What Causes an ETF’s Tracking Error?

Last week I explained the importance of monitoring an ETF’s tracking error, which is the difference between a fund’s actual performance and the returns of its index.

The most significant reason index funds lag their benchmarks is the impact of management fees and GST/HST. If your index fund has an MER of 0.25%, you should expect its tracking error to be within a basis point or two of that figure. But it’s often more than that—and sometimes it’s much less. In a series of two posts this week, I’ll look at some real examples from 2012 to illustrate the other factors that can cause an ETF’s returns to vary.

Currency hedging. US and international equity ETFs hedge currency risk using futures contracts. These are renewed every month, and if there’s a dramatic currency movement between contracts—or if the fund experiences a large cash inflow or outflow—that can show up as tracking error. The iShares S&P 500 (XSP) and Vanguard MSCI U.S. Broad Market (VUS) both had tracking errors over 70 basis points in 2012, despite MERs of just 0.24% and 0.17%, respectively.

Currency hedging can also work in the fund’s favour, however. That may help explain why both the iShares MSCI EAFE (XIN) and the Vanguard MSCI EAFE (VEF) outperformed their benchmark last year—indeed, XIN did so by 47 basis points. That was a lucky accident: over the long term one should expect currency hedging to cause a drag on returns because of its significant cost.

Cash drag. Most mutual funds keep cash on hand to pay investors who redeem their units. ETFs typically don’t do this, because their structure allows them to make redemptions without selling securities. But some ETFs may keep extra cash on hand, and this can lead to tracking error.

The BMO Equal Weight REITs (ZRE) trailed its benchmark by 76 basis points last year despite an MER of 0.62%. According to the fund’s annual report: “A small portion of cash was held within the portfolio to accommodate distributions. This portion of the portfolio can cause a drag on positive returns because the ETF was not fully invested. The underperformance caused by the ETF’s cash holdings was 0.14%.”

Poor sampling. Most equity ETFs hold every stock in their benchmark, even if there are thousands of them. It’s dangerous to do otherwise: several years ago, some of the old Claymore ETFs held only a representative sample of the stocks in their US and international funds and they got clobbered with enormous tracking errors.

But broad bond index funds can never do this: the major DEX indexes are enormous, and they include illiquid bonds the fund couldn’t buy even if it wanted to. These funds have no choice but to use sampling: they buy a smaller number of bonds that approximate the overall characteristics of the index (average term, coupon, duration, etc.).

You’re likely to see the biggest sampling errors with corporate bond funds. The iShares DEX All Corporate Bond (XCB) did quite well in 2012 with a tracking error of 52 basis points and an MER of 0.44%. However, since its inception in November 2006 its tracking error has been over 80 basis points.

Securities lending. A number of index funds engage in securities lending: that is, they lend stocks and bonds to short-sellers to earn extra revenue. While some critics consider this risky, when done responsibly it can actually benefit investors by offsetting some of the fund’s management fees. Normally these revenues are very small, although the iShares S&P/TSX 60 (XIU) earned almost $3 million from securities lending in 2012.

Typically funds keep some of these profits for themselves and pass along a portion to investors. (Vanguard is the exception in giving all securities lending revenue to unitholders.) In my opinion, any index fund that keeps revenue from securities lending should first ensure its tracking error is no higher than its management fee. Unfortunately, XIU lagged its benchmark by 20 basis points in 2012 even though it’s MER is just 0.18%.

Later this week I’ll look at other factors that may affect how closely an ETF tracks its benchmark index.

8 Responses to What Causes an ETF’s Tracking Error?

  1. Simon S April 22, 2013 at 9:40 am #

    Hi Dan,

    Thanks for another great post! I was wondering if it’s possible for you to talk about e-series fund performance a bit more in this series. It seems like you concentrate mostly on ETFs, even though a lot of people who follow your blog still invest in mutual funds, particularly e-series.

    In particular, comparing the e-series figures Ken shared in the comments on your post, it appears that two of the funds has significant error beyond their MER:

    Canadian Index e-Series: 6.9%, Benchmark: 7.2%, MER: 0.33%, Delta (D):+0.03%
    Canadian Bond Index e-Series: 3.1%, Benchmark: 3.6%, MER:0.51%, D: 0.01%
    US Index e-Series (US$): 12.6% ($CAD), Benchmark: 13.4% ($CAD), MER:0.35%, D: -0.45%
    International Index e-Series: 15.5%, Benchmark: 14.7%, MER: 0.51%, D:+1.31%

    Where D is (Fund Return + MER – Benchmark), and should ideally be zero, but is often slightly negative. A positive number indicates the fund outperformed the index when MER is accounted for.

    So, any thoughts on the discrepancies of the US and International funds? Thanks!

  2. Canadian Couch Potato April 22, 2013 at 10:31 am #

    Simon: The US tracking error is hard to figure out, because there is no explanation for it in the MRFPs. Historically the e-Series funds have had quite low tracking error, so I’m not too concerned about this, just a bit confused. International funds have some unique issues that I’ll discuss in the next post. Stay tuned. Most EAFE funds seem to show positive tracking error in 2012, which is usually the result of a short-lived anomaly.

  3. Craig April 22, 2013 at 12:38 pm #

    Hi CCP,

    I was wondering what your thoughts are regarding this Globe article that just came out today: http://www.theglobeandmail.com/globe-investor/investment-ideas/strategy-lab/index-investing/is-it-worth-making-the-shift-to-fundamental-indexes/article11446205/

    It is talking about the relative merit of cap-weighted index ETFs vs fundamental index ETFs. I know you prefer a more equal-weighted REIT fund (ZRE) compared to cap-weighted which leans too heavily to Riocan and the like – do you think it would be valuable for the more core holdings? Thanks!

  4. Canadian Couch Potato April 22, 2013 at 1:40 pm #

    @Craig: I’m pretty agnostic about fundamental indexing. The strategy is reasonable, but not essential. I would put it in the same category as giving your portfolio a tilt to value or small cap.

    I don’t think it’s the same issue as the one that comes up with REITs. The goal there is not to enhance returns (as it is with fundamental indexes) but simply to avoid overconcentration in one or two stocks. That is not usually a problem with cap-weighted core holdings.

  5. Ryan April 25, 2013 at 12:13 am #

    @CCP
    How common is it for ETF’s to change their benchmark? Can an ETF conceal or reset it’s tracking error by changing it’s benchmark?
    Thanks

  6. Canadian Couch Potato April 25, 2013 at 8:29 am #

    @Ryan: ETFs do change their benchmark indexes occasionally, but not to hide tracking error. See my most recent post, which touches on this idea:
    http://canadiancouchpotato.com/2013/04/25/when-you-can-ignore-tracking-error/

    The only exception I can think of is that S&P 500 ETFs with currency hedging (such as XSP) now use benchmarks that reset the currency hedge every month. They used to use the S&P 500 in US dollars, which is the more relevant index from the investor’s point of view, but that made them look bad because there were some years where the tracking error was well over 300 basis points.

  7. Steven J Fromm April 27, 2013 at 11:50 am #

    Very informative article. I am just learning about these type of investments and knew very little about tracking errors. Thanks for the well written post on this subject.

Trackbacks/Pingbacks

  1. Friday Links - April 26, 2013

    […] wondered What Causes an ETF’s Tracking Error?  The Canadian Couch Potato has the […]

Leave a Reply