With delinquency rates and credit card losses at record highs, a system for predicting cardholder behavior prior to bankruptcy would prove a valuable tool for banks. Now that tool is at hand.

Fair, Isaac & Co. has developed a scoring model that predicts bankruptcies up to 18 months before they happen. Using several sources to research bankruptcies, including Trans Union's consumer credit database, Fair, Isaac created "Horizon," which is now available through Trans Union's on-line, prescreen and account management delivery services, as well as Fair, Isaac's ScoreNet and PreScore Services.

Horizon not only pinpoints accounts likely to produce losses, but also those likely to prove profitable, says Fair, Isaac product manager Careen Foster. "The trick is to tease apart the cardholders likely to go bankrupt from the heavy users that will increase profits. The problem cardholders tend to have more revolving accounts, more recent bank card openings, and their total revolving balances and use of cards is higher."

Fair, Isaac's research into pre-bankruptcy cardholder behavior revealed consistent patterns that occur prior to bankruptcy. Pre-bankruptcy cardholders increased their card purchases and reduced their monthly payments beginning a full 18 months before filing for bankruptcy. Conversely, half of a sample group of bankrupts made no bank card purchases in the six months prior to bankruptcy. The conclusion: A six-month or greater lead time is necessary to see bad behavior in time to reduce losses.

Banks can cut losses by holding back credit increases and over-limit authorizations. "That's where Horizon can help," says Foster. "It can also be used to weed out problems when consumers apply for credit."

According to Moody's Investors Service, banks wrote off seven percent of credit card account balances this past May, more than one percent greater than in May 1996.

-peterson tfn.com

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.