How ETF creation and redemption actually works

The three actors

1. The fund (issuer). For Ireland-domiciled UCITS ETFs, this is typically a sub-fund of a UCITS umbrella ICAV. The fund holds the underlying basket of stocks/bonds/commodities.
2. Authorised Participants (APs). Large institutional firms — market makers, banks, prime brokers — that have signed a legal agreement with the fund. They can directly create new ETF shares (by delivering a basket of underlyings to the fund) or redeem ETF shares (by handing them back in exchange for the basket). For a typical UCITS ETF there might be 5–15 APs. Examples include Flow Traders, Optiver, Susquehanna, BNP Paribas, Goldman Sachs, JP Morgan.
3. The secondary market. Everyone else — retail investors, pension funds, financial advisers — buys and sells ETF shares on the LSE order book at market prices, not directly with the fund.

A creation walkthrough — what happens when an AP "creates"

Imagine VWRL has heavy inflows: lots of UK retail buyers, market makers running out of inventory. Step by step:

1. Market makers detect the inflow. Their inventory of VWRL is being depleted because they're selling to the market faster than they're buying. The bid-ask spread starts to widen and the market price drifts slightly above iNAV (premium of, say, 0.05%).
2. An AP sees the arbitrage opportunity. They can buy the underlying FTSE All-World basket on global markets for, say, £108.00 per "creation unit", then deliver that basket to the fund and receive 50,000 new VWRL shares (one creation unit = 50,000 shares for VWRL). The market price of VWRL is £108.05 per share, so the AP profits 5p per share on the creation = £2,500 per creation unit.
3. The AP places the order. They place buy orders for every constituent of the FTSE All-World index, in correct index weights, on every relevant exchange.
4. End-of-day settlement. The AP delivers the assembled basket to the fund (held by the fund's custodian). The fund mints 50,000 new VWRL shares and credits them to the AP's account.
5. The AP sells the new VWRL shares on the LSE. The new supply pushes the market price back toward NAV, eliminating the premium.
6. The fund's AUM has increased. The number of shares outstanding has gone up by 50,000. The underlying basket has grown. The economics for existing holders are unchanged — they own the same proportion of a now-larger fund.

A redemption walkthrough — the reverse mechanism

Now imagine heavy outflows: lots of selling. The market price drops below NAV (discount of, say, 0.05%). An AP sees the opposite arbitrage:

1. The AP buys VWRL on the LSE at the discounted market price, say £107.95 vs NAV £108.00.
2. The AP delivers 50,000 VWRL shares to the fund. The fund cancels those shares (destroys them) and hands the AP the underlying basket of FTSE All-World constituents worth £108.00 per share × 50,000 = £5.4m.
3. The AP sells the basket on global markets. They net £5.4m. Their cost was 50,000 × £107.95 = £5,397,500. Profit: £2,500 on the trade.
4. The selling pressure has been absorbed. The market price drifts back up toward NAV; the discount closes.

This continuous arbitrage activity is what keeps ETF market prices tied to NAV. As long as the underlying basket is liquid and APs are willing to engage, the gap between market price and NAV stays tiny.

Why "in-kind" matters — the tax efficiency benefit

In a traditional open-ended mutual fund (OEIC), if investors redeem in large numbers, the fund must sell underlying holdings to raise cash for the redemptions. Those sales realise capital gains, which the fund distributes to ALL remaining holders — including the ones who didn't redeem. Long-term mutual fund holders effectively subsidise the tax bill of redeemers.

ETFs avoid this with in-kind redemption. When an AP redeems, they take the underlying basket out of the fund. No sale takes place inside the fund. No capital gain is realised. Long-term holders see no taxable distribution from other people's redemptions.

This matters most in the US, where mutual funds distribute large annual capital gains and ETFs almost never do. For UK retail holding UCITS ETFs in an ISA or SIPP, the tax benefit is largely irrelevant (ISA/SIPP shelters everything). For GIA holders or in less efficient regimes, the in-kind benefit is meaningful.

Creation unit sizes — why APs deal wholesale, not retail

The minimum size for a creation or redemption is the "creation unit" — typically 50,000 or 100,000 ETF shares. At a market price of £100 per share, that's £5–10 million per creation. Far beyond any retail-sized trade.

This is why retail investors interact with the secondary market (LSE order book) rather than the fund directly. The retail trade flows are too small to be efficient at the creation unit level. APs aggregate retail order flow indirectly — they take the other side of retail trades, accumulate inventory, then create or redeem at the wholesale level when their position grows large enough.

What happens in stressed markets

The arbitrage that keeps market price tied to NAV only works if APs are willing to engage. In rare cases — March 2020 fixed-income markets, for example — underlying liquidity dried up, APs couldn't reliably price the basket they'd need to hedge, and many APs stepped back from active arbitrage. Several large bond ETFs (especially corporate bond and high-yield ETFs) traded at meaningful discounts to NAV for a few days.

The interesting thing about March 2020 is that the ETF wasn't broken — it was actually providing price discovery. The ETF market price was probably closer to the true value of the underlying bonds than the stale NAV (which used theoretical bid prices that no one was actually willing to trade at). Once the Fed stepped in to provide liquidity, the underlying bond market reopened, APs returned, and discounts closed.

Lessons for retail:

• ETF discounts in stress aren't usually a "fund failure"; they're the market telling you the underlying is hard to trade
• For long-term holders, riding through is the right move — the discount closes
• For traders, attempting to arbitrage these moments is dangerous; the underlying liquidity matters more than the ETF mechanics

The "ETF liquidity" misconception

A common retail concern: "What if I want to sell £500k of a small ETF and the order book is thin?" The fear is that you can't get out without crashing the market price.

This concern is mostly misplaced. An ETF's true liquidity is the liquidity of its underlying basket, not the visible volume on the LSE order book. If the basket is liquid (e.g. global equities), the AP can always create or redeem to absorb large orders, and market makers know this so they're willing to fill large trades at fair value even if the order book looks thin.

The empirical pattern: large institutional trades (£5m+) in UCITS ETFs almost always fill within 5–10bp of iNAV through "block trade" RFQ mechanisms or direct AP engagement, not by hitting the visible LSE order book.

For retail-sized trades (<£100k), the visible spread on the order book is the relevant cost, and underlying basket liquidity is rarely the binding constraint.

UCITS rules around creation/redemption

UCITS regulations (the EU framework that UK retail ETFs follow) impose specific rules:

• At least one AP must be active: the fund must have at least one AP committed to providing liquidity. In practice most have 5–15.
• In-kind preference: creation/redemption must be in-kind by default. Cash creations/redemptions are allowed but typically incur a fee or use the daily NAV strike rather than intraday pricing.
• Daily NAV publication: the fund must publish NAV daily, calculated using verifiable underlying prices
• Suspension rules: the fund can suspend creations/redemptions in extreme market conditions, but only under strict regulator-approved circumstances. Suspensions are very rare for liquid index-tracking ETFs.

Real-world creation/redemption volume

For a popular UCITS ETF like VWRL or CSPX, typical activity:

• Daily LSE secondary market volume: 1–5 million shares (the visible retail-plus-institutional trades)
• Daily creations/redemptions: 0.5–2 million shares (net flow that APs absorb)
• Ratio of secondary to primary: typically 3–5x. Most trading volume is on the secondary market; only the net imbalance triggers primary market activity.

In stressed periods (March 2020, late 2022 gilt crisis), creation/redemption activity can spike to dominate secondary volume as APs rebalance heavily. In quiet periods, secondary volume dominates and the fund's AUM is relatively stable.

Frequently asked questions

Related guides

• NAV vs market price + bid-ask spread mechanics
• Synthetic vs physical ETF replication
• ETF securities lending revenue
• Tracking difference vs tracking error
• ETF cost comparison calculator
• All UK Tax Drag ETF tools and guides