framework

The Two-Stage Framework

Permit transit first. Add holding second. Eight jurisdictions arrived at this sequence independently.

By Anton Titov, Founder · Plexo Institute

Published 2026-05-17

The question is not whether to regulate stablecoins. It is in what order. Permit transit first.

Reader Brief

The question is not whether to regulate stablecoins. It is in what order. Permit transit first. Add holding second. Eight jurisdictions arrived at this sequence independently.

Reading Guide

Four ideas to anchor the read.

Sequencing rejects the ban-vs-allow binary.

Permit transit in Stage 1 before holding in Stage 2. Eight jurisdictions - Singapore, UAE, ADGM, United States, United Kingdom, Brazil, EU, and Chile - converged on this sequence independently. None coordinated. Each extended an existing framework: payment services acts, central bank supervision, regulated activities, e-money regulation, or federal banking law. The common insight is that stablecoins used for payments are payment instruments; the crypto-specific regime is for trading and speculation.

Stage 1 is transit only.

The stablecoin exists for minutes during a cross-border payment. Sender pays local fiat, a licensed on-ramp converts to stablecoin, the stablecoin transits on-chain, a licensed off-ramp converts to destination fiat, and the recipient receives local fiat. No citizen holds stablecoins. No deposit substitution occurs. The license is PSP or EMI, not VASP or exchange.

Stage 2 uses four controls.

Balance caps, no yield, custodial-only holding, and aggregate reporting contain holding without banning it. The UK Bank of England consultation is the most detailed precedent, with at least 40% of backing in unremunerated central bank deposits and up to 60% in short-term government debt [10].

Without sequencing, regulators fall into the same trap.

The pattern is ignore, ban, license trading, get stuck on payments, and lose activity to regulatory arbitrage. Nigeria, India, and Turkey show the cost: less visibility, informal adoption, and a harder path back to a functional payment framework [11].

The Idea

Start with the lowest-risk use case and expand only after the supervisory infrastructure is proven.

Stablecoin regulation faces a false binary: ban everything or allow everything. Sequencing rejects both. It starts with the lowest-risk use case, transit, and adds higher-risk use cases, holding, only after the supervisory infrastructure is proven. The regulator gains data and experience at each stage before expanding the perimeter.

Diagram showing a two-stage stablecoin framework where transit permission produces evidence before limited holding rights are considered. — The two-stage framework lets regulators start with low-risk transit and expand only after Stage 1 evidence proves the system can be supervised.

Sequencing is not new.

It is how financial regulators have historically introduced new instruments. Derivatives were not deregulated overnight. Electronic trading was not permitted without circuit breakers. Mobile money in Kenya started with person-to-person transfers before expanding to savings, credit, and insurance. The principle is the same: start with the function that carries the least systemic risk. Build supervisory capacity. Expand when evidence supports it. Contract if it does not. Applied to stablecoins, Stage 1 permits the function with zero dollarization risk by design: transit. Stage 2 permits the function with containable risk: limited holding. Each stage gives the regulator data that informs the next decision. Eight jurisdictions have independently converged on this sequence [1][2][3].

Why binary thinking fails.

Regulators who wait for a comprehensive stablecoin framework before permitting any activity face a compounding problem: informal adoption grows while they deliberate. Global stablecoin payment volume reached an estimated **$5.7T** in 2024 [4]. That volume moved regardless of regulatory clarity. The question was never whether stablecoins would be used for cross-border payments. It was whether that usage would happen inside or outside the regulatory perimeter. Sequencing lets a regulator act now on the low-risk case, transit, while continuing to develop the framework for higher-risk cases such as holding, lending, and yield. Waiting for perfection means ceding the market to informal channels.

Stage 1: Transit

The stablecoin is a settlement instrument in transit, not a store of value.

In Stage 1, a stablecoin exists for minutes during a cross-border payment. Fiat enters on one side. Fiat exits on the other. The stablecoin is a settlement instrument in transit, not a store of value. Dollarization risk is **zero by design**: no citizen holds stablecoins, no deposits are substituted, and no monetary policy traction is lost [5].

Design point	Meaning
No holding	Dollarization risk stays at zero by design because users do not keep stablecoin balances.
Minutes of exposure	The stablecoin leg exists only long enough to settle the cross-border payment.
Compliance before movement	KYC/KYB, sanctions screening, Travel Rule data, and audit trail checks clear before value moves.

What Stage 1 permits: the fiat sandwich as a licensed payment service

Stage 1 covers one architecture: the fiat sandwich. 1. Sender pays in local currency. 2. Licensed on-ramp converts to stablecoin. 3. Stablecoin transits on-chain, usually seconds to minutes. 4. Licensed off-ramp converts to destination fiat. 5. Recipient receives local currency. The regulatory classification is straightforward: this is a **payment service**. The stablecoin functions as a settlement instrument, not a consumer asset. The Fiat Sandwich explains this architecture in detail. The license is a payment service provider or electronic money institution license, not a VASP or exchange license. Jurisdictions that have reached this conclusion include **Singapore**, where MAS added stablecoin activities to the Payment Services Act [3]; **UAE**, where CBUAE placed payment tokens under central bank supervision [6]; **Brazil**, where BCB classifies certain virtual asset provider activities as foreign exchange and international capital operations [7]; and **Chile**, where the Fintech Law and central bank work bring certain stablecoins into the payment perimeter [13].

What Stage 1 requires from operators: compliance before value moves

Permitting transit does not mean permitting anything. Stage 1 operators must meet the same compliance standard as any licensed payment institution.

Requirement	Standard	When it applies
KYC/KYB	Originator and beneficiary verified	Before any stablecoin is minted or transferred
Sanctions screening	OFAC, EU, and local lists	Before on-chain transfer
Travel Rule	FATF Recommendation 16 / IVMS-101 [8]	Data assembled before transfer, transmitted with or before value
Aggregate reporting	Volume, corridors, settlement times to supervisor	Periodic: daily or weekly
Audit trail	Full transaction record stored	Retained per local AML retention rules

Compliance clears **before** value moves. This is the opposite of how most peer-to-peer stablecoin transfers work today. A properly licensed Stage 1 operator produces more supervisory data than a traditional correspondent bank [9].

Stage 2: Limited Holding

Controlled expansion with built-in brakes: caps, no yield, custody, and reporting.

Stage 2 permits licensed financial institutions to hold stablecoin balances in custodial accounts, with explicit controls that contain dollarization risk. The key distinction: Stage 2 is not deregulation. It is **controlled expansion** with built-in circuit breakers.

The control design is deliberately narrow. Stage 2 does not say "let everyone hold stablecoins." It says that licensed financial institutions may hold limited balances under rules that preserve central-bank visibility and leave room to tighten or relax the system as evidence arrives.

Control	Mechanism	Precedent
Transitional balance caps	Percentage-based caps on stablecoin holdings per licensed financial institution, calibrated to average monthly payment volume	UK Bank of England proposed regime: at least 40% of backing in unremunerated central bank deposits, up to 60% in short-term government debt [10]
No yield or interest	Custodial stablecoin balances do not earn returns. This removes the better-savings-account incentive that drives deposit substitution.	GENIUS Act: payment stablecoin framework with statutory guardrails [2]
Custodial-only	Stablecoins are held only inside licensed financial institution infrastructure. No self-custodied wallets for regulated holding.	ADGM FRT: activities limited to authorised persons [1]
Aggregate reporting	Central bank receives real-time or periodic data on total holdings across licensed operators. Threshold triggers activate review.	BIS Project Aurora: cross-border supervisory data-sharing design [9]

Why no yield is the critical design choice

If a stablecoin pays yield, it becomes a deposit substitute. Citizens move savings into it. Banks lose funding. Monetary policy loses traction. The dollarization risk that Stage 1 eliminated reappears. Without yield, a stablecoin held in a custodial account is a **parking spot for payment liquidity**, not a savings product. The economic incentive to hold long term is removed. Holding remains functional, not speculative. This distinction has already become the fault line in US crypto legislation: exchanges want yield to attract deposits; banks want it banned to protect their funding base [2]. The regulatory design choice between those positions determines whether Stage 2 contains dollarization or amplifies it.

The UK model: how the Bank of England structures systemic stablecoin backing

The Bank of England's November 2025 consultation paper provides the most detailed Stage 2 architecture to date [10]. It proposes 100% backing by qualifying assets, at least **40%** in unremunerated deposits at the Bank of England, and up to **60%** in short-term sterling-denominated UK government debt. The paper also covers par redemption, temporary deviations for large unexpected redemption requests, and restrictions on reserve use. The 40/60 split is a compromise: 40% supports immediate redemption liquidity, while 60% in government debt allows the business model to generate enough return to operate. The Bank explicitly frames the proposal inside an emerging international policy direction.

The Trap Without Sequencing

Ignore, ban, license trading, get stuck on payments, then lose activity to arbitrage.

Regulators who skip sequencing and attempt to regulate all stablecoin activity at once tend to fall into a predictable five-stage pattern. Each stage makes the next one harder to escape.

Five stages of regulatory failure

**Stage 1: Ignore.** The regulator treats stablecoins as irrelevant or too small to regulate. Market adoption grows without supervision. **Stage 2: Ban.** Adoption becomes visible enough to alarm policymakers. The regulator bans crypto transactions through banking channels. Nigeria did this in 2021. India in 2018. Turkey in 2021. In every case, peer-to-peer volume increased and the central bank lost visibility. **Stage 3: License trading.** The ban proves ineffective. The regulator reverses course and licenses exchanges and trading platforms. This addresses speculation and investor protection but does nothing for payments. **Stage 4: Stuck on payments.** The regulator now has a framework for trading but no framework for payment transit. Licensed operators cannot offer fiat sandwich services because the regulatory category does not exist. The payment use case remains in informal channels. **Stage 5: Regulatory arbitrage.** Operators relocate to jurisdictions that have solved Stage 4, including Singapore, UAE, and the UK. The originating jurisdiction loses economic activity, tax revenue, and supervisory control. Sequencing avoids this trap by starting at the right point: **payments first, not trading first**.

Three countries that entered the trap and what it cost them

**Nigeria** banned crypto banking in February 2021. Peer-to-peer volume increased. Telegram groups replaced exchanges. The CBN later reversed the ban, but informal channels had already entrenched. The eNaira, launched in October 2021 as an alternative, reached 98.5% wallet inactivity within a year [11]. The country is now rebuilding regulatory architecture from scratch. **India**: The RBI banned financial institutions from servicing crypto entities in April 2018. The Supreme Court struck down the ban in March 2020, ruling it disproportionate. Two years of regulatory vacuum followed. India now has a 30% capital gains tax and 1% TDS on crypto transactions, but no dedicated payment stablecoin framework. **Turkey** banned crypto payments in April 2021 while keeping trading legal. USDT adoption continued to grow. The country processes significant stablecoin volume outside the payments regulatory perimeter. A new crypto law passed in 2024, but payment use cases remain unaddressed. In each case, the cost was the same: lost supervisory visibility, lost economic activity to other jurisdictions, and a harder path back to a functional framework.

Eight Jurisdictions, One Sequence

Evidence And Sources

This raw HTML export preserves source visibility for crawler and contractor review. Indexing decision: index, follow.