Naveed Chehrazi from the University of Texas and Thomas Weber from École Polytechnique Fédérale de Lausanne in Switzerland worked with a major credit card issuer to build their 'Dynamic Collectability Score'.

The score ranks delinquent account holders based on factors such as size of outstanding balance, mortgage status, past payment history, credit score, and external factors such as the performance of the stock market and the national unemployment rate.

The score continually adjusts as new data comes in, making it more accurate than other systems, says Chehrazi. "Each action that’s taken — from a collection phone call that goes unanswered to a partial payment that the bank receives — is factored in to revise, up or down, that person’s likelihood of future payment."

Currently banks have trouble working out who will pay back debts, say the researchers, usually involving third party collection agencies and selling the debt for pennies on the dollar.

Being able to accurately predict who will pay back makes it easier to know which accounts are worth spending money on — whether sending them to a collection agency or filing a lawsuit — based on the likelihood of repayment and the amount they can expect to recover.

Another benefit of the model is its ability to help banks work out their credit risk requirements, or the amount of money they need to have in reserve to cover future unpaid accounts.

Says Chehrazi: "The credit card collection problem is very complicated. And current bank internal scoring systems are surprisingly poor in predicting repayment behavior, given the amounts that are at stake."