Authorized Participants and Market Makers

Two hats, precisely distinguished

The market maker (MM) lives in the secondary market. Its business: post a bid and an ask in the ETF all day, capture the spread, and immediately hedge whatever inventory results (bought 20,000 ETF shares from a seller? Sell index futures or the basket against them within seconds). Its risk is being caught with unhedged inventory in a fast market; its profit is volume × spread minus hedging costs. Some MMs sign as official/registered market makers with exchanges or issuers, accepting quoting obligations (maximum spreads, minimum sizes, minimum presence) in exchange for fee rebates: the reason even a sleepy niche ETF shows a standing quote at 10:07 on a Tuesday.

The authorized participant (AP) holds an AP agreement with the fund's distributor/issuer, granting primary-market access: the right to deliver baskets and receive creation units, and vice versa. The AP's business is Chapter 31's arbitrage plus a service business: executing creations/redemptions on behalf of clients (a pension fund entering a position quietly, Chapter 88's world) for a fee.

The overlap: an MM that is also an AP can quote fearlessly, because any inventory it accumulates converts to the underlying at NAV through the window. An MM without AP status can still lay off inventory to AP partners. The ecosystem is thus layered: dozens of APs per major fund on paper, a handful active in any given fund, and a competitive ring of MMs around each order book.

Who these firms actually are

Names help make it concrete (as examples of the ecosystem, dated to the mid-2020s): specialist ETF trading houses such as Jane Street, Flow Traders, Optiver, IMC and Susquehanna; electronic market-making giants like Citadel Securities and Virtu; and the ETF desks of major banks (Goldman Sachs, JPMorgan, BNP Paribas, Société Générale among them). ICI's data has typically counted 30–50 signed APs for large US funds, with a handful doing most volume in each; Europe's fragmented venue landscape (Chapter 91) makes the specialist firms, several of them Dutch, unusually central there.

The point of naming them is not trivia; it is the structural insight: this is a competitive, multi-firm layer, not a single point of failure. Any premium or spread wide enough to be a problem is, from these firms' seats, an unclaimed paycheck.

How the sausage is priced

Watch an MM quote a Europe-listed S&P 500 ETF at 10:00 CET, US market closed, and you see the craft: it prices the basket off S&P futures (trading almost 24h), adjusts for FX, dividends going ex, and its own hedging costs, then wraps a spread around that fair value wide enough to cover the uncertainty. This is why Chapter 12's "awake underlying" rule works: when New York opens, the uncertainty term collapses and so does the spread. The MM was never guessing; it was charging precisely for how much guessing was required.

Two practical consequences for you:

• Displayed size understates real depth. MMs replenish quotes continuously and hold bigger size than they show; the true depth is the basket's (Chapter 31). Retail orders essentially never exhaust it; institutions use Chapter 88's channels.
• Quote quality is portable evidence. A persistently wide spread in a liquid-basket ETF signals weak MM competition on that listing, not fund illiquidity: often fixable by trading the same ISIN on a busier venue (Chapter 56).

The honest fragility discussion

The system's known soft spots, stated without drama:

Concentration in practice. Many signed APs, few active ones per fund, especially in bond and niche ETFs. If the top one or two pause (a risk-limit breach, a systems issue), spreads widen until others step in: minutes to hours of worse prices, historically, not failure.

Volunteers, not utilities. Chapter 31 said it: no obligation exists. The 2013 episode where a major bank paused redemptions in some muni ETFs (others continued; discounts appeared briefly) is the textbook case: the system degraded gracefully, precisely because the paused firm's profit went to competitors.

Stress behavior is repricing, not absence. March 2020, the great test: MMs kept quoting bond ETFs through the seizure, at wide spreads and visible discounts, i.e., they charged crisis prices for crisis uncertainty, and AP activity hit records. The complaint "spreads widened in the panic" is, examined closely, the system working: continuous, honest, expensive prices versus the alternative on display in the underlying bond market that week, which was often no price at all. Chapter 82 tells it fully.

The incentive alignment worth knowing: these firms earn most when volatility and volumes spike. The crisis that tempts them to step away is also their best revenue season, a cynical-sounding safeguard that has, so far, been the operative one.

A worked example

A pension fund must invest €80m in a EUR corporate-bond UCITS ETF whose screen shows €500k displayed size: hopeless, apparently.

Its dealer calls three AP/MM desks for a risk price on the full block (the RFQ world, Chapter 88). Each desk prices the underlying basket it would buy to hedge, adds its cost of executing €80m of corporate bonds, and bids. Winner executes at NAV +9bp; behind the scenes it assembles bonds over two days and runs a creation to flatten its book. The screen's €500k was never the ETF's capacity; it was the retail shopfront of a wholesale warehouse. Same machinery, Chapter 31's window, different door; and the pension's 9bp all-in beats what buying 400 bond lines itself would have cost several times over.