Abstract

Optimization of Collection Efforts in Automobile Financing Business –
A Data Mining supported Environment

Harald Kauderer

DaimlerChrysler AG
Research and Technology FTK/A

This contribution presents a project realized in the domain of risk management at a financial services subsidiary of DaimlerChrysler. The project resulted in the implementation of an innovative Data Mining approach designed to optimize the collection efforts in automobile financing. In contrast with the conventional "credit scoring" – which is mainly geared towards assessing a customer’s credit worthiness at the time when she/he applies for a loan – in this project we followed the credit life cycle downstream and focused on how to best deal with delinquent customers in an actual credit portfolio.

In traditional automotive lending "collections" generally refers to all the processes, tools and activities carried out to deal with delinquent customers. Strategies to "collect the money" usually extend from sending a letter, making phone calls to repossessing vehicles or any (sequential) combination thereof. The selection of the "right" strategy or treatment though is a function of the individual riskiness of each customer at a particular point of time.

In this project the business objective was to optimize the risk-adjusted allocation of delinquent customers/accounts to different collections treatments and levels of collector experience in a phone collections department. The resulting Data Mining solution was implemented in the finance company’s systems environment and in daily operations.