ETF Creation-Redemption Risks: Baskets, Access, and Arbitrage

Basket access and arbitrage are not guaranteed

ETF shares trade between investors on an exchange, but the number of shares outstanding can change through creation and redemption. Large broker-dealers known as authorized participants generally create or redeem ETF shares in blocks called creation units. To create shares, an authorized participant delivers the basket of securities or other assets specified by the fund, and may also deliver or receive cash. The fund issues creation-unit shares, which the participant can then sell in the secondary market. A redemption reverses that process: the participant returns ETF shares and receives the relevant basket or cash.

This mechanism connects the ETF’s exchange-traded price with the value of its underlying portfolio, usually expressed as net asset value, or NAV. If ETF shares trade materially above the value of the underlying basket, creating and selling new shares may be economically attractive. If shares trade below NAV, buying shares and redeeming them may be attractive. That broad arbitrage logic is central to ETF analysis, but it is not a guarantee that the market price will always equal NAV.

The basket is more important than the label “in kind”

A reasonable analysis can fail when it assumes every ETF uses the same creation process. Funds may use in-kind transfers, cash creations and redemptions, or a combination. They may also use baskets that differ from a simple pro rata slice of the portfolio. The fund’s prospectus and creation procedures describe the permitted methods, eligible assets, cash components, transaction fees, and the parties able to transact directly with the fund.

That detail matters because the practical cost of arbitrage depends on what must be delivered or received. A liquid equity ETF that accepts a readily tradable basket may be easier for an authorized participant to create or redeem than a fund holding less liquid bonds, foreign securities, derivatives, restricted instruments, or assets subject to settlement constraints. A cash transaction can shift the task of buying or selling underlying assets to the fund. The resulting trading costs, market impact, and timing differences can affect portfolio returns and tracking.

Custom or non-pro-rata baskets require especially careful reading. They can help a portfolio manager manage inflows, outflows, tax lots, or difficult holdings, but they also mean that an analyst cannot assume the published portfolio alone describes the transaction an authorized participant faces. The relevant question is not simply whether the ETF is in kind. It is whether the actual creation and redemption process remains workable at a cost that supports efficient secondary-market pricing.

Where premium, discount, and tracking analysis can break down

A premium or discount is the difference between an ETF’s market price and its reported NAV. It is useful evidence, but it is not self-explanatory. NAV may be based on the latest available prices for underlying assets, while the ETF trades continuously. For international securities, bonds, and other instruments that do not trade as continuously or transparently as large domestic stocks, the reported NAV may not fully reflect the price at which the basket can be traded at that moment. An apparent discount can therefore reflect valuation timing as well as impaired arbitrage.

The same caution applies to tracking analysis. An ETF can differ from its index because of its expense ratio, index-rebalancing costs, sampling rather than full replication, cash holdings, securities-lending revenue where applicable, taxes, corporate actions, and distribution timing. Creation and redemption activity can add another layer through brokerage, bid-ask spreads, settlement costs, and the cost of handling cash flows. Comparing a fund’s return with an index return without checking whether both are measured on the same tax and distribution basis can produce a misleading judgment about implementation quality.

Investors should separate three observations that are often conflated: the ETF’s quoted trading price, its reported NAV, and the estimated current value of the underlying assets. During normal trading they may be close. During a market disruption, they can diverge for understandable reasons. The analysis should examine the duration and recurrence of differences, the liquidity of the underlying market, and whether creations and redemptions remain available, rather than treating any single premium or discount as proof of failure.

The operational dependencies behind the arbitrage mechanism

Creation and redemption depend on a functioning network, not on an automatic rule. Authorized participants must be willing and able to finance positions, hedge exposure, source the basket, process settlements, and bear transaction costs. Market makers also need to quote ETF shares. If underlying markets are volatile, closed, thinly traded, or subject to trading restrictions, the cost and uncertainty of those activities can rise. Wider ETF bid-ask spreads or more persistent premiums and discounts may follow even when the fund itself continues to calculate NAV and accept orders.

Concentration is a further analytical issue. An ETF may have multiple authorized participants on paper, yet activity can be concentrated among a smaller set of firms. The useful evidence is found in fund documents and, where available, information on creation and redemption activity, shares outstanding, historical premiums and discounts, bid-ask spreads, and underlying-market liquidity. A large asset base or heavy exchange volume does not by itself establish that redemptions will be inexpensive under stressed conditions.

Fees require similar precision. The fund’s published expense ratio is a recurring portfolio cost, while brokerage commissions, spreads, creation or redemption transaction fees, and underlying trading costs arise in different places and may be borne by different parties. A low expense ratio does not ensure low total implementation cost for a buyer who trades at a wide spread or for a fund that must transact in costly markets. Nor does an in-kind process mean the fund will never make distributions; investors should review the fund’s stated distribution policy and tax reporting rather than assume a particular outcome.

A creation-and-redemption test for ETF comparisons

The risks that matter are therefore the ones that invalidate the assumptions behind a comparison: that NAV is a real-time liquidation value, that the published holdings equal the creation basket, that authorized participants will always arbitrage quickly, that index tracking captures all ownership costs, and that distributions follow automatically from the ETF structure. Those assumptions may be reasonable for some funds and market conditions, but they should be tested rather than inherited.

For a durable comparison, match the ETF’s creation method to the liquidity and trading hours of its holdings; review the prospectus for cash, in-kind, and basket flexibility; compare historical market prices with NAV across calm and difficult periods; and assess tracking after fees, distributions, and index-method differences. Creation and redemption are a powerful structural feature, but their effectiveness is conditional on the assets, counterparties, valuation methods, and market infrastructure that support them.