The fall in equity markets in the first days of February would have clearly confirmed the view of all the low volatility Cassandras and prophets of doom out there. However, rather than demonising low volatility per se, one should pay closer attention to how the market deals with it. If the market is complacent, as was the case in late January, it doesn’t take much for stocks to drop sharply. But if some precautionary measures are taken, the risk of a severe market setback is clearly reduced, and investing at low levels of volatility is far less risky than one might think. What’s more: implied volatility also provides us with a tool to re-enter the market.
Low volatility: Nothing lasts forever, but some things last for longer
Before we dig deeper into the matter of market complacency, let’s put the recent low-volatility environment into perspective. There is no doubt that, in 2017, equity-implied volatility has reached new all-time lows. And it goes without saying that this is a period of very low volatility – even after the spike in early February 2018. However, periods of low volatility happen more often than one might think. When taking a ten-day average VIX level of 13.6 as reference1 we find that there have been periods of similarly low volatility in the past, in particular in the nineties (Chart 1). Low volatility is a phenomenon that is not limited to the equity market: three-month G10-FX-implied volatility and three-month USD-swaption-implied volatility are at very low levels as well. But for them, just like the VIX, long periods of low volatility are not a new phenomenon (Chart 2).
Periods and length in days when the ten-day average of the VIX traded below 13.6
So, how do we identify potential complacency among market participants (one could also call it imprudence or overconfidence)? It should be visible from investors’ portfolios: if they are – on a broad scale – fully invested or even increasing exposure to an asset that has recently performed quite well, it indicates complacency. Unfortunately, we lack this information. We can only try to derive it from publicly available information. Estimating the exposure of the fund industry to a certain asset class is one way, closely watching sentiment surveys2 another. We present a third way: we put the price at which the market prices volatility into perspective.
And this is where it becomes tricky: volatility isn’t a classical asset like a bond or a stock, the price of which one can compare with earnings, dividends, or coupons. It is observed in the market or in its implied form (and that is what we are interested in, because it is closest to the market’s current assessment of the risk of future price changes) derived from option prices. Gaining exposure to volatility requires buying or selling an option on an asset or the sale or purchase of a future on the VIX, which only became possible in 2004 when they were introduced.
Low volatility: A bad forecaster of future higher volatility
So is low volatility per se a harbinger of future bad performance? We have already shown above that there have been very long periods of very low volatility. By comparing the VIX and the S&P 500 (Chart 3) it becomes obvious that low volatility is a bad forecaster of future high volatility most of the time. This disqualifies the level of volatility as a general indicator of complacency. It is more important to know how the market deals with the low volatility. If it is not prepared for a potential rise in volatility, then indeed a low level of volatility signals that one should be more cautious. We try to extract this information from two sources: the slope of the VIX futures curve and the implied volatility skew.
The time value of volatility futures versus the spot level puts us in a difficult situation. When comparing the spot level of the VIX with, say, the next maturing contract, the time value of the nearby contract versus the spot will determine the slope of the curve. If the contract is close to maturity, it will almost equal the spot, and if it just became the nearby contract, it is likely to trade at a decent premium to spot (at least in a low-volatility environment). To avoid any misconception, we estimate the slope of the curve between the first and the third future. This still leaves us with the problem that the average slope of the curve is highly dependent on the overall level of volatility: in an environment of high market volatility, the futures curve tends to be inverted as markets are expecting volatility to revert to the long-term average. The opposite is true in an environment of low volatility: the curve is upward sloping as markets are again expecting volatility to return to the long-term average. This means we need to take into account the current volatility scheme to determine what the ‘normal’ average slope for the prevailing level of volatility is. We therefore estimate the average steepness for the current level of the VIX by using 10% of the observations around the current level of the VIX. We find that the median performance of the S&P 500 in the next week is lower in an environment where the VIX curve is flatter than normal. This also holds for longer time periods, like four weeks or three months (Chart 4). A flatter-than-normal VIX futures curve can hence be seen as a harbinger of future subdued market performance.
A skew has been observed in equity options in particular since the 1987 stock market crash, with prices for options with strikes below the current level higher than those for options with strikes above. This means that market participants are generally willing to pay more for downside protection than upside participation. The price difference becomes visible when estimating the implied volatility of e.g. 90% and 110% out-of-the-money options. Just like for the slope of the VIX futures curve, the implied volatility skew depends to some extent on the level of the VIX: the lower the VIX, the lower the skew (Chart 5). The simple and straightforward interpretation of the skew in a low-volatility environment is: the lower the skew for a given level of volatility, the greater the complacency of market participants in this environment. However, after a prolonged period of declining equity market prices, a second interpretation becomes more important: in this case it is not overconfidence that we are experiencing, but returning confidence in the future performance of the stock market.
Scatterplot of implied vol skew (y-axis) and VIX (%, x-axis), sorted by deciles
A fairly simple model reaps the benefits of both worlds: we first identify whether the skew is lower than normal for the current level of volatility by comparing it to the average of 10% of the observations around the current VIX. If we find that it is below this average, we determine whether the stock market performance over the preceding three months was above +5% or below –5%. In the case of the former we regard a below average skew as a signal of complacency and reduce exposure to the equity market; in the case of the latter we increase it. The strategy has an average exposure of circa 36% to the equity market, and outperforms a buy-and-hold benchmark with the same equity exposure by far. In particular during the low-volatility periods from 2013 to 2015 and 2016 to 2018, it gained versus the benchmark as it predicted that confidence had returned to the market (Chart 6).
Strategy: Overweight signal: implied vol skew below average; equity market performance over the previous 3 months less than –5%; underweight signal: implied vol skew below average; equity market performance over the previous 3 months more than +5%. Benchmark: average equity market exposure of strategy (circa 36%)
It is perfectly understandable that low volatility over a prolonged period of time makes one feel uncomfortable. It is a little bit like racing a car: one knows that at some point in time one will need to hit the brakes to avoid a lethal accident. However, hitting the brakes too early might just cost the victory. While it is the skills of the driver that make the difference in racing, it is the timing skills of the investor that makes the difference in financial markets. The slope of the VIX futures curve and the equity-implied volatility skew provide us with indications of when it might be time to get out of the market – and when it is time to get back in.
1 We take this level because it is the 25th percentile of daily VIX closes from 1 January 1990 to 31 January 2018. This means that 25% of all observations were below this level.
2 See e.g. ‘Thinking Ahead’, Issue 66, November 2014, ‘Survey-based sentiment – applying a proven strategy across asset classes’.