The part that isn’t a guess

A fund’s fee is not its cost

Most funds lend their shares out to other investors and keep part of the income. That income comes back to holders, so what a fund really costs you is its fee minus what it earns lending. Look at it that way and the ranking changes.

VTI charges 0.03% a year and its lending brought in 0.0184%, so it cost $1.16 a year per $10,000 held.

IEMG charges 0.09% and VWO charges 0.06%. Both hold emerging market shares. The dearer fund is the cheaper one to own: IEMG’s lending income more than covered its fee, leaving holders 87 cents per $10,000 ahead of it, while VWO cost $1.67. Nobody comparing the two published fees would guess that. Lending income is a measurement rather than a promise, so treat its size as an estimate even though its direction is certain.

Then there is SPY. It costs $9.45 per $10,000 a year, against $1.82 to $2.94 for every other fund tracking the same index. Five times dearer for identical holdings. The reason is structural. SPY is a unit investment trust, an old legal form with no manager and no discretion, so by its own prospectus it cannot lend its shares and provides no dividend reinvestment; the dividends it collects sit in an account paying no interest, and the trustee rather than you gets the benefit.

BND is the one bond fund we audited that lends nothing at all, so its cost is exactly its fee, 0.03%.

Tax is the same size as the fee, and far less predictable

A 2020 paper in the Journal of Finance measured what US equity funds cost their taxable holders in tax: 1.12% a year for the average fund, roughly the same as its fee. The difference is spread. Tax varies about three times as much between funds as fees do. The worst quarter of funds cost over 3.90% a year in tax. The gap between the most and least tax-efficient fifth was 2.10 points a year, and the tax-efficient funds gave up nothing before tax to get there.

Structure explains most of it. When money leaves an ETF the fund can hand over shares instead of selling them, which triggers no taxable gain. A mutual fund usually has to sell, and the bill lands on everyone who stayed. In 2025, 7% of ETFs paid out a capital gain against 52% of mutual funds. Four broad index funds we checked distributed nothing at all, in every year, across a decade. Two large active funds distributed 6.62% and 7.01% of assets in a single year, and one of those years was a year the fund lost its holders nearly a quarter of their money.

Be precise about who that beats. The advantage is over active funds and over expensive index mutual funds. Against a genuinely low-turnover index mutual fund from a cheap sponsor, the difference is close to nothing, and choosing between two cheap ETFs buys you none of it.

So the account matters as much as the fund. Bonds and anything paying interest belong in a retirement account if you have one, by a factor of about four over everything else. The broad, cheap, low-turnover share fund is the one that survives a taxable account intact, which means it is the one to leave there.

The $100 you did not agree to

From 1 June 2026 Fidelity charges $100 for each purchase of a fund whose issuer refused to pay its platform fee. The published rule is 5% of the trade value capped at $100, so a $10,000 buy costs about 1% and anything below roughly $2,000 costs the full 5%. Charges apply to buying. Schwab has said it will do something comparable by the end of 2026. Check the schedule for the specific funds you want before you move an account.

Why this is the page that matters

S&P’s scorecard of active US funds is the baseline: 90.3% lost to their own benchmark over ten years, and 93.8% over twenty. Of the US equity funds alive in 2005, only 37.3% still existed in 2025. Against that, a few hundredths of a percent sounds small. It is worth more than everything else on this site, because it is certain.

What would change our mind

Annual reports showing a fund’s lending income has collapsed, or a change in how a fund is structured. SPY’s gap would close if its trust were reorganised. Mutual funds are now being allowed to add ETF share classes, which over time removes much of the distribution gap for the funds that do it.

Where these numbers come from

• Each fund’s annual census filing with the SEC, for fees and securities lending income, covering financial years ending 2018 to 2026 and read August 2026.
• Audited annual reports filed with the SEC, for capital gains distributions.
• Morningstar’s 2026 US fund fee study, and its annual survey of funds paying a capital gain.
• Sialm and Zhang, "Tax-Efficient Asset Management: Evidence from Equity Mutual Funds", Journal of Finance, 2020.
• S&P Dow Jones Indices, SPIVA US Scorecard, mid-2025.
• Fidelity’s published fee schedule effective 1 June 2026, and press reporting of Schwab’s stated plans.