Credit Scoring Models Get Analytical Adjustments

Sept. 11, 2001, changed everything - including banks' loan portfolios. "After 9/11, the whole economy changed," says Christine Pratt, analyst at Tower-Group (Needham, Mass.). "The mix in the portfolios changed, the borrowers changed, and the risk" changed.

While some lenders battened down the hatches, others took on greater risks. But banks' risk models were calibrated during better economic times and did not quickly adapt to the new reality. "People began to realize that the scoring models didn't work as well as they thought they did," Pratt says. "There is much more of an impetus to make sure that these models are constantly tested, refreshed and refined now."

Part of the solution has come through better management of customer information. "I've seen a nice uptick in the number of organizations that are looking for new collections and recovery systems, or are looking for tools to help them segment their servicing portfolios, so that they can get a better handle on what's going on," Pratt says. "People have started to say, 'We've had this same collections and recovery system for years now - it's about time we invested in some new systems.'"

For example, Intelligent Results (Bellevue, Wash.) offers a solution that incorporates unstructured data gleaned from collections agents. Banks can use this information to decide whether to sell the loan to a collection agency or hold it in the hope that it turns back into a conforming loan. "Until now, most of that information has been resident in the collectors' notes," Pratt says.

According to a survey conducted by the Consumer Bankers Association, the major vendors in the collections and recovery market include CGI Group (Montreal), which earlier this year acquired American Management Systems; Fair Isaac (Minneapolis); and Fidelity Information Services (Jacksonville, Fla.). "These are all mainframe solutions," Pratt notes.

Let's Make a Deal

A good collections and recovery system "can actually predict the right treatment for each customer," Pratt continues. In the same way that a CRM system can predict which offer is best for a customer, collections and recovery systems can use analytics to determine whether it's worthwhile to work out a payment arrangement with a customer, and at what terms.

Also, at least one major credit organization is using analytics to determine which customers are likely to struggle to make payments. "They've used that information to start a proactive customer approach," Pratt explains.

In the current market, smoother collections and recovery are important to a healthy banking strategy. "About 28 percent of the people who have loans with you can be expected to show up at collections at one point or another," Pratt says. "You don't want to give up on your customers."