How ETFs Actually Work: Creation, Redemption, and Authorised Participants

Here is a puzzle. An ETF trades on an exchange all day at whatever price buyers and sellers agree, yet its price almost never strays far from the value of the assets it holds. A closed-end investment trust, which also trades on an exchange, routinely sits at a 10% or 15% discount to its assets. What keeps the ETF tethered when the investment trust drifts? The answer is a behind-the-scenes mechanism, creation and redemption, run by specialist firms called authorised participants. It is the most important thing to understand about ETFs, and the least understood.

The basics

Every ETF has two prices that should be close but are not identical:

• NAV (net asset value): the per-share value of everything the fund actually holds, calculated from the market prices of its underlying assets.
• Market price: what the ETF's shares are trading for on the exchange right now, set by supply and demand.

When the market price sits above NAV, the ETF trades at a premium; below NAV, at a discount. In a well-functioning ETF these gaps are tiny, often a few hundredths of a percent, because of a continuous arbitrage carried out by authorised participants (APs), large financial institutions with a contractual relationship with the fund.

The mechanism, in plain terms:

• If demand pushes the ETF's price above NAV (a premium), an AP can buy the underlying basket of assets cheaply, deliver it to the fund, and receive newly created ETF shares in return, which it sells into the hot market. The extra supply pushes the price back down toward NAV.
• If selling pushes the ETF's price below NAV (a discount), an AP buys the cheap ETF shares, hands them back to the fund, and redeems them for the underlying basket, which it sells. Removing shares from the market nudges the price back up toward NAV.

Because APs profit from closing these gaps, they do it constantly. That arbitrage is the invisible force keeping an ETF's price honest.

Going deeper

The elegant part is that creation and redemption usually happen in kind: the AP swaps a basket of securities for ETF shares, or vice versa, rather than using cash. This matters for three reasons.

First, it is why the supply of ETF shares is elastic. Unlike a closed-end fund with a fixed share count, an ETF can grow or shrink its shares outstanding daily to match demand. That elasticity is precisely why ETFs avoid the persistent discounts that plague investment trusts, and why a "small" ETF holding liquid assets is not necessarily hard to trade. Tradability follows the liquidity of the underlying basket, not the ETF's own volume.

Second, in-kind transfers underpin the US tax efficiency mentioned earlier. When the fund hands out appreciated securities in kind during a redemption, it can dispose of its lowest-cost-basis shares without realising a taxable capital gain inside the fund. This is the structural reason US ETFs distribute far fewer taxable gains than US mutual funds. (For UK, Irish, and Australian investors holding inside a shelter, this effect is largely moot, as covered in the previous article.)

Third, the basket is transparent. To let APs price the arbitrage, most ETFs publish their holdings daily, which is also why ETF investors generally know exactly what they own.

Two situations test the mechanism, and beginners should know them:

Stressed markets. When underlying assets become hard to value or trade, say, during a bond-market shock, the link between price and NAV can stretch. In March 2020, several bond ETFs traded at meaningful discounts to their stated NAV for a few days. The debate that followed concluded the ETF price was often the more accurate signal: the ETF was trading continuously while the underlying bonds were barely trading at all, so the "stale" NAV was the misleading number. The mechanism bent but did not break, and prices reconverged as conditions normalised.

Synthetic and hard-to-access markets. Where the underlying assets are difficult for an AP to buy, illiquid emerging-market bonds, or markets in a different time zone that are closed while the ETF trades, arbitrage is less precise and small premiums or discounts can persist intraday. This is also where some ETFs use synthetic replication (swaps) instead of holding the assets directly, a structure we examine later in the series.

A practical implication: the bid-ask spread you see on an ETF reflects, among other things, how cheaply APs can hedge and arbitrage it. Broad, liquid funds have spreads of a penny or two; niche funds have wider spreads because the arbitrage is costlier to run. The spread is, in effect, the price of the mechanism working for you.

Common mistakes

• Judging an ETF's liquidity by its trading volume. Low on-screen volume does not mean illiquid. If the underlying holdings are liquid, APs can create shares to fill a large order. The underlying basket is what counts.
• Panicking at a discount in a crisis. A short-lived discount during market stress often means the ETF price is fresher than a stale NAV, not that the fund is broken. Forced selling into that gap can lock in a poor price.
• Trading at the open or close. Spreads are typically widest in the first and last minutes of the session, before underlying prices have settled. Trading mid-session usually gets a tighter spread.
• Assuming all ETFs hold what they track. Synthetic ETFs use swaps rather than owning the assets. The creation/redemption logic still applies, but the counterparty arrangements differ, as we cover later.
• Ignoring premiums on single-country or commodity ETPs. When the underlying market is closed or capacity-constrained, persistent premiums can appear. Paying a large premium means overpaying for the assets.

FAQ

What is an authorised participant? A large institution, often a bank or market maker, contracted to create and redeem an ETF's shares with the fund. APs run the arbitrage that keeps the ETF's price close to its NAV.

Why do ETFs track their value better than investment trusts? Because ETF shares can be created and redeemed on demand, supply adjusts to demand. A closed-end investment trust has a fixed share count, so its price can drift to a persistent premium or discount.

Can an ETF trade away from its NAV? Briefly, yes, especially in stressed markets or when the underlying assets are hard to trade. In normal conditions for liquid funds, the gap is tiny and quickly arbitraged away.

Does creation and redemption cost me anything? Not directly. You experience it indirectly through the bid-ask spread, which is tighter for funds that are cheap and easy for APs to arbitrage.

The takeaway

Creation and redemption are the engine that makes an ETF an ETF. By letting authorised participants swap baskets of assets for shares and back again, the mechanism keeps the market price tethered to the value of the holdings, makes supply elastic, and (in the US) drives tax efficiency. It also explains why an ETF's tradability depends on its underlying assets rather than its own volume, and why discounts in a crisis are often a sign of a fresh price rather than a broken fund. With the mechanics in hand, the next article steps back to the strategy most ETFs follow: tracking an index, and how well they actually do it.

Educational not advice

This article is for general educational purposes only and is not financial or investment advice. It does not account for your circumstances. Investing carries risk, including the loss of capital. Market-stress examples are illustrative, not predictive. Consider professional advice before investing.

Sources

• BlackRock and Vanguard, primers on ETF creation and redemption
• Bank of England / FCA commentary on bond ETF behaviour in March 2020
• ETFGI, ETF market structure data (2026)
• Morningstar, research on ETF premiums, discounts, and liquidity