Mortgage lenders, insurers, and regulators agree that piggyback lending - the practice of extending additional secured credit concurrent with a first mortgage - is a major reason behind the rapid growth of home equity loans.
The practice is clearly eating into mortgage insurers' business. By breaking high-loan-to-value-ratio loans into slices and keeping the first mortgage at 80% of value or lower, consumers can swap tax-deductible home equity payments for mortgage insurance payments.
With that extra credit comes added risk, of course. Just how much risk piggyback lending is introducing to the financial system is an area where consensus is harder to find.
Banking regulators are paying attention. This spring they put out general guidance on home equity lending that implicitly tried to rein in piggyback lending.
"That structure is one of the main reasons that we've seen as much growth as we've seen in home equity lines of credit," especially riskier high-leverage ones, said Barbara Grunkemeyer, the Office of the Comptroller of the Currency's deputy comptroller for risk. The guidance was prompted both by the growth of banks' home equity portfolios and by their often loose underwriting on the loans.
And the interagency guidance on first mortgages that the regulators hope to put out in the fall should touch on the risks to first mortgage holders stemming from piggybacks, Ms. Grunkemeyer said.
The mortgage insurer PMI Group Inc. spiced up the debate last week with the release of a study that claims to show lenders and others have exposed themselves to more credit risk than many realize.
That risk relates in part to the fact that housing markets considered to be the frothiest - and therefore susceptible to downward pressure on prices in the near term - are also, probably unsurprisingly, among those where piggyback loans are the most prevalent.
Indeed, all but three of the top 10 metropolitan areas that PMI considers the most likely to experience home-price declines - markets it identified in a separate report that it issues quarterly - had a piggyback component in at least half of their purchase lending.
A report last year by SMR Research Corp., said the amount of purchase-mortgage dollars involving piggyback structures more than doubled in the first half of 2004 from 2001, to 42%. About a third of all purchase mortgages had piggybacks, versus 14% in 2001, according to SMR, which crunched public-record data from First American Corp. on more than a million transactions.
"The trend is about as clear as a trend can be," said Stuart Feldstein, SMI's president. The PMI-sponsored study's author, Charles A. Calhoun of Calhoun Consulting in Washington, relied heavily on SMR's work.
The PMI paper positions itself as providing public-policy advice. It points out that the GSEs' risk-based capital standards pay no attention to piggybacks, and that banks consider a mortgage to be in the high-LTV bucket if there is a piggyback on it only if the bank also owns the second.
"It's an evolution of the market that might not have been anticipated when the regulations were put in place," Mr. Calhoun said.
Though Basel II will change things, a crackdown is needed now, said Mr. Calhoun, who has held senior research and policy analysis positions at Fannie Mae and the Urban Institute, and at one time was the Office of Housing Enterprise Oversight's deputy chief economist. "At this point there's quite a gap here," he said.
According to the OCC's annual survey of examiners' view of underwriting, released last September, 18 of the banks doing high-LTV home equity eased standards and 11 tightened. In 2002 no banks eased and 44 tightened.
That piggybacks have been a thorn in the side of mortgage insurers has been no secret. According to the Mortgage Insurance Companies of America, a trade group for six of the largest mortgage insurers, in May their primary insurance in-force was down 1.9% from a year earlier, at $599 billion.
There is some debate about whether an 80% LTV mortgage with a piggyback will perform worse than similar mortgages without one. Many analysts say they think such loans will underperform traditional 80% LTV loans but outperform 100% LTV loans, in part because the dual underwriting filters weaker borrowers.
Ken Higgins, a senior vice president at the Dominion Bond Rating Service, concurs with the first part of that thinking. "From a default frequency standpoint, we definitely expect a difference," he said.
There is evidence to support the point. According to a March report from Credit Suisse First Boston on subprime mortgage bonds, 80% LTVs with piggybacks have 30% more delinquencies than 80% LTVs without.
On the other hand, Ken Carter, the executive vice president of the wholesale home equity division at Cleveland's National City Corp., said the typical piggyback borrower's sophistication is one reason that such first mortgages may perform better than one might believe. "It's a different borrower, I think, than the average borrower," he said.
Mr. Carter said that in many cases the existence of a piggyback may not reflect anything more than the desire for future flexibility, "because it is generally no cost to them if they do it simultaneously with a purchase of a home."
Nevertheless, most of his unit's piggyback HELOCs have initial draws, and piggyback borrowers tap more of the available amount than stand-alone HELOCs, he said.
Piggybacks recently fell to about 50% of the division's loans after reaching a high of around 80%, Mr. Carter said, as demand for stand-alone seconds grew faster.
Regulators' recent guidance reflected their view that banks need to underwrite the loans more thoroughly to account for a less pristine credit environment in the future. It also said recent examinations had found "several instances of noncompliance" with previous guidelines on high-LTV lending.
Richard Brown, the Federal Deposit Insurance Corp.'s chief economist, said what many others say: that it may be futile to try to predict how current mortgage innovations will play out.
"We have been through this very benign period," Mr. Brown said, "and taking that as representative of all points of the cycle probably does not make much sense."
