FRM Fun 22 (Find the error in a practice exam question), P1

David Harper CFA FRM

David Harper CFA FRM
Subscriber
P1 only.
GARP's 2012 Practice Exam Part 2, Question 7 features a table (see below) which contains assumptions for two loan exposures. A Part 1 candidate should be able to find the error:

0916_GARP2012_P2.7.png



Question: What is wrong with the exhibit above?
 

ShaktiRathore

Well-Known Member
Subscriber
According to property of binomial distribution of EDF,
the variance of EDF is given by EDF*(1-EDF)
or volatility of EDF is sqrt(EDF*(1-EDF))
therefore,volatility of Loan A=sqrt(.015*(1-.015))=12.155%
volatility of Loan B=sqrt(.035*(1-.035))=18.378%
The above values in the exhibit are nowhere closer to these values raise doubt if these are correct values.
Otherwise using the given values of volatilities of EDF the answer comes out to be 62184 which might not be the one.
thanks
 

David Harper CFA FRM

David Harper CFA FRM
Subscriber
Yes, agreed Shakti, thank you! I submitted this error to GARP (to be honest, I am very disappointed to see this kind of mistake on a sample; variance of PD/EDF is a P1 concept--albeit applied in this context in T6. Ong, where he very clearly uses the formula just as you have written [ONg 5.3, p 113]--I feel they should know better). Thanks for confirming my view!
 

Ashwin FRM

New Member
Correction: even for multi-state, multi-year model, binomial distribution should hold once the marginal default rate for that year/state are determined - because from that particular state it can either default or not, just 2 states.

Regards.
 

caramel

Member
GAARP seems to have quite a number of errors either in their practice exams or solutions . I found atleast 2 in the 2006 paper
 

ShaktiRathore

Well-Known Member
Subscriber
Garp it seems is not particularly concerned with such type of errors.there seems to be less scrutiny in such matters where Garp need to be more vigilant and meticulous. As Caramel pointed out errors in their practice exams which is really disappointing.
Garp should revise and carefully scrutinize exercise/practice questions for their theoretical correctness before publishing them. Certainly some effort is required on their part to see that everything is correct and upholds the standards of a global exam. Its about preserving the respectand genuiness of the exam.

thanks
 

caramel

Member
seems like schweser too has committed the same error, Suppose XYZ bank has booked a loan with the following characteristics , total commitment of $ 2 , 000,000 of which $ 1, 200,000 is currently outstanding . The bank has assessed an internal credit rating equivalent to 1"% default probability over the next year. draw down upon default is assumed to be 75%. the bank has estimated a loss given default at 40%. the standard deviation of EDF and LGD is 5% and 30% respectively . Calculate unexpected loss
 

David Harper CFA FRM

David Harper CFA FRM
Subscriber
caramel: agreed, but it's (way) more than two errors per exam ... try 20%+ as a long run average, pre-2009 is really bad, ~ 30% if you include imprecise statements. (I would not bother with 2006 or 2007, myself)

@Shakti: I could not agree more, it sincerely concerns me for the long-run health of the exam. Given they only publish 45 PQ per year (20 + 25), the questions should be rock-solid and rooted in the assignments. I think it ought to be embarrassing, frankly. I can't be any more vociferous myself, I submit a constant stream of corrections and suggestions (and offers, even, to organize our extensive errata database). My submissions are sometimes acted up (e.g., last year's formalization of a default compound frequency), for which i am grateful, but most of them are received for some sort of internal review by unnamed parties of unknown duration and I don't hear back; many of the errors keep repeating, which is just plain unacceptable, given the resources available. Take this question above: it hardly requires a lengthy committee review, it's obviously wrong , the determination and fix should be quick. The standard is just too low.

It's not the mistakes per se. I make plenty of mistakes. It's the attitude toward mistakes (do they want to fix them?) and the quality standard demonstrated in action, because that impacts the reputation of credential.

Thanks,
 
Top