On 21 October, the Bank Policy Institute filed its response to the OCC's final rule on payment stablecoin yields. Three days later, its staff were already briefing reporters on what that rule would do to bank deposits. The rule itself does not take effect until 18 January 2027.
Hold that date. Fourteen months in which institutions must design, price and ship products against a standard whose central clause โ the treatment of third-party rewards โ has never been translated into plain language. No published adjudication. No reasoned opinion. A presumption clause reportedly stripped out between proposal and final text, and a lobbying apparatus now attempting to reconstruct, from silence, what the deletion meant. Underneath the chaotic surface of another compliance news cycle, something structural is being settled: who has the authority to define substance, and at whose cost the rest of the market waits to find out.
The statutory background is rarely stated cleanly. The GENIUS framework creates a category โ payment stablecoin โ whose defining legal characteristic is that it is a claim on the issuer, backed by reserves of cash and short-duration sovereign obligations. That construction is an economic machine of almost insulting simplicity: the reserve earns the risk-free rate, the holder earns nothing, the spread belongs to the issuer. The final rule builds three pillars around this machine โ reserve custody, redemption rights, risk management โ and forbids the issuer from paying yield to holders, directly or indirectly. On anti-money-laundering and sanctions compliance it says almost nothing, deferring those requirements to a separate process that has not produced text.
Between the prohibition and the deferral sits the entire competitive map of the dollar on-chain.
Consider three architectures that produce an identical consumer experience. In the first, the issuer pays the holder out of reserve income: a red line, unambiguous. In the second, the issuer shares reserve income with a distributor, and the distributor rebates a portion to the holder: a gray zone, and the actual subject of the fight. In the third, a merchant funds a discount from gross margin: most likely permissible, because it is a marketing expense rather than the time value of money. To a wallet interface, these are indistinguishable. To a regulator reading a balance sheet, they are three different industries. Note what the third path reveals: the regulator is not hostile to holders earning something. It is hostile to holders earning something out of the reserve.
When I modeled Aave v2 liquidity flows in the summer of 2020, the discipline that saved me was refusing to accept the protocol's own labels. The number that mattered was never what the interface called a position; it was which contract held the claim, and against whom it could be enforced. I withdrew exposure weeks before the anchor wobbled, not because I had a better forecast, but because I had stopped believing the dashboard.
The OCC's real technical problem is not the existence of yield. It is the observability of legal construct. The same balance accruing at 4% annualized can be a bank deposit, a distribution rebate, or a security โ and the rule classifies by legal architecture while users experience only the balance. Which is why the reported deletion of the presumption clause matters more than any individual adjudication. A presumption is a decision tree: if the payment tracks the reserve rate, treat it as indirect. Remove it, and fact-finding migrates from published rule to unpublished discretion, and every distributor must guess at the shape of a standard that shifts with each application. There is no decision tree, because there is no published reasoning from which one could be derived.
Here the conventional reading fails. The lobby pressure is widely described as an industry fighting a crackdown, but the interests are not aligned that way. The party with the strongest interest in a hard prohibition is the issuer itself, because a reserve spread competed away through channel rebates collapses toward zero โ the business degrades into a thin payment rail. The parties who genuinely want latitude are the distributors: exchanges, wallets, payment apps for whom a yield-bearing balance is the cheapest customer acquisition tool ever engineered. Large banks, meanwhile, are defending a deposit franchise, not consumer protection. Three actors, three motives, one rule.
And the securities question is not decorative. Strip away the statutory exemption and run the familiar test: money invested, in a common enterprise, with a reasonably expected profit derived from the efforts of others. A stablecoin paying nothing fails on the last two elements. A stablecoin whose annualized return is legible, advertised and anchored to the issuer's reserve management may not. The tail risk is worse than a securities label: if third-party rewards make a token economically identical to an interest-bearing deposit, the activity edges toward the statutory definition of taking deposits โ banking law, not stablecoin law.
Which explains the quiet maneuver happening in the other lane. Tokenized deposits โ bank liability represented on a ledger, interest-bearing by construction, insured by construction โ deliver the same user experience under an entirely different legal claim. The SoFi construct is the template, and its lesson is not technological. Banks are not trying to eliminate on-chain yield. They are trying to own it. Regulatory arbitrage, run in reverse, becomes a regulatory dividend. A depositor who moves from a wallet balance into a tokenized deposit notices no seam; the balance simply keeps paying.
There is also a variable nobody in this debate controls: the policy rate. Reward budgets are funded out of reserve income, so the entire economics of on-chain yield is a function of the Treasury curve, not of crypto adoption. The stablecoin reward is, functionally, a leveraged bet on the front end of the curve. A cutting cycle compresses the reward economy regardless of what the OCC writes โ and it will compress the gray-zone experiments first, because they carry compliance cost without a durable funding source.
The contrarian position, then, is uncomfortable. The prevailing assumption is that case-by-case adjudication is administrative laziness. I read it as a deliberate hedge: in a period without precedent, an agency preserves flexibility rather than locking a wrong standard into text, and the long-run cost of that caution lands on the least represented party. Distributors have the highest dependence on this rule and the weakest voice in it โ outside the bank lobbying frame, absent from the rulemaking record, reduced to filing individual applications and hoping. The most expensive outcome for them is not prohibition. It is permanent indeterminacy, because a decision tree you cannot see is a compliance cost you cannot amortize across a product roadmap. The BPI's procedural demand โ publish the reasoning, not just the result โ is therefore not proceduralism for its own sake. Under administrative law, an agency that adjudicates without explanation invites review for arbitrary and capricious action. The letters are building a record.
Fourteen months is a long time to build against a standard you cannot read. Watch whether the 2027 date holds, whether any reasoned decision is published, and whether the next fight migrates from what a token is to what a wallet lets it do โ because the deferred AML chapter will decide more than the yield clause ever could. And ask the question the rule leaves open: if two identical balances carry different rights and different risks, which one will the market actually call money?