Skip to content

FRM Part I · FRM Exam Part I · External and Internal Credit Ratings

A validation team computes the Accuracy Ratio (AR) of two rating models on the same portfolio. Model X has AR of 0.62 and Model Y has AR of 0.48. Which conclusion is best supported?

Model X discriminates better between defaulters and non-defaulters because its accuracy ratio is higher. However, AR measures only rank-ordering power, so it gives no evidence that Model X's probability of default levels are better calibrated than Model Y's.

  1. AModel X separates defaulters from non-defaulters better than Model Y, but this says nothing about whether its PDs are well calibratedCorrect
  2. BModel X has PDs that are closer to realized default rates than Model Y
  3. CModel X must have fewer rating grades than Model Y
  4. DModel X will have a lower default rate in every grade than Model Y

Explanation

AR, derived from the CAP curve, measures discriminatory power or rank ordering. A higher AR means better separation, but it is insensitive to the level of PD estimates, so calibration cannot be inferred. The other options claim things AR does not measure.

Did you get it right without looking?

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

Practise External and Internal Credit RatingsFree diagnostic mock

More External and Internal Credit Ratings questions