When two exchange-traded funds (ETFs) track the same index, most people assume they're buying the exact same thing. Check the fee, pick the cheaper one, done.
But according to Haneen Sakhi, ETF investment specialist at Xtrackers, the number that separates good trackers from mediocre ones is further down the factsheet.
"Obviously the first thing that most people look at would be the headline fee, but when you're looking at ETFs that are tracking the same benchmark, there can be differences worth checking,” she said.
"One of the main things we say to look at is tracking difference and tracking error. Tracking difference is just the gap between the performance of the benchmark and the ETF, so while that's important, tracking error tells you a lot more about the quality of replication, particularly over the long run."
Tracking difference is a snapshot, she explained: how far the fund drifted from its benchmark over one period. Tracking error measures consistency over time, whether that gap stays narrow year after year or widens when markets get rough. A fund can look fine on tracking difference over one year and still have a poor tracking error over three, which is where the divergence tends to show up.
Why one ETF tracks more consistently than another comes down to how it is built. Trackers differ in what they hold, how closely they follow the index and how far they are allowed to stray from it.
Below, Trustnet explains the main difference between tracker types (by replication method and benchmark fidelity) and why it matters for investors.
Physical replication: Full versus optimised
Most trackers hold the physical securities in the index. For benchmarks with a manageable number of constituents, that means buying everything. For larger, more complex ones, especially in fixed income, providers sample instead, building a portfolio that matches the benchmark's key characteristics rather than holding every position.
"It's very dependent on the strategy itself, but typically you're looking at a well-diversified broad benchmark with over 1,000 stocks," Sakhi said.
In fixed income, sampling means matching key rate duration, overall duration, credit ratings and sector weightings. "That's how portfolio managers end up having a portfolio that tracks very well to the benchmark," she explained.
Full and optimised replication carry no meaningful fee difference, she added; as funds grow large enough, even equity trackers tend to drift toward holding the full index.
Done well, sampling should keep tracking error low, even without holding every security, she explained.
Synthetic replication and the tax edge
Instead of holding the stocks in the index, synthetic ETFs hold collateral and enter a swap with a counterparty that delivers the index return. Providers use this structure where physical replication is impractical, in illiquid frontier or emerging markets, for example, or where it produces a performance edge.
"The synthetic ETF doesn't hold the underlying security of the benchmark itself," Sakhi explained. "It holds collateral and enters into a contract with a counterparty who delivers the index returns. The objective is the same, it's just the mechanism that's different."
The clearest edge is on US withholding tax. Irish-domiciled physical ETFs holding US equities benefit from the US-Ireland tax treaty, reducing withholding tax on dividends to 15% instead of 30%. Some synthetic ETFs tracking qualifying indices can access 0% withholding under the US Section 871(m) framework instead, a gap Sakhi put at roughly the fund's dividend yield multiplied by 15%.
"If you look at the dividend yield [for the S&P 500], it is about 1.1%. With withholding tax at 15%, you can expect around 17 basis points of savings versus an Irish physical ETF."
Similar advantages apply elsewhere, for different reasons. China's repurchase agreement (repo) market, where short-term lending between financial institutions shapes the pricing of the swaps behind synthetic ETFs, and India's capital-gains tax treatment make synthetic structures worth considering there too.
For an investor, the practical point is: two ETFs tracking the same index can end up with different returns purely because of how they are structured, before either fund manager has made a single active decision.
This is even more of a problem because of an UCITs rule by which passives that do not use full physical replication are held to the same capping rules as active funds, further limiting the flexibility of active management, as Trustnet recently revealed in its conviction cap campaign.
Hybrid replication
Some providers now split a single fund between the two approaches. Sakhi pointed to one of the firm's own products as an example: the emerging markets and US segments replicate synthetically, while Europe and World ex-US exposure is held physically, combining the tax and access benefits of swaps with the simplicity of physical holdings where there's no advantage to doing otherwise.
From pure passive to active: How far a tracker is allowed to stray
Replication method is one axis. The other is how closely a fund is required to stick to its benchmark in the first place and that varies a lot more than most investors assume.
"Essentially you start with your pure benchmark," Sakhi said. "Your typical MSCI World, S&P 500, FTSE 100 are your core and that's essentially the foundation of how most people build their portfolios."
From there, thematic and screened products, ESG funds or sector plays still count as pure passive, since they simply replicate a different, narrower index.
Smart beta funds go a step further. These are still passively tracking a benchmark, but built around factors such as value, quality or momentum, or alternative weighting schemes rather than market capitalisation.
Beyond that sits a blurrier category Sakhi grouped together as "index plus" and systematic active: funds with a defined, rules-based process for deviating from a benchmark in pursuit of extra return.
"This is really when you start moving away from pure passive, because you have a benchmark, you have a defined deviation away from that benchmark just to try and enhance and enable some of that additional return versus the benchmark itself. So that's a hybrid between your passive and active."
These funds still carry tracking error constraints and keep the benchmark as their starting point, she said, but a model, rather than a person, is deciding where to deviate.
At the far end are high-conviction active ETFs, where a manager picks the holdings and the fund itself is simply the wrapper. The split differs by region, Sakhi noticed: European flows have gone mostly into the systematic, rules-based middle category, while in the US, high-conviction active dominates, largely for tax reasons specific to the US mutual fund and ETF wrapper structure.
As for how these are used, “typically, core is built with passive, then [investors usually like] adding in some conviction, whether that's via your smart beta category, or looking at an active ETF as well, as an overlay on top of the core aspect of the portfolio.”
Tracking error matters most in the core category – funds tracking the MSCI World, S&P 500 or FTSE 100 – where fees are already wafer-thin and products are otherwise near-identical.
"In a world where everything, especially the core aspect, pretty much costs the same, even one basis point of difference matters because it can cost you performance in the long term if the tracking error isn't great," Sakhi said.
Further out on the spectrum, from smart beta through systematic active, the benchmark or the model behind it is the real differentiator, and tracking error against that chosen benchmark tells an investor less than whether it's the right benchmark, or the right strategy, to be tracking at all.
For anyone checking these points before buying, ETF issuers publish performance against benchmark on their own product pages. The gap usually only becomes visible over three years or more, Sakhi noted, as a fund that looks fine after one year can still have been tracking poorly all along.