A gold ETF has one structural feature that almost nobody prices properly: it charges you every year, forever, to hold an asset that generates no income to pay the charge from.
A gold ETF has one structural feature that almost nobody prices properly: it charges you every year, forever, to hold an asset that generates no income to pay the charge from.
That is not a criticism of ETFs. They are excellent at what they do, which is give you clean, liquid, regulated exposure to the gold price in a brokerage account. But the fee does not come out of a coupon or a dividend, because gold has neither. It comes out of your metal. Every year, a small slice of your holding is sold to pay for the privilege of holding the rest.
Over one year that is negligible. Over ten, it is visible in the price history, and we can show you exactly where.
Every ETF charges an expense ratio, deducted continuously from fund assets. For an equity fund, that fee is offset by dividends flowing in. For a bond fund, by coupons. Gold pays neither.
Economists call this a negative cost of carry: gold costs money to hold and produces nothing to cover the cost. Vaulting, insurance and the manager's fee all have to come from somewhere, and in a gold ETF the only thing available is the gold itself.
The mechanical consequence is that the number of ounces backing each share slowly declines. Your share count stays the same. What sits behind each share gets a little thinner every year.
For reference, the largest fund in the category, SPDR Gold Shares, publishes a gross expense ratio of 0.40% on State Street's own fund page. Cheaper alternatives exist, which is precisely what makes the next section possible.
Here is a test that does not require trusting anybody's fee table.
Several ETFs track the same underlying asset: the gold price. If they hold the same metal and differ mainly in what they charge, then the difference in their long-run total returns is very close to the difference in their fees. You can measure it.
Ten years of monthly closes to 21 August 2026:
Same metal. Same decade. Five to six percentage points of separation, and no investment skill involved anywhere. The gap is the fee, compounding.
The arithmetic checks out. A difference of roughly 23 basis points a year, compounded over ten years, is about 2.3% of principal, which on a position that has more than tripled works out to roughly five or six percentage points of cumulative return. Which is what the table shows.
The twenty-year version is starker, because compounding gets more brutal the longer you leave it. Over the twenty years to August 2026, spot gold compounded at about 9.85% a year while the largest ETF compounded at about 9.43%. That gap of roughly 42 basis points is, almost to the decimal, its expense ratio, extracted year after year out of metal because there was no coupon to pay it from.
What that means in practice: nothing dramatic in any single year, and a meaningful share of your holding over an investing lifetime. The fee is not the risk in a gold ETF. It is the certainty.
Strip away the technology talk and there is exactly one difference that matters over a decade.
In a gold ETF, the annual line item is negative. A fee is deducted. Your ounces per share decline.
In a yield-bearing tokenized gold position, the annual line item is positive. Yield accrues. Your claim on ounces grows.
That is a sign flip on the compounding term, and over long horizons a sign flip dominates everything else in the comparison. Even a modest positive rate, compounded, runs in the opposite direction to a modest fee, compounded.
Two things must be said immediately, because the flip is not free.
First, the yield is denominated in gold, not dollars. What accrues is a claim on more metal. If the gold price falls, the extra gold cushions the move and does not cancel it. Your position is still a gold position with all of gold's price behaviour. Nothing in this article changes that, and any comparison that treats a gold yield as a dollar return is comparing the wrong things.
Second, a positive line item requires a payer, and a payer is a credit risk. This is the honest cost of the flip and it deserves to be stated as plainly as the fee. Gold produces no income, so a yield on gold is generated by lending it, lending against it, or by an operator subsidising it. Each of those introduces a counterparty who has to perform.
An ETF has no counterparty paying you, because nothing is paying you. You are trading a certain small cost for an uncertain larger benefit plus a new risk. Whether that is a good trade depends entirely on the quality of the counterparty, which is why we wrote a whole article about ours: Is Gold Staking Safe? The Risks Most Platforms Leave Out.
We are not going to pretend this is one-sided.
If you want regulated, deeply liquid, brokerage-native gold exposure and you are untroubled by a small annual fee, an ETF is a genuinely good product and there is no shame in stopping there. Its cost is knowable in advance and it introduces almost no credit risk. That is worth something real.
If the fee drag over a long horizon bothers you, and you are willing to take a named credit risk to turn that line item positive, tokenized gold with a yield is the structure that does it. But price the credit risk properly rather than reading the yield as free money.
Two errors dominate discussion of this, and both are avoidable.
They are not comparable. A deposit account in most developed markets carries statutory deposit insurance and a prudentially regulated institution behind it. A gold yield has neither, and it is paid in a volatile commodity. The headline numbers can be similar while the products are not remotely alike.
It is not. A share in a fund holding allocated London Good Delivery bars, a token backed by vaulted metal with monthly attestations, and a token backed by reserves still in the ground are three different structures with three different risk profiles. They can all be described as "gold" and they should not be compared on fees alone. We put the major tokens side by side, including ours with its weaknesses stated, in PAXG vs XAUT vs Yield-Bearing Gold.
The fee. Every ETF publishes its expense ratio on the issuer's own fund page. Use that, not a third-party summary, because the aggregators are frequently stale.
The drag. Pull ten years of monthly closes for two funds tracking the same metal and compare cumulative returns. The gap is the fee. This takes about five minutes in any charting tool and it is the most convincing thing you will do all week.
The token. For a yield-bearing token, ask three questions:
Then ask when the last reserve attestation was and who performed it. Our longer eleven-point version is in Gold Staking Explained.
0.40% a year, per State Street's own fund page for SPDR Gold Shares. Because gold pays no income, that fee is met by drawing down metal rather than from a coupon.
IAU has historically carried a lower expense ratio, and the effect is visible in long-run returns. Over the ten years to August 2026, IAU returned 242.0% against GLD's 237.0% on the same underlying metal. Check both issuers' current fact sheets for today's figures.
No. They hold metal, which produces no income. There is nothing to distribute, which is also why the management fee has to come out of the holding itself.
For a fund charging 0.40%, roughly 4% of your holding in simple terms and slightly more once compounded. The measured difference between the largest fund and cheaper alternatives tracking the same metal was five to six percentage points of cumulative return over the decade to August 2026.
On the annual holding cost, often yes, and a yield-bearing token turns that cost into an accrual. But tokenized gold introduces counterparty and structural risks that an ETF does not have, and liquidity is usually far thinner. Cheaper is not the same as better.
Generally not, and that is a genuine advantage for ETFs. A gold ETF sits in a brokerage or retirement account without difficulty.
On the documentary evidence, a regulated fund holding allocated bars has the stronger structural position, and we would rather say that than pretend otherwise. Tokenized gold trades some of that for accessibility, continuous settlement and the possibility of a yield.
Gold ETFs, comfortably. They trade on major exchanges with tight spreads and enormous volume. Most gold tokens are far thinner, and some are effectively illiquid. Check actual market depth rather than market capitalisation before sizing any position.
It matters more, not less. A fee is a percentage of the whole position, so as the position grows the amount extracted grows with it. That is what compounding a negative number does.
A gold ETF is an honest product with a knowable cost, and that cost is a small annual subtraction from your metal, forever, because there is no income in gold to pay it from. Over ten years you can see it in the price history. Over twenty it is worth about as much as the expense ratio suggests, which is exactly what you would hope and also exactly the point.
Tokenized gold with a yield flips that line from negative to positive. That is the only structural difference worth arguing about, and it is a real one. It also introduces a payer, and a payer can fail.
The fee in an ETF is a certainty you can calculate. The yield in a tokenized product is a probability you have to underwrite. Pick the one whose failure mode you would rather own.
This article is for informational purposes only and is not financial advice.
| Fund | Cumulative 10-year return | Difference vs GLD |
|---|
| GLD (SPDR Gold Shares) | +237.0% | baseline |
| IAU (iShares Gold Trust) | +242.0% | +5.0 percentage points |
| SGOL (abrdn Physical Gold Shares) | +243.0% | +6.0 percentage points |
| Gold ETF (GLD, IAU, SGOL) | Tokenized gold with a yield |
|---|
| Annual cost or income: Negative. An expense ratio, deducted from metal | Annual cost or income: Positive. Yield accrues, subject to counterparty performance |
| Regulatory wrapper: Established, regulated fund structure | Regulatory wrapper: Varies by issuer, and rules are still developing |
| Counterparty risk: Minimal beyond the custodian | Counterparty risk: Real. The yield has a payer who must perform |
| Liquidity: Deep. Exchange-traded, tight spreads | Liquidity: Varies widely by token, often thin. Check before sizing |
| Trading hours: Market hours only | Trading hours: Continuous |
| Minimum position: One share, often several hundred dollars | Minimum position: Can be a few dollars, depending on denomination |
| Account needed: Brokerage account | Account needed: Self-custody wallet |
| What you legally hold: A fund share | What you legally hold: A token claim, structure varies by issuer |
| Held in a retirement account: Straightforward in most jurisdictions | Held in a retirement account: Generally not |
| Verification: Fund reporting and audited financials | Verification: On-chain supply, plus whatever the issuer attests |
