Tracking Difference, The Perfect ETF Metric

July 31, 2015

The vast majority of exchange-traded funds are designed to track indexes. When you buy one of these funds, your expectation is that you will get the return of the index minus the fees the fund charges.

In other words, if you buy an ETF that tracks the S&P 500 and charges 1 percent, and the S&P 500 goes up 10 percent in a year, you expect to earn 9 percent on your investment.

Expected Return = Index Return – Fees

In practice, that’s not always the return you get. Some funds track well, some track poorly and some track somewhere in between. All else equal, you want a fund that tracks its index well.

‘Tracking Difference’

To help investors find these funds, our friends at FactSet—who power our ETF data and analytics—evaluate what’s called the “tracking difference” of each fund. This data are presented free of charge on the landing pages for every ETF in the U.S., which you can access using the URL convention www.etf.com/ticker (e.g., www.etf.com/spy, www.etf.com/eem, etc.).

Tracking difference is a common-sense measure of the real-world experience investors have in a fund. To calculate it, the team at FactSet compares the performance of a fund with the performance of its index over a one-year period; they then roll things back one day and observe another one-year period; they roll things back another day and observe another period; and so on.

They do this until they have 250 individual one-year observation periods, and we then present the data on ETF.com in three ways:

  • Median Tracking Difference: the average experience of the fund
  • Max Upside Tracking Difference: the best one-year holding period
  • Max Downside Tracking Difference: the worst one-year holding period

Here’s a snapshot of what it looks like on the Efficiency tab of our fund page for the S&P 500 SPDR (SPY | A-98), the largest ETF in the world:

You can see that the fund charges 0.09 percent in fees, but on average, lags its index by 0.16 percent per year over the average measured period. Over any given one-year period, you might do a little better or a little worse: in the best period tracked, you trailed the index by 0.09 percent; in the worst, you missed by 0.21 percent.

Sometimes funds do better than you’d expect. Here’s the snapshot for the iShares MSCI Emerging Markets ETF (EEM | B-98):

You can see that, in the average one-year period, it actually earned back some of its expense ratio. It charges you 0.68 percent a year in fees but only missed the benchmark by 0.60 percent a year on average. That’s sweet!

I find tracking difference one of the most important statistics when evaluating ETFs; at least as important as the expense ratio.

After all, as the saying goes: It doesn’t matter what you pay. it matters what you get.

Reasons For Tracking Failures

There are a variety of reasons funds might track better or worse than their indexes. These include:

  • Index Changes: When a security enters or exits an index, the fund has to trade.
  • Dividend Reinvestment: How quickly and well the fund reinvests dividends; some funds—like SPY—can’t.
  • Rebalancing Costs: If the index rebalances, the fund must too.
  • Sampling: Many funds don’t hold every security in the index, but rather, a representative sample. This increases tracking risk.
  • Securities Lending: Funds can lend out securities and charge a fee, adding to performance.
  • Taxes: There may be stamp or withholding taxes for foreign-listed securities.
  • Statistical Anomalies: Various rules about how people calculate index values and fund net asset values can create artificial tracking error.

The last point can frustrate tracking analysis.

Among other issues, differences in the way funds set their net asset values and indexes strike their pricing levels can create the perception of tracking error where none truly exists. The team at FactSet has corrected as many of these errors as possible, sourcing like-for-like data wherever they can, so the data are as clean as it gets in the industry. Still, it’s not perfect.

Perfection, But Variability

I recently surveyed our database to find out how good ETF managers are at tracking their indexes. The answer is: very good.

There are currently 1,081 ETFs in our database with two years of tracking history. (There are a variety of funds—including leveraged and inverse funds—for which FactSet does not calculate tracking difference). Of the funds we have:

  • 527 trail their benchmarks, on average, after adjusting for fees
  • 35 match their index perfectly, on average, after adjusting for fees
  • 520 beat their benchmarks, on average, after adjusting for fees

Impressively, the median ETF tracked its index perfectly (after fees) over the trailing two-year period.

There was, of course, huge variability: Some funds tracked incredibly well and some missed by a country mile. But the average performance was quite good:

  • 51 percent of funds delivered the index return or better, on average
  • 73 percent of funds came within 0.10 percent of their index, or better, on average

Vanguard and iShares had perhaps the most impressive performance, with 69 percent and 60 percent of their wide ranges of ETFs, respectively, meeting or beating the return of their indexes excluding fees.

Here are the results for the 10 fund companies with the largest number of ETFs we have data on:


Median Tracking Difference Excluding Expenses
Positive Negative Mean % ≥ Index
BlackRock 158 104 0.04% 60%
Invesco PowerShares 39 84 -0.07% 32%
SSgA 63 54 0.00% 54%
First Trust* 15 57 -0.11% 21%
Vanguard 46 21 0.02% 69%
Van Eck 30 24 0.04% 56%
Barclays Capital 25 27 -0.03% 48%
Guggenheim 28 23 0.02% 55%
WisdomTree 24 20 0.01% 55%
Global X 19 17 0.03% 53%
  • First Trust makes extensive use of fair valuation in its international ETFs, which throws off this tracking data. Other firms use fair valuation as well, but the First Trust data is particularly exposed to statistical anomalies along these lines. Their actual tracking is likely better than is listed here.
  • Data as of 7/30/15

In general, the funds that did the best were funds where the issuer engaged in significant securities lending.

Take the Guggenheim Solar (TAN | D-26) as an example. The fund charges 0.71 percent per year in fees, but on average, beats its index by 2.01 percent. In other words, it’s recouping 2.78 percent per year above a rational expectation! The reason is that there is huge demand to short solar stocks, and it can charge a pretty penny loaning them out to short-sellers.

ETF investors should always be diligent when evaluating ETFs. I was hugely pleased that the data showed most funds are managed well. But there are outliers on each side, and a bit of research will pay big dividends.


At the time this article was written, the author held no positions in the securities mentioned. Contact Matt Hougan at [email protected].

Find your next ETF

Reset All