Last week we looked at two low-volatility ETF strategies based on beta and the standard deviation of daily price movements. Now let’s complete our roundup by looking at a third methodology used by MSCI, the index provider behind the new iShares family of minimum-volatility ETFs. This is one is completely different from the other two.
MSCI’s strategy is based on creating what’s called a minimum variance portfolio, an idea that goes back to Harry Markowitz’s Modern Portfolio Theory in the 1950s. What makes this strategy unique is that the individual companies don’t matter much in isolation, or even relative to the market as a whole. What’s important is their correlation with each other: the goal is to combine stocks in a way that results in a portfolio with the lowest possible volatility. Think of it like a cake recipe where you add baking powder and salt—which can be unpleasant on their own—because they taste delicious when combined with the other ingredients.
The methodology starts with a parent index that represents the broad market—such as the MSCI Canada Index—and then applies a number of rules to optimize that portfolio. Some companies are deleted from the index altogether, while those remaining are assigned whatever weight would lead to lowest overall volatility. (The specifics are complex, but you can read more on the methodology if you’re inclined.)
Once you understand this strategy, you’ll appreciate why the iShares MSCI Canada Minimum Volatility (XMV) looks a lot more like a broad-market Canadian equity ETF than its counterparts, the BMO Low Volatility Canadian Equity (ZLB) and the PowerShares S&P/TSX Composite Low Volatility (TLV). The Big Five banks are all among the top holdings, for example:
XMV Top 10 Holdings
|
||
Royal Bank of Canada | 3.3% | |
CIBC | 3.1% | |
TransCanada Corp | 3.1% | |
Bank of Montreal | 3.1% | |
Bank of Nova Scotia | 3.1% | |
Toronto-Dominion Bank | 3.1% | |
Imperial Oil | 3.1% | |
BCE | 3.0% | |
Enbridge | 2.9% | |
Husky Energy | 2.9% | |
Whereas the BMO and PowerShares low-vol ETFs are heavy on consumer retailers and utilities, these sectors play a small role in XMV. Energy, banks and materials are by far the biggest sectors in the iShares fund, just as they are in the Canadian market as a whole:
XMV Sector Breakdown
|
||
Energy | 22.6% | |
Banks | 18.2% | |
Materials | 15.3% | |
Food and staples | 7.9% | |
Insurance | 7.1% | |
Telecommunications | 6.7% | |
Utilities | 5.7% | |
Diversified financials | 3.5% | |
Real estate | 3.0% | |
Other | 6.8% |
US and international options, too
iShares also offers minimum volatility ETFs covering US equities, international equities, emerging market equities and an all-world fund. But these indexes bear less resemblance to their parents than XMV does. (The Canadian index, being smaller and less liquid that the giant US and international benchmarks, was subject to more constraints on the methodology.) For example, Apple is not even among the 122 companies in the iShares MSCI USA Minimum Volatility (XMU).
Interestingly, none of these ETFs use currency hedging, since doing so would introduce a new source of volatility and completely change the profile of the funds. That makes the iShares MSCI EAFE Minimum Volatility (XMI) a lot more interesting. There is currently no Canadian-listed ETF that tracks the well-known MSCI EAFE index (Europe, Japan and Australia) without currency hedging, which is an expensive and dubious strategy with international stocks. So XMI seems like a reasonable alternative to the more popular international ETFs, even if it does hold just 172 companies, compared with 972 in the parent index. BlackRock has even capped the MER at 0.35% until the end of 2014, making it even cheaper than Vanguard’s EAFE fund.
If you’re comfortable trading in US dollars, all of the iShares US and international low-vol ETFs are also listed in New York, and these versions have lower management fees.
After reading both low-volatility posts I’m left in a bit of confusion over which approach is best, or at least what the likely ramifications are of each of the three approaches. Do you come down firmly on one as your preference, and if so why? By my reading of the IShares model, it seems as if they’re trying to create a situation where the holdings cancel each other out and create stasis. It might be conducive to stability but does it work against growth?
Also what, if any, drawbacks are there to investing money in an ETF so early in its existence. Each of these companies is well known and stable, and certainly the companies that make up the ETFs are well established, but are their liquidity or bid/ask spread issues, or other factors to keep in mind before diverting funds to any of these?
@Trevor C: I wish I could offer a strong opinion about which low-vol strategy is the best, but I really have no idea, and I don’t place a lot of trust in index backtests. Interestingly, when I spoke to Jean Masson of TD Asset Management about this, he felt the differences would likely be quite small over any long period. He said that in his experience, despite the differences in strategies, “the returns and the volatility have been in the same ballpark.”
I don’t personally intend to use any of these strategies in my own portfolio: I just present them here in an effort to help readers understand their differences. But if someone were intending to use one of these ETFs, my advice would be to consider the sector exposure in relation to other holdings they may have. For example, if a low-vol ETF is full of real estate holdings, that would overlap with a REIT fund you might already hold. If the low-vol fund was full of dividend-paying telecoms and consumer retailers and I already owned a dividend fund with most of the same companies, I might want to avoid overlap there, too.
Regarding the risks of investing in new and small ETFs, as you say, if the underlying holdings are liquid, it should not be a significant problem, though I would keep an eye on the bid-ask spreads and always use limit orders:
https://canadiancouchpotato.com/2012/09/10/etf-liquidity-and-trading-volume/
Bugger – for a moment I thought XMI was a “proper” Canadian product, but it’s just a wrapper for a US one – meaning that, in a TFSA, I will lose withholding tax from all the securities to their locsl governments, and again to the US government – so from 100%, to perhaps 85%, to about 72%.
This is *rubbish*. Isn’t it MUCH better to buy, say, VUKE.L, HMCX.L or similar on a brokerage that allows it (examples are the FTSE 100 and FTSE 250, the two main UK indices/indexes).
@Dave: Yes, XMI has the same problematic structure that most Canadian-listed international equity ETFs have. Very few hold their stocks directly, and so there is that extra layer of withholding taxes. However, as far as I know, very few online brokerages allow you to buy securities on the London Stock Exchange. Those that do tend to charge hefty fees for the privilege.
I’m a Brit so I still have an account with a British stockbroker. Seems like I am actually better off paying the income tax from holding those investments directly, than moving them over here into a TFSA.
Questrade said I could buy shares on foreign exchanges by phone… At a cost of $195 a trade!!
@Dave: Unless your tax rate on dividends is less than 15% and capital gains are completely untaxed in your British account, I would think you are still better off holding international equities in a TFSA, no?
Well… I’m actually looking at early retirement. The UK works with a tax credit so, as far as I know, they aren’t actually withholding anything (though I don’t benefit from the tax credit either, being Canadian-resident).
Cap gains – I’m not planning on selling.
So unregistered – I lose at my taxable rate in Canada. TFSA I have to hold the US wrapper meaning I lose a guaranteed 15%.
I rather resent giving the US 15% of my dividends for nothing; I’d rather give them to the Canadian government. Plus there is double currency conversion loss – I know this is likely to be negligible.
(That is, early retirement with a fairly modest income!)
Wouldn’t we be better off using an appropriate holding of bonds to dampen the volatility in our portfolios?
@BC Doc: Excellent question. I think it’s fair to say that adding a significant bond allocation to a portfolio would dampen its volatility but also lower its expected returns. Over the last 30 years this has not necessarily been true, since over many periods bonds actually outperformed stocks. But with yields where they are now that’s probably not what one should expect.
That said, there is a DFA paper making the argument that the best risk-adjusted returns some from a portfolio of value stocks and high-quality bonds:
http://lowriefinancial.ca/whats-the-best-way-to-lower-volatility/
@BC Doc:
CPP Investment Board expects negative returns for bonds in the next few years. I will post a link to their paper if I can find it.
If CPP IB is right, increasing your bond allocation at this point in time doesn’t seem like a good tactical move.
CPP IB presentation:
http://www.osfi-bsif.gc.ca/app/DocRepository/1/eng/oca/pdf/CPP2012_Raymond_e.pdf
Jump to page 24 to see their forecasts of expected returns for various asset classes. The next page shows 68% confidence bands of expected returns.
Again, this is a tactical argument against increasing your bond allocation, not a strategic one.
I am assuming that those negative expected returns are a blend of capital gains (negative ones) and interest. But, that’s what I am having trouble wrapping my head around. If the expected return for bonds is negative in the next 5 years and one is designing a complete couch potato portfolio with 45% bonds for a 45 year-old (all held within an RRSP) why not go with laddered GICs for the next 5 years and (which should yield around 1.5%+ per annum) and then switch over to bonds around 2017 when their expected returns are positive. I do realize that those positive expected returns are likely only to be realized from a bond portfolio that one owns today and will likely be dampened significantly if one buys the bond portfolio 5 years from now, but at least the returns would be positive.
Of course, this all smacks of trying to time the market. But, at the same time, we are exiting an unusual period of 30 years where interest rates have fallen and bond prices have risen as a result, a situation which is likely to reverse itself over the next 30 years.
@GoldStone and Noel: Clearly investors need to be cautious in their expectations for bonds, but let’s remember that predictions about interest rates and bond returns have always been notoriously inaccurate. No one knows what fixed-income returns will be like in 2017, including the good people at the CPP investment board.
There is nothing wrong with using a GIC ladder (or short-term bonds) in place of a broad-based bond fund. That’s a reasonable risk-management strategy. Just remember that it could work against you, too. Investors who did this during the last four years (for the same reason people are considering it now) dramatically underperformed.
https://canadiancouchpotato.com/wp-content/uploads/2012/07/Opportunity-Cost_Bortolotti.pdf
That’s a fantastic article Dan. I was actually trying to do that simulation myself before.
But, while it exposes the folly of trying to time the market and predict things, it also suggests possible downside, while not precluding the upside, lol. As with everything, longer time horizons takes care of things.
Dan, you mentioned that you don’t intend to use any of these ETFs in your portfolio. With all the new ETFs that have come on the market lately, the lowered expectations of bonds, and the rise of the risk-doesn’t-mean-return meme etc, have you considered taking a refreshed look at your essential portfolio recommendations?
@Trevor: Not really. I think a plain-vanilla ETF portfolio should always be the starting point, which is why I favour the Complete Couch Potato. If investors want to explore other strategies, that’s fine, but I worry that it can be a distraction. I’m not even confident that the value/small tilt in the Uber-Tuber portfolio will outperform in any meaningful way. The longer I do this, the more I value simplicity.
That said, it’s worth questioning the “standard” 40% bond allocation that is used for balanced portfolios throughout the industry. A lot of expected return projections for these balanced portfolios are based on 8% to 9% bond returns, which is what we got over the last 30 years. Clearly that is not a prudent estimate for the next 30 years. Investors who need higher returns may have to start thinking about a 30% bond allocation as the starting point.
Yes, I started to get distracted with other strategies – and started to back test portfolios with additional ETFs for more weight in: small cap, value, REITs, real return bonds, emerging markets, etc… I know back-testing can be dangerous, but it is fun to explore! And sure, there are some years that a particular tilt does really well (e.g., REITs recently). But in the long-term, I’ve come full circle to stay with my basic XIC, XIN, XIU & XBB approach (from one of Daniel Solin’s earlier books). I figure the split between my stocks and bond allocation will have the biggest impact; and just focus on that. I’ve been in a couch potato strategy long enough to think that I understand the theories of adding some of the other weightings for potentially greater return without much added risk; but I tend to value simplicity even MORE as more different ETFs come out!
Noel:
For non-equity correlated component consider the simplicity of using 5 year GIC ladders at the best rates for the low duration tranche of your portfolio. The GICs are like having corporate bonds rates with government bond risk due to CDIC insurance.
Consider also the 1-5 year ladder corporate bond ETF CBO (for say a TFSA), XBB (bond universe) for medium duration and XRB (long duration real returns) and/or another longer duration bond ETF to round it out. XRB however will behave a bit differently than a pure long duration government bond ETF because of its inflation sensitivity.
@Andrew – Thanks. I just partially implemented the Complete Couch Potato for a 45yo friend at 45% bonds/55 equities and all is invested except for 3/4 of the 45% bonds (1/4 of it is already in XRB, as per the CCP model portfolio). He/we are just hesitating on the remainder (actually mainly him) so I am just trying to further educate him on XBB before we plunge. I do make everyone buy and read the Dan’s excellent book, The Moneysense Guide to the Perfect Portfolio before I am willing to sit down with them because I don’t want anyone doing anything unless they understand what they are doing. Also, it saves people running back to their brokers when the market tanks (they just keep regularly investing instead!).
There is enough room in his RRSP for the entire bond portfolio to be sheltered. I’d just go with XBB and call it a day and accept any trade-offs but he might want to play what he thinks is safe(r), even though that is undeterminable. Certainly there is the potential to have the volatility dampened. I think that a mix of laddered GICs and/or the CBO ETF along with XBB might be the best way to go for him on the remaining 3/4 of the bond component given his degree of personal risk averseness. He actually does have an extremely high ability to accept risk given his income, assets and future need for capital, but of course that doesn’t always reflect one’s degree of risk averseness or proneness.
Very useful discussion! With the current high premiums to purchase bonds it really makes little sense to purchase bonds or bond funds in a non-registered account just now! Despite the low yields GICs do make more sense in such accounts if you must have fixed income. JB makes this case clear:
https://www.pwlcapital.com/en/Advisor/Toronto/Kathleen-Clough-Justin-Bender/Justin-s-Blog/Blog-Justin-Bender/October-2012/When-Should-I-Buy-GICs
@CCP: re
“That said, it definitely worth questioning the “standard” 40% bond allocation that is used for balanced portfolios throughout the industry. ”
in the context, I wasn’t sure if you were questioning the 40% allocation as income or only specifically as bonds. I take your point that bonds may not continue to have as high returns as compared to recent anomalies due to declining interest rates over the last 10 years. But if you drop the bond allocation to 30% because you “need higher returns”
what do you insert to replace this investment room? I assume you must lose some safety if you replace bonds with equity (you never get something for nothing), but can you compromise less volatility and not quite as much expected return by using preferred shares in this slot?
For instance, in a large retirement portfolio distributed over RRSP and Taxable accounts, with “some” tolerance for risk, would this distribution of ETF’s be reasonable?:
20% Canadian equity (divided into regular and small cap)
20% US Equity (divided into regular and small cap)
20% International Equity (divided over regular and small cap)
10% Canadian Preferred Shares
15% Canadian Real Bonds such as XRB or ZRR
15% Canadian Bonds, like XBB
I have postulated Canadian Preferred because of the preferential Canadian tax treatment of dividends, but would this overbalance the Canadian equity component ?
@Noel: my query above to CCP, might also apply to your recommendation for your 45 year old friend, (ignoring for now the Income GIC/Bond distribution dilemma – I assume by 45% Bonds you mean 45% mix of Bonds and GICs). Is there any Preferred Shares in your 55% equity portion, and if so, what country?
@Goldstone: I followed your link to the CPP IB presentation, and realized with a shock that this is the CPP Investment Board — that is, the CANADIAN PENSION PLAN!! — OUR, that is yours and mine, Pension Plan!! As I understand it, correct me if I have misinterpreted the slides, those dudes have committed to active management, having considered and rejected passive management. Ouch!
@Oldie: I’m using the Complete Couch Potato portfolio with 45% bonds/55% Equities. The only change I made to it was using 50% HXT & 50% XMD instead of XIC in order to provide a small-cap ’tilt’ as outlined in https://canadiancouchpotato.com/2012/03/22/under-the-hood-ishares-sptsx-completion-xmd/ (XIU and HXT are similar but the latter uses a swap to convert dividends to equity).
I just want to keep it simple, so no preferred shares. I figured using HXT & XMD instead of XIC was complicated enough!
@Oldie: In terms of risk factors, preferred shares are generally closer to corporate bonds than to equities. Their primary risk is rising interest rates, which will cause them to lose value, just like bonds. Keep in mind also that pref shares are subordinate to bonds in the event of the issuer’s bankruptcy. I think preferred shares ETFs can have a place in a taxable account, because they offer some bond-like characteristics with equity-like tax treatment. But in a tax-sheltered account they have little going for them.
@CCP:
If you compare XMV vs XIC sector breakdown “pie chart”, XMV looks a bit more diversified then XIC (less concentrated in energy, banks and material)
Do you think XMV could be a good alternative to XIC or ZCN inside your model portfolios?
http://ca.ishares.com/product_info/fund/overview/XMV.htm
http://ca.ishares.com/product_info/fund/overview/XIC.htm
@CCP:
Also, would you consider XMW a good alternative to XWD in your “Global couch potato” model portfolio? They both have a similar MER, but XMW includes emerging markets and must be a bit more tax effective since it’s a global fund, while XWD includes 3 funds which must be balanced every year by the fund manager..
@Jas: I prefer broad-market, plain-vanilla funds for the Global Couch Potato, not funds with embedded strategies, especially those that are brand new and unproven.
@Canadian Couch Potato
I searched on ishares and did not find what is the turnover for these minimum volatility etfs but according the the MSCI web site, the turnover for the MSCI Canada Min Volume index is 21.03% compared to 4.12% for the MSCI Canada index. Source here: http://www.msci.com/resources/factsheets/index_fact_sheet/msci-canada-minimum-volatility-index-net-cad.pdf
So I imagine there is also a higher turnover in XMV than in VCE and if it is similar to their respective indexes, would that make a significant difference for taxes?
@John: In theory, sure, higher turnover could mean more distributed capital gains and higher taxes. At the end of 2013, XMV reported a reinvested distributions of over 19 cents (just under 1%), most of which was probably capital gains. That wouldn’t be confirmed until the final reporting is done in few weeks.