Does financial market volatility identify bubbles and predict subsequent crashes? In their April 2017 paper entitled “Can We Use Volatility to Diagnose Financial Bubbles? Lessons from 40 Historical Bubbles”, Didier Sornette, Peter Cauwels and Georgi Smilyanov examine price volatility before, during and after 40 financial market bubbles to determine whether realized and/or implied volatility warn of bubble conditions and subsequent crashes. They focus on the following questions:
- Does volatility tend to increase during bubble maturation?
- Does volatility surge towards the end of a bubble?
- Can investors use volatility to diagnose bubbles and forecast their collapse?
They also evaluate credit conditions before and after bubbles to estimate whether leverage generally drives them. Their approach is event-oriented and graphical, centered on index level peaks preceding crashes. Using prices, realized volatilities, implied volatilities and ratios of credit from banks to the private non-financial sector to country GDP as available before, during and after 40 bubbles involving stock indexes, singles stocks, exchange-traded funds, commodity futures and currencies from the October 1929 Dow Jones Industrial Average crash through the September 2011 gold crash, they find that:
- The literature offers no consensus on the definition of a financial bubble, nor any widely accepted methodology for ex-ante identification. Econometric models of bubbles and crashes are of little or no practical use.
- Evidence does not support belief that rising volatility gives early warning that a bubble is developing or present. About two thirds of the 40 bubbles burst after times of relatively low volatility, suggestive of “calm before the storm.”
- Volatility does reliably spike during and after bubble bursts.
- Among 29 bubbles for which the specified bank credit data are available, 16 (13) occur when leverage is relatively high (is not high).
In summary, evidence indicates that volatility is not a useful indicator of the maturation and subsequent bursting of financial market bubbles. Sometimes it increases and sometimes it decreases, but most of the time it changes little as a bubble develops.
Cautions regarding findings include:
- The authors do not address exploitability of the volatility spikes during and after bursting of financial market bubbles.
- It is not obvious that the specified broad business credit metric relates to the leverage applied during financial market bubbles in various asset classes.
