Presentation Type
Event
Student
No
Track
Other
Abstract
Understanding risky behavior associated with a given cardholder is crucial in managing a successful portfolio in the credit lending industry. Measuring this risk is done at many different time points in the customer lifecycle. This paper explores constructing a model to accurately predict the risk of a cardholder for the lifetime of the account. First, a brief introduction to the customer lifecycle will take place. Next, the structure of the data and the issues associated with it will be discussed. After this, Survival Analysis methodologies will be overviewed and applied to a case study from CAPITAL Card Services. Next, an extension of the single event framework to account for multiple events will then be discussed. Finally, an application of these models to address problems that arise in the credit card industry will be examined.
Start Date
2-5-2019 3:30 PM
End Date
2-5-2019 4:30 PM
Survival Analysis Methods to Predict Loss Rates in Credit Card Portfolios
Pheasant Room 253 A/B
Understanding risky behavior associated with a given cardholder is crucial in managing a successful portfolio in the credit lending industry. Measuring this risk is done at many different time points in the customer lifecycle. This paper explores constructing a model to accurately predict the risk of a cardholder for the lifetime of the account. First, a brief introduction to the customer lifecycle will take place. Next, the structure of the data and the issues associated with it will be discussed. After this, Survival Analysis methodologies will be overviewed and applied to a case study from CAPITAL Card Services. Next, an extension of the single event framework to account for multiple events will then be discussed. Finally, an application of these models to address problems that arise in the credit card industry will be examined.