Proposed Project for the MSU Industrial Math Students

Auto Loan Volume Response Model

Price revenue management is increasingly being employed in the financial industry to set interest rates and terms for consumer loans. A successful revenue management program optimizes price (interest rate) by formally incorporating the behavioral response of consumers into the pricing decision. The goal is to establish the optimal trade-offs between volume and return, while at the same time achieving company objectives for portfolio risk and composition. Accurate and robust volume response models are critical to price optimization, but they present an ongoing challenge in practice as economic and competitive conditions can change rapidly in the marketplace.

The objective of this project is to develop volume response models which can be used to predict the take up rate of approved auto finance applications as a function of the rates that are offered. These models will be implemented in an optimization tool that will recommend optimal rate sheet rates. A requirement of the tool is that the rate variable must have a negative slope. That is, volume or take up rate must decrease with an increase in rate. It is of particular interest to understand the causes of positive slopes in the response model as well as techniques to address the issue.

Various model forms and analytical techniques should be investigated and evaluated. (For example modeling can be done at either the aggregate or the contract level)

An additional objective is to outline best methods for evaluating competing volume response models in terms of overall volume accuracy as well as accuracy of volume distribution between risk bands.

Top of Page