[R-SIG-Finance] Portfolio VaR and Asset VaR

[Brian G. Peterson]

Jan is correct.  Value at Risk does not have the property of being 
'coherent' in the sense described in Artzner's papers.

R does have a coherent portfolio VaR available.  You can call 
portfolio_method='component' in the VaR function in PerformanceAnalytics 
which will give you the portfolio VaR and how much each asset 
contributes to the overall portfolio VaR.

Regards,

Brian


On 06/03/2015 04:43 AM, Annaert Jan wrote:
> I think this is perfectly possible. For instance, if A to E are individual
> stocks and P is, say, an equally weighted portfolio of these stocks. If
> firm-specific risk is high relative to systematic risk (which is typical),
> firm-specific risk may be to a large extent diversified away in P. As a
> consequence, VaR of P may be (much) smaller than each of the individual
> VaRs.
> HTH,
>
>
> Jan Annaert
>
> From:  Christofer Bogaso <bogaso.christofer at gmail.com>
> Date:  woensdag 3 juni 2015 05:55
>
> Let say I have a diversified portfolio of 5 assets. The individual
> Asset VaRs for them are $A, $B, $C, $D, & $E. And the overall
> portfolio VaR is $P. Assumed all VaR numbers are reported in absolute
> number
>
> It appears that P is less than all 5 individual VaRs.
>
> Can that happen? I know that P < (A+B+C+D+E). However here in my
> calculation what happened is P is less than each asset VaR.
>
> Appreciate your view.
>
> Thanks and regards,