FRM Part II · FRM Exam Part II · Credit Scoring and Retail Credit Risk Management
A card issuer finds that its scorecard, built three years ago, still ranks customers well (stable Gini) but the observed default rate in the 600-620 score band is 6%, versus 3% predicted at development. What is the most appropriate conclusion and action?
Discrimination is intact, as shown by the stable Gini, but calibration has drifted because observed defaults of 6% are double the predicted 3%. The appropriate action is to recalibrate the mapping from score to probability of default rather than rebuild the whole scorecard.
- ADiscrimination is intact but calibration has drifted, so recalibrate the score-to-PD mappingCorrect
- BThe scorecard has lost discriminatory power and must be fully rebuilt with new characteristics
- CThe population stability index must be zero, so no action is needed
- DThe scorecard is overfitted, so remove the highest-weighted variables
Explanation
A stable Gini indicates ranking ability is retained, while the observed default rate being double the predicted rate shows the PD levels are miscalibrated, e.g. from macroeconomic shifts. The appropriate response is recalibration of the score-to-PD mapping. A full rebuild is not required, and the data do not imply zero population shift or overfitting.
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.
More Credit Scoring and Retail Credit Risk Management questions
- A bank's risk committee is comparing its credit card portfolio of 400,000 small balances with a book of 40 large corporate loans. Which stat…
- A risk analyst evaluating a behavioral scorecard finds the Kolmogorov-Smirnov (KS) statistic is 45%. Which interpretation is correct?
- A lender builds a scorecard using only approved applicants who were later observed to repay or default. Rejected applicants have no performa…
- A retail lender has 10,000 accounts in the 30-days-past-due bucket at the start of a month. By month-end, 1,500 have moved to 60 days past d…
- A bank's unsecured personal loan portfolio has 20,000 accounts, each with an exposure at default of $5,000. The one-year probability of defa…
- 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…