In the last 14 months or so, Canada’s ETF providers have launched several funds based on low-volatility strategies. As we saw in my last post, the research suggests it may be possible to build a portfolio of stocks with lower volatility than the broad market without sacrificing expected returns. But exactly how do you select those stocks?
There are several ways to implement a low-volatility strategy, so before you consider any of the new ETFs, make sure you understand how they differ. Today we’ll take a look at the methodologies used by BMO and PowerShares. Next week we’ll look at the iShares strategy.
BMO looks at beta
Rather than tracking an index, the BMO Low Volatility Canadian Equity (ZLB) simply uses a transparent set of rules. You start with the 100 largest stocks in Canada and rank them according to their beta over the previous 12 months. You then select the 40 with the lowest beta: the lower the beta, the greater the company’s weight in the fund. No stock can make up more than 10%, sectors are capped at 35%, and the fund is rebalanced just once a year.
Beta can be a confusing concept. It’s important to understand it has two components: the stock’s volatility (how much its returns vary) and how closely it is correlated with the market as a whole. That means it’s entirely possible for a low-beta company to be highly volatile—as long as its wild price swings are uncorrelated with the market, the stock could still have low beta.
|ZLB Top 10 Holdings|
|Shoppers Drug Mart||3.90%|
|Weston (George) Ltd.||3.60%|
|Inter Pipeline Fund||3.10%|
One of the attractive features of ZLB is its diversification across sectors—its holdings are considerably more balanced than the broad Canadian market. As such, it would make a reasonable core Canadian equity holding.
|ZLB Sector Breakdown|
PowerShares looks at standard deviation
The PowerShares S&P/TSX Composite Low Volatility Index (TLV) uses a different strategy. Remember we said beta measures both volatility and correlation with the overall market? Well, TLV ignores that second factor and looks only at standard deviation, or the degree to which a stock’s daily price movements vary around its average. This means the fund may include companies that are highly correlated with the market (that is, stocks with relatively high beta).
The ETF’s index includes the 50 stocks in the S&P/TSX Composite Index whose price movements had the lowest standard deviation over the past 252 trading days, or 12 calendar months. Again, the companies are weighted according to their volatility, not market cap: the most stable get the largest share of the index. It’s rebalanced every quarter.
|TLV Top 10 Holdings|
|Primaris Retail REIT||2.30%|
|First Capital Realty||2.30%|
|National Bank of Canada||2.20%|
It turns out 23 of the 50 companies in TLV are also in BMO’s low-volatility fund, but the sector breakdowns of the two ETFs are very different. TLV’s holdings are dominated by real estate companies (which are classified as financials): I counted 15 of them, which is one more than you’ll find in the iShares ETF devoted to REITs. There’s no sector cap in the S&P index, and the results are pretty dramatic.
|TLV Sector Breakdown|
If you were planning to use a low-vol ETF for your core Canadian equity holding and you also have an allocation to REITs, using TLV will result in huge overlap. As always, you need to consider how any ETF in your portfolio complements the others.
I’m curious as what size of an investment would you recommend buying the stocks in a similar ratio rather then buying the ETF?
personally I’m more inclined to purchase stock, but you make some convincing arguments in favour of ETF’s
Very interesting. So I read this post and the last one and my question is – if I am understanding the underlying research correctly, it is not claiming that there is not a positive correlation between volatility and long-term return, but rather only that that correlation starts to break down at the very highest volatility. In other words, if one was seeking the best long-term return, it’s not necessarily that one would want the very least volatile stocks. Rather that the ideal would be the sort of “middle” volatility stocks – high enough to deliver increased return, but low enough not to be part of those sort of “fringe” stocks where the data says the return relationship breaks down. Is that right? So with that in mind, these ETFs (each in their own way) seem to aim to pick the very lowest volatility stocks (bottom 40 of 100 for ZLB, for example). Given the above, is that what the research is actually suggesting is beneficial?
@Mandy: The short answer is that I would almost never recommend buying individual stocks rather than the ETF:
@Danno: In general, you’re right that the most important thing seems to be getting rid of the highest-volatility stocks before you do anything else. They seem to have the worst risk-adjusted returns in most markets over most periods. But after that, it really depends on the specific market and period you’re looking at. During some periods the relationship is essentially flat: stocks in all quintiles (groups of 20%) produced very similar returns. So if that’s the case, it would make sense to look at the least volatile, not those in the middle.
Thanks for the reply. Interesting. So it seems like maybe which might be arguably more useful would be, instead of an ETF that sorts on volatility (however it’s defined) and picks the lowest quintile or two, maybe the better product would be something that does the opposite, sorts on volatility and then discards the upper quintile or so?
@Danno: Possibly. See this article by Norm Rothery in MoneySense. This link goes right to page three, where he discusses this issue:
the best way to reduce viability is through the combination of various asset classes from long and short bonds, to gold and equities from various countries. These products are plain silly.
You need to keep in mind that it is the higher volatility of returns that leads to lower compounded return. I agree with Dale, it works best at the portfolio level.
Take a look at XBB & XIU: from 2006 to 2011 they have the same simple average return: 5.7 %. The difference is in the higher volatility of returns from XIU which translate into lower ann. compounded returns over 6 years: XBB = 5.65% & XIU= 3.55%
Eric, you said you agree with Dale…
Your second paragraph seems to suggest that Low-Vol ETFs can be very useful, contrary to Dale’s statement.
Suppose you can substitute XIU with a low-volatility ETF which has the same compound and simple average returns. You get a better portfolio return with a lower volatility, without changing FI/Equity split.
These lower volatility products are certainly the ETF industry’s response to investors’ collective fear and low tolerance for volatility. They would be better served by educating Joe Investor as to the benefits of portfolio construction that just throwing out a ‘low’ volatility ETF that could just as easily correlate quite closely to the overall market in a meltdown – when everything seems to break down.
Again, the fact that the ETF industry does not conduct a massive investor education campaign in Canada is an even greater oversight. And a lost business opportunity. They keep pandering to the Advisor crowd. They are pushing on the wrong end.
And of course the banks have no interest in educating the consumer. This is marketing 101, and the ETF industry is failing quite poorly.
The US has over double the ETF penetration rate compared to Canada, with mutual fund fees that are below half of the average of Canada. There’s an easy double sitting on the table for Canadian ETF companies, and then some.
Your criticism of the industry is quite valid. However, the ETF industry didn’t invent the idea of low-volatility investing. Academics knew about low-volatility anomaly long before the industry latched to the idea.
The back-tested data shows that low-volatility strategy outperforms cap-weighted total market indexes. This is true across many asset classes and geographies. See Falkenstein’s book.
The question is, will the strategy work in real world? It wouldn’t be the first time that a back-tested strategy fails to deliver on its promise once real money is invested.
To address another point you made. It’s absolutely true that low-volatility indexes are closely correlated to the overall market in a meltdown. A high correlation between two indexes doesn’t mean they move in a lock-step. Back testing shows that low-volatility indexes underperformed in good markers and outperformed in bad markets. This under/over-performance wasn’t symmetrical. They captured more of the upside than the downside. If you can capture 80% of the upside but only 60% of the downside, you outperform in the long-term.
I don’t really know how the ETF industry would go about educating investors/savers. There are already lots of resources available on the web and at least one specialty programme on BNN hosted by a prominent ETF “educator”.
Most of my friends are in the 45-65 yr range and are Mutual Fund investors with 2%+ MERs. The fee structure and the bank products make it difficult for any other kind of message to get out to these busy working people. I have answered many friend questions on our inability to control markets and our strong ability to control our costs but to little avail. Only one of a dozen or so friends has put their faith into controlling costs and taken the time to build a decent ETF based portfolio.
It’s really weird in a way. One very good friend asked a lot of questions last week and I showed him my “return” spreadsheets and quietly explained the cost difference over time between his Mutual Funds and my ETFs. This is a pretty smart guy in his field too, but very busy in his work. he expressed an interest for more conversation soon but my guess is that he will let things sit despite his desire to retire in a few years.
“We have met the enemy” and in many cases it is us.
Three years in, as of July 2015, ZLB works like charm; Low volatility with above market return. Just check 3-month , 1/3 year charts vs. TSX Composite. And add 2% yield on top of that. What more one could ask for?
Is there a document online you guys have seen recently that confirms that this uses a 1 year beta?
If so, do you mind pointing me to it?
@Smitty: The website now says, “The Fund utilizes a rules based methodology to select the least market sensitive stocks based on five year beta.”
This post is now almost three years old and I am not sure whether this part of the methodology has changed since it was written.