University of Southampton
Using survival analysis methods to build credit scoring models
Abstract
dc:description.abstract<p>Credit scoring systems were originally built to allow organisations to measure how likely an applicant for credit is to default by a certain time in the future. In recent years the objectives of credit scoring models have shifted from choosing the customers presenting the lowest risk, towards choosing the customers offering the highest profitability. This thesis shows how using survival analysis tools from reliability and maintenance modelling allows one to build credit scoring models that assess aspects of profit as well as default.</p><p>In particular, this thesis looks at a number of extensions of Cox's proportional hazards model applied to personal loan data that make this technique a consistent and complete method for building a credit scorecard. Firstly, a new way of coarse-classing of characteristics using survival analysis methods is proposed. Secondly, a number of diagnostic methods to check the adequacy of the model fit are tested for suitability for use on loan data.</p><p>The inclusion of time-by-characteristic interactions is also proposed in order to account for non-proportional hazards and hence, extend the applicability of Cox's model.</p><p>Additionally, behavioural scoring models based on the proportional hazards approach are also developed. In conclusion, this thesis demonstrates how both behavioural and application survival analysis based models can be used to estimate the expected profit from personal loans, and can therefore be used to help lenders to move from default scoring to profit scoring.</p>
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2001
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Stepanova, Maria
Rights
- Language dc:language
- English, eng