Key takeaways
- Leave £100,000 alone for 30 years at a 6.6% gross return and a 1.00% annual fee ends at £503,236, against £640,667 at 0.20%. The gap is £137,431 — 21.5% of the lower-fee pot, and more than the sum you started with.
- That 21.5% doesn't move when you change the return assumption. At 4% a year the same pair ends at £305,433 and £239,914 — different money, the same 21.5% shortfall.
- The fees actually deducted at 1% add up to £77,674 over the 30 years, but terminal wealth is £177,088 below a zero-fee portfolio. The missing £99,414 is compounding on money that was no longer in the account.
- The stylised 1% is above what the average investor now pays. The ICI puts the 2025 asset-weighted expense ratio for US equity mutual funds at 0.40% and for index equity mutual funds at 0.05% — though the median equity share class still charges 0.99%, and the 90th percentile 1.84%.
- Vanguard's Advisor's Alpha framework claims advice can add "up to, or even exceed, 3% in net returns" — but only 44 basis points of its own table are point estimates, and 30 of those come from cutting fund fees.
The answer in money: £137,431 of a £640,667 pot
A 1% annual fee sounds like a rounding error. Here's what it does over a working life.
Take £100,000, leave it alone for 30 years, and assume it grows 6.6% a year before costs. Charge 0.20% a year and you finish with £640,667. Charge 1.00% and you finish with £503,236. The difference is £137,431, which is more than the sum you put in.
Both inputs are borrowed rather than invented. The 6.6% is the annualised real — after-inflation — return on US equities from 1900 to 2025, reported in the UBS Global Investment Returns Yearbook 2026 compiled by Elroy Dimson, Paul Marsh and Mike Staunton. Using a real return keeps every pound below in today's money. The fee convention — grow the balance by the return, then deduct the fee from the year-end balance — is the one the US Securities and Exchange Commission uses in its own investor bulletin on fees, updated 23 July 2025. Running that bulletin's example through the same arithmetic reproduces its published answers: $100,000 growing at 4% for 20 years ends at $208,413, $198,211 and $179,213 under fees of 0.25%, 0.50% and 1.00%, against the SEC's "approximately" $208,000, $198,000 and $179,000.
The currency is decorative. Swap pounds for dollars, euros or rand and every proportion below is unchanged.
The size of the gap does not depend on what markets do
This is what makes fee drag different from every other argument in investing, and the reason is arithmetic rather than judgement.
A percentage fee multiplies. Each year the account is worth its gross growth times (1 − fee). Over 30 years the low-fee pot divided by the high-fee pot is (1 − 0.002)30 ÷ (1 − 0.01)30, which comes to 1.273. The gross return has cancelled out entirely. It never appears in the ratio.
So the 0.20% investor ends with 27.3% more than the 1.00% investor whether markets returned 2% a year or 10%. Check it against the money: at 6.6% the pair is £640,667 and £503,236; at 4% it is £305,433 and £239,914; at 10% it is £1,643,224 and £1,290,733. Three very different outcomes, one identical ratio.
And because the ratio is a product of annual terms, it survives volatility too. If returns bounce around — up 20%, down 15%, up 4% — the (1 − fee) factors still multiply out cleanly and the ratio is unchanged. For a lump sum with no money going in or out, the fee gap is the one thing about a 30-year outcome that is genuinely deterministic. Everything else about it is a guess.
Getting the denominator right: 21.5% and 27.3% are not the same claim
Fee comparisons get mangled here more than anywhere else, so it pays to be exact about which balance the percentage refers to.
The £137,431 gap is 21.5% of £640,667, the lower-fee pot. It is 27.3% of £503,236, the higher-fee pot. Both are true and they describe the same money. The sentence "paying 1% instead of 0.2% costs you 21.5% of your wealth" is only correct if "your wealth" means the balance you would have had at the lower fee. Said the other way round: the cheaper investor ends 27.3% ahead of the dearer one.
There is a third denominator, and it is the one that shows what compounding costs actually do. Against a hypothetical zero-fee portfolio — £680,325 after 30 years — the 1% investor is short by £177,088, or 26.0%. But the fees deducted from that account over the three decades total only £77,674. The other £99,414 was never charged to anyone. It is the return the deducted money would have earned had it stayed invested. That is the whole mechanism: a fee is not a bill, it is a permanent reduction in the capital base doing the compounding.
At 0.20%, the same account pays £18,048 in fees and finishes £39,658 short of the zero-fee balance — 5.8% of it.
Put the two fee bills side by side and the headline number falls out. The dearer account was charged £77,674 and the cheaper one £18,048, a difference of £59,626. The gap in terminal wealth is £137,431. So under half the damage was ever charged to anybody; the rest is the return that money would have earned had it stayed invested.
Terminal wealth at five fee levels people actually pay
Rather than invent fee levels, the table below uses five that the Investment Company Institute measured across the US fund market in 2025 and published in March 2026: the asset-weighted average for index equity mutual funds (0.05%), for all equity mutual funds (0.40%) and for actively managed equity mutual funds (0.64%), plus the median equity mutual fund share class (0.99%) and the 90th percentile (1.84%). All start from £100,000 growing at 6.6% a year, with the fee deducted annually.
| Annual fee | What it represents (ICI, 2025) | After 10 years | After 20 years | After 30 years |
|---|---|---|---|---|
| 0.05% | Index equity mutual funds, asset-weighted | £188,538 | £355,468 | £670,193 |
| 0.40% | All equity mutual funds, asset-weighted | £182,039 | £331,383 | £603,249 |
| 0.64% | Actively managed equity mutual funds | £177,700 | £315,774 | £561,130 |
| 0.99% | Median equity mutual fund share class | £171,539 | £294,256 | £504,764 |
| 1.84% | 90th-percentile equity share class | £157,368 | £247,648 | £389,720 |
Two things stand out. Over 10 years the whole ladder spans £31,170 — real money, but not life-changing. Over 30 years it spans £280,473. Time is doing more work than the fee difference is; the fee just gets to compound alongside it.
The second is that the interesting comparison has moved. The classic argument pits 1% against 0.2%. The live one, for most people buying broad funds today, is 0.40% against 0.05% — and that gap is 10.0% of the lower-fee pot over 30 years, not 21.5%. The stylised numbers have been overtaken by the market. The ICI's series shows the asset-weighted average equity mutual fund expense ratio falling from 1.04% in 1996 to 0.40% in 2025 - a fall of 61.5% on those two figures, though the ICI itself quotes no percentage - with index mutual funds and ETFs reaching 52% of long-term fund assets by the end of 2025, up from 19% in 2010.
Regular savers lose a smaller share than the scary version implies
Almost every published fee illustration, including the SEC's, uses a lump sum. Most people do not invest that way.
Run £6,000 a year into the same 6.6% market for 30 years — £180,000 contributed in total — and 0.20% ends at £540,363 against £461,386 at 1.00%. The gap is £78,977, or 14.6% of the lower-fee pot, not 21.5%.
The reason is that the average pound has not been invested for 30 years. Money paid in during year 25 is exposed to the fee for five years, not thirty. Fee drag scales with time in the market, so a saver building a pot gradually carries less of it than the lump-sum illustration suggests — and a retiree drawing down carries less again. Unlike the lump-sum case, this figure does move with the return assumption: at 4% the gap is 13.5% of the pot and at 9% it is 15.6%. The clean cancellation only holds when nothing goes in or out.
None of which makes the drag small. £78,977 is 44% of everything that saver contributed.
The stated fee is not the whole cost, and the gap has been widening
Every number above uses a single all-in annual charge. Real portfolios rarely have one.
John Bogle made the point to the US House Financial Services Committee in March 2003, and it is the clearest statement of it anyone has written: "Mutual fund costs include not only expense ratios, but sales charges, portfolio transaction costs and other expenses. In fact, expense ratios represent less than one-half of the all-in costs incurred by fund investors." His own 30-year illustration sits beside the table above: $1 compounded at 10% grows to $17.50 over 30 years, and at 7.5% — the net return after the 2.5% to 3% he estimated all-in costs then ran to — it grows to $8.75, almost exactly half. "Costs matter!"
Two things have changed since. Costs are lower, though less dramatically than the usual comparison implies. Bogle's 1.36% was Lipper's 2002 simple average across stock, bond and money market funds; the ICI's 0.40% is asset-weighted and equity-only, so the two don't compare. The like-for-like figure is the ICI's 2025 simple average for equity mutual funds, 1.08% - a real decline, and a far smaller one than 1.36% against 0.40% implies. But the layering he described has, if anything, got worse, because much of the decline in stated expense ratios reflects costs moving rather than disappearing. In 2000, 46% of gross sales of US long-term mutual funds went to no-load share classes without 12b-1 fees. By 2025 it was 92%. The adviser's cut largely left the expense ratio and became a separate asset-based bill — and 68% of US households owning funds outside an employer plan hold them through an investment professional. A lower expense ratio and a higher total cost are entirely compatible.
The SEC's fee bulletin is explicit that the prospectus fee table "does not show other fees you may pay, such as brokerage commissions and other fees to financial intermediaries", nor the transaction costs a fund pays trading its own holdings. The ICI's European work puts a number on the latter: a median transaction cost of 0.10% of a fund's net assets. And the wrapper adds its own layer. The Financial Conduct Authority's investment platforms market study, published in March 2019, found charges on £5,000 held in a stocks and shares ISA ranging from 20 to 240 basis points, at least 11 different terms in use to describe a platform fee, and that most consumers who tried to estimate what they paid "made significant errors". Adding the layers together is the only version of the number the arithmetic above actually needs, and it's rarely printed in one place; LedgerTouch totals annual costs across accounts for that reason.
The strongest objection: a fee that buys behaviour could pay for itself
Everything so far assumes the fee buys nothing. That assumption is doing enormous work, and the best-argued case against it is Vanguard's.
Its 2022 paper Putting a Value on Your Value: Quantifying Vanguard Advisor's Alpha, by Francis Kinniry and colleagues, argues that following the framework "can add up to, or even exceed, 3% in net returns". The largest single component is behavioural coaching, which the paper says "may add 100 to 200 bps in net return", on the reasoning that an adviser who stops a client selling in a crash "may prevent significant wealth destruction and add percentage points — rather than basis points — of value". Take that seriously for a moment. The compounding argument runs both ways: if a 0.8-percentage-point fee difference costs 21.5% of a 30-year pot, then 3 points of added return would dwarf it, and a 1% fee that reliably buys it's the cheapest thing in this article.
Four things temper it, and the first two come from Vanguard.
The paper says plainly that the 3% "should not be viewed as an annual value-add but is likely to be intermittent", concentrated in periods of market duress. Its own summary table is more revealing still. Of the seven modules, only two carry point estimates: cost-effective implementation at 30 basis points and rebalancing at 14. Behavioural coaching is listed as "0 to > 200", asset location as "0 to 60", spending strategy as "0 to 120". The headline 3% is a stack of maxima, and it isn't a measurement — the paper describes it as a comparison of "projected results".
Third, the largest firm estimate in the table is the fee argument, not a rebuttal of it. Vanguard's 30 basis points is the distance between the industry's asset-weighted expense ratio (0.34–0.38%, depending on the stock/bond mix) and what it calls the "lowest of the low" — the cheapest 7% of funds by count (0.07–0.09%). That is the same arithmetic as the table above, dressed as advice.
Fourth, the evidence offered for the coaching number measures the problem rather than any adviser's fix. It is the shortfall of investor returns against fund returns — 0.84% a year for US large-cap blend funds over the decade to December 2021 in Vanguard's figures. That gap is smaller and more contested than the folklore suggests, and showing that badly timed money underperforms isn't the same as showing that paying someone prevents it.
The document is also marked "For institutional and sophisticated investors only", which is a fair description of its audience: it's written for advisers, by a firm that sells funds to them. That doesn't make it wrong. It does mean the estimate is produced by a party with an interest in the answer, which is a reason to read the ranges rather than the headline.
Where that leaves the objection: the fee-drag arithmetic is exact and the advice value-add is a modelled potential. They aren't the same kind of claim, and a reader who concludes that a 1% fee is worth paying for their own circumstances isn't misreading the evidence. They're weighing a certain cost against an uncertain benefit — which is what the decision actually is.
What this arithmetic can't tell you
It's deterministic, and markets aren't. The ratio result holds under volatility for a lump sum, but the balances themselves are illustrations built on one return assumption. A 6.6% real return is what US equities delivered from 1900 to 2025; it isn't a forecast, and the same Yearbook shows developed markets returning 8.5% a year in nominal dollars against 6.9% for emerging markets over the same span, so the number you pick depends heavily on where you look — and on whether it's a real or a nominal figure.
Fee levels aren't global. The ICI's asset-weighted average for US equity mutual funds was 0.40% in 2025; its equivalent for European equity UCITS was 1.11% in 2024, down from 1.49% in 2013. Cross-border European equity funds averaged 1.30% against 1.21% for single-country ones. Comparing across borders also mixes up what is inside the charge: in unbundled European share classes, distribution costs sit outside the ongoing charge and come out of the investor's pocket separately, so a lower stated figure can mean a higher total.
Averages hide dispersion. The ICI's 0.40% is asset-weighted, which is the right way to describe what investors collectively pay, but the same table shows a 10th percentile of 0.51% and a 90th percentile of 1.84% for equity mutual fund share classes. The asset-weighted average is low because most of the money sits in a handful of very large, very cheap funds, not because most funds are cheap.
And tax is ignored entirely. Whether a fee is deducted inside a fund, inside a tax-sheltered wrapper, or from taxable cash changes its real cost, and the answer differs by country and account type. The figures above assume the fee simply leaves the account.
What would change the conclusion
Three things, and one of them is already happening.
Continued fee compression narrows the practical spread rather than the arithmetic. If the realistic choice for a broad equity holding is 0.05% versus 0.40% rather than 0.20% versus 1.00%, the 30-year gap falls from 21.5% of the pot to 10.0%. The mechanism is identical; the stakes are halved. The ICI's 2025 numbers say that shift is well advanced, with 78% of index equity fund assets already sitting in the cheapest quartile of expense ratios.
A measured advice premium would change the balance of the argument. Vanguard's 3% is a projection. A study showing a realised, net-of-fee return difference between advised and unadvised investors, tracked through a full cycle rather than modelled, would move this from an uncertain benefit to a quantified one. That work doesn't exist in a form that settles it.
Horizon changes the answer more than most people expect. The same 0.2% versus 1% comparison costs 7.7% of the pot over 10 years, 14.9% over 20, 21.5% over 30 and 27.5% over 40. Someone with a decade to go and someone with four isn't having the same conversation.
The number to carry away isn't a percentage at all. It's that over 30 years, the fees a 1% investor actually paid came to £77,674 while the wealth they gave up, against a zero-fee portfolio, came to £177,088 — and the difference between those two figures is the entire argument. Whether that is worth it depends on what the fee buys, which is a question about your own circumstances rather than about arithmetic. The arithmetic is settled; the odds of picking a fund that earns its fee back are a separate question with its own evidence.