What is ETF Creation and Redemption?

Introduction

ETF creation and redemption is the process by which new ETF shares come into existence or disappear. It is the mechanism that connects the ETF shares investors trade on an exchange to the underlying portfolio of stocks, bonds, or other assets held by the fund.

That connection matters because an ETF trades in two places at once. Retail and institutional investors buy and sell shares in the secondary market on an exchange, just as they would trade a stock. But the fund itself interacts directly only with a small set of large financial institutions called Authorized Participants, or APs, in the primary market. If you understand why this split exists, the rest of the ETF structure starts to make sense: the ETF can trade all day like a stock, yet still remain tied to the value of the assets inside it.

The central puzzle is simple. A normal stock has a fixed supply unless the company issues more shares or buys them back through a separate corporate action. An ETF is different. Its share count can change day by day as demand changes. That flexibility is not a side feature. It is the design that lets ETFs absorb flows, support arbitrage, and often trade close to their net asset value, or NAV.

Why do ETFs use creation and redemption instead of a fixed share count?

A mutual fund and an ETF can both hold diversified portfolios, but they solve investor access in different ways. A mutual fund transacts directly with investors at end-of-day NAV. An ETF, by contrast, lets investors trade continuously on an exchange at market prices. Continuous trading is convenient, but it creates a problem: once shares trade freely between investors, why should the ETF’s market price stay close to the value of the securities in the portfolio?

The answer is that the ETF is not a closed pool of shares. If the ETF price drifts above the value of its holdings, APs can typically create new shares. If the ETF price drifts below that value, APs can typically redeem shares. Those creation and redemption flows change the supply of ETF shares in a way that makes the mispricing costly to ignore. The structure does not guarantee perfect alignment at every moment, but it creates a mechanism that usually pulls price back toward portfolio value.

This is the main idea worth remembering: an ETF is open-ended at the institutional level and exchange-traded at the investor level. Creation and redemption is the bridge between those two facts.

Who are Authorized Participants (APs) and why only they transact with ETFs?

ETF shares are not usually created because a retail investor clicks “buy.” When you buy 100 shares of an ETF through a brokerage account, you are almost always buying from another market participant in the secondary market. The fund does not issue those shares directly to you.

Direct dealings with the fund are generally limited to Authorized Participants. These are financial institutions, typically large broker-dealers, that have contractual arrangements with the ETF sponsor. They can transact with the ETF in large blocks called creation units. The SEC’s investor guidance describes creation units as large aggregations of ETF shares, often on the order of tens of thousands of shares. Under Rule 6c-11, an ETF is defined in part by issuing and redeeming shares in these creation units through APs while its individual shares trade on an exchange at market-determined prices.

Why restrict this process to APs rather than letting anyone create or redeem a few shares? Because the mechanism is meant to be operationally efficient and scalable. The fund needs a controlled way to accept baskets of assets, deliver baskets back out, handle settlement, and manage transfer agency and custody workflows. Large, specialized firms are better positioned to do that than ordinary investors.

There is also an economic reason. APs are not merely administrative agents. They are the institutions most able to compare the ETF price with the value of the underlying basket, hedge the difference, and act when there is an arbitrage opportunity. In other words, the ETF depends on a class of firms with both the legal right and the market infrastructure to move between the fund and the exchange.

How do Authorized Participants create new ETF shares?

Creation increases the number of ETF shares outstanding. Mechanically, an AP assembles the required basket of securities and any cash balancing amount specified by the ETF, delivers that basket to the fund, and receives a creation unit of ETF shares in return.

The intuition is easier if we make it concrete. Imagine an equity ETF that tracks a broad stock index. Before the trading day, the fund or its agent communicates the basket that corresponds to one creation unit. The AP can then buy those stocks in the market, or source them from inventory, and deliver them to the fund. In exchange, the fund issues a large block of new ETF shares to the AP. The AP can then sell those ETF shares on the exchange to investors.

Why would the AP do this? Usually because the ETF is trading at a premium to the value of the basket. If the AP can acquire the underlying securities for, say, slightly less than the market value implied by the ETF shares it will receive, the AP can create shares, sell them, and capture the spread. As APs do this, the supply of ETF shares rises, which tends to push the ETF price down toward NAV. At the same time, demand for the underlying basket rises because APs are buying those securities to complete the creation. That also tends to narrow the gap.

This is the mechanism, stripped to essentials: when ETF shares are too expensive relative to the portfolio, APs can manufacture new shares by delivering the portfolio.

How do Authorized Participants redeem ETF shares?

Redemption is the mirror image. An AP acquires ETF shares, typically in the required creation-unit size, delivers those shares to the fund, and receives a basket of securities or cash in return. The ETF shares that are returned to the fund are retired, reducing the total number of shares outstanding.

Again, the economic logic matters more than the paperwork. Suppose the ETF is trading at a discount to the value of its holdings. An AP can buy ETF shares cheaply in the market, redeem them with the fund, receive the underlying basket, and then sell or finance those assets separately. If the basket is worth more than the ETF shares the AP bought, that difference is the arbitrage profit, net of transaction and financing costs.

As redemptions happen, the supply of ETF shares in the market falls, which tends to lift the ETF price toward NAV. Meanwhile, the AP is receiving the underlying holdings out of the fund, which is the opposite flow from creation. The ETF contracts because investors as a group want out, but the contraction happens through institutional redemption rather than the fund having to redeem one small shareholder at a time.

This is why ETF liquidity can be larger than the trading volume visible on screen. The exchange is only one layer. The other layer is the ability of APs to expand or shrink the fund itself.

Why do many ETFs use in‑kind baskets rather than cash for creations and redemptions?

At first glance, it might seem easier for an ETF simply to take cash when shares are created and hand out cash when shares are redeemed. Some ETFs do use cash for some or all creations and redemptions. But many traditional ETFs rely heavily on in-kind transfers, meaning securities go in and securities come out.

The reason is that in-kind exchange pushes much of the trading activity to the AP rather than the fund. If an equity ETF receives the actual stocks it wants to hold, the fund does not need to go into the market and buy them with incoming cash. If it redeems by handing out securities instead of selling them for cash, it does not need to liquidate part of the portfolio to meet outflows. That can reduce transaction costs borne by the fund and, in many cases, improve tax efficiency.

The SEC’s investor bulletin notes that ETFs can be tax-efficient partly because many use in-kind exchanges rather than cash transactions. The basic mechanism is straightforward: when appreciated securities leave the fund as part of an in-kind redemption, the fund often avoids the kind of fund-level realization pressure that would otherwise feed capital gain distributions. The exact tax outcome depends on legal and tax rules, so this is not a universal law for every vehicle and every structure. But the broad intuition is sound: in-kind redemptions let the fund transfer assets rather than sell them.

That said, cash matters more than many simplified explanations admit. Some ETFs use cash creations or redemptions regularly. Spot bitcoin ETF disclosures, for example, have described cash creation and redemption models in which APs deliver cash and the trust or its agents handle the underlying bitcoin transactions, at least unless further approval permits in-kind treatment. Cash can be operationally necessary when the underlying asset is hard to transfer directly, hard to custody in standard fund plumbing, or subject to market-structure constraints.

How does creation/redemption arbitrage keep an ETF’s market price near NAV?

The phrase arbitrage mechanism is often used loosely, but the mechanism is specific. It is not merely that “traders keep ETF prices honest.” It is that certain traders can swap between ETF shares and the underlying basket in large size, and that convertibility creates a boundary on mispricing.

Let NAV mean the per-share value of the ETF’s assets minus liabilities. Let P mean the ETF’s market price. If P > NAV by enough to cover trading, financing, hedging, and operational costs, APs have an incentive to create shares. If P < NAV by enough to cover those same costs, APs have an incentive to redeem shares.

The important phrase there is by enough. Creation and redemption is not a magic switch. Arbitrage happens only when the gap exceeds the real-world costs of doing the trade. Those costs include bid-ask spreads in the underlying assets, balance-sheet usage, settlement friction, market impact, financing rates, and the risk that prices move before the hedge is completed.

That is why even very efficient ETFs can trade modestly above or below NAV during the day. It is also why some ETFs are tighter than others. A large U.S. equity ETF, whose holdings are liquid and easy to hedge, usually supports faster and cheaper arbitrage than a corporate bond ETF whose underlying bonds trade less frequently. The creation/redemption mechanism is the same in both cases, but the cost of using it is not.

How does daily holdings transparency support ETF arbitrage and pricing?

For APs and market makers to judge whether a creation or redemption trade is worth doing, they need to know what the ETF actually holds. This is one reason modern ETF regulation emphasizes daily holdings disclosure.

Rule 6c-11 requires covered open-end ETFs to disclose portfolio holdings each business day on their websites, before trading opens on the primary listing exchange, for the holdings that will form the basis of the next NAV calculation. The SEC has also stressed that this transparency supports intraday valuation and hedging, which in turn supports arbitrage. In a later staff statement, the SEC’s Division of Investment Management made the point even more concretely: if an ETF labels foreign-currency holdings only as “cash” rather than identifying the specific currency, that ambiguity can increase arbitrage risk and widen spreads.

The logic is mechanical. If a market maker or AP cannot accurately see the portfolio, it cannot accurately estimate fair value or hedge exposures. If hedging becomes harder, spreads widen. If spreads widen, arbitrage becomes less attractive. If arbitrage becomes less attractive, ETF prices can drift further from NAV. So disclosure is not merely for investor curiosity. It is part of the operating system of the ETF.

This also explains why ETFs generally publish more frequent portfolio information than mutual funds. A mutual fund sells and redeems directly at NAV once a day. An ETF relies on intraday secondary-market trading held together by arbitrage. That model needs more timely information.

A worked example from start to finish

Suppose a broad U.S. equity ETF holds the same stocks as its published basket, and one creation unit equals 50,000 ETF shares. During the day, strong investor demand pushes the ETF’s exchange price slightly above the value of the underlying basket. For ordinary investors, this difference may look trivial. For an AP, it may be enough to matter.

The AP buys the underlying stocks in roughly the right weights, perhaps using inventory for some names and the market for the rest. It then delivers that basket to the ETF through the fund’s operational channels. The ETF issues 50,000 new shares to the AP. The AP sells those shares on the exchange at the higher market price. The AP’s selling pressure adds ETF share supply exactly when the ETF had been rich relative to its basket, and the ETF’s portfolio now contains the securities that correspond to the newly issued shares.

A few days later, market sentiment reverses and the ETF trades slightly below the value of its holdings. The AP does the opposite trade. It buys 50,000 ETF shares in the market, returns them to the fund for redemption, receives the underlying securities, and then sells or finances those securities separately. Those ETF shares are retired. The fund shrinks, and the ETF share supply in the market contracts.

Nothing in this story requires the AP to be charitable or to “support liquidity” in the abstract. The AP is responding to price differences. The fund design turns that profit-seeking behavior into a stabilizing force.

How do creations and redemptions reduce costs for existing ETF shareholders?

Creation and redemption is not only about price alignment. It also changes who bears the cost of investor flows.

In a traditional mutual fund, large subscriptions or redemptions can force the fund itself to buy or sell portfolio securities for cash. Those trades can impose transaction costs on the fund as a whole, which means existing shareholders may bear part of the cost created by entering or exiting investors. ETFs reduce some of that pressure because many flows are handled through in-kind transfers with APs. The AP assembles or receives the basket; the fund does less portfolio trading directly.

That shift has several consequences. It can reduce realized turnover visible at the fund level. It can reduce the need to sell appreciated securities to meet redemptions. It can also let the fund use redemptions to move out less-desired positions through custom baskets, subject to written basket policies and oversight under Rule 6c-11.

Custom baskets deserve a brief note because they are important and often misunderstood. A standard basket closely reflects the fund’s portfolio or a representative slice of it. A custom basket differs from that standard composition. Rule 6c-11 allows custom baskets if the ETF adopts written policies and procedures governing basket construction and designates personnel to review them. The purpose is flexibility: a rigid pro rata basket is not always the cheapest or most efficient way to facilitate arbitrage. But the rule also recognizes the governance risk. Basket flexibility can help market quality, yet it also needs controls to limit favoritism or misuse.

When and why does creation/redemption break down during market stress?

The usual simplified claim is that arbitrage “keeps ETF prices equal to NAV.” That is too strong. A better statement is that creation and redemption often keeps prices near NAV when APs can trade, hedge, finance, and settle the relevant positions at reasonable cost.

When those assumptions weaken, premiums and discounts can widen. Corporate bond ETFs in March 2020 are the clearest example. Research from the Bank for International Settlements documented that many corporate bond ETFs traded at notable discounts to reported NAV during the stress episode. This did not mean the ETF structure had ceased to function. In part, it reflected the fact that bond-market pricing had become relatively stale and hard to observe, while ETF prices updated continuously. The ETF price may have been a faster-moving estimate of where the bonds could actually clear than the published NAV was.

That distinction matters. A discount to NAV can mean at least two different things. It can mean the ETF is mispriced relative to an accurate NAV. Or it can mean the ETF price is discovering information more quickly than a stale NAV can reflect it. In illiquid markets, those are not the same statement.

Stress also reveals another dependency: the mechanism relies on APs being willing and able to use their balance sheets. APs are limited in number, and they are not obligated to create or redeem at all times. Prospectus disclosures from large ETF sponsors make this explicit. If APs step back because hedging is expensive, financing is scarce, or operational risk is too high, discounts and premiums can persist longer than in normal markets.

Empirical research on ETF primary-market structure reinforces this point. Studies using SEC N-CEN data find that ETF creation and redemption activity is highly concentrated among a relatively small set of APs, and that ETFs with broader, more diverse AP networks tend to experience less mispricing, especially in stress periods. The reason is intuitive: the arbitrage mechanism is stronger when more institutions are able and willing to use it.

What back‑office systems and timelines make ETF creation and redemption work?

From far away, creation and redemption sounds conceptually simple: deliver basket, receive shares; deliver shares, receive basket. In practice, the process sits on top of precise operational infrastructure.

The fund or its agent prepares and communicates the Portfolio Composition File, or PCF, which specifies the basket for the next trading day. NSCC distributes this information, and ETF agents, APs, and clearing systems use it to process creations and redemptions. DTCC’s operational materials describe a timeline across T-1, trade date T, and settlement date T+1, with cutoffs for submitting instructions, status checks, and the subsequent movement of shares and component securities through NSCC’s systems and DTC book-entry settlement.

This plumbing is easy to ignore because it usually works invisibly. But it matters for two reasons. First, the arbitrage mechanism depends on reliable settlement. A creation trade is only attractive if the AP trusts that the underlying components and ETF shares will move when they are supposed to. Second, the details of what can be processed centrally affect what kinds of ETFs can use in-kind baskets efficiently. NSCC rule filings to support ETFs with option components, for instance, show how even a familiar high-level mechanism can require new messaging, clearing, and risk-allocation rules when the underlying basket includes instruments that settle in different infrastructures.

In other words, ETF creation and redemption is not just a financial idea. It is also a settlement design.

How does creation/redemption differ for bonds, commodities, and crypto ETFs?

The mechanism generalizes, but the frictions change with the asset class.

For a large domestic equity ETF, the underlying securities are usually liquid, visible, and easy to transfer in standard custody systems. That makes in-kind baskets relatively natural. For bond ETFs, the mechanism is still the same in principle, but less frequent trading in the underlying bonds means basket valuation and hedging are more difficult. For commodity-linked or digital-asset-linked exchange-traded products, custody and transfer mechanics can be even more specialized.

Spot bitcoin products illustrate this well. The familiar high-level structure remains: only APs create or redeem in large blocks, and those flows are what connect exchange-traded shares to the underlying asset exposure. But disclosures have often used cash creation/redemption models rather than fully in-kind bitcoin transfers, because the custody, execution, and regulatory pathway for handing over bitcoin directly is more constrained than handing over a basket of public equities. The mechanism is therefore the same in function even when it differs in operational form.

That is a useful test for understanding the concept. **Creation and redemption is not defined by stocks going in and out. It is defined by institutional convertibility between ETF shares and the fund’s underlying exposure. **

What SEC rules govern ETF creation/redemption and what varies by product?

The fundamental part of the concept is economic, not legal: an ETF needs a way for institutional traders to exchange large blocks of shares for the underlying exposure, so that secondary-market prices can be disciplined by arbitrage.

The legal form around that mechanism is more contingent. In the United States, SEC Rule 6c-11 modernized the framework for most open-end ETFs by allowing them to operate without obtaining individual exemptive orders if they satisfy the rule’s conditions. Those conditions include daily portfolio transparency and website disclosure of NAV, market price, premium/discount information, and median bid-ask spread. The rule also permits custom baskets subject to policies and procedures.

But not every exchange-traded product fits neatly inside that framework. UIT-based ETFs are outside Rule 6c-11 and continue under separate exemptive relief. Leveraged and inverse ETFs are excluded from relying on the rule. Other products, including many commodity or grantor-trust structures, may use exchange-traded wrappers with different legal mechanics even if the economic intuition resembles ETF arbitrage.

So the right way to think about regulation here is not that the rule invented creation and redemption. Rather, it standardized the conditions under which much of the U.S. ETF market can use that mechanism.

Conclusion

ETF creation and redemption is the institutional process that lets ETF shares expand and contract in response to demand. Authorized Participants deliver a basket of assets or cash to create new shares, or return shares to redeem a basket back out. That convertibility is what usually keeps an ETF’s market price close to the value of what the fund holds.

The idea is simple once the core problem is clear. An ETF wants to trade continuously like a stock without losing its link to an underlying portfolio. Creation and redemption is the bridge that makes those two goals compatible. It works well when the basket is transparent and tradable, when APs are active, and when settlement plumbing functions smoothly. It becomes less precise when those assumptions weaken; but even then, the mechanism explains both the ETF’s strengths and its stress points.