Journal of Credit Risk

Risk.net

A statistical technique to enhance application scorecard monitoring

Nico Kritzinger and Gary Wayne van Vuuren

  • A methodology was presented to enable the comparison of statistical performance measures more effectively between the development sample and post implementation data for the accepts population.
  • Comparing the statistical performance measures for the accept population between development and post implementation data for the application scorecard indicated significant differences. The Gini-coefficient at development was 36% with the post implementation data indicating a Gini-coefficient of 48%.
  • With the methodology and swap-set Gini-coefficient introduced, the swap-set Gini-coefficient was calculated as 45% which is a significant change from the original 36% and which compares more intuitively with the 48% Gini-coefficient of the post implementation data. The new measure is a significant change to what were generally used for application scorecard monitoring purposes for the accepts population.

Application scoring plays a critical role in determining the future quality of a lender’s book. It is therefore important to monitor the performance of an application scorecard to ensure it performs as expected. Attention has so far been focused on application scorecard modeling and assembly techniques. An area that has received less consideration is application scorecard implementation. Performance measures on the accepted population appear to change in predictive power after the implementation of an application scorecard. This paper introduces and demonstrates a statistical measure to track the performance of the accepted population after the point of implementation on a comparable basis against the development window of the application scorecard.

To continue reading...

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: