Having seen an uptick on volatility as measured by the VIX in the last few days, I find that this is a good time to talk about a phenomenon many people do not know. Volatility tends to be persistent, with low volatility leading to more low volatility, and high volatility leading to more high volatility. But more interesting still, volatility can feed on itself, with increases in volatility leading to further increases in volatility and decreases in volatility leading to further decreases in volatility. And there is a simple reason for this, which I will explain in this article.
It all has to do with how risk management is practiced today at all financial institutions. Central to the risk management is a method called "Value at risk" or "VaR". This is a method which, through statistics, tries to estimate within a given confidence range how much can an institution lose in any given day, taking into account its portfolio of investments.
How does VaR work?
How does this method work? Basically speaking, the method takes into account the volatility of each asset in the portfolio, and the correlation between each asset in a huge matrix. It can thus estimate how the portfolio as a whole will move if each asset experiences a move with a 1%, 5% level of rarity, while keeping the correlations with all the other assets in the portfolio.
Now imagine that for some reason, volatility increases for a while. This will in turn show up as an increased level of risk in all those institutions using VaR. A higher level of volatility will mean that the loss that can happen on any given day will be higher. The natural reaction will be to reduce overall exposure. This means that an increase in volatility can breed selling and thus more volatility. Worse still, if the selling happens over the whole portfolio, this will increase pressure in several assets at the same time, also increasing correlations between them. This explains how in times of volatility, correlations also usually go up.
Now, both the increased volatility and the increased correlations between assets again increase VaR. This leads to more selling, which in turn leads to more volatility and higher correlation between assets. And that is the mechanism which can easily make volatility feed on itself.
It also works in reverse
When the market is calm, there is usually a smooth drop in volatility. This leads to an ever lower VaR for the same level of exposure and thus gives managers the confidence to increase exposure. Better still, as the market is calm, the correlations between assets also drop. This works towards the same end result - a lower VaR, again allowing for more exposure.
One of the main risk management tools out there, the ever-present VaR, has a strong pro-cyclical nature, leading to increased exposure in calm markets and flashing sell signs in turbulent ones. Oddly enough, this works at least partially against sound risk management. Knowing how VaR works helps understand why the market's volatility behaves as it does.