News

Featured Publication

Industry Future Council Report 2011

Strategies for Predicting Default Study Debuts

Posted 12/18/07

Avoiding bad credit risk is a basic business performance activity for equipment finance companies, and the ability to forecast which transactions will fall into default results in a competitive advantage. A new Foundation study, Strategies for Predicting Equipment Lease Defaults, offers a more effective method for forecasting defaults. “The Foundation paper explored ways to use the models and techniques from other areas of the finance industry to provide additional insight and value to industry members,” says Melisa Carter, Foundation Trustee, from Key Equipment Finance.

The authors of the study, Levon Goukasian and Samuel Seaman of Pepperdine University, used the PayNet database as a base for their work. They found the composite PayNet Rating Score, a metric based upon conventional demographic variables and a company’s prior leasing/borrowing history, to be an outstanding predictor of credit risk in all three of the classification models investigated. They also discovered several other predictive effects worthy of note: geographic location, public/private ownership, government/non-government contracts, and ease of credit access.

Carter says, “In many ways, the study validates current industry practices as it notes that the Paynet score by itself is already a good default predictor. However, the study also shows that there are some additional factors that companies can include in the decision-making process to further improve prediction accuracy. Given the current challenging credit environment, I believe that industry members would be very positive about improving default prediction.” For example, in an industry that seeks to improve its accuracies one or two percent, the discriminate model that the study addresses heavily has yielded classification accuracies as much as 10 percent better than the competing logistic regression and neural network models.

“Their approach was very thorough, and applied not only from an academic perspective, but also addressed the real world of scarcity of data, and limitations of the models themselves,” adds Carter. “But this study could be the first in a series that explores how far the prediction accuracy could be improved." Because, one bad credit decision can negate the profits earned from many good ones, this study is sure to be a best-seller. The Default study is a great example of how the Foundation can bring tools to industry members to help them make better decisions and more realistic analyses, sums up Carter.

Donors receive this study free of charge. Non-donors may order a copy of the study from the Foundation online store for $200.00 at the Library site.