The Problem With Risk
At Foord, we have long espoused the view that volatility of investment returns is not a suitable measure of investment risk. However, the investment fraternity seems preoccupied with this flawed measure of risk in determining whether or not an investment fits a particular profile.
In financial theory, volatility is a measure of the dispersion of returns. The wider the range of returns recorded, the greater the volatility. Take a look at two asset classes, equities (jargon for listed shares) and cash. At the one extreme, equities have historically reflected the widest range of returns, while cash has the lowest. Extrapolating this logic, most investors will interpret equities as the riskiest asset class and cash the safest.
Following financial theory further, the measure of “bang for your buck” is the return earned divided by the risk taken (measured by volatility) to get there – this is known as the “information ratio”. The higher the ratio, the more return you are getting per unit of volatility taken.
The realisation has dawned on some that the volatility of cash is so infinitesimally small that it will always seem to be the greatest “risk adjusted” investment. Of course, other measures have followed to compensate for this: the Sharpe ratio measures the excess return over cash per unit of volatility taken, and the Sortino ratio measures the excess return over inflation per unit of volatility taken. Regardless, the denominator of these “risk-adjusted” measures remains historic volatility of recorded returns.
Therein lies the rub: there is no permanent correlation between volatility and return, volatility is not constant, and volatility in one period is almost useless in predicting volatility in a subsequent period. Furthermore (and fundamentally for the long-term investor), volatility is measured over a specific period (often short) thereby imputing a short-term mind-set to risk measurement for a long-term growth asset.
An examination of monthly returns on the JSE from January 1926 to January 2013 (a period of 87 years) by Darron West of the University of Cape Town provides some insight into the profound vagaries of this type of risk measurement. Over rolling monthly periods of 1, 3, 5, 10, 20 and 30 years it was shown that the returns in one period were almost entirely uncorrelated with returns in a subsequent period.
This finding augers well for the notion that “past performance is not a predictor of future performance.” However, this mantra does not appear to be articulated quite so insistently for volatility as a measure of risk, and yet the volatility of returns for the rolling monthly periods noted previously also bears little or no resemblance to the volatility of the returns in subsequent periods.
You may ask “But what of risk-adjusted returns?” Measured as “return per unit of volatility” over the same rolling periods, these too offer little or no predictive value at all, save perhaps in the case of the 30 year time horizons where the information ratio explains just short of 60% of the information ratio in a subsequent 30 year period.
To be fair, the investment industry has tried to make better sense of risk. Value investors prefer to refer to risk as “the risk of permanent loss of capital” – a forward looking concept. Proxies for this might include “drawdown” (i.e. the worst return over a period) and the frequency of negative returns over various periods. Neither of these is dependent on volatility or the dispersion of returns for their measurement, but both metrics remain backward looking.
Risk might also be the negative consequences of being wrong. This is a patently prospective notion that is not measureable. It can only be guarded against by careful management and a consideration of scenario outcomes and applying probabilities based on judgement. We use this concept widely at Foord when making investments for our clients.
So risk means different things to different people, but it most assuredly is not volatility.
It might fit neatly into a mathematical formula, but it is as variable and unpredictable as the returns it purports standardise. Measures of performance or efficiency that rely on volatility as a basis for risk measurement are inherently flawed and should not be trusted.
Yet, defective volatility metrics often form the basis of an adjustment made to a fund’s performance to account for the risk taken in generating returns. We find this practice quite absurd – we strongly believe that there is no valid way of risk adjusting returns. If you think about returns and risk clearly, it is evident that returns achieved over the long-term are after experiencing the risk. They are therefore by definition, already risk adjusted.
We are taught that the more risk we take the higher reward we should expect. This implies a linear relationship in which many believe. In our experience there is no such linear relationship – there are times when high returns can be earned with very little risk. Those who have been fooled by theoreticians into believing that volatility equals risk have suffered the consequences of lower returns.
Written by Paul Cluer, Director at Foord Asset Management.