operational VaR?

[ajsa]

Hi David,

Do you think if answer C makes sense? Should EL be deducted from 95%ile loss? or is this calc only related to Op VAR?

Thanks.


33. Find the operational VaR at a 95% confidence level given the following data.
Frequency Distribution
Probability Number
0.8 0
0.2 1
a. USD 9,000
b. USD 45,000
c. USD 91,000
d. USD 100,000
ANSWER: C
Steps to calculating operational VAR at a 95% significance level.
1. Calculate the expected loss. The expected loss = 0.2 * [ (0.75 * 20,000) +
(0.24 * 100,000) + (0.01 * 600,000)] = 9,000.
2. Next calculate the loss at the 95%ile. Start with the largest loss and work
backward. The probability of a $600,000 loss = 0.01*0.2 = 0.2%. The probability
Severity Distribution
Probability Loss
0.75 USD 20,000
0.24 USD 100,000
0.01 USD 600,000
2006 FRM Practice Exams 34
of a $100,000 loss = 0.24*0.2 = 4.8%. Therefore, the cumulative probability is
5% and the 95%ile loss equals $100,000.
3. Finally, the operational VaR equals the 95%ile loss minus the expected loss =
$100,000 – $9,000 = $91,000.
Reference: Risk Management, Crouhy, Galia, and Mark, 2001.

 

[David Harper CFA FRM]

Hi asja,

As you may suspect by now ... I think this question is bogus because either (c) or (d) is a valid answer given the information. It is not specific enough to answer.
Recall this other question, from GARP, that contradicts the above and *includes* the EL in OpRisk VaR: http://forum.bionicturtle.com/viewthread/1422/
...VaR can be absolute (absolute OpRisk VaR = UL + EL; i.e., loss from zero as baseline) or relative to the future expected value (relative OpRisk VaR = UL)

David

 

[sleepybird]

Hi David,
The link you provided above is no longer valid. Can you provide valid link?
Given the inconsistency (see http://forum.bionicturtle.com/threads/question-40-var.3522/#post-9489), I would like to know the BEST way to answer this kind of question on the actual exam: including or excluding EL?
What would be your advice?
Also, to your knowledge, was this kind of question ever on the actual exam in the past?
Many thanks.
Sleepybird

 

[David Harper CFA FRM]

Hi sleepybird,

Sorry, it's an old link (3+ years) and they are broken and I can't find the reference; e.g., this is similar but not the question.

• I don't have a definitive answer from GARP, sorry, I hope it resolves in the 2013 AIMs/method; see my input to them here.
• But I notice that the FRM Handbook Example 25.8 asks for OpRisk VaR and explicitly says "what is estimate of the VaR at the 95% confidence level, including expected loss (EL)"
  So I would certainly expect then to clarify include/exclude.
  
  Although, i will say, for Credit VaR and OpRisk VaR, i think the safe bet is to assume CVaR and OpRisk VaR include EL; e.g., such a CVaR is explicitly consistent with Basel II and, i think, such an OpRisk VaR is implicitly consistent ... thanks,

 

[sleepybird]

Hi David,
Thanks. Just to be clear:
1. OpVaR including EL = absolute OpVaR = we simply rank the loss distribution and find the VaR.
2. OpVaR excluding EL = relative OpVaR = we need to take an extra step and calculate EL and subtract it from the absolute VaR in step 1.

Is my understanding correct or the other way around? \
Thanks.

 

[David Harper CFA FRM]

Hi sleepybird, Yes, you have it correctly. Such that the original question above is really asking for a relative OpVaR; which is maybe not too surprising since the reference is Crouhy, who following an economic capital approach, would assume relative OpVaR since the assumption would be that VaR, covered by economic capital = unexpected loss; ie., excluding EL. But we can't assume this universally, imo. Thanks,