ACEMS PhD Student Jonathan Budd (pictured on the right) and ACEMS Acting Director Peter Taylor have just won the award for “Best Paper” at the Credit Scoring and Credit Control Conference in Edinburgh, Scotland, UK.
The conference at the University of Edinburgh Business School is a globally-recognised gathering of close to 400 credit risk analysts and pre-eminent academics from 40 countries. The conference influences how financial institutions decide which individuals and businesses to lend to and the lending criteria they use to make these decisions.
Out of 96 papers accepted for the conference, the paper written by Jonathan in collaboration with his PhD supervisor at The University of Melbourne, Professor Peter Taylor, finished as the best. It dealt with an issue that’s becoming increasingly important for banks: how they should set credit limits for “transacting” credit card customers, those customers who pay off their balances every month. With these customers, the bank only makes money from transaction fees paid by merchants.
The amount and availability of detailed credit card transaction data has soared recently, but there have been few attempts to utilise this data to develop new models for managing accounts. A question of interest is how a bank should determine the credit card limit of a customer, taking into account the fact that a small portion of the credit card limit has to be reserved for prudential purposes, with the bank receiving income only on the value of the purchases.
“Our approach was to directly model the behavior of an individual credit card customer, which enabled us to take advantage of the increasing availability of detailed transaction data and the information that it provides about a customer’s spending and payment habits,” said Jonathan. “This method enables us to understand what effect changing a customer’s credit limit has on both the bank’s profitability and the experience to the customer, and differs from the conventional credit scoring approaches of classification or machine learning.”
Jonathan said the model developed in this paper bears similarities to those used in inventory theory, like the “newsvendor” model. That’s the classic problem for newsvendors on how many newspapers to order on a given day, so that there are enough for every customer, with as few as possible left at the end of the day. But in this case, he’s talking about setting customer credit limits at low enough levels so banks can minimise their expenses, and the customer doesn’t hit or exceed that limit and get upset.
“The range of papers and presentations at the conference was very impressive,” said Jonathan, “and to be selected by a judging panel of representatives from across academia and industry for the Best Paper Award is a real honour.”