During the boom years, mortgage lenders paid Fannie Mae and Freddie Mac a small fee for accepting stated-income loans. Today those lenders are paying a whole lot more.

Loan-buyback requests from the government-sponsored enterprises are mounting, and alternative-A loans are at the center of many disputes. At issue are the terms of the GSEs' reduced-documentation programs and who should be responsible for a loan product that most industry experts now concede never should have been mass marketed in the first place.

Originating lenders and the GSEs are arguing about such sticking points as whether the lender should have caught borrowers who were lying about their incomes and whether having performed the requisite due diligence is a justification for rejecting the buyback demand.

"The dirty little secret was that the investors all knew this stuff was garbage, and they bought it anyway," said Michael Pfeifer, a managing partner in the Pfeifer & DeLaMora LLP law firm in Orange, Calif., which represents mortgage lenders and banks. "The buyers were just economically stronger than the sellers, but they are not innocent victims."

Representatives of Fannie and the Federal Housing Finance Agency, now the conservator of both GSEs, reject the lenders' claims. Even if the bar for diligence on alt-A loans was low, they said, lenders were still responsible for vetting the information provided by the borrower (or, in many cases, the broker who arranged the loan.)

"Lenders should be reviewing the customer's position, industry, and years on the job to determine if the stated income appeared plausible and accurate," Fannie said in a statement to American Banker. "They should also determine whether their assets and credit history support the stated income as well." (Freddie did not return a call seeking comment.)

When the housing market was humming along, the GSEs supplied guidelines explaining to lenders precisely what types of loans they would buy. After the loans went into default, the investors forced the lenders to buy the loans back — even though they were originated according to the guidelines, lenders now say.

A key sticking point in such disputes is that many stated-income loans now being pushed back to originators failed not because of problems with the original underwriting but because the real estate market tanked or the borrower lost his or her job.

"By creating the stated-income product, they essentially waived that representation by saying they don't care what the income is," said Brian Levy, a lawyer at Katten & Temple LLP in Chicago, who represents mostly small and midsize banks and mortgage lenders. "If the originating lender thought they were going to be held responsible for the income being accurate, they would have checked it."

Normally such disputes would end up in a trial, with a history of case law. But litigation is rare, though stated-income loans were originated on a mass scale from 2005 to 2008. Few investors sue the originator because of the expense and difficulty of bringing such cases and the risk of losing at trial, lawyers said.

"If the courts looked at it, they would say this is unconscionable and outrageous because the interpretations of reps and warrants are extreme," Pfeifer said. "It's a Ford Pinto problem — a product defect. They said, 'Here are the specs, build the loan exactly as we said,' and then they want them bought back."

Instead of litigation, Fannie and Freddie and other large investors have hired armies of auditors to pore over loan files finding other errors that can justify a repurchase request, lawyers say. If Fannie determines the borrower's income was inaccurate, a mortgage insurer often will rescind coverage, which prompts a repurchase request.

Originators claim they cannot be held responsible for the increasing debt burdens of borrowers whose loans were originated years ago in better economic times.

"There is usually no tie between the reason why the loan went into default and the asserted breach of the reps and warrants," Levy said. "But it's virtually impossible for the investor to prove that the reason why the loan went bad is the debt ratio of the borrower is 38% and it should have been 35% and, if it had been that, the borrower wouldn't have gone into default."

More often than not, such disputes end in protracted negotiations and stalemates.

Overall, buyback requests are increasing dramatically. In a public filing this week, Bank of America Corp., the second-largest residential lender, said it had $11.2 billion of "unresolved" mortgage buyback requests at the end of June, a 50% spike from January.

Both GSEs have reported steadily rising repurchase requests. According to their second-quarter results filed with the Securities and Exchange Commission, sellers repurchased $1.5 billion in loans in the quarter from Fannie and $1.4 billion from Freddie. Both institutions reported far larger repurchase-request volumes outstanding, however — in Freddie's case, a total of $5.6 billion.

According to conversations with lenders' representatives and presentations at Mortgage Bankers Association conferences on the subject, common triggers for the buyback demands range from a borrower's undisclosed debts to inflated appraisals and occupancy misrepresentations.

Though Alt-A loans accounted for only 10% of the GSEs' volume from 2005 to 2007, they are generating a larger share of repurchase requests. And while low-documentation loans were always considered riskier than their full-doc equivalents, Fannie was happy to guarantee them at the time.

Fannie's handbook from 2006 states, regarding one type of loan: "Lender is not required to verify the borrower's income or assets."

"All that language like that did was create a 'don't ask, don't tell' environment while giving Fannie Mae plausible deniability," Pfeifer said. "How would a broker know whether or not the borrower qualified for a fully documented loan unless they actually ran the borrower's qualifications under the full-doc guidelines? And if the borrower qualified, why go stated-income?"

A copy of the Fannie handbook details the accepted variances from the company's underwriting and documentation requirements. Among them are stated incomes, which allowed the lender to rely on an oral verification of the borrower's employment (rather than on pay stubs and tax returns) and determine that the income is "reasonable," as well as the even more bare-bones no income/no asset, or Nina, loans.

The existence of a no/low-doc channel — and the additional fees that lenders paid to participate in it — convinced many lenders that Fannie was taking on the risk of borrower misrepresentations, said Brian Chappelle of the Washington consulting firm Potomac Partners, especially since buyback requests had been exceedingly rare.

This was not always the case. Regardless of the variances in underwriting standards, and the partial reliance on borrower honesty that stated-income loans required, Fannie Mae never abdicated its right to pass back loans because of borrower dishonesty. As one recent repurchase request to a major lender stated: "The lender's contractual warranties with Fannie Mae state that the lender represents and warrants that no fraud or material misrepresentation has been committed by any party, including the borrower, and that those warranties are not limited to matters of which the lender had knowledge."

An interview with the FHFA made it clear that the agency, which is seeking to recoup as much taxpayer money as possible through repurchase requests, would consider a buyback demand on a responsibly underwritten stated-income loan to be a gray area.

"The onus really falls on the lender to demonstrate that they had a robust process to ensure the quality of the loans being delivered," said Maria Fernandez, the FHFA's associate director for credit risk. "If there was agreement on that, you wouldn't expect a repurchase."

The reasonability of a borrower's income would be, at the very least, a mitigating factor, Fernandez said. An hourly wage earner applying for a $600,000 mortgage is an example of a loan that should have flagged trouble, she said.

In some repurchase requests examined by American Banker, the trigger for the demand is precisely a lack of such "reasonability." In others, lenders verified the person's employment at a company — but not necessarily that the person had reported the correct title and salary.

Those failings are ample justification for a repurchase, Fernandez said, adding that the FHFA is confident that Fannie and Freddie are not issuing frivolous repurchase requests.

But in instances when the lender feels it met the GSE's guidelines in good faith, Fannie and Freddie are willing to hear the lender's case.

"What's difficult here is that it's not necessarily black and white," Fernandez said. "There is dialogue. I will tell you that many of these repurchase requests are tough."

That hasn't stopped some lenders and their representatives from complaining that the repurchase requests are still too draconian. Lenders contacted for this story declined to speak on the record, citing their continuing business with Fannie and Freddie and the huge volume of outstanding repurchase requests.

Chappelle said the perception that the GSEs are not playing fairly could have a major impact on the mortgage market.

If lenders perceive the GSEs as fickle in their standards — whether or not that perception is accurate — they are going to be more cautious with the new loans they make, he argued.

Already, "lenders are adding their own credit overlays to GSE loans because the GSEs are using buybacks to lay off the credit risk," Chappelle said, citing a sharp drop in the number of loans made to borrowers with credit scores below 700.

"While I am sure there is not a lot of sympathy for mortgage lenders on the buyback issue," he said, "a consequence of what lenders believe is an overzealous GSE buyback policy is that it is having an adverse impact on the housing recovery."

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.