Carrington Capital Management, a former subprime securitization specialist, salvaged an impressive amount of money from its low-ranked RMBS, thanks to the unusual strategies of its servicing affiliate.
By stockpiling foreclosed homes and declaring borrowers current — sometimes unilaterally — Carrington received payments that would otherwise have protected other investors from future losses. In a single deal that American Banker analyzed, Carrington's actions likely netted its investment fund more than $20 million.
"They appear to have managed their book for their own personal benefit in a way that screws the investors," said Laurie Goodman, an Amherst Securities Group analyst who flagged Carrington's modification practices in research notes.
Carrington is not a bank; it competes with them, and its inner workings raise unsettling questions about the mortgage market and some of the proposals to fix it.
Carrington's performance suggests that forcing securitizers to retain exposure to their own deals may not reliably guarantee quality origination and aligned incentives among investor classes, and shows how hard it is for investors to figure out how pools of collateral are being managed. It also demonstrates how the basic structure of a securitization — that the claims of some classes of investors outrank others — can be turned on its head.
Depending on how the Obama administration structures planned retention requirements, the same incentives that Carrington created for itself could be institutionalized nationwide.
"A lot of people say, well, there's not enough skin in the game," said Eric Higgins, a professor of management and finance at Kansas State University who has researched the performance of private-label mortgage securities. "But maybe it's more of a regulatory issue. … Drawing $20 million of overcollateralization out of a deal that's going to end up with a 45% delinquency rate, you should not be allowed to do that."
An internal audit viewed by American Banker shows Carrington holds similar stakes in at least 17 more deals.
While Carrington's servicing practices benefited itself, Carrington argues that its strategy also aligned the interests of both homeowners and the trusts it oversees. The company's real estate disposal strategies were reasonable, it said, and it has salvaged value by giving borrowers repeat modifications. Moreover, Carrington's bottom-ranked positions in its deals come with special discretion over default management, the company notes.
Carrington's strategy has been "to reduce losses and maintain payment streams for the trusts, by modifying loans for homeowners who are having difficulty and avoiding the sale of properties at historic market lows," company spokesman Chris Orlando said.
Evaluating Carrington's claim of good results is difficult because the company's actions, regardless of their long-term impact, have had the effect of improving the deals' short-term appearance. While Goodman and some Carrington investors anticipate that senior bondholders will eventually face greater losses due to Carrington's actions in the deal reviewed by American Banker, those losses have not materialized yet.
BIRTH OF A STRATEGY
Carrington is the product of Bruce Rose.
A former airline mechanic and pilot, Rose moved into finance at the beginning of the 1980s. While working at Solomon Brothers, he became familiar with subprime through helping to finance the early shops producing such securities, according to a Reuters profile. In 2003, he left Citigroup to start his own mortgage securities hedge fund. Carrington Capital Management's business was to package subprime debt into private-label securities, retaining the highest-yielding, lowest-ranking portion of the deals for itself.
"From its inception, [Carrington] had a consistent philosophy: maximize asset value over the long term, focus on return over volume and align interests and incentives," Orlando said. "That's where you see the risk-retention concept come into play."
That concept led Carrington to insist on a special feature of deals in which it invested: Despite holding the lowest-ranked slices of the securitization, Carrington claimed a high degree of authority over handling troubled collateral. Among its privileges was the right, as Carrington would note later in court, to "direct the servicer regarding the disposition and sale of properties whose mortgages go into default."
Neither the ratings agencies nor investors appear to have objected to this detail, and Carrington had an important asset: close ties with a subprime origination star, New Century Financial. The Irvine, Calif., firm provided seed funding to Carrington in exchange for a stake in its fund, and it often supplied loans to Carrington and serviced its deals.
The collapse killed off New Century and many of the subprime originators Carrington worked with, and subprime investment funds were overwhelmed by losses, litigation and redemption requests.
A March 2007 BusinessWeek article with the headline Who Will Get Shredded? declared that Carrington "may have gotten badly burned" by its ties to the then-floundering New Century and "seems particularly vulnerable."
Instead of retreating, Carrington waded further into the industry. In May 2007 it paid $188 million to purchase New Century's servicing rights and infrastructure out of bankruptcy. In the company's investor letter that June, Rose heralded the purchase.
"Bottom line is that CCM bought the servicing platform for the price of the financial assets only," he wrote. "[D]irect and absolute control of the credit performance of the New Century-originated assets on [Carrington Investment Partners'] balance sheet are now firmly in the hands of CCM."
That "absolute control" was more valuable than anyone outside the company could have guessed at the time. Three years later, despite concentrated investments in the lowest tranche of deals that are approaching 50% delinquent or foreclosed, Carrington is still standing. How the company handled its servicing duties in deals like one called CARR 2005-NC3 helps explain why.
Carrington has said that its investments led it to be choosy about the deals it assembled, but the 2005-NC3 doesn't reflect that. According to the deal's offering documents, the average FICO score for borrowers in the $1.7 billion pool was only 617. Nearly half were stated income. The loans, originated and serviced by New Century, were skewed toward California and Florida.
"This was a terrible pool of loans," said Higgins, who reviewed the 2005-NC3 prospectus at American Banker's request.
Out of that pool, Carrington owned the credit enhancement (CE) tranche, a bottom-ranking, 3% slice of the deal created to act as a buffer against loss for more senior investors. With a face value of $53 million, the CE class was supposed to receive leftover cash after it has been assured that other investors will be paid.
Within two years of its origination, however, a surplus of cash in the deal looked like a stretch. With borrowers already struggling when Carrington bought New Century in May, Carrington's CE class earned no money at all. It didn't get paid in June 2007, either. Or July.
While the face value of Carrington's position was still nominally worth $50 million, trustee remittance reports suggested that its asset was sliding toward worthless. Serious delinquency rates were at 10% and climbing, constricting the deal's net cash flow. Instead of paying off speculative investors like Carrington, the trust's diminishing proceeds were increasingly being retained for senior investors who had paid for triple-A-rated protection.
But while investors were bound to take a bath on the repossession and sale of subprime collateral, the losses couldn't be booked if repossessed homes never sold. And in many Carrington serviced deals, including 2005-NC3, foreclosed homes didn't do much selling.
"We try to be the best sale on the block, not the first sale on the block," Rose said in a January 2009 Reuters profile. Predicting the housing market would soon recover, he said Carrington wasn't in a rush to clear out inventory.
Foreclosed homes have not historically tended to rise in value, though that was the bet that Rose committed Carrington's fellow investors in 2005-3 and numerous other deals to making. At its peak, 18% of the properties in the portfolio were foreclosed properties in the hands of Carrington. In some of its deals, Carrington sought to rent out the properties.
"The senior investors would clearly prefer the liquidation," Goodman said, because unoccupied homes lose value over time and rental income is unlikely to offset the damage. While Rose's bet that loss severities would decrease with time was largely wrong, Goodman said, the strategy "worked to the benefit of the CE holder and the junior investors."
Loss severities from 2005 NC3 and other Carrington deals suggest Goodman is correct that the delay was harmful to the trust as a whole. In 2007 the average monthly loss severity on REO sold was 32%. In 2008, that rose to 50%. And in 2009 and 2010, losses exceeded 70%. While Carrington notes that its current severities compare well to other subprime losses, Goodman and investors argue that Carrington passed up a chance to get rid of collateral at points when its loss severities were lower.
A growing REO mountain wasn't the only surprise to be found in 2005-NC3's remittance reports: an relatively obscure data point called "negative curtailments" was growing.
Normally a positive number, curtailments track unscheduled partial paydowns of a loan balance, such as when borrowers start sending in an extra $100 above their required monthly mortgage payment. But negative curtailments — unexpected increases in a loan's balance — were now on the rise, according to remittance data compiled by a private investor and checked by American Banker.
Troubled subprime loans were getting more valuable.
Carrington had become an ardent practitioner of capitalization modifications, in which it tacked whatever missed payments a borrower had incurred on to the top of a loan's unpaid principal and declared the loan to again be current. Reviled by many borrower advocates for saddling a struggling borrower with a higher principal and potentially higher payments than the ones he or she failed to make previously, the modifications nonetheless buy both parties a little time.
Carrington bought a lot of it. A delinquent borrower would be given a mod, perhaps make a few payments, and then often fall back into delinquency. Then Carrington would sometimes give that borrower another mod, perhaps coupled with a rate reduction, repeating the process. In some instances, it appears to have applied three modifications over three years, allowing the loans to negatively amortize.
Carrington's focus on capitalization mods was one of several practices that drew a lawsuit from then-Ohio Attorney General Richard Cordray, whose office alleged in a 2009 complaint that many Carrington modifications did not amount to a "good faith" effort to help borrowers stay in their home over the long term.
Carrington said it could not comment on the Ohio suit, which is still pending.
"The idea of a modification program is that you lower the payments or the principal amount to the point where the borrower can afford to live in the house," Goodman said, contrasting that with Carrington's activities. "All you're doing is putting the delinquent money into the balance of the loan, and declaring the guy current on an accounting statement" Goodman said. "You're just stalling foreclosure, because you haven't improved the borrower's situation."
Orlando rejected that allegation, citing what he said was a 20% average payment reduction on loans the company modified. He also pointed to Carrington's outperformance in the Home Affordable Modification Program (HAMP): When HAMP ceased reporting servicer-by-servicer outcome data last year, Carrington had one of the best temporary to permanent modification conversion ratios in the industry.
But while CoreLogic data compiled by Goodman's ASG showed that serious re-delinquency rates of Carrington serviced deals are only slightly worse than its peers, those results benefit from the repeat nature of many modifications: some borrowers were effectively permitted to skip tens of thousands of dollars of payments on multiple occasions.
Moreover, the large mods most helpful to Carrington's CE position appear to have had an extraordinarily poor track record. Using the high activity month of October 2008 as an example, a review of the top 20 largest modifications in 2005-NC3 shows that all but one is in foreclosure or has been seized — and that the sole exception hasn't made a payment in over a year.
The best days for Carrington's 2005-NC3 holdings were still to come. While a mod at any point had the potential to boost Carrington's payout, a key window of opportunity for 2005-NC3 had arrived in March of 2008. Called a "stepdown," it obligates the trust to release tens of millions of dollars in excess principal to investors — assuming that the deal is healthy enough to do so.
The deal wasn't. In March of 2008, 2005-NC3 needed a delinquency rate of less than 24% to allow for a principal payout, but the deal was near 32% despite Carrington's modifications. Six months went by, each worse than the last. Carrington's CE tranche again stopped paying out.
Then, in September, Carrington initiated a high volume of modifications, throwing the remittance data's failing statistics into reverse. After averaging under $6 million of capital modifications per month all year, Carrington reworked $18 million of loans in October, according to CoreLogic data compiled by Goodman's ASG. In November, it surpassed $26 million. And in December, it peaked at $36 million. (Other types of modifications also grew.)
That did it. Remittance reports show that delinquencies, which were over 34% in September, dipped to 25.6% in December, squeezing by the current trigger of 26%. 2005-NC3's overcollateralization fell from $53 million to $31 million, rendering the deal — at least on paper — significantly overcollateralized. Between principal and increased interest payments, Carrington's CE position pocketed $18 million over the next five months.
Ironically, during that period, senior investors received nothing. Money owed to the CE class was consuming every dollar that was then available.
Carrington argues that those benefits were incidental to a servicing job that benefited the trust as a whole.
"Any examination of performance should look at total payments to the trust not just certain classes," Orlando said. "If you measure the strategy based on that primary obligation, you find that deal performs as well or better than [peers.]"
Carrington provided American Banker with several comparisons seeking to demonstrate this outperformance, including a common ABX index of contemporary subprime deals that showed Carrington has so far maintained better-than-average cash flows in 2005-NC3 and recognized fewer losses.
The results suggested that Carrington's approach has worked, Orlando said.
That Carrington would have chosen to service its portfolio on behalf of the entire trust does not easily fit with its court filings arguing that its "special rights" trump those of other investors.
At the same time that Carrington was collecting the payout from hitting its stepdown, it was preparing a suit against American Home Mortgage Servicing, the Wilbur Ross-owned successor to failed servicer Option One. The feud boiled down to a dispute over whether American Home had violated Carrington's special rights regarding the disposition of collateral in four deals in which Carrington held a stake.
"Carrington's special rights … take precedence over the interest of other certificateholders," Carrington asserted in its complaint, citing contractual language that "the rights granted to the CE Holder may be inconsistent with, and adverse to the interests" of its fellow investors. "The CE holder has no obligation or duty to consider [other investors'] interests," it wrote.
In its complaint, Carrington alleged that American Home, short on cash and seeking to avoid advancing principal on bad loans, had begun selling its portfolio of foreclosed homes at fire-sale prices to lower its financing costs. Such tactics, alleged by borrower advocates and some investors to have occurred in other instances, can save a servicer money while at the same time unnecessarily precipitating foreclosures.
Regardless of American Home's intent, however, Carrington argued that its special rights gave it the final say. "CCM implemented the rights, and negotiated them up front," Orlando said of Carrington's general CE privileges. "It was priced into the deal and clearly disclosed."
American Home filed a counterclaim, alleging that Carrington was simply trying to delay the recognition of losses so as to benefit its CE class. In some instances, American Home alleged, Carrington had forbidden the sale of homes even after potential buyers met or exceeded their listing prices.
"On Jan 22, 2009, when [a property in Sacramento, Calif.] had been on the market over 450 days, American Home told Carrington of an offer of $600,000, above the listed price of $599,900," American Home's attorneys wrote. "Carrington e-mailed American Home the next day stating that the offer should be declined and the house relisted for $799,000. This property still has not sold after almost 500 days on the market."
While many of the discovery documents in the court case are either sealed or redacted, American Home has argued that Carrington illegitimately sought to steer money intended for more senior investors to itself.
"Plaintiffs refuse to produce a witness to testify about their strategy of 'slowing foreclosures and REO sales as a means of enhancing cash flows' to Carrington," American Home's attorneys wrote, quoting a document that is not in the public record. "Plaintiffs say that Carrington's strategy — admitted in a document produced by plaintiffs — is not relevant to the upcoming hearing.
As lucrative as Carrington's handling of 2005-NC3 appears to have been, it couldn't last forever. After a 20-month streak in which the deal paid out an average of $1.4 million per month to its CE class, the 2005-NC3 fell below its overcollateralization target of $31 million in August of last year. Cash flow to Carrington again shut off, this time for good.
It was just at that point that Carrington's servicing strategy again changed. After repeatedly insisting in both the press and court documents that it was irresponsible for a servicer to sell off property in a depressed market, Carrington began to do just that.
In the first seven months of 2010, Carrington had sold an average of four homes per month out of the roughly 350-home stockpile that 2005-NC3 owned. In August it sold 37 — and has averaged 48 for every month since. Carrington's advances — the money that it must borrow to pay investors on behalf of nonpaying customers — fell by half.
Over the next nine months, as Carrington cleared out collateral at high loss severities, the remaining value of 2005-NC3 plummeted. In the six months prior to August 2010, the deal balance had declined by just $6 million. In the six months thereafter, it fell by $46 million. Investors senior to Carrington are set to begin recognizing losses within at most a few months.
Carrington said this abrupt switch was precipitated by a change in the company's real estate outlook, not changed incentives.
"Several factors changed in the first half of 2010, extending the time frame for an anticipated recovery and creating a window prior to shadow inventory hitting the market," said Orlando, citing macroeconomic factors and events such as the European debt crisis. While the loss severities for 2005-NC3 and other Carrington deals worsened over time, Orlando notes that it still compares well to many others in the ABX's subprime index.
While Carrington's specific servicing actions in deals like 2005-NC3 has not previously been reported, the idea that Carrington's servicing poses a conflict of interest is not new. In addition to Goodman's research on the subject and American Home Servicing's legal briefs, Carrington's performance has drawn scrutiny from the press. A May 2009 article in Housing Wire concluded with a business professor bluntly advising senior Carrington bondholders to pursue a cease and desist order.
Some investors were already thinking along those lines. In April of 2009, Declaration Capital Management, a McLean, Va., subsidiary of John Hancock Financial Services, posted a request on the Web site of a Carrington trustee seeking to organize a conference call of investors to "determine whether certificateholders together should give notice of a servicer event of default."
"Our feeling was that letting things pile up in REO and then trying to rent homes was largely just for stalling the inevitable, and they were doing it because they owned junior portions of the capital structure," one Carrington investor said.
Despite the benefits Carrington derived from its New Century purchase, the company has been through a rough few years. A copy of an internal audit obtained by American Banker showed that by 3Q08 Carrington had taken $250 million in realized and mark-to-model losses on its $1.1 billion of holdings. Its audit materials warned of both a possible cash crunch and the threat that lenders might cut off current lines of funding. These dire circumstances were the backdrop for Carrington's foray into servicing.
While Carrington's grazing of the 2005-NC3 trigger is unique, its incentives were broadly similar across well over a dozen deals. Mods and stockpiled foreclosures postponed the extinguishment of the CE class, allowing Carrington to continue receiving payments even as the broader deal veered into troubled territory.
Among the lessons which can be drawn from Carrington's story is an old one: buyer beware.
While senior investors could not have foreseen that Carrington would end up servicing deals it held, the company's close relationship with New Century was known and 2005-NC3's offering documents warned that Carrington might seek to advantage itself at higher-ranking investors' expense. Though Higgins said investors and ratings agencies rarely paid attention to such details, they should have.
Another lesson is that "skin in the game" may not assure quality securitizations or aligned incentives in servicing. But as Carrington rightly notes, the exposure it held is roughly the sort that the Obama administration is expected to mandate as part of structural reforms in the mortgage market.
Carrington "pioneered the skin-in-the-game risk-retention concept," Orlando said. "It closely approximates what they're talking about in risk-retention reforms right now."