Skip to content

FRM Part II · FRM Exam Part II · Credit Scoring and Rating

A lender's application scorecard was developed on accepted applicants only. Management plans to apply it to all through-the-door applicants, including those previously rejected. Which technique is designed to address the resulting sample selection bias, and what does it do?

Reject inference addresses the bias. Because rejected applicants have no observed repayment performance, the technique infers likely good or bad outcomes for them and adds them to the development data, so the scorecard better represents the full through-the-door population. Resampling methods do not correct for missing outcomes.

  1. AReject inference, which assigns likely good/bad outcomes to rejected applicants so the development sample better reflects the full applicant populationCorrect
  2. BBootstrapping, which resamples accepted applicants to enlarge the sample and reduce variance of coefficients
  3. CCross-validation, which splits accepted applicants to test for overfitting in the scorecard
  4. DCutoff tightening, which raises the minimum score so only applicants resembling the accepted sample are approved

Explanation

Models fit on accepted accounts suffer selection bias because rejected applicants have no observed performance. Reject inference infers outcomes for them (for example by augmentation or parceling). Bootstrapping and cross-validation use only accepted accounts and do not fix the bias.

Did you get it right without looking?

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

Practise Credit Scoring and RatingFree diagnostic mock

More Credit Scoring and Rating questions