Mortgage Industry Follows Travel, Insurance, Autos In Tailoring its Rates
The Wall Street Journal
Page A1 (Copyright (c) 2001, Dow Jones & Company, Inc. 01/05/2001
When Mark Kegelman went shopping for a mortgage recently, he didn't encounter the sort of standardized pricing traditionally offered by home lenders.
The 31-year-old pharmaceutical salesman, who was buying a new home in Cincinnati, clicked on an innovative customized-lending service offered by Priceline.com. In a matter of hours, he was offered an attractive $188,000, 30-year, fixed-rate mortgage at a 7.65% rate -- one of thousands of mortgage permutations available from the online lender. Priceline 's computers determined that people in the Cincinnati area tend to keep their homes a long time, meaning they pay interest for long periods and generate more profit for lenders. That, together with factors such as Mr. Kegelman's healthy $80,500 down payment, prompted Priceline to make him an offer better than any other he received, so he took it.
Priceline is just one of a group of pioneer lenders and loan brokers, most operating on the Internet, that are using new computer programs to customize mortgages for individual home buyers. These systems are appearing at a time of heightened mortgage demand, with Wednesday's surprise Federal Reserve interest-rate cut expected to drive mortgage rates lower.
Home-loan rates have tumbled for months, and they are already low enough to ignite a miniboom in refinance activity. The rate for an average 30-year, fixed-rate loan fell to 7.07% this week, the lowest since May 1999, according to Freddie Mac, the giant mortgage -purchasing company.
If the novel customized- mortgage approach is adopted broadly, it could shake up the $1 trillion-a-year home-lending industry. The change could mean that far fewer people would flat-out be denied a home loan. Borrowers with excellent credit records could enjoy savings of tens of thousands of dollars over the course of their mortgages. Lenders probably would push rates up for some borrowers with mixed financial histories. But even many people with thoroughly shaky financial pasts might get a shot at home loans from mainstream lenders -- although at high prices.
Customized mortgages are part of an individualized-pricing trend sweeping the economy. Airlines began the practice in the 1980s and now program their computers to change fares as often as several times a day, depending on who's buying and how fast a flight is filling up. Lately Priceline , even as it has struggled financially, has helped accelerate the shift, with its services allowing consumers to name their own price for hotels and airline tickets. Ever-more-powerful computer software, along with increasing use of the Internet, have driven the move to protean pricing for cars, credit cards and insurance, as well.
The personalized- mortgage strategy is drawing upstart lenders and stirring interest among some powerful investors, including Microsoft Corp., General Motors Corp. and Electronic Data Systems Corp. Major banks, including J.P. Morgan Chase & Co. are also exploring the personalized- mortgage strategy.
But big lenders are moving slowly. Chase and others say they worry that customization could put minority-group members or working-class borrowers at a disadvantage.
There is reason for lenders to fear criticism and possible lawsuits. The data fed into the new mortgage -calculation systems may not always be as fair and precise as they seem -- a problem that can crop up with any sort of high-tech variable pricing. Mortgage -industry officials estimate that only a tiny percentage of borrowers -- probably less than 1% -- are now receiving fully customized home loans.
For generations, "prime" mortgage lenders have offered attractive applicants a one-size-fits-all rate they could take or leave. People denied loans in the prime market could go to "subprime" lenders that offer a limited range of much-higher rates.
In the near future, this sparse range of offerings could be replaced by an almost endless array of mortgage rates and accompanying fees. Customization could also spark more competition among lenders, driving down prices for valued customers such as Mr. Kegelman of Cincinnati.
And proponents of the new approach say that it would encourage mainstream lenders to offer loans -- albeit at high rates -- to more marginal borrowers. That could make it unnecessary for those borrowers to rely on subprime lenders, some of which have been accused of shady practices and onerous payment terms.
Priceline declines to say how much money it has lent through the customized- mortgage service it started last year.
But IndyMac Bancorp Inc., which offers a competing customized service, reports that by the third quarter of 2000, borrowers received mortgages totaling $1.95 billion using the new approach. That amounted to 85% of the mortgage money the Internet-based company lent that period. Fourth-quarter statistics aren't available yet. IndyMac, the 28th-largest mortgage lender in the nation, according to Insider Mortgage Finance, an industry newsletter, has 10 conventional branches in California.
A third outfit exploring the frontier of mortgage lending is HomeAdvisor Technologies Inc., a joint venture whose investors include J.P. Morgan Chase's private-equity arm and GM's GMAC-Residential Funding Corp. Last year, the venture's Web site, HomeAdvisor.com, launched a service that gives member lenders the ability to offer more-precise methods of analyzing mortgage applicants. So far, two of the dozen or so financial-service companies participating in the HomeAdvisor system are offering individualized pricing: Homecomings Financial Network Inc., a unit of GMAC-Residential Funding, and Wendover Financial Services Corp., a unit of computer-services giant EDS.
Several factors have contributed to the current flux in the mortgage market. The first is the increasing popularity in the lending world of "credit scores." These numerical ratings, mostly generated by credit-reporting agencies, typically are based on a scale of 350 to 900, with 900 indicating a borrower almost certain to repay. The criteria determining a credit score -- past borrowing and repayment patterns -- are similar to those in traditional credit ratings. But rather than a mere listing of missed payments and debts outstanding, credit scores are supposed to offer the lender a precise, numerical means of ranking applicants.
First widely used by credit-card companies in the 1980s, credit scores gained credibility with mortgage lenders in the mid-1990s. This happened after Fannie Mae and Freddie Mac, the congressionally chartered companies that promote market liquidity by buying and selling home loans, concluded that credit scores accurately predict borrower behavior.
Fannie Mae and Freddie Mac in the mid-1990s introduced a second new element to the mortgage market: "automated underwriting" software that can crunch credit scores and other factors -- such as a borrower's cash reserves -- and in a matter of minutes, spit out a decision on whether a loan should be approved.
Most lenders quickly adopted the automated approach, either by creating their own software or linking up with that of Fannie Mae and Freddie Mac. While most banks still employ mortgage -loan officers to oversee the process, applicants no longer wait days, let alone weeks, to find out if they qualify for a home loan.
In the late 1990s, some mortgage lenders started using automated underwriting and credit scores to separate borrowers into a handful of "buckets," each with a distinct interest rate. The next step was to begin offering individual prices to each borrower, which is what Priceline , IndyMac and a few others are now doing.
These futuristic lenders combine automated-underwriting, credit scores and such additional factors as geography, the amount of taxes on the house to be purchased, whether it is a primary or secondary residence and whether a borrower is a potential repeat customer. In the future, other ingredients are expected to be added, including the borrower's line of work. Teachers and doctors are thought to be good credit risks, for example, so they might obtain better deals.
An appraisal of the house is still part of this new-fangled loan process, but in some cases, even that is done with computerized databases of property values.
Mr. Kegelman, the Cincinnati drug salesman, filled out his online application for a Priceline loan one morning last summer. He typed in the address of the $268,500 two-story brick colonial house he wanted to buy, the name of his employer, his income, his Social Security number, and other information. Then he hit "submit" and left for work.
Priceline officials say their system first zipped his application to Washington, D.C., where Fannie Mae's automated-underwriting computers determined that Mr. Kegelman was a desirable borrower. Next, Priceline 's computers in Jersey City, N.J., ran the application through a series of digital tests, which determined an appropriate mortgage rate, down to five one-hundredths of a percentage point. The system also ran through possible calculations of "discount points," a common type of up-front fee. Finally, the system figured Mr. Kegelman's closing costs, based on tax and fee levels common in his area.
When Mr. Kegelman got home from work that night, his computer had his 7.65% rate -- below the average at the time of more than 8% -- and Priceline demanded no discount points. That translated into a monthly payment of about $1,632. "I was pretty much oblivious" to the inner workings of the system, says Mr. Kegelman, who never met anyone in person from the lender. "All I knew was I got a good rate."
Dennis Cohen of Merrick, N.Y., had a similar experience. The 39-year-old owner of a commercial roofing business recently borrowed more than $200,000 from Priceline to buy a two-story, five-bedroom home overlooking a bay off Long Island. His personally tailored 15-year, fixed-rate mortgage came with an attractive interest rate of 7% and a quarter of a point in fees, which translates to monthly payments of about $3,000. That was less than what several traditional lenders had offered him, Mr. Cohen says. Priceline even threw in two free round-trip plane tickets to any of the contiguous 48 states.
"Once I saw [ Priceline 's] prices," he says, "I was hooked."
While customization may please some borrowers, the new approach doesn't eliminate potential hazards in the existing mortgage market. For starters, the seeming precision of a three-digit credit score may be undermined by something as simple as a mistaken identity.
Consider the experience of Crispin Delgado, a paint-factory employee in Richmond, Calif., and his wife, Rosa, who sought to refinance their home last year through a conventional loan broker.
When they obtained a mortgage to buy the house in 1996, credit agencies had pegged their credit scores at higher than 700, according to their loan broker. But when they went back to the broker to refinance , they were told their scores had fallen to as low as 552 -- a drop the Delgados say must have been an error. Scores below 600 generally disqualify borrowers from obtaining loans in the prime market. Since they were refinancing, and therefore interested only in a rate below what they already had, the Delgados ruled out the high-price subprime market.
Under a customized- mortgage system, they probably could have obtained a loan from a prime lender, but only at a rate above what would have made sense for a refinancing. Customization software would have relied on the same lower credit scores that the Delgados say are mistaken.
The couple says the credit agencies confused Mrs. Delgado, who is 45, with their 27-year-old daughter, who has the same first name and who declared personal bankruptcy in 1998. After the Delgados explained the mistake, the credit agencies promised to adjust their scores, the couple says. But when their broker later rechecked the scores, he found that some were still as low as 582.
Thwarted, the Delgados abandoned their plan to use cash from the refinancing to help their 24-year-old son buy his own house.
Even when credit scores are correct, they may hurt some consumers by excluding human judgment.
Ralph Sweat, a retired street-maintenance worker in Lumberton, N.C., is far from an ideal credit risk. He declared personal bankruptcy three years ago, when the textile mill where his wife worked closed and the couple was temporarily unable to pay their bills. For much of last year, Mr. Sweat looked for a mortgage to buy a new house in Lumberton, but he says he was turned down by three conventional lenders.
He eventually did get a 30-year loan, however, at the fixed rate of 10.5%, from Self-Help Credit Union, an unusual nonprofit lender in Durham, N.C., that caters to borrowers with credit problems. Self-Help staff members say they spent more than 40 hours studying his recent utility bills, rent payments and other records to determine that he had kept current on his debts for the last two years.
Under an individualized- mortgage system, Mr. Sweat probably would have paid several percentage points more -- about what he would have paid had he gone to a subprime lender. That's because, as they are now structured, customized-loan systems generally don't include information about an applicant's recent utility bills or rent payments. And the speedy high-tech approach typically doesn't allow for painstaking human attention.
Some consumer advocates warn that if customized mortgages rely on automated-underwriting systems developed by Fannie Mae and Freddie Mac, minority applicants could suffer. The automated systems' statistical models incorporate data from millions of home loans made over the past 20 years. But during that period, blacks and other nonwhites received a disproportionately low percentage of home loans, at least in part because lenders generally avoided certain low-income neighborhoods, according to statistics gathered by the Federal Reserve and the U.S. Department of Housing and Urban Development. If customized- mortgage software relies on the historical data to disfavor the same areas, lenders, in theory, could perpetuate discriminatory patterns.
Fannie Mae and Freddie Mac officials maintain their automated systems are more fair than the traditional human-review process, because the computers don't consider race at all. Still, HUD is reviewing the Fannie Mae and Freddie Mac systems for potential racial bias.
Whatever that review reveals, some mortgage -industry veterans are wary of making too many distinctions among potential customers. Westar Mortgage , a lender and broker in Austin, Texas, so far is sticking to more traditional pricing methods, says President Scott Norman. "Lenders are going to tell you you're in a `good' county vs. a `bad' county?" he muses. "It just seems so unfair." |
