Of all the questions surrounding the aggressive mortgage lending of the past few years, the toughest to answer may be this: where has the risk gone?
A look at the mortgage assets on a lender's balance sheet tells only part of the story. Two similar-looking portfolios do not always carry the same risks, and exposure to a loan does not always end when it leaves the books. Risk that has been offloaded to another party often gets reloaded.
In such an environment, asking questions about things like reserves for forced loan buybacks, or valuations for servicing rights or interest-only strips, may be as important as asking about loan-loss set-asides. Even knowing what guesses to make takes insight into an evolving secondary market.
Dozens of interviews with industry executives and observers over the last few months identified numerous reasons why the industry's conventional way of thinking about mortgage risk no longer applies.
1) After a loan is sold, the lender may well have to buy it back.
Among of the biggest headaches in mortgage banking are the representations and warranties that loan sellers make to buyers that the loans are free of such things as inflated appraisals or underwriting mistakes. If a loan turns out to have such defects, the buyer can force the seller to repurchase it, and the buyer may have this right for the life of the loan.
With the government-sponsored enterprises and other buyers accepting riskier loans, the odds are that these obligations will rise, several executives said.
"It's kind of a simple math equation: The more loans that go down, the more likely it is that there are going to be things that the counterparty is going to come back to discuss," said Willie Newman, an executive vice president at ABN Amro Mortgage Group Inc.
NetBank Inc. executives said on a conference call last week that in the first quarter it increased its provisioning for repurchases of nonconforming loans, because it was forced to take back more of its recently originated loans in the last three quarters.
That trend raises an important question, Steven F. Herbert, its chief financial officer, said on the call. "Will the fully seasoned frequency of repurchases be higher because the early repurchase activity is higher than in the past, or is the industry getting better at putting loans back faster?"
Executives say Fannie Mae and Freddie Mac typically force more buybacks later in loans' lives than private securitizers. The two GSEs hold more bargaining power than most loan buyers when forcing buybacks and can seize servicing rights from sellers. Their use of "retro-appraisals" to prove an inflated appraisal can be tough to combat, lenders say. (The Federal Housing Administration is said to be most willing to force indemnifications for even minor mistakes.)
But some say private conduits, which securitize many of the riskiest loans, could become more aggressive in this regard if bad performance irks bond investors enough to push trustees to go through files with fine-tooth combs.
Of course, large and midsize lenders can also turn around and try to force the small lenders and brokers that supply their loans to make good on their own promises about loan quality. But the small lenders may not have the means to do so.
In certain loan programs and individual bulk sales to the GSEs, usually the higher-risk ones, lenders agree to formal recourse, meaning they share the losses. The parties usually do not publicize such agreements. Additionally, lenders that sell mortgages to the Federal Home Loan banks retain credit risk in the form of the promise of future payments for good loan performance.
2) In securitizations, lenders can also retain credit risk - though they haven't done it as much lately.
They retain risk through "excess spread" residuals - interest payments above what goes to bondholders, or payments from things like prepayment penalties - or lower-rated security classes.
However, a healthy appetite for collateralized debt obligations among foreign investors, pension funds, and hedge funds has let many securitizers pass lower-quality bonds to CDO underwriters. (Through retranching and diversification, CDOs produce higher-rated securities from lower-rated ones.) The small risk premiums investors demand for lower-rated bonds also shows there is plenty of appetite for them on their own, some observers said.
Charlie Ryan, who manages Accredited Home Lenders Inc.'s securitizations, said it rarely holds on to lower-rated bonds, because of the strong pricing. It does retain "excess spread," even though there are also potential buyers for these interests, he said.
Since last year New Century Financial Corp. has been selling nonprime loans to a fund run by the Greenwich, Conn., hedge fund Carrington Capital Management LLC. New Century capitalized the fund with $25 million and then sold stakes to other investors, according to securities filings.
The fund securitizes the loans and holds on to the residuals. In the fourth quarter when the fund raised additional capital and New Century's ownership stake dropped to 38%, the subprime lender took the fund's assets off its balance sheet, decreasing the value of residuals on it by 26%, to $152 million.
3) Mortgage insurance is often reinsured - by the lender.
If things get bad enough, lenders may have to pay claims on captive reinsurance deals with mortgage insurers (which usually cover high-loan-to-value loans sold to the GSEs).
In a standard "deep-cede" arrangement, a reinsurer might agree to take up to 7% of the losses after the first 3% for a year's worth of loans, in return for 35% of the monthly premiums.
In an insurance arrangement that covers 25% of losses on $1 billion of loans, the lender's captive provides between $7.5 million and $25 million of coverage. In another common type of deal - the 4%-10%-40% arrangement - the reinsurer starts taking losses when they exceed $10 million and stops at $35 million.
A lender also might have given in to pressure in recent years from mortgage insurers to move to "quota share" deals, in which losses are shared dollar-for-dollar and based on premium sharing, minus insurer expenses. The reinsurer would be at risk earlier but would avoid bigger hits on larger losses, up to catastrophic levels not covered in the other deals.
4) Triple-A isn't all it's cracked up to be.
If the rating agencies have underestimated the amount of losses for underlying loans, triple-A bondholders could face downgrades or, perhaps, losses.
Not all triple-A rated mortgage bonds are the same, though. According to Scott Simon, a managing director at Pacific Investment Management Co., the giant bond fund manager has been one of the investors asking for "super-senior" tranches. Thumbing their noses at rating agencies, investors like Pimco are looking for even more protection than that provided by other triple-A classes on similar deals that take earlier hits from principal shortfalls.
At a recent conference, Mr. Simon said it does not cost much to get the added protection, which is needed because rating agency "models are all insanely screwed up" because so many potential defaults became refinancings or home sales in recent years.
Rating agency analysts say they are well aware of that problem. At the same conference, they said they are doing their best to be conservative with new products and combinations of risk factors. (And in the past year, despite competitive pressures among raters, credit enhancement requirements have risen for a variety of products.)
"Everything that has been happening in the last five or seven years in real estate and mortgage performance has actually been better than expected, not worse than expected - the conditions ratings should address," said Ken Higgins, a senior vice president at the Dominion Bond Rating Service, in an interview.
"The less history" that exists for a product, "the higher the 'uncertainty factor' should be," he said. For instance, Dominion multiplies expected loss frequencies for any interest-only loan by a factor determined on a sliding scale based on when principal payments begin. If they begin within the first three years, Dominion assumes about 1.3 times the defaults that its models would predict otherwise.
Glenn Costello, a managing director at Fitch Inc., said it not only relies on extremely positive recent performance data, but it also looks back to the 1980s in making judgments. Particularly for alt-A pools, Fitch has been conservative about the layering of risks, and that philosophy has cost it market share in the sector, he said. "We'll hit some of these loans so hard, we'll assume the probability of default is 100%."
Michael D. Youngblood, the head of asset-backed securities research at Friedman, Billings, Ramsey Group Inc., said, "The evidence to date is that the rating agencies are very conservatively assessing all of the newer innovations in the mortgage area."
5) Simple LTV ratios don't say enough about risk.
Regulators' desire for banks to slice and dice their loan and securities books into different risk buckets shows that taking comfort from broad disclosures can be perilous.
A bank with an average first-mortgage loan-to-value ratio of 70% might seem well protected. However, since that figure is an average, a chunk of the loans could still have LTV ratios of 95%. And because of the sharp rise in piggyback home equity products used in lieu of down payments, some borrowers may be more leveraged (and hence more likely to default) than the first-mortgage ratios would suggest.
There is little doubt that banking companies have been increasing their direct exposure to housing debt. According to the Federal Deposit Insurance Corp., mortgage assets made up 71% of the growth in banks' interest-bearing assets in the fourth quarter, the most recent period for which data is available. First-mortgage debt on the books climbed 3%, mortgage-backed securities rose 5.9%, and home equity loans surged 6.7%.
When borrowers are not forced to get mortgage insurance, lender-paid insurance - paid for by charging higher rates - can cover the risk. For example, while Golden West Financial Corp. makes option ARMs with LTV ratios of up to 95% that allow for negative amortization, any loan with a ratio above 80% is insured, according to Russ Kettell, its president and chief financial officer.
But the rise of piggyback products is making insurance less prevalent and leaves lenders holding more risks in the form of second loans. (Piggyback second mortgages can also be insured; American International Group Inc.'s United Guaranty Corp. provides such coverage.)
6) Piggybacks don't provide as much protection as mortgage insurance.
When piggybacks are used instead of mortgage insurance, there is less protection, even when the first mortgage is sold. Most insurance covers losses of more than 20% of the home's value at the time of origination. For example, Thornburg Mortgage Inc. requires the typical 25% coverage for 90%-LTV loans; insurance on a $555,555 home with a $55,555 down payment on an interest-only loan will cover $125,000 of losses.
Joseph Badal, a senior executive vice president at Thornburg and its chief lending officer, said that even though there is no difference in its guidelines either way, in a piggyback scenario, "we're going to be a little less liberal if a borrower's expecting an exception to guidelines," which it often will do otherwise.
7) Traditional pipeline hedges don't work as well with the hot new products.
A shift in the secondary market can also lead to problems with any loans a company has committed to make (or buy) or has made (or bought) with the intention of selling.
At each quarter's end the loans and commitments must be marked to market, so a market shift can lead to losses. Hedges might be in place, but they can be imperfect.
A lender hedging its pipeline of nonconforming loans by using forward sales in the "to be announced" market will find a mismatch if agency security prices remain strong while the price it can get for its loans drops. This is just one example of how pipeline hedging "basis risk" can grow with the proliferation of nonstandard products. (A lender with firm sales agreements would fare better, with the buyer taking the hits.)
8) Escaping losses could make matters worse with your regulator.
For example, regulators that see a lender with too many foreclosures - and not enough losses - might start to think the underwriting focused too much on collateral, rather than repayment capacity. In guidance last year, the Office of the Comptroller of the Currency reiterated that such underwriting is its definition of predatory lending. It also listed a set of practices that might be predatory if not used judiciously. Many of the practices, such as negative amortization and stated incomes, are increasingly popular among lenders.
