VaR – Good for a Manager, But Bad for the Markets as a Whole – Call it the “Paradox of VaR Risk Management”

So if there’s a big market sell-off and as a response the VaR overreacts and shoots up, then many investors are kind of forced to sell because they have to stay within their VaR limits and this selling will then be done in an already collapsing market…rigorous use of VaR measures undermines the stability of markets.

It’s the type of risk management practice that works well as long as it is not needed; just like Bernanke observed after the credit crisis about their standard models that proved to be “successful for non-crisis periods”. 

-Harold de Boer
Transtrend Opalesque Roundtable series ’17 Netherlands

What may be good risk management for any one manager or small set of managers can be bad for the market as a whole. The popularization of VaR risk management techniques sews the seeds of its own destruction. Call it the “paradox of VaR risk management”.

If every manager adjust his positions based on changes in VaR, there will be a feedback loop between the reaction to volatility changes and market moves. If there is a volatility shock, positions will be closed and market prices will be forced lower. This shock will lead to more endogenous trading that will further increase the price impact of the initial shock. Everyone cannot be following VaR strategies or the price link with volatility will be extremely negative and create more volatility.

Another way of thinking about this volatility feedback loop is through the impact of higher portfolio leverage during the current low volatility period. A portfolio that is leveraged to reach a volatility target based on longer-term asset volatilities will need to delever if there is a volatility shock. The chance of a volatility leverage shock increases if volatility stays well below historical means. This can be viewed as a Minsky moment for volatility or leveraged portfolio management.

This shock impact will also increase if there is no government or monetary support to dampen volatility. We can expect the link between returns and volatility to be higher if our VaR story is true, but we cannot measure the forward sensitivity of this feedback loop. The market will only know it to be true after the next market crisis.