ETF Premium and Discount: Why the Price You Pay Differs From NAV

You bought 200 shares at 2:14 in the afternoon. The quote was $52.14 bid, $52.18 ask, and you took the ask, so the order cost $10,436. That night the fund published its net asset value: $52.11 a share, which values the same 200 shares at $10,422.

Fourteen dollars vanished between the click and the accounting. That difference is a premium — 0.13% of the trade — and on a broad US equity fund it would be gone by lunchtime tomorrow. On a fund holding Tokyo-listed stocks whose home market closed eleven hours before your fill, those same fourteen dollars may not be a cost at all.

Kill one idea before going further: NAV is not the correct price and market price is not the mistaken one. Both are estimates. NAV is a once-daily accounting number built from the last recorded prices of the holdings. Market price is a live number set by people willing to put money down right now. When the two disagree, the useful question is which one is stale.

The number, and what it costs in dollars

Premium or discount is the gap between market price and NAV, as a percentage:

(Market price − NAV) ÷ NAV × 100

Positive is a premium — you paid more than the holdings were worth. Negative is a discount. For your trade: ($52.18 − $52.11) ÷ $52.11 = 0.1343%, or $14.00 on $10,436.

Run the same percentage across order sizes and the arithmetic gets uncomfortable fast. On a $2,000 order that gap costs $2.69. On $50,000, $67. On $250,000, $336. Identical fund, identical click, identical percentage — the dollar consequence scales in a perfectly straight line, and the amount of attention most people pay does not scale with it at all.

There’s a second cost hiding inside the first one, and it isn’t premium. Your fill was at the ask, $52.18, while the midpoint of the quote sat at $52.16. That two-cent difference is half the bid-ask spread: 200 × $0.02 = $4.00, or 0.04% of the trade. The remaining $10.00 is the genuine gap between the market’s live opinion ($52.16) and last night’s marks ($52.11) — 0.10%.

So the headline “0.13% premium” is really two separate mechanisms sharing one ticket. Four dollars of spread, which you’d pay buying anything from anyone. Ten dollars of price-above-NAV, which is the market saying the underlying is worth more than the accounting says.

Now flip the tape forward. Suppose Tokyo opens the next morning up 0.2% and NAV catches up to $52.21. Your $10.00 premium was not a fee. It was correct information, arriving before the accounting could reflect it. You were early to a price move, not overcharged. This distinction gets missed constantly, and it is the whole reason premiums on international funds cannot be read the same way as premiums on domestic ones.

Why the gap exists at all

Three mechanisms produce it, and they behave nothing alike.

Measurement times differ. NAV uses the closing prices of the holdings. For a fund holding Japanese equities, those closes happened at 2 a.m. New York time. By the time you traded at 2:14 p.m., a full US session of news had landed that Tokyo has not reacted to yet. The ETF price reflects that news because buyers and sellers are pricing it in real time. NAV does not, because nothing has re-marked those shares. What looks like mispricing is the ETF market doing price discovery while the underlying market sleeps.

Marks go stale on thin holdings. Corporate bonds, munis, and bank loans don’t trade every minute. A bond that last printed at 10:30 a.m. gets carried at that price into the 4 p.m. NAV even if the world changed at 2 p.m. On a stressed afternoon the ETF trades at a discount to a NAV assembled from prices that no longer exist anywhere.

Arbitrage costs money. Authorised participants create and redeem ETF shares in large blocks, delivering or receiving the underlying securities. When the ETF trades above NAV, an AP buys the basket, creates shares, sells them into the market, and keeps the difference. That flow drags price back toward NAV. But the AP pays its own trading costs, borrow costs, and carries risk while the trade is open. If assembling the basket costs 0.20%, nobody is arbitraging a 0.10% premium. It simply sits there.

That third mechanism explains why funds holding illiquid things run persistently wider gaps than funds holding Apple and Microsoft. It isn’t sloppiness on anyone’s part. It’s the cost of the underlying market, passed through.

What structural width looks like by asset class

Two questions determine how wide a fund’s gaps run: how liquid the holdings are, and whether the holdings’ home market is even open while the ETF trades. Crossing those axes gives a rough map.

These bands are structural expectations derived from the logic above, not measured averages from any dataset. Treat them as what to expect, not what was recorded.

The last row deserves more attention than the other nine combined. When an issuer stops creating new shares — because of a regulatory limit, a position limit in the futures it holds, or plain capacity constraints — the arbitrage that anchors price to NAV switches off. The fund becomes a closed pool of shares priced purely by demand for those shares. Premiums in that condition have historically reached double digits, and buyers at those levels have taken heavy losses when creations restarted and the premium collapsed into nothing. This is the one condition where “the gap is small, ignore it” stops being true, and it is not hypothetical.

The comparison that reorders your priorities

Expense ratios are annual. Premiums are per trade. People argue endlessly about the first and click straight past the second.

Take a fund charging three basis points. On a $10,000 position that’s $3.00 a year — genuinely cheap, and the number everyone quotes. Now pay a 0.05% premium entering and sell into a 0.05% discount exiting. That round trip cost 0.10%, or $10.00. One clumsy entry and exit cost more than three years of the fee you chose the fund for. Widen it to 0.20% round trip and you’ve spent nearly seven years of management fees in two clicks.

Scale it up and the gap gets louder. A 0.30% round trip on $50,000 is $150. On a fund charging 0.07%, the annual fee on that position is $35, so the round trip cost more than four years of it.

None of this replaces the expense ratio — it stacks on top, and so does the spread. Three costs, three mechanisms, one ticket. The SEC’s plain-language page on mutual fund and ETF fees and expenses covers the ongoing-charge side, which is a separate drag entirely from the price-NAV gap.

If you want a ceiling on the damage, invert the arithmetic. To hold premium cost under $25, divide $25 by the order size: at $20,000 you can tolerate 0.125%; at $50,000, 0.05%; at $250,000, one basis point. Your 0.13% fill clears the bar at $20,000 and blows through it at $50,000.

Does settlement timing change any of this?

US securities moved to a T+1 standard settlement cycle, so trades settle one business day after the trade date rather than two. The SEC’s announcement of the rules shortening the settlement cycle covers the change and its stated aim of cutting credit and market risk in clearance.

Faster settlement does not narrow the gap. NAV is still struck once a day, and Tokyo still closes when Tokyo closes. What changes is how fast cash from a sale becomes usable and how long counterparty exposure stays open — which touches the arbitrage side of the trade more than your side of the screen. Plumbing, not pricing.

Two readings people get backwards

“A discount means the fund is on sale.” Sometimes. A bond ETF at a 1.5% discount on a chaotic Thursday can mean you’re buying $10,000 of bonds for $9,850. During stress it more often means NAV is stale: the bonds were last marked before the selloff, and the ETF price at $9,850 is the honest current estimate while the accounting is the fiction. Buy that discount and NAV may drop to meet you by the next close. The 1.5% bargain becomes 0.0%, and you paid a spread for it. There’s a clean test after the fact — if the discount closed because NAV fell rather than because price rose, it was information, not a sale.

“The gaps are too small to matter.” On a broad US equity fund traded mid-session, largely true; two basis points on $10,000 is $2.00. On anything in the lower half of that table, run the numbers before deciding. The phrase does a great deal of unexamined work, and the order size that makes it true on one fund makes it false on the next.

Where this arithmetic breaks

Everything above assumes the mechanism works and the inputs are clean. Several real conditions violate that.

Creations get suspended, and the anchor is cut. Gaps of 5%, 10%, 20% become possible and can persist for weeks. No table bounds this.

NAV itself is an estimate for funds holding securities with no recent trade. Those funds apply fair-value pricing methodologies, and two funds holding similar assets can strike different NAVs on the same day. Comparing your hard fill price against a soft accounting number produces a soft answer.

Trades near the open and the close distort the reading. In the first and last few minutes, spreads widen and quotes get unreliable. A premium measured against a 9:30:05 fill overstates the structural gap and understates how much of it was bad timing.

Ex-dividend dates manufacture phantom premiums. NAV drops by the distribution amount on that date; compare a pre-distribution price against a post-distribution NAV and you’ll invent a gap that never existed.

Currency moves matter on international funds. Part of your observed gap is the exchange rate moving after the foreign market closed — a real economic difference, not a pricing failure, and it can flip sign overnight.

Reported NAV can lag or be revised. A same-day comparison against a preliminary figure can be off by a basis point or two, which is fatal precision when the whole gap is three basis points.

What this does not tell you

It doesn’t tell you any specific fund’s actual premium and discount history. Issuers publish that themselves, usually as a count of trading days falling in each band, and it varies enormously by fund, by year, and by market condition. Nothing here substitutes for reading a particular fund’s own disclosure.

It assumes one trade at one price. Real fills at size arrive in pieces at different prices, so your effective premium is a weighted average you’d have to compute from your own execution report.

It says nothing about whether any fund or asset class suits you. Mechanics only. And it excludes commissions, account fees, taxes on realised gains, and the tax treatment of distributions — any of which can swamp a few basis points of premium.

The asset-class bands are structural expectations, not measurements. A fund can sit outside them for reasons specific to what it holds.

FAQ

Is a 0.10% ETF premium a lot?

Ten dollars on a $10,000 purchase, $100 on a $100,000 purchase. Whether that matters depends on order size and trading frequency. Buy once at $10,000 and hold for decades and it disappears into the compounding. Trade $100,000 in and out monthly, paying 0.10% each direction, and that’s roughly $2,400 a year in gap cost alone — before spreads, before the expense ratio.

Why doesn’t my ETF price match NAV exactly?

NAV is computed once, after the close, from the last available prices of the holdings. Market price updates every time someone trades. They’re snapshots of different moments. A few hundredths of a percent of daylight on a liquid US equity fund is the normal state of affairs, not a malfunction.

Do premiums and discounts always close?

Where arbitrage works freely, they usually compress within hours or a day or two, as authorised participants create or redeem shares to capture the gap. Where the underlying is expensive or impossible to trade, they run wider and stay wider. Where creations are suspended outright, they can persist indefinitely — that’s the case that has produced double-digit premiums.

Should I buy an ETF trading at a discount?

That’s a decision about your own circumstances, not something arithmetic answers. What the arithmetic does say: a discount only benefits you if NAV is accurate and price is wrong. On bond funds during stress it’s frequently the reverse. Check which side moved before treating a discount as an advantage.

How much does a 1% premium cost on a $25,000 purchase?

$250 — that’s $25,000 × 0.01. You also end up with fewer shares. At a $52.11 NAV you’d expect about 480 shares; at a 1% premium, about 475. To keep that trade’s premium cost under $25, you’d need the gap inside 0.10%.

Is the premium the same thing as the bid-ask spread?

No, and they stack. The spread is the distance between the best bid and offer right now, and you pay roughly half of it on entry. The premium is the distance between the midpoint and NAV. The trade at the top of this article showed a 0.13% total gap that split into 0.04% spread and 0.10% premium to mid — two costs, two causes, one ticket.

This article is general information, not financial advice. See our disclaimer.

Also worth reading: Ex-Dividend Date: Why the Stock Price Drops by (Roughly) the Dividend

Also worth reading: Dark Pools: Why Your 50,000-Share Order Hides

More in Market Mechanics · all topics · calculators · how this was checked