It’s report card time in the investment fund world. Every year at this time, all mutual funds and ETFs are required to file a Management Report of Fund Performance, a document that contains a wealth of useful information. (They’re available from the SEDAR website.) If you’re an index investor, one of the most important things you can learn is a fund’s tracking error.
Tracking error is the difference between the performance of the fund’s underlying index and that of the fund itself. Remember, the goal of an index fund is to deliver the returns of a particular asset class, as measured by an index. If a fund does a consistently poor job of that, it’s time to start looking for another option.
iShares released its MRFPs yesterday, and I looked up the tracking errors for a number of their most popular ETFs:
Canadian equity ETF | Ticker | Fund | Index | Error |
iShares S&P/TSX 60 | XIU | -9.23% | -9.08% | -0.15% |
iShares S&P/TSX Capped Composite | XIC | -8.93% | -8.71% | -0.22% |
iShares S&P/TSX Completion | XMD | -8.33% | -7.85% | -0.48% |
iShares S&P/TSX SmallCap | XCS | -16.66% | -16.43% | -0.23% |
iShares DJ Canada Select Dividend | XDV | 3.49% | 4.09% | -0.60% |
iShares S&P/TSX Capped REIT | XRE | 20.95% | 21.67% | -0.72% |
Global equity ETF | Ticker | Fund | Index | Error |
iShares S&P 500 | XSP | 1.07% | 1.71% | -0.64% |
iShares MSCI EAFE | XIN | -12.71% | -12.12% | -0.59% |
iShares MSCI Emerging Markets | XEM | -17.09% | -16.40% | -0.69% |
iShares MSCI World | XWD | -3.57% | -3.20% | -0.37% |
Fixed income ETF | Ticker | Fund | Index | Error |
iShares DEX Universe Bond | XBB | 9.38% | 9.67% | -0.29% |
iShares DEX All Government Bond | XGB | 9.82% | 10.20% | -0.38% |
iShares DEX All Corporate Bond | XCB | 7.51% | 8.24% | -0.73% |
iShares DEX Short Term Bond | XSB | 4.42% | 4.65% | -0.23% |
iShares DEX Real Return Bond | XRB | 17.87% | 18.35% | -0.48% |
iShares U.S. IG Corporate Bond | XIG | 10.40% | 11.85% | -1.45% |
Sizing up the performance
Canadian equities: You should expect a Canadian equity index fund’s tracking error to be about as large as its MER—perhaps a few basis points more. The iShares ETFs have exceeded expectations here: XIC, XIU, XMD and XCS all have tracking errors lower than their annual fees. That means the fund managers not only tracked the indexes perfectly, they even managed to add a little value, probably through activities such as securities lending and arbitrage. This is top-shelf performance.
US equities: Tracking foreign equities is always more difficult due to withholding taxes. But it’s currency hedging that creates the biggest drag on fund’s like XSP. The fund’s tracking error was –0.64%, despite an annual fee of just 0.25%. But there’s more: XSP’s index accounts for the currency hedging strategy, which is reset each month, and this index returned 1.71%. However, the S&P 500’s actual return in US dollars was 2.11% last year. So the true underperformance of XSP was –1.04%. Over the last seven years, this performance gap has averaged more than 2%. This is why I don’t recommend currency hedging. (Note that the two international iShares ETFs that do not use hedging—XEM and XWD—both had tracking errors lower than their MERs.)
Fixed income: The two largest bond funds, XBB and XSB, also had tracking errors lower than their MERs, another excellent result. The corporate bond ETFs did not make out so well. It is actually much more difficult to track a corporate bond index than many investors realize. As a result, XCB’s tracking error has averaged –0.82% over the last five years, despite an MER of 0.42%. XIG, which tracks US investment-grade corporates with currency hedging added, shows a huge tracking error of –1.45%, but it outperformed its underlying ETF in US dollar terms (10.40% versus 8.89%). I admit that one has me baffled.
I’ll have more tracking error reports to share next week when all the documents are available.
I’m a novice, please help me out Dan. Why do some funds have a minus sign and others not? As always the info you provide is thought provoking!
@John: The funds with a minus a sign lost money in 2011. :)
So what would be the point of buying XIN and XSP? Would a portfolio not suffice with XIC and XWD and XBB for diversification?
@alni: The only reason to buy XIN or XSP is to get the currency hedging. Some people are willing to endure the long-tern drag in order to eliminate currency risk from the portfolio, and that’s fine if they understand the cost. But XWD, in my opinion is a better choice as it is approximately half US and half intentional, all in a single ETF with no hedging. That’s what I recommend in the Global Couch Potato.
Your posts on tracking errors are some of my favourites, thank you for compiling the results and presenting them with such clear explanations.
One aspect that I have been reviewing for international equities was choosing between iShares XIN and the equivalent TD eFunds (TDB905 or TDB 911 if I do not want hedging) tracking the MSCI EAFE indexes. Tracking error is of the main criteria that will help me in deciding with allocation, but to my surprise it seems as iShares and TD do not track the exactly the same index.
For example for TDB911 the website mentions that their benchmark statement is: MSCI EAFE ND (C$). It would appear the ND means net of dividends (I saw this information in the funds report a while ago and I am having a tough time locating the report again, so I am not a 100% sure at this moment about the interpretation of ND). I do not think that the benchmark iShares uses subtracts dividends. This would then distort any comparison between the two competing funds when using tracking errors.
Are you aware of many funds that have a benchmark that are net of dividends? If so would you have any advice on the best way to compare competing funds with slightly different benchmarks?
@Philippe: These are great questions.
First, you are correct that ND means “net dividends.” All indexes used as benchmarks for investment funds assume reinvested dividends, which is really the only accurate way to measure performance. Usually the S&P indexes used the term “total return” or TR, but the two terms mean the same thing.
Your second question is also a great one. I hinted at it in my comments about XSP in the post. Funds that use currency hedging often benchmark their performance against an index that measures returns in Canadian dollars, assuming a currency hedge that is adjusted monthly. From the fund’s perspective, this makes sense, because no manager can hedge currency perfectly, and since most funds reset the currency hedge every month, it makes some sense for them to evaluate their performance against a benchmark that does the same.
However, as an investor, I probably don’t care about this. If I buy a foreign equity fund that uses currency hedging, I probably expect my returns to be more or less the same as those that would be received by local investors. In other words, if I buy an S&P 500 fund, I want to get the same returns as if I bought the stocks in US dollars. There can be a fairly large difference here. In 2011, the S&P 500 CAD-Hedged Index returned 1.71%, whereas the S&P 500 in US dollars returned 2.1%. As far as I’m concerned, the tracking error should be measured against the 2.1% figure.
The same concept is at work with the MSCI EAFE indexes. Funds that use currency hedging (such as XIN) track the MSCI EAFE Index in local currencies, assuming a monthly currency hedge. Unhedged funds (such as TDB911) track the MSCI EAFE Index in Canadian dollars. In 2011, the former index returned –12.1%, while the latter returned –10.0%.
I hope this makes sense. I think I will do a post about this in the coming days, because it is quite confusing. :)
Do you know if Blackrock’s Canadian ETFs are allowed to invest in securities outside of their mandate in order to try to eliminate tracking error/compensate for MER? I think it’s crazy that HYG (tracking a high yield *bond* index) in the U.S. is allowed to hold a small percentage of assets in Apple shares, though I understand why they’re doing it–to try to juice returns and compensate for the higher MER in comparison to JNK.
Hi Dan,
I’ve also been reading your comments about currency hedging with great interest. Currency hedging seems to be a complex issue. I’ve found a few papers on the Internet that tend to say that, over the long term, it make no big difference. Yet on the shorter term (less than 15 years), it could make a relatively big difference to an investor. My usual view of long term, for stocks, being 5 years or more, I find waiting 15 years for purchasing power parity to recover a little long. So, here’s my question:
Are you completely sure that the tracking penalty due to hedging (relative to the unhedged index) is a permanent one? In other words, would it be possible for currency hedging to give a positive boost, instead of a negative one, if the currency movement between the CA$ and US$ was reversed? Maybe this kind of hedging simply reduces currency movements, instead of completely eliminating them?
What’s you opinion?
@Chris: I am not aware of any iShares ETF that invests outside its mandate: the risk of even greater tracking error is just too high. I have heard that this occasionally goes on in bond index funds (higher weighting to corporates in a DEX Universe tracker, for example), though I have never been able to find an actual example.
Where did you hear that HYG holds Apple shares? The fund’s holdings are disclosed on the website every day, and there is no mention of any holding other than JNK.
@CCP Fan: The increased tracking error in currency hedged funds occurs no matter which way the loonie moves: it is never a positive tracking error. You can still do better with a currency hedged fund during a period where the Canadian dollar rises rapidly, of course. It was a huge help in the mid-2000s. But over the long run, the loonie would have to soar to great heights for it to work in your favour.
During the 10 years from 2001 through 2010, the loonie went from 66 cents US to above par. Yet the currency-hedged Global Couch Potato outperformed the non-hedged version by only 84 basis points a year. That’s not a trivial difference, but I think it’s a lot smaller than one would have expected considering the huge run-up in the currency.
https://canadiancouchpotato.com/2011/04/18/the-couch-potatos-10-year-report-card/
In a large portfolio, there is always the option of hedging half the currency. Nothing wrong with that.
@CC: You’re right… I misread this article and assumed they were referring to HYG: http://finance.yahoo.com/news/shares-are-too-tempting-not-to-bite-.html
It looks like they’re referring to Blackrock’s mutual fund counterpart to HYG instead.
@Chris: That’s still pretty disturbing, but not too surprising in an actively managed fund.
Thank you for your answers to my questions, they have been very helpful.
What I conclude from your explanations is that fund managers use a benchmark that is more relevant to them in terms of parameters they can influence. Fair enough.
I remain somewhat puzzled however as to why some TD eFunds of foreign equities use benchmarks that are net of dividends. Yes the fund manager can not control foreign withholding taxes, but the last time I looked the EAFE index had a dividend yield of around 3%. Not including dividends in this in the benchmark does not seem appropriate.
I had not even realised how complicated matters were with funds that are hedged and their “alternate” benchmarks. Thank you for elaborating on the topic and I will be looking forward to your future post explaining these points in more detail.
@Philippe: Glad I could help. To clarify, “net dividends” means “including dividends.”
Ok thank you for clarifying, net dividends would then presummably be after withholding taxes, which is a fairer way to measure the performance of the fund manager.
Hi there,
First off, I want to thank you for your informative articles. They are a boon to Canadian investors everywhere!
You mentioned that the US ETFs seem to have fared poorly due to currency hedging, and that XEM and XWD did not suffer from such a fate. Can you name any US index ETFs which would not have such hedging, or that you would recommend in terms of MER efficiency?
*Sorry, I forgot to specify: Any US index ETFs that do not have such hedging AND are available on the TSX?
The former Claymore funds were interesting too. Half were entered in SEDAR as Claymore and the other half as iShares, so searching “fundamental index” is the easiest way to find all the RAFI ones.
Anyway, looks like tracking error has improved significantly on CIE since they switched away from representative sampling. -13.1% for the NAV, compared to -12.2% for the index. With a MER of 0.73% and TER of 0.05%, that’s only a difference of about 0.1%. (Market value decrease was a bit more, but I assume the NAV is a fairer comparison to the index. (?))
@Frank W: Thanks for the comment. The one fund worth looking at is the iShares (formerly Claymore) US Fundamental Index ETF, which is available in an non-hedged version with ticker CLU.C. Otherwise your best bet is to use US-listed ETFs and find a way to reduce the costs of currency conversion.
@Nathan: Yes, I noticed that too, and it holds for CLU as well. I am definitely looking at changing my Uber Tuber portfolio to recommend CIE and CLU.C instead of the PowerShares version. My initial reluctance to recommend the Claymore ETFs was due to the extreme tracking errors, but they seem to have plugged that leak, and with BlackRock managing the funds going forward I have more confidence in them.
Regarding NAV versus market price, I think you could make an argument for either them being the better way to measure tracking error. However, most fund providers use NAV only, so when making comparisons with Claymore I always use NAV to be consistent.
Hi Dan,
Thank you for the information. Are there any Canadian ETFs that track the Russell 2000 unhedged that you know of? I’ve been unable to find one that does.
@Frank: No, sorry. Your choice is the hedged iShares ETF or a US-listed ETF.
Dan,
For the Uber Tuber portfolio, have you considered switching PRF/CLU plus VXF to a combination like (VBR or VIOV) plus (VOE or VTV) plus VTI instead? Looks like you could get the same Fama-French factor loadings (about 0.3/0.2 value/small for your current portfolio) with significantly lower fees and same or greater diversification.
Definitely agree that CIE is the best choice for international small/value tilt though, aside from possibly DFA if available. The offerings from vanguard and others are much sparser than on the US side, and since CIE is Canadian domiciled you avoid the double withholding taxes (which isn’t an issue for US holdings, as you know).
Actually, decided to take my own advice and analyze the various alternatives in this post:
http://www.bogleheads.org/forum/viewtopic.php?f=10&t=93931
A combination of 2/3 PRF, 1/3 VXF (as in the Uber-Tuber) has a Fama-French factor weighting of about 0.3/0.12 value/small since 2006 (calculated more accurately than my previous estimate), and a combined MER of 0.35%.
One could achieve exactly the same factor loading, and thereby the same theoretical risk-reward, with (for example) 17% VBR, 42% VOE, 34% VTV, and 7% VTI, for a total MER of 0.14%. Or, if we don’t care about getting *exactly* the same factor loadings, you could simplify it to 20% VBR, 40% VOE, 40% VTV, with the same MER and 0.33/0.14 value/small; or even, 50% VBR, 50% VTV with a MER of 0.18% and value/small loadings of 0.38/0.21.
(For folks willing to ‘do the math’ though, my suggestion would be to start by deciding on factor loadings appropriate to your own situation, then figuring out the mix of funds to best achieve those loadings. Robert T on the bogleheads forum has a fantastic summary of all aspects of this approach here (although you have to extrapolate the Canadian part): http://www.bogleheads.org/forum/viewtopic.php?t=7353&highlight=collective)
One could get similar performance to the US part of uber tuber with half the fees by substituting PRF and VXF with a 50/50 split of VBR and VTV. Or, for a large portfolio, you could get an ever closer match by adding VOE and/or VTI.
@CCP: Thanks for your detailed answer!
@Nathan: Thanks for the analysis. If you want to swap PRF and VXF with VTV and VBR, that would be fine, but the last thing I would want is to add more ETFs to an already complex portfolio.
By the way, what software or data are you using to determine the factor exposure of each ETF?
No problem! I agree, simplicity is good. Although once portfolios get to a certain size, it may be worth it to some to add an extra fund on if it means saving ten or twenty basis points.
I got the Fama-French factor data from Ken French’s website here: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. ETF data came from yahoo finance. (In historical returns, their ‘adjusted close’ column gives total return.) Finally, I used OpenOffice Calc to hold the data and R to do the regressions. All freely available. There is a great, easy-to-follow tutorial here on how to put it all together: http://www.calculatinginvestor.com/2011/04/19/fama-french-tutorial/
I also made a minor addition to their script to print out the data in a nicely formatted way: http://www.closetindexer.com/files/Personal/ff/ff_regression.R (Tutorial above describes how to use. Then just paste in the actual name and MER of the fund before posting the results to your favorite investing forum or whatever. :) )
@Nathan: This is a great resource, thanks. I will take a closer look at it and see if my little brain can figure it out! :)
Glad to help. When I get around to it, I’m planning on building a web front-end to automatically grab the data from yahoo finance and run R, so all you would have to do is enter the fund or ETF ticker symbol as found in yahoo finance and the rest will be automatic. I’ll let you know if/when I get it together. (On the other hand, I think it can be helpful to actually go through the process to really understand how the regression works. It’s really neat. :) )
Hi Dan,
Just wondering if you are planning to do a similar analysis for Vanguard’s ETFs. I quickly looked at one of them (VWO) and seemed to me that the tracking error was enormous. I am curious if the rest of their ETFs follow the same trend?
Cheers!!!!
Cesar
@Cesar: I looked up the report, and you seem to be right, though that’s really surprising. But it’s also worth noting that VWO had a positive tracking error of +1.51% last year, with the reporting years ending October 31. I think the issue is the one I describe here:
https://canadiancouchpotato.com/2010/04/23/international-tracking-error-part-1/