Skip to main content

Creation / Redemption Mechanism

Updated Jun 24, 2026 at 2:35pm

Research Draft Medium 1,246 words

The creation/redemption mechanism is the plumbing that lets an exchange-traded fund (ETF) trade like a stock all day while remaining a fund whose share count expands and contracts on demand. A select group of large broker-dealers — Authorized Participants (APs) — can exchange a defined basket of the fund's underlying securities (or cash) directly with the fund issuer for large blocks of newly minted ETF shares (a "creation unit"), or hand shares back for the basket ("redemption"). This primary-market access is what distinguishes an ETF from a closed-end fund: supply is elastic, so secondary-market price is tethered to the value of the underlying holdings (net asset value, or NAV) by arbitrage rather than by sentiment. The core tension is that this tether is conditional — it holds only as long as APs find it profitable and operationally feasible to act, which is precisely when markets are calm and weakest when they are not.

How it works

Authorized Participants. APs are institutional broker-dealers that sign an agreement with the ETF sponsor. They are the only entities that can transact directly with the fund, and per the Investment Company Institute (ICI) they "receive no compensation" from the fund for this — their incentive is arbitrage profit and the market-making spreads it supports. APs are distinct from the market makers and ordinary investors who trade ETF shares on exchange; an AP may be its own market maker or transact on behalf of one.

Creation units. Shares are created and redeemed only in large blocks. The ICI states creation units generally range from 25,000 to 200,000 shares; industry explainers cite a similar range (often "at least 25,000," sometimes up to 500,000). To create, an AP delivers the fund's published creation basket and receives a creation unit, which it can hold or break up and sell on exchange. To redeem, the AP returns a creation unit and receives the redemption basket. Baskets are published daily, giving the market transparency into what the fund holds.

In-kind vs. cash. The defining feature is that exchanges are typically in-kind — actual securities for shares, not cash. For broad equity ETFs the creation basket is essentially a pro-rata slice of the portfolio (near-100% alignment). For bond ETFs it is not: the BIS finds the largest bond ETF's baskets represent only about 3% of holdings on average, and corporate-bond ETF baskets cover roughly 20% (creation) to 35% (redemption) of holdings, drawn as a representative or custom sample because trading every illiquid CUSIP daily is impractical. Cash creation/redemption is permitted (and common for hard-to-source assets), but it forfeits part of the tax benefit below.

How it's used in practice

The mechanism's economic purpose is arbitrage that keeps price near NAV:

  • Premium (price > NAV): an AP buys the cheaper underlying securities, delivers them to create new shares, and sells the shares into the richer market — pocketing the spread. New supply pushes the price down toward NAV.
  • Discount (price < NAV): an AP buys the cheap ETF shares on exchange, redeems them for the more valuable basket, and sells the securities. Removing supply pushes the price up toward NAV.

Because creation is the channel through which all new ETF supply enters the market, an ETF's true liquidity is the liquidity of its underlying basket, not just its on-screen volume — a heavily traded basket can absorb large inflows even if the ETF itself shows thin secondary volume.

Tax efficiency is the other major consequence. In-kind redemptions let a fund hand low-cost-basis securities out the door without selling them, so no taxable capital gain is realized inside the fund. This is the structural reason US equity ETFs distribute far fewer capital gains than comparable mutual funds, which must sell holdings to meet cash redemptions.

Adoption, debate & evidence

The mechanism is universal — every US-listed ETF operates this way, and the SEC ETF Rule (6c-11), adopted September 2019, standardized the use of custom baskets across the industry. It is genuinely effective in normal conditions: APs step in whenever the premium/discount is wide enough to cover their transaction and balance-sheet costs (there is no fixed industry-wide trigger — the threshold varies by fund and basket liquidity), and the BIS reports bond-ETF tracking error historically averaged about 0.7 basis points across the sector.

The honest caveat is concentration and conditionality. There are usually only three to five active APs per ETF (BIS), and in fixed income the concentration is severe: per Bank of Canada / SEC data, three APs handled ~82% of all fixed-income ETF creations and redemptions in 2019, versus eight APs for ~80% of equity ETF activity. APs are not obligated to act; they will sit out when arbitrage is unprofitable or capital-constrained.

The clearest stress test was the March 2020 "dash for cash." Per Invesco, investment-grade corporate-bond ETFs traded at average discounts to NAV of about 3.4% in mid-March, with some products exceeding 8% (academic work puts the IG-ETF average for the week of 15 March at roughly 3.36%), and the BIS reports tracking error spiked above 200 basis points for some funds — gaps the in-kind mechanism did not immediately arbitrage away. Peer-reviewed work (Pan & Zeng; and the Journal of Banking & Finance dash-for-cash study) attributes this to APs' regulatory capital and balance-sheet constraints: the arbitrage relationship weakened most for ETFs holding illiquid bonds and those tied to capital-constrained APs. (Some argue the bond ETF's price was the better real-time signal and NAV was stale — an unresolved debate, but either way the tether visibly stretched.)

Strengths & limitations

Strengths: elastic supply keeps price near fair value, in-kind transfer delivers structural tax efficiency, and daily basket disclosure makes the vehicle transparent. Limitations: the tether depends on willing, well-capitalized APs; it is weakest for illiquid underlyings (high-yield bonds, EM, thinly traded small-caps) and during system-wide stress exactly when investors most need it. The #1 misconception is treating the ETF wrapper as more liquid than its holdings — in a crisis the ETF can only be as liquid as the basket an AP must transact, and persistent premiums/discounts are the warning sign that arbitrage has stalled.

Sources

Disputed: whether March-2020 bond-ETF discounts reflected ETF mispricing or stale NAVs is debated; both camps agree the in-kind arbitrage tether weakened materially.