Canada's 1% Redemption Ceiling Is a Tail-Risk Instrument, Not a Consumer Right

One percent.

That is the only hard number in the entire document, and it is the number that tells you who is expected to absorb the cost of a bank run.

What reached me was not the proposal. It was a six-point summary — CryptoSlate's relay, re-summarized by The Catalyst newsletter — carrying an explicit warning that the full text requires the original. No proposal number. No named issuing authority. No link to a Department of Finance Canada consultation paper, no OSFI guidance, no Bank of Canada press release. In my line of work that is not a source. That is a rumor with a citation trail.

But the phrasing of those six points is specific enough to reverse-engineer the machinery underneath, and the machinery is where the risk sits. Treat everything below as structural inference from wording, not confirmation. I will flag confidence where it changes the conclusion.

The first thing I did was look at the number. A regulator who writes "1% fee ceiling" instead of "redemption at par" has already told you that redemption will be intermediated.

Context: six points, and a philosophy nobody is reading correctly

Reconstruct the known set. A proposal would cap stablecoin redemption fees at one percent. There would be no minimum redemption amount. Reserve-market liquidity could constrain how much can exit per day. Access to banking rails is named as a key condition for the whole thing to work. The definition of "covered holders" is undefined. The scope of stablecoins covered — CAD-only, or USD-denominated as well — is undefined. Effective date: unknown.

Six data points. Zero primary documents.

Here is why that matters more than usual. Canada does not currently have a federal stablecoin-specific regime. What it has is a patchwork: the Retail Payment Activities Act, whose registration regime the Bank of Canada has been administering since 2024, covering payment service providers but not the design of the asset itself; and Canadian Securities Administrators guidance on value-referencing crypto assets, which treats stablecoins as securities-adjacent and imposes custody and disclosure conditions rather than redemption rights. The Bank of Canada's retail digital dollar work was set aside in 2024 on the grounds that the evidence for a CBDC was not there. That left a genuine policy vacuum on the question of what a stablecoin holder is legally owed.

Compare the three frameworks that actually exist. The EU's MiCA, in force since mid-2024, requires e-money token issuers to redeem at par, at any time, on request of the holder. The US GENIUS framework routes through issuer licensing, reserve composition rules, periodic attestation, and again, par redemption. Hong Kong's Stablecoins Ordinance runs a dual track: licensing plus segregated reserve custody.

Every one of those regimes regulates the issuer. Canada's proposal, on the evidence available, would regulate the exit. That is the actual news value here — not the one percent. A G7 member contemplating a holder-rights framework instead of an issuer-obligation framework is a genuine divergence in regulatory philosophy, and philosophies have consequences that outlive the consultation period.

The crypto-native read of this will be "Canada softens on stablecoins." That read is wrong, and it is wrong in a way that will cost people money.

Core: the architecture the wording implies

Start with a deductive step that requires no documents at all. If issuers redeemed at par, there would be no fee to cap. A ceiling of one percent exists only because someone is expected to charge for the conversion — a licensed payment service provider, an on-ramp, a bank partner acting as agent. The regulator is not writing an obligation onto the issuer's balance sheet. It is writing a price control onto a service market it assumes will exist.

That is a fundamentally different bet on where friction should live. MiCA and GENIUS put the cost of exit inside the issuer, funded by reserve carry. This puts it outside, in a market, and asks the market to make a living under a fixed margin. Confidence: medium, because the fee could in theory be a cap on issuer charges. But issuers do not charge for redemption. Intermediaries do.

Now do the arithmetic on the ceiling itself. In calm markets, exiting a stablecoin costs somewhere between 10 and 50 basis points through an OTC desk, and roughly 10 to 40 through a centralized exchange withdrawal. A one percent ceiling is two to ten times the prevailing price. It binds nothing. It is decoration.

In a de-peg, the picture inverts. Secondary-market discounts on a stressed stablecoin have run from two percent to ten percent and beyond. I spent two weeks in May 2022 reverse-engineering the UST stabilization mechanism from on-chain data, building a simulation that showed the death was mathematically overdetermined once the liquidity buffer fell below a computable threshold. The model published hours before the final cascade. What it also showed is that during the unwind, the "official" redemption channel was functionally closed while the secondary market traded at sixty cents. At that moment, a one percent ceiling is not a protection. It is a hard constraint below market price.

A price ceiling below market price does not create capacity. It creates a queue.

This is the single most important structural fact about the proposal: the one percent number is not a consumer pricing rule. It is a tail-risk instrument, aimed precisely at the moment when the market's own price signal is telling capital to stay away.

History repeats, but the signature changes. Every price control on a scarce good during a shortage produces the same signature: formal price stability, informal rationing. The instrument is always marketed as protection for the small holder. The mechanism always transfers the shortage to the small holder, because the large holder has an OTC desk and a relationship manager and a lawyer who reads the standing definition.

Read the next two points together, because they are the contradiction. Universal access — no floor, so a fifty-dollar redemption is a right. And an explicit acknowledgment that the reserve side may cap how much can leave per day.

You cannot have both, on the days that matter. If daily capacity is finite and demand is not, allocation happens by some mechanism. That mechanism is either price, queue, or discretion. The proposal removes price. So it is queue or discretion. Both are rationing. Neither is disclosed.

Here is the engineering version, and this is where I have ground truth rather than inference. In 2020 I ran a volatile 3pool position on Curve with fifteen thousand dollars of my own capital and learned what "liquidity" means when everybody wants the same side of the trade. The pool did not fail because the math was wrong. It failed because the exit was a shared resource with fixed capacity, and capacity was consumed in order of arrival. Slippage is a queue with a price attached. Strip out the price and you have a queue with no signal. I lost forty percent of principal to that lesson, partly to impermanent loss and partly to slippage on the way out. It remains the cheapest forty percent I have ever spent, because it replaced theory with a rule: never confuse a claim on a shared exit with a guarantee of that exit.

Impermanent is a promise, not a guarantee. So is a redemption right with an undisclosed daily cap.

The precedent nobody is citing: money market funds

There is a regulated product that already solved this exact problem, and it solved it in the opposite direction.

US money market funds under Rule 2a-7 were amended after 2008 to allow, and later to require, liquidity fees and gates when weekly liquid assets fell below a threshold. The Reserve Primary Fund breaking the buck in September 2008 taught the same lesson the redemption proposal is now ignoring: when a vehicle promises par and holds assets that cannot always be sold at par, the promise is a liquidity mismatch wearing a stable price. The 2010 and 2014 amendments responded by making the mismatch explicit and contractual — boards could impose a fee, and they could gate. The 2023 amendments made fees mandatory in stress.

The design principle there was the reverse of this proposal. Regulators did not cap the fee at a low level. They made the fee a variable instrument that rises when liquidity tightens, precisely so that the price signal recruits exit liquidity instead of suppressing it. They paired it with a gate whose trigger is disclosed in advance.

Canada's proposal, as summarized, appears to do the opposite: fix the price low, leave the gate undisclosed. That combination is how you convert a price event into a duration event. Holders who arrive after the cap fills do not get a worse price. They get a later date. In a de-peg, later is worse.

The banking rail is the sequencer

Every Layer 2 I have audited carries the same structural joke. The marketing says decentralized sequencing. The architecture says one node, a privileged key, and a fallback operator. Two years of roadmaps, and block production is still a single entity behind a load balancer, with a governance forum post standing in for a security model. I have made this point repeatedly and I will make it again: the decentralization is a slide, not a deployment.

The redemption proposal has the identical shape. On paper, the holder has a right. In practice the right executes through a licensed payment provider, which executes through a bank account, which executes through a bank whose internal risk committee has veto power it will never have to explain. The bank is the sequencer. If that bank decides crypto-adjacent flows are not worth the compliance overhead — and Canadian banks have a documented history of de-risking exactly this flow — the right is a smart contract with no executor.

Verify the code, trust the ledger. Here there is no code to verify. That is not a rhetorical flourish; it is the literal gap. A redemption right that depends on a discretionary banking relationship is a promise, not a guarantee, and a promise is only as good as the counterparty's incentive to honor it on the worst day of the decade.

I learned the counterparty version of this in November 2022. I was not in FTX. I was holding stablecoins on Celsius and reading the contagion map, and I moved fifty thousand dollars of USDC into a multi-sig hardware setup in Auckland over three days, deliberately, while peers panic-sold into the same frozen exits. The lesson was not about timing. It was that the right to withdraw is a function of the custodian's solvency and the rail's availability, and neither of those is on your side of the screen.

The economics do not clear at one percent

Model the redemption service provider. It must hold pre-funded fiat, carry compliance capital, staff KYC and AML, integrate with a bank, and absorb the risk that every holder arrives on the same afternoon. Its gross margin is capped at one percent.

In normal times that is workable. Volumes are modest, float income helps, and the ceiling rarely binds because the market price is lower anyway. In a run, its cost of liquidity spikes — it is bidding for the same dollars everyone else wants — while its revenue is legally frozen. The rational response is not to serve the queue. The rational response is to narrow the queue: extend processing times, tighten KYC, decline large tickets, quietly close the account to new holders.

This is the cost-bearer mismatch that quietly kills well-intentioned consumer rules. You can cap the price. You cannot cap the risk. The risk reappears as a non-price barrier, and the barrier always falls hardest on the unsophisticated holder the rule was written for. Risk is the price of admission. Forbid the price and you do not remove the admission — you make it discretionary.

Trace the incidence. Reserve assets — T-bills, bank deposits — earn carry for the issuer. The issuer earns the spread. The redemption provider earns a capped one percent. The holder holds a right. The proposal's binding constraints land on the middle link, and the one percent ceiling is a protected gross margin for a licensed intermediary. The no-minimum clause is a volume mandate for that same intermediary. The banking-access condition is a moat, because only entities with existing bank relationships can play. The clear winners are RPAA-registered payment service providers and the banks that clear for them. The losers are non-compliant OTC desks whose spread gets competed away, and issuers who inherit compliance cost without inheriting the fee.

The issuer side hides the second-order damage. A stablecoin issuer's revenue is reserve carry. Impose localization of reserve custody, or a higher high-quality-liquid-asset ratio, and you compress net interest margin directly. Canada is a small market. Compress the margin enough and the rational issuer does not comply — it geofences. The rule then protects Canadian holders of a product no one is willing to issue into Canada.

And the scope trap. Canadian retail overwhelmingly holds US dollar stablecoins, not CAD stablecoins. The CAD-denominated stablecoin market is, in market-cap terms, a rounding error — verify it on DefiLlama in under a minute, and please do, because the number is the whole argument. If the proposal covers only Canadian-regulated issuers, it governs a market that barely exists. If it covers any stablecoin sold to Canadian residents, it asserts extraterritorial reach over offshore issuers holding offshore reserves, and the enforcement mechanism is a bank rail the issuer may not need. Regulatory scope that does not match the asset actually held is scope that does not bind. Same error the omnichain narrative made: contracts deployed everywhere, users caring nowhere.

Contrarian: the ceiling may make holders less safe

The consensus read will be that this is Canada moving toward consumer protection — a G7 signal, mildly bullish for adoption.

The contrarian read is that the one percent ceiling may make Canadian stablecoin holders less safe in exactly the scenario the rule was written for.

A de-peg is an information event. The discount is the market telling holders how much redemption capacity exists and what it costs to reach it. That price is ugly, and it is also the incentive that recruits new liquidity into the exit — desks bid the discount, capital arrives, the gap closes. Suppress the price with a ceiling and you suppress the recruitment signal. Capital that would have provided exit liquidity at four percent has no reason to show up at one. The queue lengthens, the discount persists, and the holder who most needed out is holding a claim on a capped, capacity-limited window.

There is a second blind spot. Everyone is debating the fee. Nobody is debating the definition of "covered holder." That undefined term is where the entire policy will be fought. Does it include beneficial owners behind an omnibus account? Holders on a foreign exchange? A smart contract? A trustee? Each answer changes who has standing to demand redemption, and therefore changes the size of the liability the intermediary is underwriting. A fee cap is a headline. A standing definition is a balance sheet. In 2017 I audited early ERC-20 implementations and found a replay weakness in the transferFrom path that only manifested across chains with identical identifiers — the fix was a domain separator, not a change in intent. The lesson was structural: a right that executes differently depending on which environment it lands in is not one right. It is several, and only one of them was written down.

Silence before the volatility spike is never empty. It is usually a drafting committee.

Takeaway

Three things to watch, in order. Whether a daily redemption cap or queue is disclosed in the actual text. Whether OSFI issues companion guidance on bank access for redemption flows. And how "covered holder" is defined. If the cap exists and the definition is narrow, the one percent ceiling is a consumer right that functions only when it is worthless. If the definition is broad and the banking rail is guaranteed, Canada will have built the first G7 framework that regulates the exit instead of the issuer — and the rest of the G7 will copy the architecture, not the number.

The market whispers. The blockchain shouts. A consultation paper, today, does neither. The question is not whether the ceiling is one percent. It is who is standing in the queue when it fills.