Track 2 Session 2

10:50 to 11:50 a.m. Wednesday, October 13, 2010

Portfolio Lifetime Loss Prediction in the Consumer Lending Industry - A Comparative Study of Parametric vs. Other Hazard Models

In the consumer loan business, a reasonably accurate and dynamic assessment of a customer’s future payment behavior is necessary to manage their credit risk effectively. This necessitates estimation of the timing when the loan might experience a default event. To have better forecasts of future losses, one needs to develop a dynamic model that reflects the economic changes that may occur during the term of the loan. This presentation shares a comparative analysis of various modeling approaches to estimate the timing of the consumer loan default by controlling product, customer and market conditions. The following topics will be covered: 1) A brief introduction to the consumer lending business and industry loan lifetime loss estimation practices. 2) Pros and cons of parametric and other hazard modeling approaches for estimation of a loan default incident timing. 3) Model simulation results. 4) Potential modeling errors and their impacts. 5) Sharing best practices across the industry.

Key Words: Parametric, Semi-Parametric, Default, Loss, Consumer Lending

Balakerthy Punyakoti, Naresh Chakkula and Dheeraj Awasthy

HSBC Technology and Services