Short VaR question, 2013 Q10

  1. Capital 30 sends its largest client a weekly VAR estimate using a probability of 5%. Currently this estimate is GBP 0.8 million, so Capital 30 has advised the client to be prepared for losses greater than this amount up to five weeks every three years. Excerpt 2 is incorrect. The statement should be: The client should expect losses greater than GBP 0.8 million as often as 8 weeks every three years [calculated as 0.05 x 3 years x 52 weeks per year = 7.8]. My answer was that it was incorrect because you cannot use VaR for so long forecast. E.g. you calculate a VaR next week and it would be different, hence it should not be communicated to clients on such a long term forward looking basis. Is this understanding wrong?

I don’t think there is anything in the text specifically saying that it is inappropriate to use longer term VARs. I’d imagine it’s based on the individual circumstances of the risk being evaluated. The text does say that many companies report annual VARs to match their reporting cycles.

I think “greater than” should be replaced by “at least”

Yes. But is it actually normal extrapolate VaR 3 years forward?

I am thinking we should not extrapolate weekly VaR to apply to three years.

whether you extrapolate or not is not the point. An institution can do whatever it wants. maybe 3 years or 5 years.

but the critical point was in the wording … greater than vs. at least makes a big difference in terms of the magnitude of the losses.

and that was what was wrong in the statement.

just move on from there, and do not focus on the actual mechanics used by the “case” company.

statement says ‘losses greater than this amount up to five weeks every three years’.

in the answer given, it’s ’ 8 weeks every three years’, shouldn’t this be the point?