packaged as asset-backed securities. The story quoted a practitioner who ascribed the surge to a parallel boom in the issuance of a kind of mortgage insurance. The insurance in question is the credit default swap, a common enough item in the corporate and sovereign debt markets but a late arrival in the mortgage market. Nowadays, a qualified investor can buy a CDS on a particular mortgage-backed bond and even a specific particular tranche of that security. In the language of Wall Street, the CDS buyer is a "buyer of protection." The cost he pays is an interest rate, and the party to which he pays it is the seller of protection. "With the advent of the synthetic market," observed the Reuters expert, "there are tremendous amounts of home equity risks being traded, much of which is driven by the CDO desire to sell protection in their structures."
This last comment explains more than it might seem. To see it for the revelation that it is, a layman may need to pause to catch his breath and review some basic nomenclature. Recall, to start with, that a CDO is a pile of debts refashioned into a security. It is structured in slices, or tranches, from supposedly bulletproof (triple-A) to admittedly perilous (speculative-grade or not-rated). It is highly leveraged, with a single dollar of equity supporting as much as $100 in debt.
There are at least two kinds of CDOs. The first is the cash variety, which is stocked with bonds or tranches of asset-backed securities. The second is the synthetic kind, which is created by selling protection on the bonds or ABS. How can a CDO be built from credit options? Consider that the seller of protection has the same credit exposure as does the buyer of bonds--in case of a credit event, he is on the hook. The rage to create synthetic CDOs is, on balance, a good thing for the prudent readers of Grant's. The booming supply of CDS lowers the cost of protection they buy, or can (and should) buy. Synthetic CDOs are believed to be widely marketed to the trusting financial institutions of Europe and Asia.
In this essay about derivatives, our view is itself partially a derivative. The entity from which it is derived is Pennant Capital Management, a New Jersey long-short equity hedge fund. Alan Fournier, a paid-up subscriber to Grant's, is the managing member. Fournier says that Pennant is expressing a bearish view on housing in the CDS market by buying protection on the weaker tranches of at-risk mortgage structures. At the cost of $14.25 million a year, the fund has exposure to $750 million face amount of mortgage debt.
"I come to this as a student of subprime lending and the housing sector," Fournier tells colleague Dan Gertner. "We were actually long the subprime lending stocks until four or five months ago. We have been short the housing stocks since last summer. The dynamics of those two industries are sort of colliding here in what I think will be a very significant home-price decline. That is the backdrop."
As a buyer of protection, Fournier writes checks to the sellers of protection. The prices he's paying are remarkably low, both he and we judge. They range from 190 basis points a year for the so-called better loans to 220 basis points a year for the riskier ones. He keeps writing checks to the sellers unless and until there is a "credit event," an interruption in the payment of principal and interest by the home buyers to the lenders. If and when trouble strikes, it's the sellers of protection who start writing checks to the buyers.
The odds of a credit event heavily depend on the structure of the mortgage security, or tranches of mortgage security, on which one is buying protection. As a rule in an asset-backed deal, principal and interest come in at the top of the credit ladder and cascade down, while losses come in at the bottom of the credit ladder and infiltrate up. At the penthouse are triple-A assets; at the ground floor, triple-B-minus-rated ones; in the basement are the unrated assets, including what is called an "overcollateralization" tranche. "What has happened over the last four or five years," Fournier observes, "is that home prices have been rising so rapidly that not only did you have the shock absorber of overcollateralization in the loan, but you also have the 10% accretion in values of homes per year that created additional equity to create very solid credit performance for these securities historically."
Yet, even in the best of times, subprime mortgages suffered losses of 4% to 5% a year. In what are no longer the best of times, the damage is bound to be greater. Overcollateralization today runs to about 5% per CDO, Fournier says. Is it so hard to imagine losses equal to, or in excess of, 5% in a national housing bear market? Losses over and above the overcollateralization shock absorber would eat first into the lowest-rated investment-grade tranche, i.e., the triple-B-minus layer, which typically accounts for 2% or 3% of assets. They would next undercut the triple-B-rated tranche, which accounts for another 2% or 3% of assets. If the losses kept coming, the higher-rated tranches would follow the lower-rated ones to the mark-to-market chopping block.
But it would require no national catastrophe to deliver outsize returns to the discriminating CDS buyer. The sharp corrections already under way in the boomier real estate markets might suffice to wreak havoc in a geographically concentrated CDO. Fournier says he invests security by security. He likes "high Florida exposure, high California exposure, high second-lien exposure. You look for equity take-out loans, because those appraisals tend to be overstated, a high percentage of stated-income loans (a.k.a. liars' loans), and you build yourself a portfolio of credits from weak underwriters that are ultimately likely to be impaired.
"Most people start with the assumption that house prices don't go down," Fournier goes on. "I think they will. I think if they only went down 2% or 3%, it would be remarkable. This paper has been experiencing 4% to 5% cumulative losses during a home price environment where we've seen 10% annual increases. In theory, if we just went flat, you would see 14% to 15% losses in these same portfolios, all else being equal. All else isn't equal, obviously. We have oil prices up, we have $400 billion of ARMs adjusting up this year, another $1 trillion reset next year, and the whole idea that people will simply refi their way out of trouble is no longer going to be an option. The guys that write this paper--the subprime lenders--view these guys that are having these resets as future business, 'because we will just write them a new loan.' It is not going to work if home prices are not going up, and the fed funds rate is not back to 1%."
Prompted by Fournier, Gertner delved into one of the myriad of mortgage-backed structures on which a professional investor can buy or sell protection (administrative complexities bar the amateur, even a rich and sophisticated one, from doing the same for his or her own account). The ACE Securities Corp. Home Equity Loan Trust, Series 2005-HE5, is the specimen under examination. The trust, which came into the world in August 2005, is no outlier but a fairly standard item of the hundreds of billions of dollars' worth in the market today. It was created from a pool of first and second liens of varying credit characteristics (4,666 of the loans conformed to Freddie Mac loan limits, which earned them the imprimatur, "Group I"; the balance of the loans may or may not so conform and are designated "Group II"). At inception, the trust had a par value of a little more than $1.4 billion; 17.8% of the loans were fixed-rate, the balance adjustable.
Simplicity is not the trust's outstanding design feature. It holds 20 tranches, with the bulk of the dollar value in triple-A loans but--as the diagram points up--tens of millions of dollars in loans in the lower realms of investment-grade and an equity pool in the sum of $11.5 million. These tranches are the cannon fodder of a hypothetical real-estate bear market. Realized losses on the mortgages held in the portfolio would be absorbed, first, by that net monthly excess cash-flow account; second, by the CE certificates (for "credit enhancement"); third, by the class B-3 certificates, and so forth, until housing Armageddon, when not even the A-1 tranche would be left undamaged.
Studying the architecture of this edifice of home equity loans, Gertner notes a striking lack of diversification. At the time of creation, no less than 34.5% of the principal balance of the mortgages was exposed to California, 11% to Florida and 10.4% to New York. Interest-rate reset dates were bunched in May-June 2007, when more than 90% of the ARMs in the portfolio are expected to be adjusted. Forty-odd percent could be adjusted by two percentage points, while 59% could be adjusted by as many as three percentage points. Subsequently, the loans can be adjusted between one and two percentage points every six months.
"Of course," notes Gertner, "the rate could be adjusted down as well as up, but looking at the reference rate--for the most part, six-month Libor--an upward reset seems much more likely. First payments for the loans in the trusts occurred between September 2004 and August 2005, between which dates six-month Libor climbed to 4.1% from 2.2%. Today, with Libor at 5.4%, a three-percentage-point reset is possible, and a reset of more than one percentage point is probable. Naturally, interest rates could fall by the middle of next year. But a weak economy--if that were the reason for the drop--would add another hurdle to the already obstacle-littered real-estate playing field."
At the time of closing, 29% of the loans were of the interest-only kind (70% had the traditional principal amortization feature and 1% were balloon loans). As to the purpose of the loans, almost half were earmarked for cash-out refinancing. As to documentation, 58% had the works, with most of the balance showing only "stated documentation" (cross your heart, Mr. or Ms. Mortgage Applicant, please, not your fingers).
To date, the trust has given a good account of itself, with not one credit event blackening the record of the first year. In the 13 months since launch, the natural churn of the U.S. housing market has reduced the outstanding principal balance of the trust by $414 million, to $1.023 billion, and the number of loans by 1,935, to 5,277. Because the junior tranches are supporting a lower dollar value of senior debt, effective credit support for the high-rated debt has ratcheted up. All of which is to the good.
But termites are busily gnawing at the mortgage foundations. At last report, which was August's, 8.8% of the principal was delinquent and 4.2% was in foreclosure--$90 million and $43 million, respectively. For perspective, just $66 million of principal buffer stands between the two lowest-rated mezzanine tranches, M-9 and M-10 on the diagram, and some future loss.
Yet, according to Fournier, credit protection on those very two tranches is available for only 220 basis points a year. Is it so hard to envision the circumstances in which delinquencies and foreclosures on the California and Florida segments of the trust's portfolio would move drastically higher? We can hardly imagine circumstances in which they wouldn't.
"What I have done," Fournier tells Gertner, "is put together a portfolio of this stuff. I have $750 million of this stuff shorted. My cost is 1.9% [the previously cited $14.25 million a year]. My return could be $750 million." As risks and rewards go, we judge, not bad.
SELECT YOUR TERM:
Browse 32 years of Grant’s archives by topic or article.
Grant’s hosts semiannual investment conferences in New York.
Questions/Problems with the site?
Questions regarding your subscription?