Consumer lenders, operating in a market of tight margins and increasingly high risk, are leaning more heavily than ever on credit scoring and related technologies.
Credit scoring - assigning to a potential customer a number indicating creditworthiness - is not new. Credit card enterprises, among others, have used the process for years.
As consumer loan delinquency rates and personal bankruptcies mushroom, banks are looking to credit scoring and related processes to help them make good loans in the first place.
"I think a lot of banks are moving toward fully automating credit decisions and letting systems make decisions, as opposed to people making decisions," said Stephen Darsie, senior vice president at CCN Inc., an Atlanta developer of decision-support software. "There's a lot of ways to do it, but it all relates back to having a scoring system."
Banks generally are reluctant to talk about how credit scoring and automated decisions affect their bottom lines.
But Mr. Darsie said credit automation can let banks raise origination volumes 7% to 15% "without any corresponding increase in delinquencies or loss rates."
Other forces behind the rising use of credit scoring include laws such as the Community Reinvestment Act and mergers and acquisitions.
The top 100 banks spent about $5 million on third-party credit scoring software in 1994 and are likely to quadruple such spending by 1999, according to William Bradway, an analyst at the Tower Group, Wellesley, Mass.
Investments in proprietary software at those institutions are probably similar to the expenditures on third-party packages, he said.
The providers of third-party credit scoring systems include CCN and Fair, Isaac & Co. of San Rafael, Calif.
Though each company's technology is different, the credit scoring process is fairly standard.
Credit files, supplied by bureaus such as Equifax Inc., Trans Union, or TRW, give banks core information about a lending applicant's history.
Bankers use this information and data supplied by the applicant and other third-party sources to determine the "three C's:" collateral, capacity, and character.
The recent problems with delinquencies and bankruptcies are attributable to the last of the three, experts said.
"What we've seen in the last few years is an erosion in character," said Collin McKenny, president of card services at Star Banc Corp., Cincinnati.
Ms. McKenny said she believes character - defined as willingness to pay - is perhaps the most important of the three in determining whether the bank has made a good credit decision.
The bank counts on credit scoring to find customers who, despite "personal adversity," will pay their debts. "It's more an art than a science, and that's why it's as challenging as it is," she said.
"The art is being able to identify the person with character," she continued. "If I figure out how to do it, I'll patent it and make millions."
Since 1989, Star has taken an aggressive approach - automating credit approval operations through an artificial intelligence system it developed with Cybertek Cogensys of Dallas.
The system, called Galaxy, digests credit application data, pulls in credit bureau information, and makes a lending judgment after comparing the application and credit score against the bank's data base of customers and rejected applications.
"If it flat-out has never seen those facts before, or cannot make a decision because the probability is too low, then it will kick it out on a referral," Ms. McKenny said.
When an application is referred, a credit analyst steps in to make a decision - but not before keying in information to Galaxy that can "train" it to deal automatically with a similar future application.
Star's Galaxy system processes up to 10,000 applications a month, in as little as "a matter of seconds," she said.
In 1992, the system referred 15% of all credit card applications. This rate dropped to 5% last year. "It's getting smarter," Ms. McKenny said, adding that the technology had helped the company cut its department staff in half, to 15.
"It's had a huge productivity impact," she said. "We've even automated the decline process so that if customers do not meet the unsecured threshold they are automatically offered secured cards as a counteroffer."
In addition to the bank's credit card operations, Galaxy is used for other consumer loans and may even be used by the bank's mortgage department soon.
Ms. McKenny is pleased at the bank's experience with artificial intelligence but admitted it had been difficult to gather enough data for the system to be effective.
Star plans to sell its system to other banks, with the help of Cybertek Cogensys.
"We plan to 'market our brains,' so to speak," Ms. McKenny said. "We think there might be some opportunity there early next year."
She said she finds credit scoring invaluable in the bank's effort to predict consumer credit quality. The scorecards play a large role in the bank's effort to price its services according to the risk posed by a particular credit.
Risk-based pricing has taken on new importance with the rising presence of subprime lenders.
"It's probably been the single greatest growth area in the whole consumer credit market," said CCN's Mr. Darsie. Subprime lenders, which also rely heavily on scoring technology, often grant loans to people with damaged credit, bankruptcies, or other credit problems, he added.
In addition to helping lenders in decision-making and pricing, credit scoring is used by many banks in portfolio management.
Harris Bankcorp, Chicago, recently began using scoring to manage its home equity loan portfolios.
"We use credit bureau data to look at a customer's performance in general," said Judith Shulist, vice president. The bank's system sifts credit bureau data for delinquencies in other obligations, such as unsecured cards.
Harris also runs new checking account customers through Equifax's credit scoring system, called Decision Power, to preapprove customers automatically for overdraft protection.
Last year Harris also began to buy customized credit bureau information in search of preferred customers for customized direct mailings.
"It's a great way to save marketing dollars," Ms. Shulist said.
Observers said banks' mortgage areas could benefit the most from greater use of credit scoring and related technology.
Decision-support vendors like CCN and Fair Isaac are scrambling to design software that helps streamline mortgage lending.
Last year, the Federal National Mortgage Association and the Federal Home Loan Mortgage Corp. approved the technique. More recently, the Federal Reserve published a report showing the benefits of its use.
Larry E. Rosenberger, president and chief executive officer of Fair Isaac, said it has taken longer for the mortgage industry to embrace credit automation because there are "fewer sample points" against which applicants can be compared than in credit card or auto lending.
"There is a great degree of skepticism" among mortgage bankers, he said. "I think that for a long time the mortgage industry thought the technology would not add value to their decisions. I believe that's changed pretty dramatically recently."
The opportunities "are great on the mortgage side," agreed Yasmine D. Anavi, vice president of credit risk for Chase Manhattan Corp.'s national consumer finance business.
"The point in making consistent credit decisions is very important," she said. "That's a big advantage in using credit scoring models: It's fair, objective, consistent, and fast."