LoanPerformance Predictive Tool Targets Mortgage Firms
American Banker
Friday, March 2, 2007
Mortgage lenders increasingly are using technology to weed out or "scrub" loans that get sold in the secondary market.
A new product from LoanPerformance, an analytics and mortgage data unit of First American Corp. of Santa Ana, Calif., is one of the tools they are using to gauge a borrower's likelihood to prepay a mortgage in the short term.
Richard Harmon, the head of LoanPerformance's customer intelligence group, said the company wanted to create a simple technology that would allow mortgage lenders to take a list of borrower names and addresses and, without using Fico scores, determine whether a borrower had a propensity either to move, to refinance, or to pay off a mortgage.
Prepayment risk is a major factor in valuing mortgage securities and servicing portfolios, particularly as interest rates rise, he said.
PreTell, which LoanPerformance announced Feb. 20, analyzes borrowers' payment records and uses First American's property records and household demographic data to score 22 different mortgage products, from 30-year fixed-rate loans to subprime loans, Mr. Harmon said. After sifting through 900 pieces of data, it produces a score ranging from 1 to 1,000 that can be used to predict a borrower's likelihood of prepaying a mortgage, he said.
New Century Financial Corp., in Irvine, Calif., began testing PreTell in late 2006 to find out if there was a correlation between loan quality and the performance of loans sold to the secondary market.
David Hurt, a senior vice president of business development at LoanPerformance, said in an interview Thursday that both whole-loan sellers and buyers could use the PreTell scores to assess the risk of loan pools.
"The relative duration of payments could be extremely meaningful to their P&L in deciding which loans they want to keep in their portfolio, or put in their servicing portfolio," Mr. Hurt said. "It's something that they could arbitrage."
LoanPerformance has been marketing the product largely as a customer retention tool. Mr. Hurt said the servicing industry spends millions marketing to customers "but their campaigns are so broad-based that they don't deliver meaningful returns for the kind of money they're spending."
"Instead of doing mass mailings using huge customer lists, the idea is that servicing companies can reduce their mailing volume significantly, because now they're targeting borrowers who have a propensity to sell and need another mortgage product," he said.
Mr. Harmon said PreTell does not use credit data "in any way"; it allows institutions to make offers to customers without doing a costly credit analysis beforehand. Some companies also have been limiting their use of credit data in identifying risk because they want to avoid fair-lending requirements, he said.
Charles Stone, a vice president of customer initiatives at General Motors Corp.'s GMAC Mortgage, said that in a test PreTell saved GMAC $100,000 in mailing costs in the first half of 2006 by affording "better visibility into our servicing portfolio."