One portfolio, two places to trade
An exchange-traded fund, or ETF, pools assets but gives its shareholders an exchange price that changes during the trading day. Net asset value, or NAV, is the portfolio’s assets minus liabilities, divided by shares outstanding. An investor who sells on an exchange receives the available market price, which can be above or below NAV. This article concerns U.S. ETFs registered under the Investment Company Act, rather than exchange-traded notes or commodity trusts with different structures. [1]
The important distinction is between trading an existing share and changing the fund’s size. A buyer and seller can exchange shares without moving a single bond or stock into or out of the portfolio. A burst of exchange turnover therefore does not by itself establish that the fund has received new money or sold underlying assets. The secondary market is a transfer between investors; the primary market changes the share count.
How the primary market links the prices
Authorized participants, or APs, are financial institutions with agreements permitting them to create or redeem large blocks of ETF shares. In a creation, the AP delivers the specified basket of securities, cash or a combination and receives shares. Redemption runs in reverse. Ordinary exchange investors generally do not have the right to hand one share to the fund and demand its slice of assets. [2]
The economic incentive is easiest to see when the same portfolio can be acquired more cheaply through one route. If the ETF is expensive relative to the deliverable basket, creation can increase the supply of shares. If it is cheap, purchasing shares and redeeming them can remove supply. But the comparison includes trading, financing, hedging and processing costs. A visible gap is not automatically a realizable profit. [2]
AP status and market-making are different functions, even when one firm performs both. The AP accesses creation and redemption; a market maker quotes prices to other traders. The mechanism depends on participants finding transactions worthwhile. It is an economic link between markets, not a promise that every displayed price will equal a calculated portfolio value.
Hypothetical creation: the gross gap is not the profit
Assume a creation unit contains 50,000 shares, the deliverable basket costs $5,000,000, and the AP can sell all the new shares at $100.10. Proceeds would be $5,005,000, giving a $5,000 gross difference. If basket execution costs $2,000, fund transaction charges $1,000 and financing and hedging cost $1,500, the remaining amount is $500. These are invented amounts, not a fund quotation.
If the executable share price falls to $100.08 before the sale is secured, proceeds decline by $1,000 and the same trade loses $500. A quoted ten-cent premium can therefore attract activity without being an enduring ten-cent profit. Speed, certainty of execution and the size of available quotes matter as much as the arithmetic at a single screen price.
The example also explains why a small premium may persist. No one needs to make a mistake for two prices to differ by less than the cost of connecting them. Conversely, a wider premium may encourage more creation until additional shares, changing basket prices or competition remove the opportunity. The outcome depends on executable prices at the relevant size.
Spread, premium and tracking difference answer different questions
The bid-ask spread is the difference between the price available to a seller and the price available to a buyer. A premium or discount compares the share price with NAV. Tracking difference compares the fund’s return with its benchmark over a period. They are not interchangeable measures of quality. Portfolio expenses reduce assets; a shareholder can also incur a trading cost outside the fund. [1]
For a hypothetical ETF quoted at $49.95 bid and $50.05 ask, the spread is ten cents, or 0.20% of the $50 midpoint. With a $50 NAV, the midpoint shows no premium, yet someone buying at the ask pays 0.10% above NAV. A purchase followed immediately by a sale at unchanged quotes loses $0.10 per share before other charges. This is a trading-cost example, not a forecast of investment performance.
Now suppose a benchmark earns 8.00% over a year and a fund earns 7.80% on a comparable total-return basis. The tracking difference is minus 0.20 percentage points. That annual result says nothing definite about the spread on a particular afternoon. Mixing price returns with total returns, or different valuation times, can manufacture an apparent discrepancy.
Bond baskets make the connection less mechanical
BIS research published in March 2021 explains why bond ETF creation and redemption baskets can contain only a subset of portfolio holdings. Individual bonds differ in availability, trading size and remaining maturity. Basket flexibility helps a manager handle these constraints, but it also means an AP does not always exchange a miniature copy of the full fund. Basket uncertainty and dealers’ other business can weaken simple price-convergence incentives. This is research interpretation, not a rule applicable identically to every ETF. [3]
There are competing implications. An in-kind redemption can transfer securities rather than require the fund to sell them immediately. At the same time, an AP receiving difficult-to-trade bonds can demand compensation through the price it is willing to pay for ETF shares. Flexibility can help the portfolio while leaving the exiting shareholder with a meaningful market discount. Both effects can occur in the same transaction. [3]
A discount can reveal information and still be costly
A BIS study dated April 14, 2020 found that corporate-bond ETF discounts during the March 2020 turmoil reflected several forces: bond valuations incorporated news more slowly, dealer risk-taking diminished, and portfolio reallocations followed policy announcements. Its interpretation cautions against treating NAV as an immediately executable exit price for every bond. These are findings about that historical episode, not evidence that a particular ETF is mispriced today. [4]
Consider an invented bond fund with a published NAV of $100, current estimated realizable basket proceeds of $98.80 and shares trading at $98.90. The apparent discount to published NAV is 1.10%, yet redeeming a share into that basket would lose ten cents before fees. Alternatively, if reliable executable basket proceeds really were $100, the same share price would present a very different opportunity. The two cases look identical in a headline price-to-NAV statistic.
The decisive evidence includes valuation timestamps, actual basket terms, executable underlying prices, spreads at meaningful order sizes and completed creation/redemption activity. A persistent gap can indicate valuation lag, expensive intermediation or disruption. Distinguishing those explanations is more informative than declaring that exchange trading either guarantees or has failed simply because price and NAV diverge.
Sources
- SEC Investor.gov, Exchange-Traded Funds; current explanatory page checked October 4, 2026Official sourceBack to text: ↑1↑2
- SEC, Updated Investor Bulletin: Exchange-Traded Funds, February 23, 2023Official sourceBack to text: ↑1↑2
- BIS, The anatomy of bond ETF arbitrage, March 2021; authors’ researchSourceBack to text: ↑1↑2
- BIS Bulletin 6, The recent distress in corporate bond markets: cues from ETFs, April 14, 2020Source · PDFBack to text: ↑