ETF Tracking Difference vs Tracking Error: Reading ETF Factsheets

Two numbers, one question: what did the fund actually deliver?

Two ETFs can advertise the same index and the same expense ratio and still hand you different returns. The figures that explain why are not the headline yield or the star rating. They are the tracking difference and the tracking error — one tells you the size of the fund's gap to its benchmark, the other tells you how reliable that gap is.

They are routinely confused, partly because some providers use "tracking error" loosely to mean the gap itself. The distinction matters if you are comparing index funds you intend to hold for years: a fund that lags by 0.10% a year in a straight line is a different proposition from one that lags by an average of 0.10% but swings 0.60% either side of its index.

Tracking difference: how big is the gap?

Tracking difference is simply the fund's total return minus the index's total return over the same period. It is expressed in percentage points and can be annualised. Morningstar's explanation is a good reference point: a tracking difference of −0.5% for a year means the fund's net asset value (NAV) lagged the index by half a percentage point over that year (Morningstar).

A simple worked example. Suppose an index returns 8.75% in a year and the fund's NAV returns 8.62% with all distributions reinvested. The tracking difference is −0.13 percentage points, or −13 basis points. If the same fund returned 8.81%, the tracking difference would be +0.06 points — positive, meaning the fund beat its own benchmark.

Positive tracking difference is not a manager's genius. It usually comes from securities lending revenue, a better withholding-tax rate than the one the index vendor assumes, or fair-value pricing of foreign holdings.

Tracking error: how stable is that gap?

Tracking error is the annualised standard deviation of the differences between fund and index returns — normally calculated from daily returns and scaled to a year by multiplying by the square root of 252 trading days (Fidelity). It is a risk measure, not a performance measure.

Vanguard's own explainer gives the practical reading: if tracking error is 50 basis points, roughly two-thirds of the time the fund's tracking difference should sit within 50 basis points of its average tracking difference (Vanguard). A low tracking error means the gap is predictable; a high one means it jumps around, which makes the fund's future return less certain even if its average cost is low.

What drives each metric

Fees and total cost

Cost is the main driver of tracking difference. Among funds tracking the same index, more expensive funds almost always show a larger negative gap, and fee cuts show up in tracking difference over time (Morningstar). But the stated expense ratio is not the whole bill. Index rebalancing forces the fund to trade, and it pays spreads and commissions that the index does not; currency conversion, stamp duties and swing-pricing adjustments can add more. Our breakdown of TER vs total cost of ownership covers the full stack.

Cash drag and dividend timing

An index is a frictionless calculation. A fund is not: it holds a cash buffer to meet redemptions, pay distributions and absorb subscriptions in kind. Every day that cash sits uninvested is a small drag, and it shows up as negative tracking difference in a rising market. Dividend timing compounds this. Indexes that assume instant reinvestment of dividends, and funds that receive them days later, create a structural gap that widens with the dividend yield of the underlying market.

Sampling risk

Not every fund buys every holding. Full replication is practical for large, liquid indexes; for broad or illiquid universes, providers use sampling or optimisation, holding a representative subset. Sampling usually lowers transaction costs, but it introduces dispersion: the sample can drift away from the index between rebalances, and the resulting gap pushes tracking difference and tracking error in either direction (Vanguard). This is why emerging-market and small-cap index funds typically report higher realised tracking error than developed large-cap funds.

Tax drag and withholding tax

For dividend-paying indexes, tax is often the largest single explainer of tracking difference — and it is a domicile question, not a management question. For non-US investors, dividends paid by a US-domiciled ETF are generally subject to 30% US withholding unless reduced by a treaty (IRS). An Irish-domiciled UCITS ETF holding the same US stocks typically suffers around 15% withholding at fund level under the US–Ireland treaty, with no further Irish withholding for non-resident investors (State Street). As a simple illustration, on a hypothetical dividend yield of 1.3%, a 15-percentage-point difference in withholding works out to roughly 0.2 percentage points of return a year (1.3% × 15%).

That helps explain why a European-domiciled S&P 500 tracker can post a smaller negative tracking difference than a US-domiciled fund tracking the identical index. Note the jurisdiction: this comparison applies to non-US investors in the EU, UK and Asia — a US-resident investor faces an entirely different tax picture.

Securities lending, and where a positive number comes from

Many physical ETFs lend part of their portfolio and pass the revenue back to the fund. Lending income can offset fees and flatter tracking difference, which is legitimate but not free: the fund takes on counterparty and collateral exposure, and revenue is not guaranteed in stressed markets. The mechanics are covered in our piece on ETFs and securities lending.

Don't confuse the NAV gap with the trading gap

Tracking difference is measured against NAV. What you experience depends on the price you pay. An ETF's market price can sit above or below NAV, and the creation/redemption arbitrage mechanism is what keeps the two close. FactSet's study of US-listed ETFs found that among funds trading more than $250,000 a day, the median trailing 12-month daily premium or discount was 0.02%, and 68% closed within 0.10% of fair value (FactSet). The SEC's investor bulletin describes the same mechanism and notes that historical premium/discount data is published by the fund (SEC Investor.gov).

The practical consequence: a tightly tracking fund with a wider spread can cost you more to buy than a looser tracker with a tight spread. Our explainer on creation and redemption covers how that gap opens.

Where the numbers are published

In the EU, the disclosure is prescribed. The ESMA guidelines on ETFs and other UCITS issues require an index-tracking UCITS to include in its prospectus the anticipated level of tracking error in normal market conditions, and to state the size of the tracking error at the end of the period in its annual and half-yearly reports. The annual report must also explain any divergence between anticipated and realised tracking error and disclose and explain the annual tracking difference (ESMA guidelines, 2014/0011).

That gives European investors a genuinely useful comparison: what the fund promised, and what it delivered. Vanguard's EU report for the period ended 31 December 2024, for example, lists an anticipated tracking error of 0.200% and a realised tracking error of 0.052% for its FTSE Developed World UCITS ETF, alongside a tracking difference of −0.09 percentage points (Vanguard Funds plc semiannual report). That combination — small average gap, low variability — is what careful index investors are looking for. In the US, there is no single equivalent tracking mandate, so fund websites and factsheets are usually the only place to find both numbers side by side.

Check three things before trusting any figure:

• The measurement period. A one-year tracking difference is noisy; three- to five-year figures are more informative, and inception-date figures can be dominated by a single unusual year.
• Gross or net. Confirm dividend treatment (gross vs net index returns) and whether the fund return is NAV-based with income reinvested.
• Currency. A fund reporting in EUR while the index is USD embeds an FX effect unless the report says otherwise.

Risks and costs

No tracking metric is risk-free, and the cheapest-looking number often carries the most hidden risk.

• Positive tracking difference is not repeatable. Lending revenue, favourable withholding rates and fair-value pricing timing all fluctuate, and tax treaties can change. A fund that beat its index in one year may lag in the next.
• Replication method changes the risk, not just the cost. Swap-based (synthetic) structures can deliver a tighter gap to the index but add counterparty and collateral risk, which the ESMA guidelines require to be disclosed in the prospectus.
• Sampling adds dispersion. A sampled portfolio can drift from its benchmark between rebalances, which is a structural source of tracking error that no fee cut can remove.
• Small funds cut both ways. Tracking error statistics from a short history mean little, and small funds can trade with wider spreads and face a higher probability of closure or merger.
• Switching has its own cost. Realising a capital gain in one fund to save a few basis points a year in another is frequently value-destructive. In the EU, local capital gains tax rates and rules vary by residence; in the US, ETFs are generally more tax-efficient than mutual funds but sales still trigger gains in taxable accounts.
• Currency-hedged share classes track a hedged index, not the plain one. Hedge roll costs are a permanent drag, which is why hedged funds show structurally worse tracking difference.
• Premiums and discounts on the exchange are a separate cost. Check the median bid-ask spread and premium/discount history alongside the tracking figures.

Bottom line

Tracking difference answers "how much did the fund lag or beat its index?" and tracking error answers "how consistently?" Read them together, always over three to five years, and always in the same currency and with the same dividend treatment on both sides. Then check that the reason for a surprisingly good number — lending, tax-treaty access, a favourable valuation timing choice, or heavy sampling — is a risk you are comfortable holding for a decade. The fund with the lowest headline expense ratio is not always the one with the smallest or most stable gap to its index.