Heard in the Media: The VIX volatility index is low, portending trouble ahead
Bottom Line: The data does not indicate that extreme low VIX readings imply immediate short term trouble is ahead
Low Vix Impact on Portfolios: Neutral, leans positive
The VIX (the Chicago Board Options Exchange Volatility Index) is back in the market headlines. The VIX is often used to measure the market’s expectations of volatility. In this case, the VIX is at historical lows, which tends to bring out the fear mongers. Note, when the VIX is at high levels the same fear mongers headline all financial media.
Two articles came out in the last week that are worth a read. Both present a slightly different perspective and both reference arguments about what the VIX means. I highlight and discuss each article then I present our own data based approach that seeks to filter out noice and opinions.
Perspective A: Excessive stability breeds instability down the road
In a Bloomberg article titled “Unusually Low Volatility Can Be Dangerous for Markets,” Mohamed A. El-Erian writes, “The lower the market volatility, the less likely a trader is to be “stopped out” of a position by short-term price fluctuations. Under such circumstances, traders are enticed to put on on bigger positions and assume greater risks.” Very well, what about average investors?
“Patient long-term investors can also be influenced by unusually low volatility, whether they know it or not,” El-Erian continues, “Expected volatility is among the three major inputs that drive asset-allocation models, including the “policy portfolio” optimization approach used by quite a few foundations and endowments. As the specifications of such expectations are often overly influenced by observations from recent history, the lower the volatility of an asset class, the more the optimizer will allocate funds to it.”
El-Erian then focuses on policy concerns, and he does not directly make the claim that the low VIX by itself will lead to unpleasant market outcomes. He thinks the VIX is currently disconnected from the real world geopolitical risks, a whole other matter.
The focus on traders is of little to no concern for the vast majority of investors out there. Daily trading accounts for a minuscule portion of total market capitalization. This has almost no bearing on investors with longer-term horizons.
El-Erian claims that institutions and endowments can be influenced through investment policy optimization approaches. They certainly can. However, El-Erian is equating a short-term volatility index with institutional multi-year cycle adjustments. For our concerns with individual investors, there is little connection.
He correctly states that asset allocation optimizers use volatility as a key input, along with expected returns and correlations. Most long term investment policy optimization models, however, are unlikely to materially change based on a short term low VIX reading for two key reasons.
First, most policy models aimed at retail investment programs in the mass affluent and high net worth markets use long-term average market volatilities. The VIX goes back to 1990. In 2013, the total historical average for the entire VIX period was 20.0. Today it is 19.4. Even if asset allocation models used the VIX as the volatility input (they typically use historical realized market volatility), the impact of the change in average is highly questionable. More importantly, in order for an asset allocation optimizer to prefer equities based on volatility changes alone, bond volatility would have to remain constant or increase. But bond volatility is also lower. Because volatility is lower in both stocks and bonds, a lower VIX by itself in insufficient for asset allocation optimizers to prefer stocks.
Second, changing investment policy models for investment firms is an extremely costly undertaking. All the inputs into multiple technology, reporting, compliance and monitoring systems cost a lot of money. Furthermore, advisors would then have to transact, incurring commission costs and taxes, to align portfolios with the adjustment to the optimization models. Asset allocation models are imprecise by definition. Advisors should only adjust to changes in modeling output when the change is material, and advisors have conviction that clients will benefit. Again, a lower VIX is insufficient to overcome these hurdles.
Perspective B: Debunking the VIX
In a piece last week “Time to Stop Calling the VIX the ‘Fear Index’, Barry Ritholtz adds the VIX to his series of market-indicator debunkings. Ritholtz offers a few compelling points to properly frame what the VIX means, ending with the conclusion that “The bottom line is simply that the VIX tells us very little about what the market is likely to be doing tomorrow, or next month, or even next year. It certainly is not the reliable prophesy of doom some perma-bears make it out to be.”
Our Perspective:
Like most data sets, the VIX is most relevant at extremes. But high and low extremes are different. When the VIX registers readings in the top 10th or 20th percentile of all readings, the market sells off concurrently, as shown in the following graph. In this scenario, the more extreme the VIX reading, the more likely a market snap back in the ensuing period.
But what about low levels like today? The VIX ended the week last Friday (5/12/17) at 10.4, which places it in the lowest 1% (0.70%tile) of all weekly readings going back to 1990. I am using weekly data to filter out day-to-day noise. The 50th %tile would be about 17.55. If complacency is defined by low volatility and complacency leads to trouble down the road, we want to study what happens to the market at some point after extreme low VIX readings are breached.
I use forward 26 week returns for the S&P 500, the most widely used U.S. stock market benchmark. I use 26 week returns because a half-year of market action is noteworthy for most investors, even if they have long-term policy allocations.
Average 26 Week Forward Return 1990-May 2017 = 4.22%
Average 26 Week Forward Return when VIX < 10th %tile = 5.31%, an increase of +1.1%
Percent of all Forward 26 Week Returns > 0 = 72%
Percent of 26 Week Forward Returns > 0 when VIX < 10th %tile = 84%, an increase of 11.5%
The data indicates that the lowest VIX readings do not imply an increase in the probability of below average returns. In fact, at the lowest 10% of all VIX readings, average 26 week forward returns are higher than average, with a higher probability of returns greater than 0% (Batting Average), and with lower volatility. The same trends apply to forward weekly returns of 9, 13 and 52 weeks.
Every data based analysis has limitations. The first and most important is alway the potential bias of the person doing the study. Biases lead to selective use, consciously or subconsciously, of data. Second, the VIX only goes back to 1990, so we are not study the history of market returns and volatility. Third, the VIX is only one data set. I would argue that the VIX is a concurrent indicator of other data influences. In my view, the VIX is a dependent variable, it is influenced by other factors.
Bottom Line: The data does not indicate that extreme low VIX readings imply immediate short-term trouble is ahead. Don’t let the fear mongers steer you off course.