Skip to content

FRM Part II · FRM Exam Part II · Credit Scoring and Retail Credit Risk Management

During validation of a credit card behavioral scorecard, the model shows an AUC of 0.78, stable over time, but actual default rates in every score band are consistently about 40% higher than the model's predicted PDs. Which conclusion is most appropriate?

Discrimination is acceptable but calibration is poor. A stable AUC of 0.78 shows the scorecard still ranks risk well, yet observed defaults are consistently about 40% above predicted PDs, so the score-to-PD mapping should be recalibrated rather than the whole model being rejected.

  1. ADiscriminatory power is acceptable but calibration is poor, so the PD mapping should be recalibratedCorrect
  2. BDiscriminatory power has failed, so the model should be rejected on rank-ordering grounds
  3. CThe model is well calibrated because the AUC is stable
  4. DThe PSI must exceed 0.25, which proves the model is miscalibrated

Explanation

An AUC of 0.78 that is stable indicates the model still rank-orders risk well. Systematically higher observed default rates than predicted show a calibration problem, fixable by recalibrating the score-to-PD mapping. AUC does not measure calibration, and PSI is not implied by the data given.

Did you get it right without looking?

One question tells you little. A timed set on Credit Scoring and Retail Credit Risk Management shows your real accuracy, how long you take and where you lose marks.

Practise Credit Scoring and Retail Credit Risk ManagementFree diagnostic mock

More Credit Scoring and Retail Credit Risk Management questions