Theories Challenged by Low-Volatility Gains
It's time to re-examine a commonly held investment theory about volatility in the asset management industry.
Surveying a wide body of industry research, Dana D'Auria, research director at Symmetry Partners, says that the theory that higher risk leads to the potential for higher returns can be challenged.
"What the data is telling us is that lower volatility stocks tend to have higher return," D'Auria says. "In particular, what's driving that is the highest volatility stocks have really low returns — they have abysmal returns.
"If you rank order stocks by their volatility it ends up being that the highest volatility stocks underperform the lowest volatility stocks, both on a risk-adjusted and even on a raw-return basis."
Investors have been seeking out low volatility strategies and funds. The biggest volatility-tracking fund, the U.S.-focused iShares MSCI Min Vol ETF (USMV), has seen inflows for six of the last seven weeks and totaled $238 million last month, the most since July, according to Bloomberg News.
Recent political events have shown how quickly sentiment can turn on volatility strategies. Reflecting turmoil in the White House, the CBOE Volatility Index (VIX) has surged by almost half in one day, Bloomberg reports.
Speaking with Money Management Executive, D'Auria, noted among the 2016 Women in Asset Management awards, outlined why low volatility performance challenges accepted industry notions.
Why has there been so much success of low volatility and have you been surprised in any of your research on this?
Certainly there are return chasing elements that suggest low volatility has just had good performance. Like any of these factors, as they become more well-known, you have more product proliferation and more people moving into them.
There were a few seminal papers in 2006 and 2009 that came out and really established low-volatility strategies. Those papers probably have led some to assume there is more interest in this type of investing. So, certainly you have seen more money flowing into it.
Maybe the most surprising thing, on an elemental level, is that risk and return in this context don't seem to be related. In fact, low volatility seems to outperform high-volatility. The theory has always been that higher risk leads to the potential for higher returns. What the data is telling us is that lower volatility stocks tend to have higher return. In particular, what's driving that is the highest volatility stocks have really low returns — they have abysmal returns. If you rank order stocks by their volatility, it ends up being that the highest volatility stocks underperform the lowest volatility stocks, both on a risk-adjusted and even on a raw-return basis.
Okay — why are low-risk stocks returning less?
The short answer is that we don't know exactly. We don't have a consensus on why this happens. We do have some theories.
One theory is that it is not really a reliable factor, so to speak. There are papers that suggest that the phenomena is really related to just how you measure volatility — what your look-back period is and how you're weighting your portfolios when you put them together. Then there is a paper out there that suggests that if you X-sized the smallest high-volatility stocks, all of a sudden the effect disappears. That's one theory — that this is just an artifact of the data and a type of data mining. Then there are other papers that say, "We looked at that idea and we still think it works."
Why are there so many differences in opinion regarding the performance of low-volatility?
When you do a thorough reading of the academic literature in any of these factors, you find a lot more disagreement than you might think, based on the fact that there are so many products out there. There is not consensus around a lot of this stuff; however there is significant consensus that high-volatility does underperform. There are still differences of opinions here.
There's potential that any of these factors can be an artifact of the data at the time. A lot of the factors have been examined enough that you can say this is probably not just how we measured it since we see so many different measurements that work. We also see it in different time frames and in different countries.
For a lot of these factors there is a spectrum. There are the factors that are really highly established and there are factors that are decently established, and they probably are true, but they don't have as much behind them.
What's the most current research that you have looked into on this?
There is a paper that literally just came out attacking the some 447 factors that have been found and how many of them have actually passed the test for high significance, and it's not that many.
The importance of low volatility, or any factor where you would say that there's some evidence that maybe it's just data mining — from our perspective you would want to when you're building your portfolio to consider the obvious evidence both supporting the factor itself, but then you would also want to think of the factor in the context of the other factors you are going to capture.
There's an element here, and it mirrors this idea that not only do you want to look at a factor in isolation; you want to say, "Conditioned on the fact that I am going to have certain factors in my portfolio, is this a factor that while on a standalone basis maybe has very good evidence and good interplay with the factors that I already have in a portfolio? That can make a big difference in how you want to engineer the portfolio.
How would a firm suggest an adviser use low-volatility in a portfolio?
One way you might use it would be that if you have other factors that would raise your risk, raise the volatility, and raise the beta, low-volatility can act as a substitute for market, where you essentially go after market-like returns, but with lower risk. So, in that case, you would pull the beta — you would pull the volatility of the overall portfolio down.
What other theories are there for why low-volatility is becoming so popular?
There are a bunch of different theories, but I would concentrate on four. One of those would be: it's possible that low-volatility stocks covary with something important. It may be the other way around: high-volatility stocks covary with something important that makes it such that if they underperformed, even though they have more risk, possibly in some way they alleviate some other risk. This would be a question of; do high-volatility stocks protect you when you need it? Right when the market is doing poorly are high-volatility stocks paying off? These are the types of theoretical questions that could covary with something else.
Theory number two is are we seeing low-volatility outperforming high-volatility, counter to intuition, because in fact there is some other risk — not standard deviation risk or volatility risk - that we care about that high-volatility is actually better on, and that might explain why we would see this.
That is definitely counterintuitive.
Theory numbers three and four are both behavioral- and structural-based. Number three is along the lines of the lotto effect, or option-like compensation for investment managers. These are the ideas where people prefer more than they should something that they think might deliver a high payoff.
So all things equal; people are bidding up the high-volatility stocks. That's why they are not paying off as well. Why are they bidding up the high-volatility stocks over what they should? It's because they think high-volatility also means positive skew or they think it means they have a chance at a big payoff. In that case, they look at the high-volatility stocks like a lotto ticket and they overbuy it relative to its value, and that idea can go across not just regular investors but also across money managers where maybe they're assessment of how they are going to be paid is asymmetric so maybe they look at it and say, "Slow and steady will not win the race for me. Either I take off or I don't. Either I distinguish myself relative to other managers, or not." So, they bid up high-volatility stocks.
Peer pressure can be hard to ignore.
Then the fourth theory looks at this from the perspective of why low-volatility does better on a risk adjusted basis — meaning, why are low-volatility stocks much lower than the rest of the stock returns? They should be because they are lower volatility; so on a return-per-unit-of-risk, if they were all equal, then lower-volatility would have a lower return because you would then be dividing by a lower number.
So, on a risk-adjusted basis, why did they outperform? This theory says that it's because there are limits to things that would fix it, like arbitrage or leverage. So, the in-the-know-money maybe would like to take advantage of the fact that low-volatility stocks have the better Sharpe ratio, but they can't because there are limits to how much you can lever up a high-Sharpe ratio but low return investment.
Would you say then there is a way to truly reduce risk without diminishing expected returns?
I'm a data-driven person and I have read enough of these papers, all of which are very established academics with high-level work going on here, and they all make a good case. I think there is a case for both low-volatility and any other factors. There is probably an amalgamation of the facts, or a compilation.
I once heard Kenneth French say something like, "How much of the value factor is risk versus behavior?" He came up with some number, and he was joking. His point was that there is probably a little of both. I would say that any of these effects, you could do a test on them and say what's really causing it and you would probably find a number of different factors.
Have there been any alternative philosophies that suggest high-volatility is outperforming?
There are certainly plenty of theories suggesting that we shouldn't get the result that we have been getting. If you go back to the capital asset pricing model, it suggests that the return of the stock is a function of its market beta — how much it covaries with the market. So if it has a market beta over one, then it's going to do even better than the market and vice versa — when the market is down it will do even worse.
So, that theory says that risk and return are related, but what it says is that only systematic risk is really rewarded. So, the risk you would take in your covariance with the overall market is rewarded, but any individual risk that's not related to how the markets are moving should not be rewarded. The idea behind that is that idiosyncratic specific risk of a stock shouldn't pay off because you can diversify that risk away, so you shouldn't even be taking that risk. But, there are those other theories that say that even idiosyncratic risk will pay off because the reality is we don't all have perfect information that investors are not able or not willing to diversify to the extent that they should. There are papers that would say that you in fact would expect to see idiosyncratic risk pay because the theory that says only systematic risk pays is the assumption that's incorrect.
What we do in fact see is that volatility, whether it's measured systematic by beta, whether it's measured idiosyncratic, which is after you account for the systematic part, or if you measure raw; all of them give you the same basic result, which is, they don't show the return to risk relationship that you would expect.