See here (via here).
This is a timeless issue. The specific risk model under discussion isn’t the central issue. It never is. The central issue is that financial-risk models whose effectiveness depends on the accuracy of their assumptions about the distribution of securities-price movements eventually blow up. This is why “portfolio insurance” failed in (helped to precipitate) the 1987 crash, why Long Term Capital Management blew up, why Fannie Mae’s risk estimates vastly understated the real risk and why countless other “value at risk” schemes cause more problems than they mitigate. In simple terms, these schemes assume that in the event of portfolio losses you will be able to sell off your portfolio incrementally without incurring further large losses. In practice, the very fact that your portfolio is experiencing an extreme decline in value means there are no buyers except at lower prices and that further losses are probably inevitable: if the life boats are all on one side of your supposedly unsinkable ship you may still capsize if the passengers move there en masse. This is human nature and can’t be hedged away by invoking clever math, though clever people keep making this mistake (and will keep making it, because human nature doesn’t change).
In the long run the only reliable way to limit the risk in your market portfolio is to structure it so that you don’t lose money if the impossible happens. But this is expensive (insurance usually is), and it’s always tempting to lower your costs, and raise your short-run returns, by assuming you don’t have to worry about 100-year floods. The problem is that 100-year floods occur in financial markets every five or 10 years.
BTW, this is also why the notion of “stress testing” banks is fatally flawed. You cannot assess the risk of loss in a financial portfolio by asking what happens under conditions of moderate, i.e., likely, financial stress. If there is a systematic fatal weakness, however improbable, in your financial system the markets will eventually find it and the system will blow up.