Measures of market volatility have traded well below their longterm historical averages in recent years, partly because extremely lax central bank monetary policy anaesthetised investors against the prospect of big market corrections. In early February, however, equity markets corrected sharply as attention focused once again on big upward moves in volatility. Since equity volatility tends to be mean-reverting, we suspect that it reflects a shift away from the complacently low levels which prevailed throughout 2017. Nor is there much we can do to offset the move, since volatility hedging strategies tend to be very expensive. All we can do is ride out the storm.
When giving lectures on non-linear mathematics, the mathematician Stan Ulam used to start by claiming he was embarrassed by the inappropriateness of title. After all, nearly all mathematics concerns the study of non-linear phenomenon and only that which is trivial can bedescribed in purely linear terms. So it is in markets, where financial asset prices never run in straight lines – at least not for long – and the processes driving price movements tend to be driven by complex non-linear processes.
As a result, we are unable to consistently forecast asset returns which tend to show a significant degree of volatility. The higher the volatility, the greater the degree of risk associated with the outcome, and finance theory has long used the variability of returns as a measure of markets’ risk perception. But today we can go even further. Financial markets have developed a series of instruments such as futures, swaps and options designed to reduce the degree of exposure to unforeseen market moves. By measuring the prices which investors are prepared to pay to hedge against market volatility we can put a price on risk perception.
Risk, uncertainty and volatility are inherent features of everyday life – there is a stochastic element to everything we do. Although we can put a price on financial risk, measuring it in a wider context is much more difficult. However, one widely used measure is the Economic Policy Uncertainty Index1 which tracks newspaper coverage of policy-related economic uncertainty and disagreements amongst economic forecasters as a proxy indicator. This particular measure of risk spiked sharply in 2016, in the wake of the Brexit referendum and the election of Trump, but has recently corrected sharply downwards as we head back towards the pre-2016 average (Chart 1).
Number of standard deviations from the mean (1997–2015)
However, market measures of risk appear to have run in the other direction. Given the elevated nature of global risk and asset price valuations, one of the enduring puzzles in financial markets of late is why measures of market volatility remained so low. The most closely watched equity volatility series is the Cboe Volatility Index, or VIX, which measures the weighted average of put and call options on the S&P 500 as an estimate of 30-day expected equity volatility. In November 2017, the VIX fell to a historical low based on daily data back to 1990, suggesting that investors had little fear of a significant market sell-off. The MOVE Index of bond market option volatility, calculated by Bank of America Merrill Lynch, also registered a historical low around the same time on a time series of comparable length (Chart 2).
In early February, equity markets corrected sharply lower, and attention turned to the spike in measures of option volatility, notably the VIX. We did point out in ‘Thinking Ahead’ in January 2018 that ‘one of the biggest risks on most investors’ radar screens is the possibility of a more severe-than-anticipated market correction. The combination of elevated valuations and the prospect of additional Fed tightening are certainly issues to bear in mind, particularly in view of the apparent lack of market concern regarding risks. Indeed, measures of equity option volatility such as the VIX have traded at historical lows during 2017, which does not appear justified by mounting geopolitical risk.’
The data series on which we base our analysis only extends back some three decades. In order to generate a longer time series, we have constructed a historical proxy based on the coefficient of variation of the S&P 500 (the standard deviation relative to the mean). Over the calibration period 1990 to 2017, the synthetic index tracks the VIX pretty well, with an average error close to zero. Backcasting all the way back to the beginning of the twentieth century, the synthetic index averages 20.4 over the period 1901 to 1989 compared to an average of 19.3 for the VIX series measured between 1990 and 2017 (Chart 3). This would appear to support our suspicion that recent equity volatility levels were unprecedentedly low.
In order to explain this, we need to identify prevailing conditions which are historically unusual. One of the most obvious differences to the past is the level of interest rates, which have spent much of the last decade at all-time lows. This has inflated the discounted present value of earnings which has supported ever rising equity markets. In effect, investors have discounted the prospect that equity markets will fall. A second possibility is that the rise of smart-beta and risk-premia strategies has introduced additional distortions into the market that previously did not exist. An increase in the supply of products offering such strategies depresses realised volatility via hedging activity, and as more are issued, the more volatility is depressed.
But both of these factors are likely to prove less supportive in future. On the one hand the degree of monetary support is about to come to an end. The Fed has raised interest rates five times during the course of this cycle which has pushed the yield on two-year Treasury notes above the dividend yield on the S&P 500. Such a trend will eventually erode the belief that equities are the only game in town with the result that volatility begins to slowly edge up. To compound the problem, since the smart-beta and risk-premia products operate on the basis of leverage, small increases in volatility tend to be magnified with the result that measures such as the VIX can snap back sharply. This might explain why equity volatility began to edge up slowly during January 2017 before surging in early February to levels only surpassed once in the preceding six years.
Whither equity volatility?
An examination of recent trends raises a number of key questions. First, is volatility mean-reverting? If so, neither the extremely low levels of 2017 nor the elevated levels of early-February 2018 will be sustained. Second, if market volatility measures do move back towards more ‘normal’ levels, how quickly is this likely to occur? And third, is it possible that the trend volatility level has changed?
With regard to the first question, the post-1990 evidence does suggest that equity volatility is mean-reverting although it can diverge from the mean for a considerable period of time. With regard to the second issue, on average since 1990 each period of over- or undervaluation relative to the mean lasted for 17 months, which suggests that the period of adjustment is relatively slow. It is also notable that 88% of the time since 1990 the VIX was within one standard deviation of the mean (although on only 38% of occasions was it within half a standard deviation). One standard deviation represents a seven-point move in the VIX, which is a relatively tolerable move. It is only when we see the kinds of spikes associated with the bursting of the tech bubble between 1999 and 2002, or the post-crisis period of 2008-09, would high equity volatility threaten to derail the markets more seriously.
However, there is a risk that an extended period of low volatility sows the seeds for a period of higher volatility. Lower volatility during periods of economic upswing tend to result in higher risk-taking and excessive leverage, with the result that even small price declines can force investors to dump asset holdings, depressing prices further and generating higher volatility. This triggers a second round of price declines and volatility spikes which could turn into a self-reinforcing spiral. But as it currently stands, despite the sharp spike in equity volatility in early February, the forward volatility curve is pricing in a decline back to levels close to the long-run average over a five-month horizon (Chart 4). This downward sloping volatility curve is not indicative of a market which is expecting a significant change in risk conditions.
VIX forward curve
As for the third question of whether there has been a shift in the trend level of the VIX, and therefore a shift in investor risk tolerance, the jury is still out. We will probably only know after a prolonged period of tighter monetary policy. The four most dangerous words in finance are ‘this time it’s different’. Any data series which shows strong mean-reverting trends should be treated as such until we have overwhelming evidence to the contrary.
Profiting (or otherwise) from volatility risk
In recent years investors have woken up to the possibility of including volatility indices in their personal portfolios, and numerous exchange traded funds (ETFs) and exchange traded notes (ETNs) have been set up to facilitate such interest. These indices normally move inversely with the market so when volatility is falling, the ETFs/ETNs will generally rise. One such note is the VelocityShares Daily Inverse VIX Short-Term ETN (XIV) which increased by a factor of 13 between January 2011 and January 2018 but collapsed by 95% in early February as volatility spiked up (Chart 5). Such moves highlight the risks associated with inverse volatility indices: they work well when volatility is relatively stable but are subject to huge declines when volatility jumps sharply, as was the case recently. Moreover, the underlying index, in this case the VIX, is theoretically unlimited to the upside which highlights the extent of downside risk.
Clearly, they are not suited for personal portfolios unless the individual in question has a huge risk appetite. But given the damage that volatility spikes can cause, is there a case for institutional investors to use similar instruments for hedging purposes? Instruments such as the iPath S&P500 VIX Short-Term Futures ETN (VXX) and the iPath S&P500 VIX Medium-Term Futures ETN (VXZ) have been designed with such a use in mind.
Whilst there are some advantages, volatility hedging strategies have proven to be more costly than they are worth in recent years. In order to understand why this is the case, we need to look at the structure of the volatility futures curve. As Chart 4 indicates, at the end of 2017 the curve was upward sloping – a situation which is typical of the past decade – implying that the next contract on the curve is more expensive than the current one. An investor paying for protection by buying the next futures contract upon expiry of the current one has to pay a premium. Continually rolling over in this way means that any potential gains are eroded by the costs of protection. In other words, when volatility is low it is not worth paying for protection unless we expect a huge spike.
We illustrate this with a strategy which assumes investors took out protection by buying the VXX and VXZ indices. Chart 6 illustrates that over the period January 2017 to end-January 2018, investing in the VXX would have generated a loss of 70%. The VXZ strategy would have produced a slightly smaller loss of around 50% since the medium-term segment of the forward curve is generally flatter, resulting in lower roll costs and therefore smaller losses. Note, however, that both the VXX and VXZ curves turn upwards towards the end of the period, reflecting the fact that the volatility insurance policy kicked in. But despite this, both curves remain in negative territory. At the same time, the S&P 500 still ended the period 15% higher than at the beginning of 2017, despite the most recent losses, so the rational investment choice would have been not to hedge.
Percentage return since 01/01/2017
In graphical terms, the strategy only pays off when the VXX and VXZ curves lie above the S&P 500 curve. This would have worked, for example, had we taken out protection at the start of 2018. But as a general rule, a buy-and-hold volatility investment strategy does not pay off over a long horizon although it may be useful for investors who are moving in and out of the market on a high frequency basis.
Volatility is the curse of market investment. It exists because lack of perfect foresight means that investors respond to the prospect of market gains or losses in unpredictable ways. If, for example, we have enjoyed a period of strong asset price growth – as has been the case since 2009 – the rational response to changes in any of the background conditions which have supported markets is to sell in order to lock in price gains. Thus, if investors fear that higher interest rates will reduce the scope for equity price gains, this will be manifest in a reduction in prices and higher volatility as risk perception rises. Such impacts might well be magnified as leveraged positions are unwound. Unfortunately, predicting risk is virtually impossible since we cannot assess how investors will respond to a given set of circumstances at different points in time, so we just have to live with it. But taking risk is an inherent part of financial life: to paraphrase Muhammad Ali, ‘(S)He who is not courageous enough to take risks will accomplish nothing in finance.’
1 This can be found at www.policyuncertainty.com/index.html.