## Introduction

## Source details

**Canonical URL:** [Introduction](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025186-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025186-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025186-source-pdf.pdf.json)

---

### Major themes covered
- Consensus Mechanisms
  - Nakamoto Consensus: PoW in the Bitcoin Network
  - Hybrid BFT / Nakamoto: PoS in the Ethereum Network
  - BFT Consensus: Proof-of-History (PoH) and Tower BFT in the Solana Network
  - Supervisory Considerations for Consensus Mechanisms
- Scalability Solutions
  - State channels
  - Rollups
  - Sidechains
  - Supervisory Considerations of Scalability Solutions
- Conclusion
- References

### Overview of DLT and blockchains
- DLT enables a single, sequenced, standardized, and cryptographically secured record of activity to be safely distributed to, and acted on by, a network of participants.
- The record can contain transactions, asset holdings, or identities.
- A blockchain is a type of DLT organizing data in a chain of blocks; blockchains can be public or private, permissioned or permissionless.
- Permissionless networks use crypto tokens to compensate validators; permissioned networks generally do not.

### Consensus mechanisms: frameworks and purpose
- Generating consensus addresses the Byzantine General’s Problem (BGP) where distributed participants need to agree on the common state of distributed ledgers.
- Two broad frameworks aim to be Byzantine Fault Tolerant (BFT):
  - Nakamoto Consensus: consensus based on which chain has the most computational work (longest chain rule).
  - Traditional BFT models: supermajority voting and agreement on the state of the network.
- The paper focuses on Nakamoto, BFT, and hybrid frameworks underpinning Bitcoin, Ethereum, and Solana.

### Nakamoto Consensus — PoW in Bitcoin (key mechanics and statistics)
- PoW solves the BGP via hashing puzzles and the longest chain rule.
- Transaction processing steps (selected):
  - Sender specifies transfer; converted into standard byte format and hashed twice to produce a transaction ID using SHA-256.
  - Sender signs the transaction using their private key.
  - Valid transactions enter a memory pool (mempool) of unconfirmed transactions.
- Miners group transactions into blocks; each block has capacity of 1MB (typically supports 2000-4000 transactions).
- Segregated Witness (SegWit) separates digital signatures outside the main block structure, allowing for more information per block and potentially doubling block capacity depending on transaction complexity (base block size remains 1MB).
- PoW hashing process:
  - Input is the block header (reference to previous block, timestamp, difficulty target, Merkle root, nonce).
  - Miners iterate the nonce until a valid hash (below a target, usually with a series of leading zeros) is found and broadcast as proof-of-work.
- Rewards and incentives:
  - Miner reward is locked for 100 blocks to ensure settlement finality.
  - Transaction fees and block rewards incentivize miners; block rewards halve every four years.
- Network timing and difficulty:
  - Difficulty adjusts every 2016 blocks (roughly every 2 weeks) to target block production roughly every 10 minutes.
- Examples and variants:
  - Litecoin and Dogecoin use Scrypt rather than SHA-256.
- Concentration and energy:
  - Four mining pools (Foundry USA, AntPool, ViaBTC, and F2POOL) currently control roughly 75% of the hashrate of the Bitcoin network.
  - Energy consumption remains a challenge; the energy mix has evolved toward renewables in some locations.
- Settlement finality (PoW example):
  - Finality often considered after 6 block confirmations; this implies roughly 1 hour on Bitcoin, about 15 minutes on Litecoin, and about 6 minutes on Dogecoin.
  - Economic infeasibility to reverse transactions can be up to 100 block confirmations (point at which coinbase reward is paid).

### Hybrid BFT / Nakamoto — PoS in Ethereum (key mechanics and statistics)
- Ethereum’s PoS (Gasper) combines Nakamoto-like selection with BFT-style finality:
  - Validators are randomly selected based on value staked and stake duration.
  - A proposer is selected, a block is proposed, and finality is reached if two-thirds of validators confirm the block.
- Key supervisory risk updates since 2022:
  - Wealth accumulation via staking: unclear whether staking yields wealth concentration; evidence is mixed and PoS implementations affect outcomes.
  - Nothing-at-stake problem mitigations: slashing is now widely used; on Ethereum penalties include a loss of 1/32 of staked ether up to 1 ether, suspension for 36 days incurring an additional penalty of approximately 0.07 ether, and a correlation penalty based on total ether of all slashed validators in the previous 36 days.
  - Locked liquidity: in some networks (e.g., Cardano) almost two-thirds of tokens are staked. Liquid staking developments:
    - Some estimates suggest roughly a third of all staked ether is locked in liquid staking protocols.
    - Liquid staking protocols can concentrate validation in a small number of professional validators.
    - Leveraged re-staking estimated leverage in the Ethereum network over recent years has been under 4 times (based on referenced performance analysis).
- Settlement finality (PoS example):
  - On Ethereum, technical finality is considered after 2 epochs (about 13 minutes).
  - Solana’s technical finality can be as little as a second.

### BFT Consensus, Proof-of-History, and Tower BFT — Solana (key mechanics)
- PoH is an ordering mechanism: sequential hashes (based on SHA-256) create an internal clock producing verifiable pre-ordering of transactions.
- Tower BFT (a TBFT variant of PBFT) leverages PoH to reach consensus with reduced communication overhead.
- Solana specifics:
  - Validators stake SOL; validator schedule is known in advance and determined by staked SOL.
  - Validators repeatedly create sequential hashes; transactions are placed within slots and use the PoH sequence as a cryptographic timestamp.
  - Consensus uses voting by supermajority; accumulated votes build a “tower” pushing a chain toward economic and legal finality.
  - The Solana network is more centralized; validators are known (by public key) and the network exhibits higher centralization relative to some other public networks.

### Supervisory considerations for consensus mechanisms (findings and risks)
- Supervisory focus:
  - Understand consensus mechanisms to develop risk-based supervision where technology may be deployed at scale or generate unique risks.
  - Focus remains on the end product/service rather than technology itself, but device-specific risks matter for monitoring and inspections.
- Key supervisory risks highlighted:
  - Market integrity risks (market manipulation, market abuse), consumer protection (digital literacy gaps), and potential financial stability risks if networks scale rapidly or increase links to wider financial markets.
  - Settlement finality is probabilistic for PoW and PoS; time to considered final varies across networks and architectures (see times above).
  - Selfish mining (PoW): dishonest miners hide mined blocks to build a private chain and later override the public chain.
  - Maximal Extractable Value (PoS and others): validators or operators can reorder transactions to extract value via front-running, back-running, sandwiching.
  - Embedded supervision, regulators-as-a-node, and similar proposals are novel; their necessity and proportionality remain uncertain given fragmentation, need for specialist expertise, limited on-chain data, and wallet pseudonymity.
- Regulatory posture and emerging differentiation:
  - BCBS categorizes crypto tokens deployed on public permissionless networks as riskier than those on permissioned networks, subject to greater prudential requirements.
  - Legislative and regulatory developments may distinguish truly decentralized “mature blockchains” from more centralized networks, affecting supervisory treatment.
  - Regulators should supervise firms providing services on networks (banks, BigTech, crypto exchanges, custodians), not blockchain networks themselves.

### Scalability solutions (state channels, rollups, sidechains) — mechanics and statistics
- Scalability trilemma: trade-offs among security, decentralization, and scalability; Bitcoin handles about 7 transactions per second (growing SegWit adoption allows potential doubling).
- Layering approach: layer 1 is the main blockchain; layer 2 protocols run operations off-chain with periodic settlement on layer 1.
- State channels:
  - Allow multiple off-chain transactions between participants with only final state settled on-chain; opening/closing channels incurs layer 1 fees.
  - Lightning Network (Bitcoin) launched to public in 2018; estimates suggest as of late 2024 almost 15% of bitcoin payments are made using the Lightning Network.
  - Raiden Network is a state channel solution on Ethereum for ERC-20 tokens.
- Rollups (most popular on Ethereum):
  - Optimistic rollups:
    - Bundle transactions off-chain; assume validity; include a challenge period during which fraud proofs can be submitted.
    - Between January 2022 and June 2024 total value locked and bridged between optimistic rollups and Ethereum was approximately $186bn.
    - Popular optimistic rollups: Optimism, Base, Arbitrum.
  - Zero-knowledge rollups:
    - Use zero-knowledge proofs (SNARKs/STARKs) to prove validity off-chain; no challenge period; users can withdraw funds almost immediately assuming proofs are sound.
    - Between January 2022 and June 2024 the total for zero-knowledge rollups was around $20bn.
- Sidechains:
  - Independent parallel blockchains interoperable with main chain via two-way pegs; issue wrapped tokens (wBTC, wETH).
  - Example: Rootstock (Bitcoin sidechain) with 30-second confirmation and Ethereum-compatible smart contract functionality; Polygon PoS is a prominent sidechain with its own token (POL).
  - Sidechains are not traditional layer 2s and can use different consensus mechanisms and participant sets.
  - Bridge vulnerabilities: Ronin Bridge hack (March 2022) resulted in the loss of over $600mn worth of crypto tokens.

### Supervisory considerations of scalability solutions (findings and risks)
- Benefits:
  - Scalability solutions improve speed and reduce transaction costs, potentially making public permissionless blockchains more viable for financial services.
- Risks and complexities:
  - Additional architectural complexity: supervisors must understand off-chain protocols and any weaker inherit security compared with main chain.
  - Security and operational resilience: sidechains and bridges introduce central points of failure and targets for cyberattacks; centralizing features of some layer 2s can increase cyber risk.
  - Market integrity amplification: central roles (operators, sequencers) in rollups can front-run transactions, exploit order flow information, and extract rents.
  - Migration and liquidity fragmentation: products or services tied to less-efficient or diminishing networks risk being stranded; listing across multiple networks can split liquidity and shallow markets.
  - Network weakening risks: a diminishing network may be more susceptible to 51% attacks and other compromises.
- Supervisory response options:
  - Monitor concentration metrics (mining pools, validator concentration, liquid staking providers).
  - Combine on-chain and off-chain data from public and private repositories as good practice.
  - Use policy sandboxes (UK Digital Securities Sandbox, EU DLT Pilot Regime) to address settlement frictions; efficacy remains to be seen.
  - Clarify regulatory perimeter where staking-as-a-service may meet securities definitions; jurisdictions vary in exemptions and clarifications.

### Conclusion and supervisory imperatives
- The growth of blockchain-based products and services, both in crypto and wider financial markets, requires supervisors to take a more active role in understanding blockchain networks.
- Complex products, growing interlinkages, and the entrance of firms with novel business models are likely to challenge supervisors’ ability to identify where innovation may be beneficial and where it might cause risks to consumer protection, market integrity, and financial stability.
- Supervisors must understand the key components and tradeoffs of blockchain networks to ensure market participants have effective risk management frameworks and that financial authorities can develop robust regulatory frameworks and conduct effective supervision.
- It is not clear whether DLT is the future of financial services or whether it will remain a technology that supports some functions rather than replacing existing financial plumbing; supervisors should actively monitor developments while allowing the private sector to drive technological development.
- Key supervisory actions emphasized:
  - Continue horizon scanning and reflection on emerging risks.
  - Use a combination of on-chain and off-chain data provided by commercial and public bodies (including regulatory reporting) to improve risk monitoring.
  - Consider policy sandboxes and product testing to improve knowledge and potentially tailor regulation for DLT use cases.
  - Be aware of tradeoffs introduced by layer 2 protocols and their implications for transparency, auditability, and commercial viability.

*IMF WORKING PAPERS — Blockchain Consensus Mechanisms: A Primer for Supervisors (2025 Update) — Introduction*

### Introduction ...........................................................................................................

### Introduction

### Major themes covered
- Consensus Mechanisms
  - Nakamoto Consensus: PoW in the Bitcoin Network
  - Hybrid BFT / Nakamoto: PoS in the Ethereum Network
  - BFT Consensus: Proof-of-History (PoH) and Tower BFT in the Solana Network
  - Supervisory Considerations for Consensus Mechanisms
- Scalability Solutions
  - State channels
  - Rollups
  - Sidechains
  - Supervisory Considerations of Scalability Solutions
- Conclusion
- References

### Key factual notes from the source (numeric fidelity preserved)
- Theoretical maximum TPS is larger in all networks than the average TPS between January and July 2025.
- Average TPS referenced covers the period "between January and July 2025".
- Fees can spike during periods of network congestion.
- The use of layer 2’s has resulted in reduced fees generally.
- Data have been generated from on-chain and commercial data providers.
- A caution: some measurements are not directly comparable given these networks may be used for different use cases. Example given: "As Bitcoin evolves into being used more as an asset class than a peer-to-peer electronic cash network, TPS becomes less important."

### Structural layout of the chapter/section
- Introduction (page or section marker shows "6")
- Detailed subsections on Consensus Mechanisms (starts at "7")
  - Nakamoto Consensus: PoW in the Bitcoin Network (7)
  - Hybrid BFT / Nakamoto: PoS in the Ethereum Network (9)
  - BFT Consensus: Proof-of-History (PoH) and Tower BFT in the Solana Network (11)
  - Supervisory Considerations for Consensus Mechanisms (12)
- Scalability Solutions (starts at "15")
  - State channels (15)
  - Rollups (16)
  - Sidechains (17)
  - Supervisory Considerations of Scalability Solutions (18)
- Conclusion (19)
- References (20)

### Implicit supervisory focus (as indicated by headings)
- Evaluation of different consensus mechanisms for supervisory relevance and risk considerations.
- Assessment of scalability solutions (state channels, rollups, sidechains) and their supervisory implications.
- Comparative measurements and data considerations for supervisors, including limits of comparability and data sources.

*IMF WORKING PAPERS Blockchain Consensus Mechanisms: A Primer for Supervisors (2025 Update), INTERNATIONAL MONETARY FUND*

### Introduction

### Introduction

### Overview of DLT and blockchains
- DLT enables a single, sequenced, standardized, and cryptographically secured record of activity to be safely distributed to, and acted on by, a network of participants.
- The record can contain transactions, asset holdings, or identities.
- A blockchain is a type of DLT organizing data in a chain of blocks; blockchains can be public or private, permissioned or permissionless.
- Permissionless networks use crypto tokens to compensate validators; permissioned networks generally do not.

### Consensus mechanisms: frameworks and purpose
- Generating consensus addresses the Byzantine General’s Problem (BGP) where distributed participants need to agree on the common state of distributed ledgers.
- Two broad frameworks aim to be Byzantine Fault Tolerant (BFT):
  - Nakamoto Consensus: consensus based on which chain has the most computational work (longest chain rule).
  - Traditional BFT models: supermajority voting and agreement on the state of the network.
- The paper focuses on Nakamoto, BFT, and hybrid frameworks underpinning Bitcoin, Ethereum, and Solana.

### Nakamoto Consensus — PoW in Bitcoin (key mechanics and statistics)
- PoW solves the BGP via hashing puzzles and the longest chain rule.
- Transaction processing steps (selected):
  - Sender specifies transfer; converted into standard byte format and hashed twice to produce a transaction ID using SHA-256.
  - Sender signs the transaction using their private key.
  - Valid transactions enter a memory pool (mempool) of unconfirmed transactions.
- Miners group transactions into blocks; each block has capacity of 1MB (typically supports 2000-4000 transactions).
- Segregated Witness (SegWit) separates digital signatures outside the main block structure, allowing for more information per block and potentially doubling block capacity depending on transaction complexity (base block size remains 1MB).
- PoW hashing process:
  - Input is the block header (reference to previous block, timestamp, difficulty target, Merkle root, nonce).
  - Miners iterate the nonce until a valid hash (below a target, usually with a series of leading zeros) is found and broadcast as proof-of-work.
- Rewards and incentives:
  - Miner reward is locked for 100 blocks to ensure settlement finality.
  - Transaction fees and block rewards incentivize miners; block rewards halve every four years.
- Network timing and difficulty:
  - Difficulty adjusts every 2016 blocks (roughly every 2 weeks) to target block production roughly every 10 minutes.
- Examples and variants:
  - Litecoin and Dogecoin use Scrypt rather than SHA-256.
- Concentration and energy:
  - Four mining pools (Foundry USA, AntPool, ViaBTC, and F2POOL) currently control roughly 75% of the hashrate of the Bitcoin network.
  - Energy consumption remains a challenge; the energy mix has evolved toward renewables in some locations.
- Settlement finality (PoW example):
  - Finality often considered after 6 block confirmations; this implies roughly 1 hour on Bitcoin, about 15 minutes on Litecoin, and about 6 minutes on Dogecoin.
  - Economic infeasibility to reverse transactions can be up to 100 block confirmations (point at which coinbase reward is paid).

### Hybrid BFT / Nakamoto — PoS in Ethereum (key mechanics and statistics)
- Ethereum’s PoS (Gasper) combines Nakamoto-like selection with BFT-style finality:
  - Validators are randomly selected based on value staked and stake duration.
  - A proposer is selected, a block is proposed, and finality is reached if two-thirds of validators confirm the block.
- Key supervisory risk updates since 2022:
  - Wealth accumulation via staking: unclear whether staking yields wealth concentration; evidence is mixed and PoS implementations affect outcomes.
  - Nothing-at-stake problem mitigations: slashing is now widely used; on Ethereum penalties include a loss of 1/32 of staked ether up to 1 ether, suspension for 36 days incurring an additional penalty of approximately 0.07 ether, and a correlation penalty based on total ether of all slashed validators in the previous 36 days.
  - Locked liquidity: in some networks (e.g., Cardano) almost two-thirds of tokens are staked. Liquid staking developments:
    - Some estimates suggest roughly a third of all staked ether is locked in liquid staking protocols.
    - Liquid staking protocols can concentrate validation in a small number of professional validators.
    - Leveraged re-staking estimated leverage in the Ethereum network over recent years has been under 4 times (based on referenced performance analysis).
- Settlement finality (PoS example):
  - On Ethereum, technical finality is considered after 2 epochs (about 13 minutes).
  - Solana’s technical finality can be as little as a second.

### BFT Consensus, Proof-of-History, and Tower BFT — Solana (key mechanics)
- PoH is an ordering mechanism: sequential hashes (based on SHA-256) create an internal clock producing verifiable pre-ordering of transactions.
- Tower BFT (a TBFT variant of PBFT) leverages PoH to reach consensus with reduced communication overhead.
- Solana specifics:
  - Validators stake SOL; validator schedule is known in advance and determined by staked SOL.
  - Validators repeatedly create sequential hashes; transactions are placed within slots and use the PoH sequence as a cryptographic timestamp.
  - Consensus uses voting by supermajority; accumulated votes build a “tower” pushing a chain toward economic and legal finality.
  - The Solana network is more centralized; validators are known (by public key) and the network exhibits higher centralization relative to some other public networks.

### Supervisory considerations for consensus mechanisms (findings and risks)
- Supervisory focus:
  - Understand consensus mechanisms to develop risk-based supervision where technology may be deployed at scale or generate unique risks.
  - Focus remains on the end product/service rather than technology itself, but device-specific risks matter for monitoring and inspections.
- Key supervisory risks highlighted:
  - Market integrity risks (market manipulation, market abuse), consumer protection (digital literacy gaps), and potential financial stability risks if networks scale rapidly or increase links to wider financial markets.
  - Settlement finality is probabilistic for PoW and PoS; time to considered final varies across networks and architectures (see times above).
  - Selfish mining (PoW): dishonest miners hide mined blocks to build a private chain and later override the public chain.
  - Maximal Extractable Value (PoS and others): validators or operators can reorder transactions to extract value via front-running, back-running, sandwiching.
  - Embedded supervision, regulators-as-a-node, and similar proposals are novel; their necessity and proportionality remain uncertain given fragmentation, need for specialist expertise, limited on-chain data, and wallet pseudonymity.
- Regulatory posture and emerging differentiation:
  - BCBS categorizes crypto tokens deployed on public permissionless networks as riskier than those on permissioned networks, subject to greater prudential requirements.
  - Legislative and regulatory developments may distinguish truly decentralized “mature blockchains” from more centralized networks, affecting supervisory treatment.
  - Regulators should supervise firms providing services on networks (banks, BigTech, crypto exchanges, custodians), not blockchain networks themselves.

### Scalability solutions (state channels, rollups, sidechains) — mechanics and statistics
- Scalability trilemma: trade-offs among security, decentralization, and scalability; Bitcoin handles about 7 transactions per second (growing SegWit adoption allows potential doubling).
- Layering approach: layer 1 is the main blockchain; layer 2 protocols run operations off-chain with periodic settlement on layer 1.
- State channels:
  - Allow multiple off-chain transactions between participants with only final state settled on-chain; opening/closing channels incurs layer 1 fees.
  - Lightning Network (Bitcoin) launched to public in 2018; estimates suggest as of late 2024 almost 15% of bitcoin payments are made using the Lightning Network.
  - Raiden Network is a state channel solution on Ethereum for ERC-20 tokens.
- Rollups (most popular on Ethereum):
  - Optimistic rollups:
    - Bundle transactions off-chain; assume validity; include a challenge period during which fraud proofs can be submitted.
    - Between January 2022 and June 2024 total value locked and bridged between optimistic rollups and Ethereum was approximately $186bn.
    - Popular optimistic rollups: Optimism, Base, Arbitrum.
  - Zero-knowledge rollups:
    - Use zero-knowledge proofs (SNARKs/STARKs) to prove validity off-chain; no challenge period; users can withdraw funds almost immediately assuming proofs are sound.
    - Between January 2022 and June 2024 the total for zero-knowledge rollups was around $20bn.
- Sidechains:
  - Independent parallel blockchains interoperable with main chain via two-way pegs; issue wrapped tokens (wBTC, wETH).
  - Example: Rootstock (Bitcoin sidechain) with 30-second confirmation and Ethereum-compatible smart contract functionality; Polygon PoS is a prominent sidechain with its own token (POL).
  - Sidechains are not traditional layer 2s and can use different consensus mechanisms and participant sets.
  - Bridge vulnerabilities: Ronin Bridge hack (March 2022) resulted in the loss of over $600mn worth of crypto tokens.

### Supervisory considerations of scalability solutions (findings and risks)
- Benefits:
  - Scalability solutions improve speed and reduce transaction costs, potentially making public permissionless blockchains more viable for financial services.
- Risks and complexities:
  - Additional architectural complexity: supervisors must understand off-chain protocols and any weaker inherit security compared with main chain.
  - Security and operational resilience: sidechains and bridges introduce central points of failure and targets for cyberattacks; centralizing features of some layer 2s can increase cyber risk.
  - Market integrity amplification: central roles (operators, sequencers) in rollups can front-run transactions, exploit order flow information, and extract rents.
  - Migration and liquidity fragmentation: products or services tied to less-efficient or diminishing networks risk being stranded; listing across multiple networks can split liquidity and shallow markets.
  - Network weakening risks: a diminishing network may be more susceptible to 51% attacks and other compromises.
- Supervisory response options:
  - Monitor concentration metrics (mining pools, validator concentration, liquid staking providers).
  - Combine on-chain and off-chain data from public and private repositories as good practice.
  - Use policy sandboxes (UK Digital Securities Sandbox, EU DLT Pilot Regime) to address settlement frictions; efficacy remains to be seen.
  - Clarify regulatory perimeter where staking-as-a-service may meet securities definitions; jurisdictions vary in exemptions and clarifications.

*IMF Working Papers — Blockchain Consensus Mechanisms: A Primer for Supervisors (2025 Update) — Introduction*

### Conclusion

### Conclusion

### Supervisory imperative and overall findings
- The growth of blockchain-based products and services, both in crypto and wider financial markets, requires supervisors to take a more active role in understanding blockchain networks.
- Complex products, growing interlinkages, and the entrance of firms with novel business models are likely to challenge supervisors’ ability to identify where innovation may be beneficial and where it might cause risks to consumer protection, market integrity, and financial stability.
- Supervisors must understand the key components and tradeoffs of blockchain networks to ensure market participants have effective risk management frameworks and that financial authorities can develop robust regulatory frameworks and conduct effective supervision.
- It is not clear whether DLT is the future of financial services or whether it will remain a technology that supports some functions rather than replacing existing financial plumbing; supervisors should actively monitor developments while allowing the private sector to drive technological development.

### Importance of consensus-mechanism understanding
- Understanding how blockchain networks generate consensus is an important requirement to improving knowledge of the products and services delivered on this technology.
- Since the last publication, some risks have remained, others have not been realized, and new risks have emerged; supervisors must continue horizon scanning and reflection on emerging risks.
- A broad understanding of consensus mechanisms enables supervisors to analyze products, services, and firms operating on the technology and take appropriate steps to protect markets and consumers to avoid a repeat of the Global Financial Crisis.

### Supervisory approaches and horizon scanning
- Embedded supervision experiments are growing but feasibility is currently unclear and may not be appropriate for most supervisors unless the market grows considerably.
- Regulatory (product testing) sandboxes have been used to improve knowledge and have fed into the development of domestic guidance and rules, but outcomes have generally been mixed.
- Policy sandboxes that aim to support new regulation tailored for DLT (rather than testing against existing rules) have shown promise and could provide solutions around settlement finality and recording ownership of products and services deployed on blockchain-based infrastructure.
- Using a combination of on-chain and off-chain data provided by commercial and public bodies (including regulatory reporting) is considered good practice to improve risk monitoring.

### Layer 2 protocols and supervisory complexities
- The growth of layer 2 protocols generates new supervisory complexities and may alter the balance of some benefits provided by large public networks (such as transparency and auditability) while making the technology more commercially viable and scalable.
- Supervisors should be aware of the various tradeoffs of layer 2s to understand their potential market impact while also allowing for development.

*IMF WORKING PAPERS Blockchain Consensus Mechanisms: A Primer for Supervisors (2025 Update)*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025186-source-pdf.pdf_