Key takeaways
- Over the five years to the end of August 2026 the STOXX Europe 600 returned 38.3% as a price index, 57.4% as a net total return index and 62.3% as a gross total return index.
- The 24.0 points between the price and gross versions are 38.5% of the entire five-year gross return, so dividends did more than a third of the work.
- Withholding tax accounts for 4.9 of those 24.0 points, and it is the only thing separating the gross series from the net one.
- MSCI deducts 0% from UK dividends and 30% from US dividends. Across 2024 the gross to net gap was 0.03 points on UK shares and 0.54 points on MSCI World.
- For the ten years to the end of September 2026, MSCI World in sterling annualised at 13.11% gross and 12.56% net, a spread of 0.55 points a year.
Price return vs total return, and the third version almost nobody quotes
When a headline says a market is up for the year, that is nearly always the price return: share prices, nothing else. The total return version of the same index keeps the dividends and puts them back in. There's a third version as well, net total return, which puts the dividends back after a tax deduction. One basket of shares, three published indices, and over five years they can sit 24 points apart.
Take the STOXX Europe 600, which covers the 600 largest companies by free float market capitalisation in developed Europe. Over the five years to the end of August 2026, in euros, the price version returned 38.3%. The net total return version returned 57.4%. The gross version returned 62.3%. Three numbers, published by one administrator, on one date, for one list of companies.
So price return vs total return isn't a rounding question. Dividends are 24.0 points of that 62.3% gross return, which is 38.5% of the whole figure. Compare a fund against the wrong version of its own benchmark and that's the size of the error you've built in before you start.
Three versions of one index, read side by side
Everything below comes from the STOXX factsheets for August 2026, in euros. STOXX publishes the price version under the symbol SXXP, the net version under SXXR and the gross version under SXXGR.
| Period to end-August 2026 | Price | Net total return | Gross total return |
|---|---|---|---|
| 1 year | 18.4% | 21.3% | 22.0% |
| 3 years | 42.1% | 53.5% | 56.4% |
| 5 years | 38.3% | 57.4% | 62.3% |
| 5 years, annualised | 6.8% | 9.6% | 10.3% |
| 10,000 euros of exposure, 5 years | 13,830 | 15,740 | 16,230 |
The bottom row is the same arithmetic in money, before any fund charge, spread or dealing cost. On these index returns the dividend reinvestment is worth 1,910 euros and the withholding deduction costs 490 euros. The chart above plots the five-year row for this index and for the DAX.
Watch how the gaps scale. Over one year the dividend effect is 2.9 points and the tax effect is 0.7 points. Over five years those compound to 19.1 and 4.9 points. Nothing in the method changed. The same annual drag simply had five years to work.
What each version does on the day a dividend is paid
The mechanics sit in the providers' own calculation guides, and they're simpler than the names suggest. The STOXX Calculation Guide of September 2026 puts it in one sentence: "Dividend payments are invested in return indices only and neglected in price indices."
In the corporate actions table of that same guide, a regular cash dividend reduces the adjusted price of a gross return index by the full dividend, and a net return index by the dividend less withholding tax. Price indices "are not affected". S&P Dow Jones Indices describes the identical split in its index mathematics methodology: "In a total return index changes in the index level reflect both movements in stock prices and the reinvestment of dividend income."
FTSE Russell's calculation guide, in its August 2026 edition, works a three-day example that makes the arithmetic visible. The capital index runs 3,190, then 3,200, then 3,220. On day three an ex-dividend adjustment of 5 index points lands. The total return index runs 1,000.00, then 1,003.13, then 1,010.98. Across those two days the capital index gained 0.94% and the total return index gained 1.10%. Five index points of dividend, 0.16 points of extra return.
One detail hides in the same table and it catches people out. A special cash dividend adjusts the price index too. A price index therefore ignores the ordinary dividend and not the extraordinary one, which is exactly the opposite of what most readers assume. If you're reconciling an index against a holding of your own, that's where the stub usually is.
Net total return is a tax stance, not a fact about the shares
Gross total return has one definition: reinvest the whole declared dividend. Net has as many definitions as there are tax positions, and the published series picks the harshest of them.
MSCI says its net series "approximates the minimum possible reinvestment of regular cash distributions", reinvesting the dividend after deducting tax "by applying the maximum rate of the company's country of incorporation applicable to institutional investors". FTSE Russell uses the same stance, the rate for institutional investors "who are not resident in the same country as the remitting company and who do not benefit from double taxation treaties". S&P Dow Jones Indices applies "the rate to non-resident institutions that do not benefit from double taxation treaties".
The rates MSCI publishes for that stance, set out in its August 2025 calculation methodology, vary enormously by country of incorporation. The United Kingdom is 0%. Germany is 26.375%. The United States is 30%. Switzerland is 35%. A net index for a world portfolio therefore carries a blended haircut that depends on where the dividends are paid, not on who holds the fund. That is also why free float adjustment and country weights feed straight into the size of the net to gross gap.
Almost nobody actually pays the maximum rate. The IRS treaty table, last revised in May 2023, shows 15% on ordinary dividends from US corporations paid to a resident of Ireland or of the United Kingdom, against 30% for "Other Countries". A fund resident in Ireland holding US shares therefore sits in the 15% column and not the 30% one. The published net series is a floor, the gross series is a ceiling, and a real fund sits between the two.
FTSE Russell makes that explicit by publishing more than one net. Alongside the maximum-rate series it calculates net-of-tax indices for a US Regulated Investment Company and for a UK pension fund, plus custom versions for an Irish pension fund, a Luxembourg or Netherlands UCITS and a UK UCITS. Same shares, same dividends, different owner, different index level.
Why the gap is 0.03 points on UK shares and 0.54 on a world index
Two things set the size of the tax wedge: how large the dividends are, and where they're paid from. MSCI's factsheets make the contrast sharp, because the annual figures for one index are published in both a gross and a net version.
In 2024 the MSCI United Kingdom IMI index returned 9.12% gross and 9.09% net, in sterling. The gap is 0.03 points, and the methodology says why. MSCI's withholding table puts the United Kingdom at 0%, and its own footnote explains that with no UK withholding tax the gross and net amounts of the cash dividend reinvested are equal. What residue there is comes from constituents incorporated outside the UK and from property income distributions on UK real estate investment trusts, which the same MSCI table docks 20%.
In the same calendar year MSCI World returned 21.33% gross and 20.79% net in sterling, a gap of 0.54 points. Over longer windows the drag is steady rather than dramatic. For the ten years to the end of September 2026, MSCI World in sterling annualised at 13.11% gross against 12.56% net. Compound that spread across the decade and the net series finishes 4.8% below the gross one.
Yield explains most of the direction. At the end of September 2026 MSCI World yielded 1.52% and the MSCI United Kingdom index yielded 3.17%. The UK index has roughly twice the dividend and almost none of the tax. The world index has half the dividend and nearly all of the tax.
Germany publishes both versions, which is why the DAX looks strong
Most markets quote the price index and bury the total return one. Germany does the reverse. STOXX, which administers the index, describes the DAX as "primarily calculated as a performance index and is therefore one of the few major country indices that also takes dividend yields into account".
The quoted DAX is the gross total return version of an index of 40 companies. STOXX publishes the price return version of the same index separately, under the symbol DAXK. Over the five years to the end of August 2026 the price version returned 43.2% and the gross version 65.8%, a gap of 22.6 points, or 3.2 points a year. Over the single year to that date it was 7.0% against 9.9%.
This is where cross-border comparisons go wrong. Any table that sets the quoted DAX beside another market's price index is putting a total return series next to a capital-only one, and the German column carries 3.2 points a year the other one cannot. The arithmetic says nothing about German companies. It says the two columns are not the same measurement.
The case against quoting gross total return at all
There's a fair objection to all of the above: the gross series describes an investor who doesn't exist. Nobody receives the full declared dividend on a foreign holding. MSCI is candid that its gross series "approximates the maximum possible reinvestment of regular cash distributions" and that the amount reinvested "does not include tax credits". It's an upper bound, not an experience.
That objection has real force when the question is whether a fund did its job. A fund measured against a gross index is being asked to beat a tax position it cannot occupy, because nobody reinvests a dividend the taxman has already taken a slice of. The reply is that the net series is an equally artificial lower bound, built on a no-treaty rate that a well-structured fund avoids.
Both are conventions, and neither is anyone's return. The honest reading is that the published gross and net series bracket the answer, and for a world equity index the width of that bracket is the 0.55 points a year above. Fund-level costs are a separate layer again, which is the subject of tracking difference vs tracking error.
What this comparison cannot tell you
These are index figures, not fund figures. None of them is reduced by an ongoing charge, a spread, a platform fee or stamp duty, and the index licensing fees a tracker pays for the right to copy the index are inside the fund's costs, not inside the index.
The windows are short and they are not forecasts. Five years to August 2026 covers one particular stretch of European markets. A five-year window ending in a drawdown splits price and dividend very differently, because falling prices shrink the capital leg while the dividends keep arriving. The 38.5% dividend share is a fact about that sample and nothing more.
Currencies are mixed here on purpose, because that is how the providers publish. The STOXX and DAX series are in euros and the MSCI series in sterling. Withholding is a percentage of the dividend rather than a fixed sum, so the direction of the gross to net gap holds in any reporting currency, but the headline returns don't compare across the two.
The withholding rates are published policy, not your tax position. MSCI's table is the maximum institutional rate by country of incorporation, and the IRS cautions that its own treaty table "is not a comprehensive guide to all eligibility requirements for every treaty rate of tax or exemption listed". What a UK investor actually bears depends on the fund's domicile, the treaty and the asset class, and a reader living off the income has a different question again, closer to natural income vs total return.
What would change the conclusion
If dividend yields fell towards zero, the three versions would converge and the distinction would stop mattering. That has not happened: MSCI World still yielded 1.52% at the end of September 2026, and the UK index 3.17%. The 24.0-point five-year gap on the STOXX Europe 600 is a direct function of those yields, and it shrinks or grows with them.
If withholding tax were abolished or made fully reclaimable, net and gross would collapse into one series. The UK already shows what that looks like, with a 0.03-point gap in 2024 against 0.54 points on MSCI World. The reverse is also possible. A rise in statutory rates would widen the net to gross gap without anything happening to a single share price.
The one number worth checking is which version your own benchmark is. A factsheet that quotes a price index while the fund itself reinvests income shows a flattering gap every year, and the gap is the dividend stream rather than skill. Find the words price, net or gross on the factsheet before reading the comparison, because the same index under a different label is a different measurement.