RAROC practice question II

[john.ophof]

Inv A and B have:

A 5 Million return both trades have 100000 face amount volatility 14
B 6 Million retun volatility 16

Based on 99% confidence level RAROC who is better?

Again here return / var in stead of Economic capital in denominator was this switched because of readings?
If we do this correct must we do like this.
Return / ((var (99%) - var (50%) )

John

 

[David Harper CFA FRM]

John,

can you help me with a reference to the question, where is this so i can give a close look. I think *maybe* these are doing a market-risk-type RAROC where the mean EL is effectively zero, but i'd like to look at the source so i can be careful about my response

as i've mentioned on other threads, RAROC has different flavors (variations) - the key thing is to ensure ratio consistency (i.e., is the profit-flow metric in the numerator claimed by the capital base in the denominator. This is the first principle of all performance ratios.). This RAROC probably meets that test but has been defined differently than the credit RAROC we study in the assigned Crouhy.

David

 

[john.ophof]

It is 2004 FRM question, RAPM method and maybe you are right since there is no holding period in the question it can be that the EL is 0.
If you have VAR with a confidence level you must have a mean loss too isn't it. Since it is based on a distribution.

John

 

[David Harper CFA FRM]

I don't think i have that source question. Anyway...

The absolute VaR = -mean + (volatility)(deviate as F[confidence])

and for short periods (daily), we assume (simplifying assumption) that mean = 0, such that relative VaR is just a special case of absolute VaR:
relative VaR = (volatility)(deviate as F[confidence]); i.e., implictly assume mean = 0 for short period

But i think the first-order question is, is this market VaR or credit VaR? The above is implicitly market VaR. If it is credit VaR, I do not *think* you can justify excluding EL from EC.

(although you can calculate RAROC differently, in theory, you can include EL in the numerator and, to be consisent, use just VaR() in the denominator. I haven't seen this used, but this would be a fair type of "gross RAROC;" i.e., profit including EL/VaR is okay itself b/c it is consistent. This goes back to: profit/capital is itself vague)

David

 

[vkaul1]

Liquidity VAR has a confidence level term as 1.95 for 95% confidence in the second term instead of 1.65 for 95% confidence in VAR.

 

[David Harper CFA FRM]

vivek,

why? it does not make sense as a two-tailed confidence, the confidence concerns widening of the spread not narrowing. Should be one-tailed; e.g., 1.645 @ 95%

David