False Tracking Error = Bad ETF Decisions
October 24, 2012
Here’s how I know the ETF Revolution has long since passed, and what we’re living in now is the new ETF normal: The questions from advisors are getting a lot smarter.
I used to get emails about how creation and redemption worked. Now I get questions about tracking error.
Unfortunately, most people think about tracking error all wrong.
Here’s a perfect example. Take two funds that have been in the headlines a lot these past few weeks, the Vanguard MSCI Emerging Markets ETF (NYSEArca: VWO) and the iShares MSCI Emerging Markets Index Fund (NYSEArca: EEM).
Now imagine you’re a Sophisticated Investor. You know a few things: You know expense ratio matters. You know spreads matter. You know tracking error matters.
So you pop up your Bloomberg, and here’s what you see:
Even on trading, Vanguard wins on expenses. But Holy Meatballs Batman, what are those guys down in Pennsylvania doing!? A tracking error of 4.433 percent?
And at this point, many advisors will make a critical mistake, assuming that the Vanguard fund is horribly mismanaged. It’s not an unreasonable assumption, if in fact this was an accurate tracking error number. But it’s not.
Remember, academic tracking error is the annualized standard deviation of daily return differences. If the index is up 1 percent today, and VWO is up 0.95 percent, well, that’s -.05 percent to add to the series. Take that whole series, plug it into your stats package, get the standard deviation, annualize it, and there you go.
There are a few reasons this is all a terrible idea. First of all, imagine that VWO was actually missing its mark by 0.05 percent, day in and day out. Well, the standard deviation of those daily differences will be zero. It’s enormously consistent.
Of course, if the index stayed perfectly flat all year, you’d lose a cumulative 12 percent of your investment in that “perfect” index fund. And because expense ratios are assessed daily, no matter how big, it will never show up as “tracking error.”
Second, “annualizing” a daily number is almost a useless exercise. A 0.01 percent daily standard deviation would equal about 0.16 percent annualized. A 0.05 percent daily standard deviation would give you about 0.79 percent over a year.
Neither of those figures actually has any bearing on whether you’ll be ahead of or behind your expected index returns, and since almost all tracking error is mean-reverting, in both cases your expected return likely centers somewhere around the index’s return, minus expense ratio.
In short, these kinds of tracking error measurements have essentially no bearing on actual investor experience. What most investors care about is holding period returns. So that’s what we look at here at IndexUniverse. We pose a simple question: What’s the difference between the index’s return over the past year, and the ETF’s?
Grab that number for today, then go back to yesterday and get the same one-year numbers. Keep doing that until you have a year’s worth of rolling returns, based on two years of data. Here’s what you get for these two funds:
When we first started doing these calculations and presenting them to folks a few years ago, we made the point of calling this “Tracking Difference” to separate it from academic tracking error. We got a lot of funny looks, but recently the Investment Company Institute adopted precisely the same language, so we think it’s sticking now.
And what does it tell us? It tells us that your expectation for any given year should be that VWO trails its index by about its expense ratio (26 basis points vs. a 22 basis point expense ratio). EEM will meanwhile actually beat its own expense ratio, trailing the index by just 0.48 percent, which is likely due to good securities lending.
However, with EEM, your range of expected outcomes is much, much wider, from trailing the index by 1.22 percent to beating the index by 0.52 percent. That’s a pretty wide range of outcomes.
So purely from the “tracking” perspective, VWO gets the nod here, exactly the opposite of what that “tracking error” statistic shows.
So where does that big number come from? Accounting issues. Vanguard, like quite a few ETF issuers, chooses to publish a “fair value” net asset value (NAV). That means instead of taking the price of some South Korean company after markets closed there, you adjust the price of that company based on certain proxies such as how the currency has moved, how futures have traded, and so on.
It’s essentially what the market does all day with VWO—after all, it will trade up or down even though most of its constituents are closed for trading. It’s not right or wrong—it’s a fund accounting choice, designed to get the NAV of the fund closer to reality.
EEM makes a different choice—to publish a NAV more aligned with how the index provider determines the level of the index. So on any individual day, EEM will report a NAV that’s much more closely tied to the reported change in the MSCI Emerging Markets Index.
These kind of accounting issues pop up all over the ETF landscape. Many international indexes mark all currency conversions at 4 p.m. GMT. Funds that track those indexes often mark their currency conversions at 4 p.m. Eastern time. If the currency moves a lot in the intervening five hours, you can get wild swings in perceived “tracking.”
In the bond markets, index providers and issuers can use different pricing services, leading to apparent tracking differences even if the fund holds precisely the same portfolio as the index.
The moral of the story is simple: Always consider your investments from your actual holding expectations, not just a surface statistic.
At the time this article was written, the author had no positions in the securities mentioned. Contact Dave Nadig at email@example.com.