The relationship between risk and reward is one of the most fundamental in finance: to get higher returns, say the traditional models, you must accept more risk. There’s just one problem with that idea: it’s not borne out by history. Over long periods, in most markets around the world, stocks with the highest volatility have not been the best performers—and during some periods, the least volatile companies have done significantly better. This low-volatility anomaly has been known since the 1970s, and has gained renewed popularity with the appearance of a number of ETFs designed to capitalize on it.
In the last year or so, BMO,  iShares and PowerShares have all launched low-volatitly ETFs in Canada, and each them uses a quite different strategy. I’ll review these later in the week, but first, I’d like to share an excerpt from my recent interview with Jean Masson, managing director at TD Asset Management and an expert on low-vol strategies. Dr. Masson is the portfolio manager for the  TD Canadian Low Volatility Class and the TD Global Low Volatility Fund, as well as two institutional funds based on similar strategies.
Can you explain what the data tell us about the relationship between volatility and returns in the equity markets?
JM: Most investors do not like to take risk for no compensation. But somehow, if you look at the history of the stock market in Canada over the longest series we have, you will not observe higher returns on more volatile equities. In the US we have accurate data starting in 1926, and since that time the relationship between risk and return is basically flat—whereas the theory suggests it should be positive. You can sort the performance based on beta or on standard deviation and you get the same result.
When markets are strong, you do see the positive relationship between risk and return, and that is true in Canada, in the emerging markets, in the US, everywhere that I have seen. Let’s say you believe there should be a 4% equity risk premium above bond yields—so these days that might be 6% or 7%—and returns end up being above that expectation. In those environments, it’s typically the riskier equities that do really, really well. But when equity returns are within expectations, then riskier equities do terribly and less risky equities do relatively better. And when the returns of the overall market are not very good, the relationship is actually inverted.
So the theory says less volatile equities should have lower expected returns, but all of the evidence we have shows they do not. If there is no compensation for risk as far back as our data go, and if we think there will be no compensation in the foreseeable future, then our thinking is you might as well minimize expected risk.
And by doing so you can expect higher returns?
JM: We have never said if you reduce volatility you will necessarily have better returns. What we’re saying is that you can take less risk and not sacrifice expected return. Some people are more aggressive and say you can have your cake and eat it too: you can take less risk and have better returns. That is what we’ve observed in the last 10 or 15 years, but I think that’s a little aggressive going forward. Most retail investors chase returns, and low-vol has recently done better, but that’s the wrong reason to use the strategy.
I guess that is the distinction between a low-vol strategy and a value strategy. The former attempts to provide better risk-adjusted returns, but not necessarily returns superior to the benchmark. Whereas a value strategy does attempt to outperform, and not necessarily with less risk.
JM: I agree. The traditional goal of active management is to give you a higher expected return for the same risk. Value managers might build a portfolio of stocks that trade at low price-to-book or price-to-earnings ratios, or high dividend yields, or whatever. They try to find stocks that look cheap relative to their potential. And historically the value style has produced slightly higher returns for about the same risk—let’s say 14% to 15% standard deviation, similar to the index in terms of risk, with slightly higher returns. Low-vol is about giving a similar return to the index while taking a lot less risk—let’s say 30% less risk or so.
Is it fair to say that you could buy a low-volatility fund and simply hold it for your investment lifetime? In other words, low-vol is not a tactical strategy.
JM: That’s right. In the long run the returns compared with the overall market are competitive, and if things turn out badly you won’t go hungry. To me, the core should be low-vol for the long-term: it’s not designed for moving in and out.
Anyone who’s invested in RIM in the past 2 years would be really interested in an option to lower risk without diminishing potential returns. Great read, thanks!
So the goal of a low volatility strategy is to have similar return, but with less risk, risk being defined as volatility. Compared to an indexing approach, costs will be higher, although the difference may be small. My guess is that taxation will be greater. Dividend yield will be higher with a low volatility strategy? That’s a guess, but if it’s true, there will be less ability to defer tax and foreign dividend tax will be at one’s marginal rate. Also, there will be increased turnover, most likely resulting in increased capital gains, in a low volatility strategy compared to an indexing strategy. Will the returns of a low volatility strategy be similar to those of an indexing strategy, after taking into account taxes and costs?
Also, risk is being defined as volatility here. Warren Buffett says that risk is the reasoned probability of losing purchasing power over the time that the investment is held. Joel Greenblatt gives a similar definition, but also states that another definition of risk is that of underperforming another investment strategy over a defined period of time. I’m not sure that a low volatility strategy addresses either definition of risk.
Finally, let’s assume that risk is volatility. If you’re using an indexing strategy, and want to decrease risk, a low volatility strategy will do that. But instead of using a low volatility strategy, why not buy more government bonds? In a bull market, government bonds will most likely have a negative correlation with the stock market. The same can’t be said of a low volatility strategy.
Finally, there’s been marked increase in interest in low volatility strategies in the last few years. If low volatility is a free lunch (less risk for the same return) and interest in this strategy is increased, then it may no longer work. I’m not saying that this will happen. As an analogy, the value premium still exists, and it is in part at least, a free lunch. But the low volatility strategy is comparatively new, and whether it will stand the test of time is still unknown.
@Park: All good points. Any time you deviate from a total-market cap-weighted index you run the risk of higher turnover and therefore higher expenses and more potential for capital gains. Though to be fair, ETF providers have largely been good about keeping these to a minimum, usually by imposing rules about the frequency of rebalancing.
Whether volatility equals risk is on some level a semantic argument. I think we can all agree that few investors embrace volatility, and that if it were possible to achieve the same returns with lower standard deviation, that would be a good thing all around.
Regarding using bonds to dampen volatility in a portfolio, this is an important point that has been discussed by other researchers. It might be possible to achieve a similar risk-return tradeoff by combining equities with high-quality fixed income. You may find this article interesting:
http://lowriefinancial.ca/whats-the-best-way-to-lower-volatility/
I’ve only taken time to scan a couple of technical papers on this subject, but I haven’t figured out yet whether they look at the future performance of low vs. high volatility stocks. By this I mean that if we divide stocks into low and high volatility based on their results in years 1 to n, do we then look at their performance over years 1 to n or do we look at their performance in the years after n. It is the latter scenario that reflects the way that we can really invest our money.
I like Park’s comments above because I have never been comfortable with defining risk entirely as a function of volatility alone. I pay much attention to it but would include credit risk, the risk of inflation and interest rate risk as prominent. This line of reasoning then encapsulates Park’s comment about W. Buffet in that risk is related to purchasing power over time. Risk to me is therefore the potential for a bad outcome at some future point in time, like not having enough to retire on.
For example a portfolio may exhibit lower volatility over a given time period if it had a tilt toward, say being inflation sensitive in a rising inflation environment, or a lower duration portfolio tilt in a rising interest rate environment.
Great discussion. I like and use the idea of investing in lower volatility stocks when valuations are above average. I thinks it a great way to invest in the emerging and frontier markets. John Templeton and Mark Mobius have built great track records with a bottom up investment philosophy that packed Templeton funds with low volatility stocks.
@Michael James: In my newbie amateurish blog post (http://myetfportfolio.ca/2012/11/23/volatility-what-a-drag/), I reference S&P’s research on this topic (http://us.spindices.com/documents/research/low-volatility-effect-comprehensive-look-201208.pdf).
The way I understood section 3.1 of S&P’s article it is they use trailing price changes to determine security weights going forward and re-balance at a prescribed interval, hence they would have to look at performance for ‘year n’ onward. Otherwise the measurement wouldn’t really make sense…one would need to be able to predict low volatility stocks in advance. I would assume this would have to be true for any technical article that looks at this issue?
I guess the underlying assumption becomes that historical low volatility security will continue to have low volatility in the future. Probably a reasonable assumption?
Lots of the supporting research for low vol can be found at http://www.lowvolatilitystocks.com/resources/articles-on-low-volatility-stocks/ . That low vol does produce higher returns with lower volatility (std dev) is little contested. As to why it does, now that’s crucial to knowing if it can/will continue. There seem to be several hypotheses – leverage aversion aka default risk, behavioural biases amongst institutional investors towards high vol stocks, removal of price as the factor for weighting a portfolio (i.e. cap-weight indices overweight expensive assets). An interesting book I’ve started reading – The Missing Risk Premium
: Why Low Volatility Investing Works by Eric Falkenstein – turns the issue around by saying high volatility /risk doesn’t work.
@CanadianInvestor and Mazhar: Thanks for the links. I found Falkenstein’s book pretty tough reading, not so much because it’s too technical, but because it’s just not that well written. For a clearer explanation of Falkenstein’s work, I’d recommend Norm Rothery’s 2011 article from MoneySense:
http://www.moneysense.ca/2011/03/01/risky-business/
@Mazhar: That’s the issue I was wondering about. It may be true that low volatility stocks in one time interval tend to remain low volatility stocks in the next interval, but I haven’t read any study that confirms this. Perhaps these studies exist, but I haven’t really looked yet.
CCP, thanks for the Rothery link, good article. There’s a sure-fire marketing idea near the end that would surely appeal to Canadians – “…taking the middle way with a moderate amount of risk appears to hit the sweet spot “. Those ETF names with low vol or low beta sound far too technical or esoteric. Yes, unfortunately Falkenstein’s book needed a better editor but for some familiar with finance lingo it’s possible to read around it.
From the research I have read, you can think of high beta stocks as using leverage, and there is a cost to leverage. A stock with a beta of 2 if like a stock with a beta of 1 + a call option. The drag on performance is the imputed option premium. Sometimes it pays off (when markets outperform expectation), but on average it underperforms slightly.
http://seekingalpha.com/article/1031231-why-low-volatility-etfs-are-a-sign-of-the-times
@Park: the article you linked claims that low-volatility ETFs can be very sensitive to the rising interest rates, due to their heavy exposure to utility stocks. This argument applies to the ranking-based ETFs like SPLV. SPLV picks the 100 least-volatile stocks in the S&P in the past 12 months. SPLV is indeed very heavy in utilities.
In my opinion, this argument does not apply to the low-volatility ETFs that follow MSCI low-volatility indexes. MSCI uses stock correlation matrices to minimize volatility at the portfolio level. When they run their optimizations, they control for sector weights. The weights have to stay within 5% of the sector weights in the parent index.
MSCI methodology is described here:
http://www.msci.com/eqb/methodology/meth_docs/MSCI_Minimum_Volatility_Methodology_Jan12.pdf
Love couch potato investing!! Thank you for all of your hard work!
I do practice some personal Value investing as well (love Buffet’s style) but only in select stocks.
Derek
@Canadian Couch Potato: In trying to understand the implications of the premise that low volatility stocks do not outperform high volatility stocks over the long term, I am struck by the apparent discrepancy of logic against the fact that elsewhere in this website we are also advised that low capitalization stocks have a projected higher expectation of return over the long term than large cap stocks, and that this expectation of higher return is our “reward” for tolerating the higher “risk” (which I assume means volatility) of the small cap stocks. This other principle is supported by Fema and French, and in fact the Uber Couch Potato Portfolio modelled on this website contains contains substantial components of Low Cap Canadian, US and International Equity Index Funds.
In trying to detect an internal logic that does not violate either of these two principles, I wondered if the higher projected return of the small cap funds had nothing to do with, (or , more precisely, was not positively correlated with the volatility of) the individual funds), but rather that the higher volatility was an inevitable characteristic of small cap stocks, but not the reason for their higher profit. If this were so, could we not conceivably detect a class of small cap stocks (say, culled from an Index of Small Cap US Equity) that was further characterized by lower volatility, as defined by some appropriate rational mathematical parameter, as ZLB has done for the selection of the low volatility sector taken out of the Canadian Large Cap Equity Index as a whole.
Even if this were true, though, doesn’t the principle of “transparency of the market” foil your plans for future profit? i.e., the market has already figured out the value premium of these low-beta stocks making up your low beta index, and has already pushed up the price of these components, making your low beta index unacceptably expensive to purchase?
@Oldie: These are good questions. The data from the U.S. clearly show that small-cap stocks, which are generally more volatile, have outperformed large-caps over most periods going back to the 1920s. However, my understanding is that if you look only at small-cap stocks, or only at large-cap stocks, then within those categories you will see a much lower correlation between volatility and returns.
In other words, if you take a large sample of large-cap stocks and divide them into five quintiles according to volatility, the relationship between risk and return is not strong. I have seen the data confirming this, at least over some periods. I have not seen similar data regarding small-cap stocks, so I’m not sure how they would break down. But my guess is that if you looked at the most volatile small-cap stocks, you would not see that they deliver the highest returns in the category.
It is probably worth mentioning that I hope investors don’t feel the need to act on all of this information. I think the low-vol research is interesting, and all of the Canadian ETFs based on this idea use completely different strategies, so I thought it would be worth decoding them. But I’m certainly not trying to imply that everyone should rush out and embrace a low-vol strategy (or a small-value strategy, for that matter). For most people, simple is better.
Just had a look at the TD Cdn fund and wonder why a) the fund is rated as “medium” risk (shouldn’t it be less?) b) why the MER box is blank (a great incentive to buy into a fund when you are not even told what it will cost) c) no info on holdings (buy something with a name but we’ll not tell you what you are actually buying?). The fund holds only TD Canadian Low Volatility Pool O-Series but there is no further info to be found. The link to the prospectus doesn’t even work. Nice going TD. I’m surprised it has even managed to attract $3.5 million in investor money.
@CanadianInvestor:
a)Med risk based on 100% Equity,
b)new fund can’t state what the MER is capped at 2.23%
c)retail fund is available in Corporate Class structure, so it holds the no fee O-Series
prospectus: http://www.tdassetmanagement.com/Download/Inv_PartA_Final.pdf
fund facts: http://www.tdassetmanagement.com/Download/TDB2720E.pdf
disclosure: I like this fund and the research Robeco and TDAM have done on Low Volatility.
Bobby, I still have the same problems:
a) fund risk is “medium”, same as for their standard Canadian equity fund that tracks the TSX Composite; yet, if this fund works properly as other low vol funds do, it will have a LOT less risk; just for fun, go into Yahoo Finance and graph BMO’s low vol ZLB against XIC or XIU and you will see what I mean
b) maybe my eyes aren’t sharp enough or my search doc function isn’t working properly but neither of the docs you link to says anything about MER capped at 2.23% or any other specific number; even if it is that number, that is way too high to look attractive to me compared to ETF low vol alternatives
c) now something is available: the info on holdings is this usual partial one with top ten holdings dated end of March – http://www.tdassetmanagement.com/Download/2720-I%281%29.pdf