Published on: 2026-08-13
Updated on: 2026-08-13
Two funds, one index, one calendar year. The index closes up 10.00%. The first ETF returns 9.94%. The second, tracking the identical benchmark, returns 9.86%. Nobody swapped the index and nobody bought a different stock, yet eight basis points went missing somewhere between the benchmark and the investor.
Eight basis points sounds like a rounding error, and across a single year it behaves like one. Stretch it over twenty years and a six-figure position and the arithmetic stops looking harmless.
The industry has a name for the missing eight points: tracking difference. It records what replication actually delivered once a fund has paid its fees, settled its taxes, crossed its spreads and collected income the index never sees.
Cheaper can still finish behind. A fund charging 0.08% can beat one charging 0.05% when lending revenue and tax recovery are working.
The fee is the opening figure. Withholding tax, replication, cash drag, rebalance execution and lending income all move the final number.
Two metrics, two jobs. Tracking difference sizes the shortfall; tracking error measures how steadily the fund got there.
Read three and five years. A single year flatters funds that caught a kind rebalance or a rich lending market.
Check the index first. Near-identical names sometimes hide different index families, which makes any return comparison worthless.
Tracking difference subtracts the benchmark’s total return from the fund’s total return over a defined period.
Tracking difference = ETF return - index return
A negative figure means the fund lagged; a positive one means it finished ahead. Because an index pays no custody fees, executes no trades and holds no cash, the normal outcome for a physical equity ETF is a small negative number sitting close to its expense ratio.
Tracking difference is often confused with tracking error.
| Metric | Question it answers | Best used for |
|---|---|---|
| Tracking difference | By how much did the fund beat or lag its index? | Comparing realised cost on one benchmark |
| Tracking error | How steadily did the fund track over time? | Judging replication consistency |
Tracking error takes the annualised standard deviation of those return differences and reports consistency. Two funds can post an identical annual shortfall of 0.15%, one arriving in a straight line and the other swinging between 0.02% and 0.40%. For a long-term holder, the size of the shortfall carries more weight than its rhythm.
The published fee accrues daily against net asset value and behaves as a constant drag. Figures published by the Investment Company Institute in March 2026 put the 2025 average expense ratio for index equity ETFs at 0.14% and index bond ETFs at 0.09%.
Core broad-market equity funds price well below those averages, several at 0.02% to 0.03%. Fee compression has narrowed the spread between rival products to a few basis points, pushing the explanation into less visible parts of the fund’s machinery.
Full replication holds every index constituent at index weight. Optimised sampling holds a representative subset chosen to mirror the benchmark’s sector, size and risk characteristics, an approach used where an index contains thousands of names, some thinly traded.
Sampling lowers transaction costs and accepts a small chance the subset drifts from the whole. Synthetic replication uses swap agreements to receive the index return, removing sampling risk and substituting counterparty exposure.
Dividends crossing borders are generally taxed at source. The rate a fund pays depends on its domicile, the treaty network attached to it and its legal structure. Index providers calculate total return on an assumed tax treatment, so a fund that keeps more dividend income than assumed recoups part of its fee without owning a different security.
Where offering documents permit, a fund may lend portfolio holdings for a fee, most of which is credited back to the fund. In securities with strong borrowing demand, that revenue offsets part of operating costs and occasionally more. Lending policy is the usual explanation when the more expensive fund tracks more closely.
Dividends land on payment dates and subscriptions arrive in cash; neither invested the instant it appears. An index reinvests notionally and immediately. In a rising market, uninvested cash lags; in a falling market it cushions.
An index swaps constituents at a stated price on an effective date and pays nothing to do so. A fund has to trade, crossing spreads and absorbing market impact, often alongside every other fund following the same methodology on the same day. Skill in scheduling those trades is worth basis points.
Funds holding international securities often trade while the underlying markets are closed. Comparing a fund’s close with an index struck at a different local close introduces a timing artefact, one that usually washes out over longer periods.
Most comparisons stop at the expense ratio: 0.05% beats 0.08%, so the cheaper fund wins. Realised returns are less obliging.
| Metric | Fund A | Fund B |
|---|---|---|
| Expense ratio | 0.05% | 0.08% |
| Index return | 10.00% | 10.00% |
| Fund return | 9.88% | 9.91% |
| Tracking difference | -0.12% | -0.09% |
Fund B charges three basis points more and finished three ahead. Lending revenue, cleaner rebalance execution, better withholding tax recovery or tighter cash management can each account for the reversal, and often several do.
The expense ratio states what the manager charges. Tracking difference reports what ownership cost once charges, taxes, trading and offsetting income are counted. One is a quoted price. The other is the receipt.
Confirm the benchmarks are genuinely identical. Different index families produce different exposures, and any return difference traces back to that.
Compare three-year and five-year tracking difference over matching periods, with the expense ratio as an opening reference.
Check the replication method: full, sampled or synthetic.
Note the fund domicile and how dividend income is taxed at source.
Check whether securities lending is permitted and how much revenue returns to the fund.
Compare bid-ask spreads, since the investor absorbs that cost at the point of trade.
Look at fund size and trading history, which influence spreads and large-order execution.
Because funds operate in the real world. Fees, withholding tax, trading costs and cash balances pull returns below the index, while lending income and efficient tax treatment push them back up.
Tracking difference measures how far a fund finished from its index. Tracking error measures how variable that distance has been.
There is no universal threshold. A smaller and steadier shortfall is preferable, though the figure only means something against funds tracking the same index over the same period.
Yes. Securities lending revenue and favourable withholding tax treatment can exceed a fund’s costs in a period, producing a positive tracking difference. Check the benchmark methodology and pricing convention before reading it as durable skill.
An index is an idea. A fund has to buy that idea in a live market, hold it through corporate actions, pay tax on its income, meet redemptions and publish a price every day. Each obligation costs something, and one or two earn something back.
Those frictions are permanent. What separates one fund from another is how well each is managed, and that competence never appears on a factsheet. Tracking difference sits at the end of the chain as one number, which is why it belongs beside the expense ratio whenever two funds share a benchmark.
Before your next purchase, pull the three- and five-year tracking difference for every fund on your shortlist. For more on fund costs and structure, explore the ETF guides from EBC Financial Group.
