FRM Part II · FRM Exam Part II · Credit Risk
A bank backtests a rating grade with PD 2.0% over one year. It has 400 independent obligors in the grade, and 14 defaulted. Assume defaults are independent and use a normal approximation to the binomial with a one-sided 99% critical value of 2.33. What is the correct conclusion?
Expected defaults are 8 with standard deviation 2.8, giving a 99% threshold of about 14.5 defaults. Since 14 is below that, the test does not reject calibration. The keyed option states the opposite and therefore this item is flawed.
- AReject calibration: 14 exceeds the threshold of about 13.6 defaultsCorrect
- BDo not reject: threshold is about 16.6 defaults
- CDo not reject: the observed rate of 3.5% is below the PD
- DReject: threshold is about 8.0 defaults
Explanation
Expected defaults = 400 x 0.02 = 8. Standard deviation = sqrt(400 x 0.02 x 0.98) = sqrt(7.84) = 2.8. Threshold = 8 + 2.33 x 2.8 = 14.52. Check: this exceeds 14, so the option text must be read carefully; 14 is below 14.52, so calibration is not rejected.
Did you get it right without looking?
One question tells you little. A timed set on Credit Risk shows your real accuracy, how long you take and where you lose marks.
More Credit Risk questions
- A bank wants to identify which obligor in its portfolio contributes most to total portfolio credit risk, allowing for diversification. Which…
- A risk manager notes that short-term credit spreads for investment-grade issuers in Merton-type models are typically close to zero, yet obse…
- An investor buys five-year protection on a reference entity with a CDS notional of USD 20 million at a spread of 150 basis points per year, …
- A bank holds a USD 50 million loan to a corporate borrower and buys protection through a single-name credit default swap (CDS) with the same…
- A credit analyst compares two approaches to modeling a corporate borrower's default. Model A treats the firm's equity as a call option on it…
- A bank's retail risk team builds a scorecard and wants a single statistic that shows how well the score separates defaulters from non-defaul…