Stablecoins take the payment fee, not the deposit

The ABA's case against stablecoins has been a deposit case from the start: let dollar tokens circulate widely enough, the association warns, and community banks lose the balances that fund their lending. An op-ed this week complicates the first half of that warning, because the balances, so far, are staying put.

If balances are sticking, the leak the ABA has been describing is not the one that is open: a deposit balance and the payment instruction that moves money through the account are separate assets, and only one of them is insured, relationship-anchored, and slow to move. The instruction travels at the speed of a merchant's checkout page and a payroll vendor's next software release, which creates an ordering problem for anyone watching deposits—by the time a balance leaves a community bank, the decision to move it was likely made much earlier, somewhere else in the customer's stack, by someone who was not thinking about the bank at all.

Stickiness of the kind the op-ed describes is likely a fact about switching costs rather than affection: a business keeps its operating account where payroll clears and receivables land, and changing that means re-papering standing instructions, reissuing account details to counterparties, and moving the people who reconcile the account at month end. None of it is impossible, but all of it is the sort of work a controller postpones until something forces the issue—which is how a stablecoin takes the instruction without ever taking the balance.

That is the leak that costs, and it costs in the currency community banks feel first: the workflow generating fee income is the payment rather than the idle dollar, and a payment that settles on a token rail does not generate the fee income it generated before. The arithmetic is unattractive in a specific way—the bank keeps the account, the compliance burden, and the examiner's attention while giving up the transaction revenue that justified carrying the relationship. This leak opens without a single depositor deciding to leave.

The split is easy to miss because both assets sit on the same statement: a commercial customer's average balance shows up in the deposit base and the funding plan, while the instructions that move money in and out of that balance show up in the fee schedule, if they show up anywhere. The balance is the easier number to cite and the number the industry cites when it talks about itself, because payment volume is dispersed across customers and products and its economics arrive as fee lines that are harder to argue about in public.

The op-ed's contribution is narrow and useful: it puts deposit behavior somewhere the argument can be tested, even though one dataset does not settle a decade-long question and nothing in it rules out balances eventually following the instructions. What it supports for now is the claim that the balance sheet and the payment workflow are running on different clocks, and the ABA's warning has been written to the slower one.

The distinction decides which remedies have a chance, because a deposit-flight diagnosis produces deposit-shaped rules—what must back a token, what an issuer must hold against it, what an issuer may promise the people holding it—and every one of those reaches the balance sheet while none of them reaches the instruction. A fully reserved, plainly disclosed dollar token still settles a payment, and settlement is where the economics sit. If community banks want the revenue attached to moving money, the argument worth making is about who may initiate and settle a dollar payment and how much of the settlement chain a bank is allowed to own.

For the warning to describe a live threat rather than a distant one, several things have to hold at once: businesses willing to keep working balances in tokens rather than at insured institutions, counterparties accepting token settlement without a lag or a discount, accounts payable and payroll absorbing a new set of reconciliation chores, and auditors signing off on them. The op-ed's data suggests the first condition is not yet true at the balance level, but the other three can turn first, and one of them turning is enough to move the instruction.

The ABA can fairly answer that today's stickiness says nothing about the default a decade out, since deposits come from business formation and the next cohort of merchants and payroll processors is being built by people who assume dollars move as tokens. A generation of operating accounts that never opens at a bank is a slower and larger version of the problem the association is describing, and no amount of present-day balance stability prevents it.

That argument is about a decade, and it postpones the question rather than answering it: the workflow leak arrives at the near end of the timeline, where payment volume can migrate while balances hold and the fee income goes with the volume. Read closely, the op-ed's data is evidence against the deposit case rather than for it, since the balances it describes are holding while the revenue attached to them is leaving anyway.

A community bank has no obvious way to win back an instruction once a customer's software stack has been rebuilt around another rail, because the switching costs that keep the deposit in place do not extend to the payment—the customer never has to close anything, the instruction simply stops arriving, and the fee stops with it.

Suppose the association wins the fight it has been having: community banks would then face a dollar instrument whose backing rules have been tightened and whose capacity to carry a payment was untouched. The members would have preserved the balance and conceded the instruction, which is the trade the payment data suggests is already under way.

Nothing on this week's Washington calendar resolves the payments question: Lummis revised the CLARITY market-structure bill with three narrow changes five days ahead of a 60-vote cloture motion that still turns on the DeFi anti-money-laundering exemption, and the timing of a revision that late suggests the votes were not yet assembled. Whatever the whip count, the vote matters a great deal to who supervises crypto intermediaries and how much compliance burden they carry.

Whether cloture succeeds or fails, a merchant's settlement instruction reaches a bank or bypasses one on the same terms the next morning, because market-structure legislation draws a line between two federal market regulators while the payments question is about where a dollar instruction is allowed to sit and who is permitted to move it. Nothing in that fight reaches the instruction, which leaves the competitive question where it started: with the institutions holding the accounts and the firms building the rails that carry the instructions.

There is a version of this argument in which the banking industry's instinct is right and only the timing is wrong: if tokens become the default settlement asset for commerce, balances follow instructions eventually, and a community bank that conceded the second loses the first too. But the hedge against that future is a position in the flow, not a rulebook for reserves.

The evidence that would settle the argument is not hard to describe and largely missing from the week's material: count the share of business payments that initiate and settle without touching a bank, then track it against the deposit balances the op-ed puts at issue. The first number decides whether community banks have a durable role in payments; the second is the one the industry has been spending its political capital to defend.

The association is defending the cheaper asset. A balance can be repriced, borrowed against, or replaced at a cost the market publishes every day; a payment instruction has no replacement market, and once a customer's payments have moved, the account left behind is a legacy balance with a compliance bill attached. Given a choice between rules that shelter the balance and rules that keep the instruction, community banks get more from the second set—and that is the set this week's legislative calendar does not contain.

The trajectory argument is real, and the deposit data does not refute it—it delays it, which is the input that matters most to a bank choosing where to spend its next compliance dollar. The workflow case pays out first because reserve rules, however well drafted, do not hold a payment instruction at a bank.