The ETF industry's next moat is quoting capacity

Canada's ETF market recorded CAD125 billion of activity this year, on PWD's tracking, and the same market produced a corporate decision that deserves a second look: National Bank sold its research business. Those two facts don't explain each other, but they do clarify a hierarchy: research is a fee stream attached to people who can be hired, published, and valued, while quoting stays on the balance sheet as capital committed against inventory, with no buyer for that risk who is not already carrying it.

The rest of the product calendar sharpens the point: the ETF launch machine is listing complex products faster than the desks that quote them can price the creation baskets behind them, and quoting capacity sits on a small number of balance sheets, which together make quoting the binding constraint. Distribution can be rented from a platform, index construction can be licensed from a provider, and a listing can be papered in a quarter; market-making capital cannot be conjured on the afternoon a fund needs a two-sided quote, and that is the input an accelerating launch calendar keeps asking for.

The industry still argues about the wrong scarcity: shelf space and fee schedules are outputs of a system that runs on committed capital, while the input is a desk willing to hold inventory in a product it did not design on a day when the sector that product tracks is moving hard, and nothing in the launch data suggests the number of desks willing to do that has grown alongside the number of funds asking.

Ranked on defensibility, the research sale reads differently. A research franchise travels, with analysts, coverage, a publishing rhythm, and the client relationships that pay for all of it; a quoting operation does not travel, because its value is a standing willingness to warehouse risk, and willingness sits with the balance sheet rather than with any team that could be sold out of it. Canada's activity figure and its corporate action point the same way, toward the part of the business that cannot be handed to a buyer.

What CAD125 billion of activity means for quoting depends on how many firms are making the prices, and if that activity runs through a handful of desks every incremental listing adds load to the same small group, with the marginal fund quoted by a book more crowded than the one before it. Concentration of that kind produces no crisis and no headline. It produces a price.

The funds at the center of the flow will always have a desk's attention because they earn it, while the funds at the edge of the queue get whatever attention is left and the cost of being an afterthought shows up in the gap between a fund's market price and the value of what it holds on the days the underlying moves. Issuers count listings, but the market counts that gap.

There is no published figure for how much of that load any one desk will take, and issuers have not been asked to publish one either; what they can observe is the behavior of the funds they already run, from the spread on the small funds to the size that can be created or redeemed without moving the basket to the days when the quote is simply wider. None of that shows up in a fee comparison, which is where most product decisions are still made.

Whether the research sale followed from the economics of research or something else in the franchise, the ETF reading is the same: the sell-side's portable assets are the ones being repriced, and the assets that stay consume balance sheet, which is capacity that does not fit on a launch budget line.

Ultrashort government bond ETFs took 94% of August's government bond ETF flows and 82% of the year's, and the category sells duration and price, so a fund competing on price is asking for the smallest possible compensation for the balance sheet standing behind it. That flow is also a rate-cycle trade with an expiry built in; when the front end stops paying, the money that made the category large has little reason to stay.

Active ETFs took 37% of industry flows on a shelf that is three years old, and the number says more about the wrapper than the strategies inside it because the money arrived faster than the track records could have been established. The quoting consequence runs opposite to what the wrapper debate assumes: an index fund's creation basket is a rulebook anyone can rebuild from a file, while an active fund's basket is what the manager owns, changing on a decision rather than a rule, and desks price decisions more expensively than formulas.

Put those two facts side by side and the category that has taken almost all of the government bond money is the one that pays least for a market maker's presence, while the shelf that has taken more than a third of everything is the one that needs the most judgment behind its quotes. The industry is paying for quoting capacity in the place where it needs it least and asking for it in the place where it is hardest to supply.

In the ALPS REIT fund, healthcare names take 21.65% of the book, which makes the creation basket and the investment thesis the same object: a concentrated sector position whose returns depend on occupancy and leverage figures the sector reports. A desk asked to quote that in size is quoting healthcare real estate, and it will want to be paid for it.

Reckoner's two active CLO ETFs take the same idea into structured credit, pitched as a trade on tranche design and leverage and handed, by the terms of that pitch, to the rate cycle. Whatever selection skill ends up inside those funds, their early trading will be read as a test of whether a quoting desk can price a structure when the underlying is busy, which is a different exam from the one the manager sits.

The Japan product making the rounds fits the pattern at smaller scale: a 14% AI sleeve on a payout fund, where the sleeve does the selling and the caveat attached to the pitch is the more useful half. Tilts and sleeves are the shelf's answer to a crowded market, and every one of them narrows the basket a market maker has to price.

Concentration in quoting is the issuer's problem because the issuer is the party with something to lose and no contract that solves it. A dealer quotes when the product is easy and widens when it is not, which describes a balance-sheet business rather than a service level, and the issuer that launches into that arrangement has handed its own liquidity to firms that rank it against every other fund on the same book.

There is a second-order effect that matters more than the first: concentration of quoting capacity does not reduce the number of launches; it reduces the number of desks available to each one. Product creation is cheap and getting cheaper, the cost of a launch is booked at the issuer, and the cost of quoting it is booked at somebody else's desk, so an industry that measures itself in listings will keep producing listings and the strain lands with whichever firms are still willing to make a two-sided market in them.

The trade we would make runs the other way: committed quoting capacity should be bought the way shelf space is bought, with seed capital, economics that make a fund worth a desk's attention on a quiet afternoon, and a relationship that holds through the first day the sector gaps. Launching a complex product without that arrangement is buying a marketing expense and booking it as a product.

The alternative is to accept the concentration and price it with fewer launches, larger ones, seeded with enough capital to interest a desk, and sold on the premise that the fund will trade well rather than on the premise that it exists. It is a slower product strategy and likely a better one, because the funds that survive a quoting squeeze are the ones somebody was willing to make a market in.

The verdict shows up in trading rather than in filings: watch the CLO launches and the concentrated REIT book through their first stretch of real volatility in credit and healthcare real estate, and compare each fund's market price with the value of what it holds on the days its sector moves. That spread is the honest measure of quoting capacity in this market, and the next complex launch will price it whether or not anyone in the product meeting is watching.