Mortgage executives are showing growing confidence in credit scoring as a tool to help their businesses.
At a recent housing and mortgage conference in Park City, Utah, several speakers described their increasing involvement with or reliance on credit scoring.
One of them, William D. Dallas, president and chief executive of First Franklin Financial Corp., San Jose, Calif., said he was now using credit scoring exclusively to underwrite and price loans, all of which come from brokers.
First Franklin has converted from being a prime lender to subprime because of the narrow profit margins in the prime market. And Mr. Dallas said credit scoring is preferable to relying on information in customer applications.
"Applications are just full of fraudulent statements," he said. But his credit scoring system, which uses the Federal Home Loan Mortgage Corp.'s Loan Prospector and Standard & Poor's score-based credit gradings as well as some internal features, relies almost entirely on information from third parties.
Mr. Dallas also said he believes many loans underwritten as prime are actually of B quality and have been mispriced, undercutting the profitability of the A market. Scoring reduces this problem, he said.
Other speakers said scoring technology had outgrown the controversy over possible issues of fairness to minorities and was now regarded as an effective way to guard against bias charges.
Jim MacLeod, senior vice president for field operations at Mortgage Guaranty Insurance Corp., Milwaukee, said it had developed its own scoring model, which supplements consumer credit data with additional loan and borrower information and market or property information. He said MGIC's model increases the predictive value of the mortgage scores.
He said the lowest (least favorable) quintile of credit scores as calculated using Fair, Isaac & Co.'s model, called Fico, had eight times more foreclosures than the highest quartile. The MGIC model, however, had 34 times more foreclosures.
Mr. MacLeod said he expected the use of mortgage scores to spread to many more lending functions. They are applicable to portfolio management, useful in monitoring the performance of brokers and correspondents, and in quality control.
