While publishing its last quarterly results, Morgan Stanley, the American bank, announced it had « modified its VaR model in order for it to be more receptive to the recent market conditions » (L’Agefi)... What’s behind this weird wording ?
The acronym VaR stands for Value at Risk. The function of that formula is to measure the market risks of a portfolio of financial instruments. In order to evaluate its risks in real-time, a bank enters into this model all of the financial assets it is holding on its balance sheet, including their amounts as well as their characteristics, and then gets as a result a maximum risk of losses. It helps the bank in controlling its level of risk and helps in determining the amount of liquidity it has to keep on hands.
On paper, the idea is quite appealing; the Bâle Committee has even adopted it in its banking regulations. But there is a root problem : this mathematical formula is based on normal law, otherwise named the Gauss Curve, or the Bell Curve, which does not account for extreme events (a brief explanation : this law functions well for unrelated events, like flipping coins, but not for inter-related events, as on markets, which produce a concentration of risks). Benoît Mandelbrot, the mathematician, first and then Nassim Taleb, in his book Le Cygne noir (The Black Swan), have denounced this recourse to the normal law that blinds us to those extreme events that are very common on the financial markets.
One way to compensate the normal law’s weak capacity of accounting for extreme events would be to use historical data that span many years, so as to integrate former shocks into our calculations. Keeping in mind the recent crashes helps in remaining prudent with those calculations.
But Morgan Stanley just did the very opposite ! Up to now, its VaR was based on a four-year span, meaning it included the September 2008 crash, and now it’s base on a one-year span only. And it appears that, over the last twelve months, everything is fine and dandy : no crash, a bullish stock market (thanks to central banks’ quantitative easing), GDP growth and unemployment going down in the USA (even if the numbers are « distorted »).
As a result, the cost of risk diminishes and the bank can then diminish its regulatory holdings in order to invest still more on the markets. In so doing, it increases its leverage or, if one is thinking like a good manager, its real risk. But what really counts for those regulators that have agreed to this modification is the amount of risk calculated by the VaR...
And then Morgan Stanley announces record upside results... so everything is alright ! And they are not the only ones playing this dangerous game, all the banks play it. Everything will be fine... until the day when a huge shock ultimately proves those calculations to be wrong. Then the banks will claim that they have scrupulously respected the banking regulations... before asking for yet another bailout from their States.