|
The Recession Question Has prosperity become credit scoring's Achilles heel? That critical issue is now being debated among bankers, regulators and outside experts. Today's scoring models rely heavily on data gathered during a decade of booming economic growth. Will those models hold up during a future recession? "The scorecards that are being used today are creatures of the '90s," says Jeffrey Brown, director of risk analysis for the Office of the Comptroller of the Currency. "They have not been tested with the experience of a real economic downturn, so in many cases we don't know what will happen in a recessionary environment." The ramifications of this debate extend far beyond a small circle of statisticians. Scoring models used to predict loan repayment are constantly evolving to accommodate new products and changes in borrower behaviors. They work best when they incorporate enough experiences with troubled credits to allow lenders to exercise their judgment within realistic parameters. Yet in some of the newer scoring arenas, such as small business or home-equity lending, the models may not incorporate sufficient negative data to work at peak efficiency. Even in areas where scoring has a long history, such as credit cards, the age of the information relating to prior periods of default stress raises questions about the predictive strength of the models. "It's unlikely that anyone today is using a scoring model that has the same level of negative data as those constructed during the 1980s," says Yasmine Avani, director of risk management for North American cards at Citibank. In a recession, "it's difficult to say how resilient and effective some of those scores will be." Her advice: monitor portfolio performance closely to ensure that the results come in as predicted. Regional variations must also be taken into account. Michael James, group executive vice president for Wells Fargo & Co.'s small-business and consumer lending group, notes that California took longer than the rest of the nation to rebound from the early '90's recession. Scoring models used in that region had to be adjusted accordingly. The scoring debate won't be resolved until a recession finally arrives. But Brown says the issue does highlight a glaring, if surprising, weakness of scoring models: for all of their ostensible power, no one has yet come up with a useful gauge for how their performance is impacted by the economy. "If somebody could find a clever way to figure economic variables into the equation, that would be really useful to all of us," Brown says. Close this window to return to the article. |