ETF expense ratios are the most underestimated number in retail investing. They look small enough to ignore — 0.03%, 0.50%, 0.75% — and the difference between them feels like statistical noise. It isn’t. Stretch those numbers across a 30-year investing horizon and the gap between a “cheap” ETF and a “reasonable” one is the price of a house. The math is unforgiving in a way most fee disclaimers don’t bother to translate into dollars, so we’ll do that here, then walk through the four hidden costs the headline number quietly leaves out.

What an ETF expense ratio actually is

An expense ratio is the annual percentage of fund assets that the fund company charges you for running the fund. If a fund has a 0.50% expense ratio and you have $10,000 invested, the fund company collects $50 a year. That’s the simple version. The slightly trickier version is how it collects: the fee is extracted daily, in tiny slivers, directly from the fund’s net asset value. You never see a line item on your brokerage statement that says “fund fee.” Your balance just grows a little less than the index does, by an amount that approximates the expense ratio each year.

The fund company is required by the SEC to disclose the expense ratio in the prospectus and the summary document. Across the ETF universe, the numbers span a wide range. ICI’s 2025 Fact Book pegs the asset-weighted average for index equity ETFs at 0.14%. For actively managed ETFs across all asset classes, the average is around 0.49% per SEC data. Blend the two — heavily weighted toward index because index ETFs hold most of the money — and you get something close to 0.17% as the rough industry midpoint. So when an ETF charges you 0.50% or 0.75%, you’re not paying “a little bit more than average.” You’re paying roughly three to five times more than the average dollar in ETFs.

Why ETF expense ratios actually matter (the compounding math)

Here’s the part fee disclosures rarely show you. Imagine you invest $100,000 in an S&P 500 fund and the gross return averages 7% a year for 30 years — roughly the long-run real-plus-inflation number people use as a planning assumption. At a 0.03% expense ratio (what VOO, IVV, and SCHB all charge today), you end up with about $754,000. The fee took ~$7,000 out of what would have been a fee-free $761,000.

Now run the same scenario at a 0.50% expense ratio — the kind of number you’d see on a “smart beta” ETF or some sector funds. You end up with $661,000. The fee took $93,000 more out of the same starting balance, against the same gross return. At a 1.00% expense ratio — common on thematic ETFs and some active products — the final balance drops to $574,000. The fee gap from cheapest to priciest is $180,000. The two funds did identical work. You picked the wrong one, and you paid for a house.

This is the dynamic John Bogle described as the “tyranny of compounding costs” in The Little Book of Common Sense Investing: “The miracle of compounding returns is overwhelmed by the tyranny of compounding costs.” The math is symmetric. The same exponent that makes returns explode upward also makes fees explode upward. Most retail conversations about ETFs spend ten minutes debating which one tracked the index by 0.2% better last year and zero minutes asking which one charges 0.3% more every year forever. The second question is several orders of magnitude more important.

The hidden costs the expense ratio doesn’t show

Here’s where the simple “lower fee = better” rule gets more interesting. The expense ratio is the sticker price on the ETF. There are at least four other costs that don’t appear in the ratio but show up in your real-world return. Think of them as dealership fees.

1. Bid-ask spread

An ETF trades like a stock, which means there’s a bid (what buyers offer) and an ask (what sellers want). On a huge ETF like VOO or SPY, the spread is typically a single penny on a ~$500 share — basically nothing, around 0.002%. On a small or thinly-traded ETF (under $100M in assets, or a narrow international slice), the spread can be 0.10% to 0.30% — and you pay it on both entry and exit, so a round trip is double that. Across a buy-and-hold lifetime that’s not catastrophic, but it absolutely matters if you’re rebalancing or dollar-cost-averaging into a thin fund. The smaller the AUM, the wider the spread tends to be.

2. Tracking error

This is the gap between the fund’s actual return and the index’s return. A well-run index ETF tracks within a few basis points of its benchmark. A poorly-run one — usually because of internal trading costs, sampling shortcuts, or imperfect cash management — can lag by 0.10% to 0.30% on top of its stated expense ratio. The fund company doesn’t broadcast this. You have to read the annual report or pull the data yourself. A 0.05% fund with 0.20% of tracking error costs you the same as a 0.25% fund with perfect tracking.

3. Premium or discount to NAV

An ETF’s market price can drift slightly above or below its net asset value during the trading day. For mainstream funds, this gap is tiny — measured in fractions of a basis point. For esoteric ETFs that hold illiquid underlying assets (international small-cap, frontier markets, certain bond funds), the premium or discount can run ±0.50% intraday, occasionally more during stress. If you buy at a premium and sell at a discount, you’ve eaten the spread on top of everything else.

4. Tax-cost ratio

If you hold an ETF in a taxable account, the fund’s capital gains distributions are a real cost. Morningstar publishes a “tax-cost ratio” for every fund — the percentage of return lost annually to taxes on distributions, assuming the highest federal rate. For a typical broad-market index ETF, this is around 0.10% to 0.20%. For a high-turnover active ETF or one with frequent rebalancing, it can be 0.40% or more. The expense ratio is what the fund charges. The tax-cost ratio is what the fund’s behaviour costs you on top.

Add those four together and you get something closer to total cost of ownership: the headline expense ratio plus the spread you pay to trade plus the tracking error you eat versus the index plus the tax drag in a taxable account. A 0.05% ETF with tight spreads, near-zero tracking error, and a 0.10% tax-cost ratio costs you about 0.17% a year all-in. A 0.05% ETF with a 0.20% spread, 0.15% tracking error, and a 0.35% tax-cost ratio costs you 0.75% a year. Same expense ratio. Five times the actual drag.

When “needlessly cheap” is also a thing

Here’s the diminishing-returns piece that nobody tells you. Once you’re below about 0.05%, the marginal saving from going cheaper is essentially nothing. The difference between a 0.03% ETF and a 0.05% ETF over 30 years on $100,000 at 7% gross return is roughly $3,000 — meaningful, but not enough to switch funds and trigger taxable capital gains if you already own the slightly more expensive one. Past a certain point you’re optimising to the third decimal place while ignoring the things that actually matter: AUM, tracking record, your existing cost basis.

The real money is not in the 0.03% vs 0.05% argument. It’s in catching the 0.50% or 0.75% ETF that’s marketed as “low-cost” because it sits next to a 1.5% mutual fund in the comparison. Smart-beta funds, sector funds, thematic ETFs, and most actively-managed ETFs sit in this range. Some of them earn that fee through genuine alpha or differentiated exposure. Most don’t. That’s the question worth twenty minutes of your time before you buy: is this fund actually doing something the cheap index can’t, and do I believe it can keep doing it net of fees?

Vanguard’s most recent fee cut in February 2026 — 84 share classes across 53 funds, with the average ratio across their lineup dropping to 0.06% — pushed the floor lower yet again. For comparison, you could explore how VOO and VTI stack up, or look at the brokers that offer the cheapest ETF lineups in their core fund families. The headline takeaway is that “cheap enough” is now closer to 0.05% than 0.20% for core US equity exposure. If your fund is materially above that, the burden of proof is on the fund.

A four-step framework for sizing up an ETF’s true cost

Before you buy any ETF, run through these four checks. The whole exercise takes about ten minutes per fund and avoids almost every fee-related mistake retail investors make.

1. Pull the expense ratio from the fund issuer’s page (Vanguard, iShares, Schwab, SPDR) or Morningstar. Compare it to the cheapest credible benchmark in the same category — for US equities that’s VOO/VTI/SCHB at 0.03%, for total international it’s VXUS at 0.05%, for US bonds it’s BND at 0.03%.
2. Check the AUM. Anything above $1 billion usually has tight spreads. Anything below $100 million is a yellow flag — the spread you’ll pay to trade may quietly eat the fee saving.
3. Read the tax-cost ratio on Morningstar (it’s free). If you’re holding in a taxable account and the tax-cost ratio is above 0.30%, the fund’s turnover is costing you more than the expense ratio.
4. Find one independent reason the fund deserves to charge what it charges. “It tracks the same index as VOO but costs 0.40%” is not a reason. “It uses a quality screen that has historically reduced drawdowns by 15% in recessions” might be. If you can’t articulate the reason in one sentence, pay less.

This framework works because it forces the fee question to compete with everything else about the fund. Most investors anchor on whichever fund their brokerage’s research tab puts at the top and never compare it to the category benchmark. The benchmark comparison is the part that catches the “low-cost” thematic fund that’s actually charging 0.55% for the privilege of holding the same forty stocks you could get in a $1B index ETF for 0.07%.

Where ETF expense ratios are heading

The long-run trend is unambiguous. Index equity ETF expense ratios have dropped roughly 33% over the past nine years according to ICI data, landing at 0.14% as the asset-weighted average in 2025. For perspective, equity mutual funds — a longer-dated product where ETFs barely existed at scale before 2003 — fell from 0.99% in 2000 to 0.40% in 2025. The trajectory has been steady, and the floor keeps dropping as Vanguard, BlackRock and State Street keep locking each other in a price war on core index products.

The interesting wrinkle is what’s happening at the other end of the spectrum. Most new ETF launches in 2024 and 2025 have been thematic or actively managed — categories that charge 0.40% to 0.85%. So the industry-wide average is being held up by a growing satellite of premium-priced funds even as the core index ratios fall. Two ETF investors can be paying wildly different real costs depending on which side of that divide their portfolio sits on. The investor who owns the cheap index core has never had it cheaper. The investor chasing themes is paying a premium that, statistically, mostly underperforms the boring benchmark after fees.

The same fee math applies to superannuation and other long-horizon retirement vehicles — 0.5% sounds invisible until you compound it across a working lifetime, at which point it’s the difference between retiring at 60 and retiring at 67. Fees are the rare investing question where the right answer is mechanical, knowable, and almost free to act on. Use that.

The bottom line

For the core of a long-term portfolio — broad index equity, total international, broad bond — low expense ratios are the single highest-confidence “free lunch” in investing. The decision is mechanical: pick the cheapest credible fund (large AUM, tight spread, low tracking error), and stop optimising past the third decimal place. For the satellite of your portfolio — thematic ETFs, smart-beta, active strategies — the expense ratio becomes one input among several, but if you’re paying 0.75% you should be able to say in one sentence what the fund does that an equivalent 0.05% index can’t. If you can’t, you’ve just paid for the marketing. Look at the alt text on the chart in your fund’s annual report instead of the colour of the logo on its homepage. That’s where the cost actually lives.