At last week's Performance Measurement Forum meeting in Rome I mentioned how during this most recent economic crisis the Value at Risk metric has demonstrated how little value it provides: what firm's use of this measure provided them with any degree of accuracy? And yet the measure clearly has its supporters.
I am wrapping up an article for the New York Society of Security Analysts' (NYSSA) journal on this topic: specifically the benefits and shortcomings of VaR. In a nutshell, the measure on the surface seems like an excellent one as it offers a very intuitive view of a portfolio's risk: "the most you can lose is $5 million..." Simple. Easy to grasp. And, it's a forward looking measure as opposed to one that says what the risk was. How better to report risk? There are, of course, two other bits of information that go along with such a report: "...over the next week at a 95% confidence level."
The addition of the time period only enhances VaR's value. BUT, the confidence level can be a bit confusing. What does it mean? Well first, the 95% shows us that this is the worst possible loss that can occur within this range; the missing 5% means that it can, in reality, be worse ... perhaps a lot worse.
I won't go into more detail on this topic here; you can "read all about it" when my article is published. Suffice it to say, the VaR critics have been having a field day since the most recent market downturn hit a year ago.