Investors need a more accurate method of measuring and assessing volatility, according to Hermes Investment Management.
“Normal statistics will understate the risk of a large or catastrophic event.”Historical equity prices do not conform to the usual measure of volatility—standard deviation and normal distribution—according to analysis by Eoin Murray, head of Hermes’ investment office, and Ian Johnson, a performance analyst.
“Given the time-varying nature of markets, it is not unreasonable to question the legitimacy of using a normal distribution to measure risk in environments of prolonged high volatility,” wrote Murray and Johnson. “Accordingly, normal statistics will understate the risk of a large or catastrophic event. If we know that tail events occur more often than expected, it would be prudent to try to identify a method to predict the likelihood of such outcomes.”
The pair analyzed returns and prices based on a Hermes US equity portfolio and compared these to the VIX volatility index’s long-term average of 20%. They showed that, just as average returns vary for different periods of time, so does sigma, the likelihood of an outlier result.
The portfolio tested by Murray and Johnson had a “fatter tail” than in normal distribution models, the research showed, indicating that outlier results were more common than demonstrated by traditional models.
“Being aware of the propensity for tail risk to change unexpectedly as market conditions evolve provides us with a strong foundation for developing more astute ways of managing extreme downside risk,” the authors wrote.
The VIX’s January spike “put investors on notice that the most recent low-volatility era could be coming to an end,” according to Murray and Johnson. “Based on our research, we now know that a volatility regime change is likely to challenge the most-entrenched statistical assumptions. How we implement this new knowledge in our portfolios is the next big challenge.”
At the height of the sub-prime mortgage crisis in 2007 and 2008, several commentators referred to standard deviation when attempting to explain how unlikely the market crash was. As Goldman Sachs’ former CFO David Viniar told the Financial Times in August 2007, “We were seeing things that were 25-standard deviation moves, several days in a row.” This has since been estimated to represent a once-in-a-100,000-years event.
Read Eoin Murray and Ian Johnson’s paper, “Fattail Frequencies and the New Non-Normal.”
Related: Why Volatility Is Good for (Selling) Smart Beta & The Death of LDI