## ftnea2025001

## Source details

**Canonical URL:** [ftnea2025001](https://www.imf.org/-/media/files/publications/ftn063/2025/english/ftnea2025001.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/ftn063/2025/english/ftnea2025001.pdf.md)
- [Structured JSON version](/-/media/files/publications/ftn063/2025/english/ftnea2025001.pdf.json)

---

### Key definitions and conceptual framing
- "A digital token is "an asset or a representation of an asset on a digital ledger that is shared, trusted, and programmable.""
- Sharedness: a ledger is considered shared once it has more than one user; degree of sharedness increases as more users join or ledgers merge or interoperate.
- Trust:
  - Trust in ownership: a token is not transferable without its owner’s consent and, once transferred with such consent, recipients credence that ownership is lasting and cannot be suddenly reversed.
  - Trust in orders: instructions on the ledger will be predictably executed, confirmations will be accurate, truthful, and final.
  - Trust can be attained to different degrees; higher levels of trust tend to attract more users and increase sharedness.
- Programmability: the ledger stores code-based instructions (smart contracts) to create assets or financial applications, enforce rules (including compliance constraints), enable composability, and atomicity.

### Ledger forms, token types, and models to attain trust
- Token creation distinctions:
  - Native tokens: issued directly on a shared, programmable, and trusted ledger; exist solely on the ledger.
  - Non-native tokens: virtual representations of existing assets that exist off-ledger and are represented on the ledger.
  - Migration: replacing a non-native token and its off-ledger asset by a native token on the ledger; tokens may migrate between ledgers.
  - Ownership trust for non-native tokens depends on credence in parity between the off-ledger asset and its on-ledger representation.
- Models to attain sharedness, programmability, and trust:
  - Model (1) Single ledger: issue (native) assets on a ledger on which all (potential) transacting parties (or their intermediaries) have accounts.
  - Model (2) Common ledger: create representations of assets (non-native) on a ledger accessible to all parties (for example, escrow certificates transferable among participants).
  - Model (3) Compatible ledgers: standardize technology across ledgers so they can run the same programs and offer coordinated transfers (interoperability).
- Models to achieve trust (Box 1 continued):
  - Single-operator systems: trust stems from credibility of the operator, supervision, and legal recourse.
  - Permissionless DLT: decentralized nodes verify and validate; validator incentives via newly minted coins and transaction fees. Examples: Bitcoin and Ethereum.
  - Permissioned DLT: validators are onboarded by a central authority.
  - Technical note: the agreement algorithm is called a “consensus mechanism.”

### Risks to trust across ledger types
- Permissionless DLT risks:
  - Consensus attacks: under proof-of-work, a successful attack requires control of more than 50 percent of the computational power.
  - Transaction sequencing and fee-based prioritization enable front-running and “sandwich” attacks.
  - Oracles present vulnerabilities and account for a significant fraction of cyberattacks.
  - Imposing penalties for regulatory noncompliance is more challenging with decentralized governance.
- Permissioned DLT risks:
  - Trust may fail if network authorities modify records or programming functions at their discretion; mitigants include reputational costs and regulatory penalties.
- Centralized ledger risks:
  - Greater risk that a single authority can modify ledger state or programming functions without consensus.
  - Security depends on operator credibility and on server backups; permissioned distributed ledgers with multiple copies may achieve more security relative to a centralized ledger with no backups.
- Data governance and regulatory context:
  - Traditional trading platforms face regulatory requirements on data usage (examples cited: GDPR; California Consumer Privacy Act).
  - US regulators cited: Securities and Exchange Commission; Commodity Futures Trading Commission.

### Effects of tokenization across the asset lifecycle
- Framing: effects stem from changes in sharedness, programmability, and trust and apply across tokenization models unless otherwise noted.

A. Asset Issuance
- Current practice: registrars maintain bondholder/shareholder records; charge transaction fees.
- Tokenization effects:
  - Shared ledger: every asset is linked to owner accounts—potentially obviating separate registrars and saving issuers’ transaction costs.
  - Native tokens: may reduce need for custodian services.
  - Programmability: smart contracts can automate auction rules and embed data to reduce information search costs.
  - Example: 2022 BNP Paribas tokenized renewables project finance bond embedding term sheet and ESG data.

B. Asset Trading
- Frictions addressed: counterparty risks, search frictions, data processing costs.
- Tokenization effects:
  - Smart contracts can enable simultaneous settlement (delivery versus payment) by locking funds and assets until conditions met.
  - Programmability enables bundling multi-step transactions to minimize counterparty risk from sequential steps.
  - Limits: cannot remove counterparty risk for future-obligation contracts (derivatives); responses include:
    - Use an intermediary (CCP or trusted institution) to guarantee repayment.
    - Lock programmatic collateral for contract duration (may be costly and reduce incentives to trade derivatives).
  - Instantaneous settlement implications:
    - CCH clearing delays can currently take up to several business days (shortened to one day in the US since May 2024).
    - Faster settlement can improve capital allocation and reduce collateral lock-up times.
    - Tradeoffs: may require prepositioning of assets and funds, could limit short-selling or pre-trade financing, and may partially reveal trading history with adverse liquidity consequences.
  - Search frictions: direct investor access on shared ledgers can reduce reliance on broker-dealers in specialized or low-liquidity OTC markets and potentially reduce bid-ask spreads.

C. Asset Servicing and Redemption
- Current friction: transaction costs transmitting interest, principal, dividends due to registrar recordkeeping fees.
- Tokenization effects:
  - Shared ledger links assets to owner accounts—potentially eliminating separate owner records.
  - Programmability: embed servicing rules (specifying amounts and timing) to automate collection and distribution of dividends and interest and reduce servicing costs.

### Empirical estimates and quantitative findings (Box 2 and studies)
- Broad industry estimate:
  - Boston Consulting Group: by 2030, asset tokenization could reach "$16 trillion or 10 percent of global GDP."
- Project Agorá reference:
  - On April 3, 2024, BIS together with seven central banks announced plans to explore integration of tokenized commercial bank deposits with tokenized wholesale central bank money in a public–private programmable core financial platform (Project Agorá).
- Allen and Wittwer (2023):
  - Counterfactual welfare gains to institutional investors increase by up to 27 percent with no switching costs between brokers in a centralized ledger scenario.
- Pintér and Üslϋ (2022):
  - Estimated welfare losses due to decentralized asset trade: 7.8 percent in government bond markets; 12.2 percent in corporate bond markets.
- Onyx (J.P. Morgan) and Apollo (2023) fragment (verbatim):
  - "This is estimated to result in a 24-basis point reduction (from an average of    1.09 percent to ... 0.85 percent) in the portfolio management fee to investors (assuming a full pass-through of managers’ benefits to investors)."
- Leung and others (2023):
  - Underwriting fees of issuing a tokenized bond are, on average, 0.22 percentage points lower relative to conventional bonds.
  - Investors accept a yield spread 0.78 percentage points lower for tokenized bonds.
  - Bid-ask spreads of tokenized bonds are 0.035 percentage points lower than conventional bonds.
  - These reductions correspond to: 25.8 percent of average underwriting fees; 23.9 percent of average yield spreads; and 5.3 percent of the average bid-ask spread of matched conventional bonds.
- Liu, Shim, and Zheng (2023):
  - Yields for newly issued blockchain-based ABS are on average 25 basis points lower than for traditional ABS (variation depends on underlying asset type and familiarity of trading parties).

### Externalities, systemic risk channels, and market-structure implications
- Shock transmission mechanisms:
  - Reduced issuance cost can lower cost of leverage and incentivize higher leverage.
  - Programmability can facilitate rehypothecation via smart contracts, potentially increasing intermediary leverage.
  - Shift of retail deposit funding to tokenized financial assets could make banks more reliant on wholesale funding, which can dry up during shocks.
  - Increased sharedness raises interconnectedness through broader investor access and composed assets depending on multiple underlying assets.
  - Shared ledgers remove frictions between ledgers, increasing trading speed and velocity of shock transmission.
  - Programmability can enable automated trading behavior that increases the risk of extreme volatility (flash crashes) via chains of contingent smart contracts.
- Market infrastructure investment externalities:
  - Operating a ledger requires multilayered security; if operators bear only part of operational risk costs they may underinvest.
  - A widely shared ledger can create concentration risk and a single point of failure for cyberattacks.
  - In permissionless ledgers, larger user bases can increase settlement-asset value, attracting validators and raising security; private-ledger concentration risks remain salient.
- Network externalities and liquidity:
  - Sharedness increases liquidity by bringing more participants together, standardizing contracts, and reducing search frictions.
  - Fractionalization can enhance retail access but may introduce recovery risks on broker bankruptcy and intermediary dependence.
  - Ledger proliferation and noninteroperability can fragment liquidity and increase transaction costs; empirical liquidity bifurcation evidence exists.
- Market power and switching costs:
  - A privately owned dominant ledger could extract monopoly rents and monetize privileged data access.
  - Network effects, setup costs, and returns to scale can act as barriers to entry.
  - Shared ledgers could lower switching costs (current ACATS process: typically between three and six business days), intensify competition, and exert downward pressure on fees—but intermediaries may persist due to demand for intermediation.

### Retail investor internalities and consumer protection
- Direct ledger access can increase internalities when retail investors do not fully understand risks.
- Programmability makes constructing complex products easier, compounding internality costs, especially for derivatives.
- Permissionless DLTs complicate enforcement (e.g., reversal of fraudulent transactions), potentially increasing retail costs.
- Regulatory design and consumer protection are key to limiting internality risks.

### Policy and research implications
- Further research is needed to quantify benefits and risks associated with tokenization’s impact on market inefficiencies.
- Regulatory adaptation may be required to harness benefits and mitigate adverse effects; standard-setting bodies and financial sector regulators and supervisors are the remit for such changes.
- Key regulatory concerns highlighted:
  - Data governance and privacy compliance.
  - Operational security and incentives for adequate investment.
  - Market-power abuse by ledger owners and fee structures.
  - Consumer protection for retail participants and enforcement mechanisms for fraud and manipulation.

*Source: Annex 1. Ledger Design and Risks to Trust, from the IMF Fintech Note "Tokenization and Financial Market Inefficiencies."*

### Annex 1. Ledger Design and Risks to Trust ..............................................................................

### Annex 1. Ledger Design and Risks to Trust

### Key definitions and conceptual framing
- A digital token is "an asset or a representation of an asset on a digital ledger that is shared, trusted, and programmable."
- Sharedness: transacting parties can possess, acquire, and transfer assets on the ledger; a ledger is considered shared once it has more than one user and its degree of sharedness increases as more users join or ledgers merge or interoperate.
- Trust:
  - Trust in ownership: a token is not transferable without its owner’s consent and, once transferred with such consent, recipients credence that ownership is lasting and cannot be suddenly reversed.
  - Trust in orders: instructions on the ledger will be predictably executed, confirmations will be accurate, truthful, and final.
  - Trust can be attained to different degrees; higher levels of trust tend to attract more users and increase sharedness.
- Programmability: the ledger stores code-based instructions (smart contracts) to create assets or financial applications, enforce rules (including compliance constraints), enable composability (reuse and combination of applications), and atomicity (multiple steps executed as a single inseparable transaction).

### Forms of token creation and distinctions
- Native tokens: new tokens issued directly on a shared, programmable, and trusted ledger; they exist solely on the ledger.
- Non-native tokens: virtual representations of existing assets (for example, financial assets held by a custodian) that exist off-ledger and are represented on the ledger.
- Migration: replacing a non-native token and its off-ledger asset by a native token on the ledger; tokens may migrate between ledgers.
- Ownership trust for non-native tokens also depends on credence in parity between the off-ledger asset and its on-ledger representation.

### Models to attain sharedness, programmability, and trust (Box 1)
- Model (1) Single ledger: issue (native) assets on a ledger on which all (potential) transacting parties (or their intermediaries) have accounts.
- Model (2) Common ledger: create representations of assets (non-native) on a ledger accessible to all parties, for example by placing assets in escrow and issuing escrow certificates transferable among participants.
- Model (3) Compatible ledgers: standardize technology across ledgers so they can run the same programs (smart contracts) and offer coordinated transfers (interoperability).
- Programmability distinction: token-ledger programmability merges code and database (internal programmability), guaranteeing coherence and facilitating bundling and contingency in smart contracts; this contrasts with traditional models where external users pass instructions to a separate database operator.

### Examples and illustrative points
- Ether on Ethereum is cited as an example of a digital token deployed on a distributed ledger that is shared, trusted by users, and programmable; Ethereum supports other ledger assets, simultaneous exchanges, and composability (for example, executing an interest rate swap and a foreign exchange forward at once).
- The features (sharedness, programmability, trust) are continuous rather than dichotomic; incentives can exist to migrate tokens to ledgers offering greater degrees of these features.

### Observations relevant to risks to trust (as framed by the note)
- Trust is a necessary condition for ledger utilization; Annex 1 specifically addresses the relation between models to attain trust and the conditions under which trust can fail.
- Reversals of illicit or fraudulent transactions can be an exception to immutable-transfer assumptions where an authority with credible ability to identify fraud can reverse transactions; such oversight may increase trust for bona fide users relative to ledgers where owners alone control assets.
- Current trading platforms already exercise quasi-regulatory powers (rulebooks, market abuse monitoring, risk management) and are typically overseen by regulators, illustrating existing institutional arrangements that affect trust.

### Relevant numeric and project references from the note
- The Boston Consulting Group estimates that, by 2030, asset tokenization could reach "$16 trillion or 10 percent of global GDP."
- On April 3, 2024, the Bank for International Settlements, together with seven central banks (Bank of France, Bank of Japan, Bank of Korea, Bank of Mexico, Swiss National Bank, Bank of England, and the Federal Reserve Bank of New York), announced plans to join forces with a large group of private financial firms convened by the Institute of International Finance to explore integration of tokenized commercial bank deposits with tokenized wholesale central bank money in a public–private programmable core financial platform (Project Agorá).

*Source: Annex 1. Ledger Design and Risks to Trust, from the IMF Fintech Note "Tokenization and Financial Market Inefficiencies."*

### Box 1 (continued)

### Box 1 (continued)

### Models to achieve trust: single-operator, permissionless DLT, permissioned DLT
- Single-operator systems: a single entity is responsible to keep records of ownership, execute transactions, and confirm them. Trust stems from a mix of credibility of the ledger operators, supervision, and legal recourse.
- Permissionless DLT: relies on a decentralized network of nodes to verify and validate transactions; information on the state and programming of the ledger is shared across the network so no single operator can unilaterally alter the state or programming functions of the ledger. Validator nodes are incentivized through rewards based on newly minted coins and transaction fees. Examples: Bitcoin and Ethereum.
- Permissioned DLT: a hybrid model wherein the ledger’s validators are onboarded by a central authority.
- Technical note: in DLT systems the agreement algorithm is called a “consensus mechanism,” which allows nodes (validators) to agree on the status of the network and maintain consistent copies of a single data set.
- Conceptual tensions mentioned:
  - Budish (forthcoming) argument: for systems that operate at scale, centralization in a single operator reduces the costs of creating trust.
  - Blockchain Trilemma (Vitalik Buterin): scalability trades off with decentralization and/or secureness.
  - Databases with internal programmability (smart contracts) are well suited to both single-operator and DLT systems; external programmability complicates execution across multiple operators.

### III. Inefficiencies in Financial Asset Markets and the Role of Intermediaries
- Lifecycle stages of a traded financial asset: issuance, exchange, servicing, and redemption.
  - Issuance: creating and selling new financial assets (examples: IPOs, private placements).
  - Exchange: buying and selling on the market (secondary market).
  - Servicing: payment of interest or dividend on financial assets.
  - Redemption: returning assets to issuer in exchange for cash or other assets.
- Market frictions (subcategory of market inefficiencies): information asymmetries, search problems, transaction costs, and counterparty risks.
  - Asymmetric information: one party has more relevant information (example: IPO issuer vs potential buyers).
  - Search frictions: impediments to matching parties (example: issuer not knowing which investors are interested).
  - Transaction costs: costs of trade and opportunity cost of time involved.
  - Counterparty risk:
    - Immediate trades: risk arises from potential delay between payment and delivery enabling second-mover cancellation.
    - Derivative trades: additional counterparty risk from failure to meet obligations between contract writing and future transactions.
- Intermediaries addressing frictions:
  - Investment banks: due diligence, underwriting, acting as a “credibility bridge” to mitigate asymmetric information.
  - Central securities depositories (CSDs) and centralized clearinghouses (CCHs): address counterparty risk by recording assets, managing settlement, and only delivering assets once payment is made.
  - Central counterparties (CCPs): become counterparty to all transacting parties in derivatives; use margin, collateral, and guarantee funds to cover defaults.
- Externalities, internalities, and market power:
  - Externalities: network effects (positive) can deepen markets and increase liquidity; socialization of losses (negative) can incentivize excessive risk-taking by systemically important intermediaries with implicit/explicit bailout guarantees.
  - Internalities: uninformed or imperfectly rational agents (example: retail investors not fully grasping risk).
  - Market power: firms influencing prices or terms (example: barriers to move funds across institutions enabling higher-than-competitive fees).
- Framing for the note:
  - The note examines how tokenization may affect market inefficiencies relative to prevailing structures.
  - Effects of tokenization emanate from changes in sharedness, programmability, and trust of digital token ledgers and apply across tokenization models unless otherwise noted.
  - Tokenization may reduce some frictions more than intermediaries or make it cheaper to address frictions, but may not eliminate the need for intermediaries entirely; pass-through of cost reductions to investors depends on market competition.

### IV. Tokenization and Frictions in the Lifecycle of an Asset — qualitative effects across stages
- Objective: qualitative discussion of how tokenized ledgers change frictions versus prevailing market structures; Box 2 (not reproduced here) summarizes empirical literature.

A. Asset Issuance
- Current practice: registrars (often banks or trust companies) maintain recordkeeping of bondholders and shareholders due to economies of scale; registrars incur operational costs and charge transaction fees.
- Tokenization effects:
  - Shared ledger: no separate record of asset owners is needed because every asset is linked to the account of the owner that purchased it; registrar services may not be required, saving issuers’ transaction costs (Cohen and others 2018).
  - Native tokens: custodian services may not be required, potentially generating cost savings.
  - Programmability (smart contracts): could improve auction efficiency (example: smart contracts automatically executing auction rules once predefined conditions are met; Omar and others 2021) and enable data to be wrapped into the token asset to aid transparency and reduce information search costs for potential investors.
  - Example: in 2022, BNP Paribas distributed a renewables project finance bond in tokenized form, embedding bond term sheet and environmental, social, and governance data in the token; the bank argued this improved transparency for this small-scale project finance initiative.

B. Asset Trading
- Frictions: counterparty risks, search frictions, and data processing costs.
- Current intermediaries (CCHs, CSDs) mitigate counterparty risk but impose fees (staff, equipment, office space, and for CCPs costs associated with absorbing counterparty risk).
- Tokenization effects:
  - Shared and programmable ledgers can enable simultaneous settlement for immediate trades through smart contracts that lock funds and assets and execute the exchange only when both are ready, ensuring delivery versus payment and potentially yielding transaction cost savings (Benos and others 2022).
  - Bundling transactions: programmability enables packaging multi-step transactions into a single conditional execution to minimize counterparty risks from consecutive transactions.
  - Limits: sharedness and programmability cannot remove counterparty risk for contracts where obligations occur in the future (derivatives). Two potential routes on tokenized ledgers:
    - Use an intermediary (for example, a CCP or trusted financial institution) to guarantee repayment.
    - Lock all assets in programmatic collateral for the contract’s duration (would require providing collateral at contract initiation and locking it until expiration), which reduces counterparty risk but may be perceived as costly and reduce incentives to write and trade derivatives.
  - Instantaneous settlement: a shared ledger can enable settlement immediately once a trade is agreed, reducing transaction opportunity costs caused by synchronization delays across multiple ledgers. Example timing change: CCH clearing delays can currently take up to several business days (shortened to one day in the US since May 2024).
    - Faster settlement can improve capital allocation and reduce collateral lock-up times, aiding liquidity management.
    - Tradeoffs: instantaneous settlement may require prepositioning of assets and funds (prepositioning potentially by intermediaries), could limit short-selling or pre-trade financing, and may partially reveal trading history with possible adverse consequences on market liquidity (Lee 2021; Lee, Martin, and Müller 2022; Lee, Martin, and Townsend 2024).
  - Search frictions: shared digital ledgers with direct investor access can reduce reliance on broker-dealers in specialized or low-liquidity OTC markets and potentially reduce trading costs such as bid-ask spreads.

C. Asset Servicing and Redemption
- Current friction: transaction costs in transmitting interest, principal, or dividend payments due to registrars’ recordkeeping and fees.
- Tokenization effects:
  - Shared ledger: no separate record of asset owners is needed because every asset is linked to the account of the owner that purchased it.
  - Programmability: smart contracts can embed asset servicing rules (for example, specifying how many money tokens to service and when, conditional on funds being available), enabling automation of collection and distribution of dividends and interest and reducing servicing costs.

*Source: FINTECH NOTES Tokenization and Financial Market Inefficiencies (Box 1 continued).*

### Box 2. Estimates of Tokenization’s Impact on Financial Asset Market Frictions

### Box 2. Estimates of Tokenization’s Impact on Financial Asset Market Frictions

### Summary of empirical studies on tokenization and market frictions
- The measurement of the impact of tokenization is characterized as a novel field of research.  
- The box summarizes findings from several studies that quantify gains from financial asset tokenization.

### Allen and Wittwer (2023) — centralizing OTC bond markets (sharedness feature)
- Data and setting:
  - Use transaction level data on the trading of sovereign (federal) bonds in Canada.
  - Institutional investors’ transactions can be executed either OTC through brokers or directly on a centralized ledger that is owned by the brokers.
- Key findings:
  - Broker market power plays a central role in explaining the observed market structure.
  - Only a subset of institutional investors uses the ledger; costs of ledger use remain relatively high.
  - Counterfactual with no switching costs between brokers:
    - Welfare gains from trade to institutional investors increase by up to 27 percent.
    - Gains arise because of both increased competition between brokers and investor ledger entry.

### Pintér and Üslϋ (2022) — structural search model on UK sovereign and corporate bond markets
- Data and setting:
  - Estimate a structural search model on transaction data from the UK sovereign and corporate bond markets, which trade OTC rather than on a centralized ledger.
  - Identification uses cross-market variation of trading outcomes for clients active in both government and corporate bond markets.
- Key findings:
  - Estimated welfare losses due to decentralized asset trade:
    - 7.8 percent in government bond markets.
    - 12.2 percent in corporate bond markets.
  - Principal sources of welfare losses:
    - Settlement delays are the primary driver in the corporate bond market.
    - A combination of settlement delays and broker market power in the government bond market.

### Note on OTC market role
- Over-the-counter markets may facilitate trading in nonstandardized instruments or complex derivative transactions requiring capital allocation from broker-dealers.

### Onyx (J.P. Morgan) and Apollo (2023) — back-of-the-envelope estimate on programmability (automated portfolio deployment of cash)
- Focus:
  - Quantification of a friction reduction related to programmability: automated portfolio deployment of cash and reduced cash holdings by improving continuous reinvestment by portfolio managers.
- Reported estimate fragment (verbatim from source):
  - This is estimated to result in a 24-basis point reduction (from an average of    1.09 percent to

*International Monetary Fund — FINTECH NOTES Tokenization and Financial Market Inefficiencies (Box 2).*

### 0.85 percent) in the portfolio management fee to investors (assuming a full pass-through of managers’

### ftnea2025001 - 0.85 percent) in the portfolio management fee to investors (assuming a full pass-through of managers’ benefits to investors).

### Empirical evidence on issuance, pricing, and liquidity
- Leung and others (2023):
  - Underwriting fees of issuing a tokenized bond are, on average, 0.22 percentage points lower relative to conventional bonds.
  - Investors accept a yield spread 0.78 percentage points lower for tokenized bonds.
  - Bid-ask spreads of tokenized bonds are 0.035 percentage points lower than those for conventional bonds.
  - The reductions correspond to: 25.8 percent of average underwriting fees; 23.9 percent of average yield spreads; and 5.3 percent of the average bid-ask spread of matched conventional bonds.
- Liu, Shim, and Zheng (2023):
  - Use data on 5,000 ABS issued in China and coarsened exact matching techniques.
  - Yields for newly issued blockchain-based ABS are on average 25 basis points lower than for traditional ABS.
  - The yield differential varies depending on the type of underlying asset and the familiarity of key trading parties.
- Caveats:
  - Studies are not exhaustive of tokenization literature and focus on empirical studies of financial asset tokenization; definitions of tokenization differ across papers and this note applies its own consistent definition.
  - Some studies (notably small-sample ones) have important limitations and results should be interpreted with caution.

### Shock transmission externalities (leverage, funding liquidity, interconnectedness, speed)
- Mechanisms by which tokenization can amplify shock transmission:
  - Reduced cost of issuance can decrease the cost of leverage and incentivize higher leverage (programmable and shared ledgers reduce issuance costs; see Box 2 and Section IV.A).
  - Programmability can facilitate rehypothecation via smart contracts, potentially increasing intermediary leverage (FSB 2024).
  - If retail deposit funding declines because depositors shift to tokenized financial assets, banks could become more reliant on wholesale funding, which can dry up during shocks (Brunnermeier and Pedersen 2009).
  - Increased sharedness can raise interconnectedness through broader investor access and composed assets whose returns depend on multiple underlying assets.
  - Shared ledgers can remove frictions (“inadvertent dams”) between ledgers, increasing trading speed and the velocity of shock transmission.
  - Programmability can enable automated trading behavior that increases the risk of extreme volatility (flash crashes) through chains of contingent smart contracts (Kirilenko and others 2017).
- Implications:
  - Faster shock propagation increases difficulty for policymakers to mitigate effects and costs associated with externalities.
  - Greater interconnectedness and leverage can raise the potential for domino-default dynamics among intermediaries.

### Market infrastructure investment externalities
- Increased sharedness and programmability can raise the social costs of underinvestment in operational security:
  - Operating a ledger requires multilayered security: encryption, authentication, access controls, monitoring (Lavayssière and Zhang 2024).
  - If ledger operators bear only part of total costs from operational risk events, they may underinvest, magnifying aggregate failures.
  - A widely shared ledger can create concentration risk and a single point of failure for cyberattacks.
  - In permissionless ledgers, larger user bases can increase the value of the settlement asset, attracting validators and raising security, but private-ledger concentration risks remain salient.

### Network externalities and market liquidity
- Sharedness increases market liquidity by bringing more buyers and sellers together, standardizing contracts, and reducing search frictions, which lowers transaction costs and bid-ask spreads.
- Fractionalization of assets can enhance retail investor access:
  - Tokenized ledgers enable asset fractionalization by design; intermediaries (e.g., Robinhood, Fidelity) currently provide limited fractionalization (equities only, with limits on bonds and international equities).
  - Fractionalization can increase positive liquidity externalities but may introduce recovery risks on broker bankruptcy and intermediary dependence.
- Risks from ledger proliferation:
  - Multiple noninteroperable ledgers can fragment liquidity, increase transaction costs, and produce market segmentation—reducing sharedness and network externalities.
  - Empirical evidence of liquidity bifurcation exists (Allen and Wittwer (2023); see Box 2).

### Knowledge externalities and innovation incentives
- Shared programmable ledgers create common infrastructure that can lower costs for developers and help innovations attain critical mass.
- Open code access affects incentives:
  - Open code facilitates improvements and lowers development costs but can reduce private returns to R&D because code can be copied.
  - Licensing or patenting can protect returns but may limit open-source benefits.
- Larger shared user bases strengthen private incentives to invest in ledger-related innovation.

### Retail investor internalities
- Direct trading by retail investors on ledgers can increase internalities when investors do not fully comprehend asset risks:
  - Retail investors may forgo advisory services due to search costs, increasing exposure to products they do not fully understand and reducing diversification.
  - Programmability makes constructing complex assets easier, compounding internality costs, especially for derivatives.
  - Permissionless DLTs complicate enforcement (e.g., reversal of fraudulent transactions), potentially increasing costs to retail investors.
- Regulatory design and consumer protection are key to limiting internality risks.

### Market power inefficiencies and switching costs
- Current switching frictions (time, effort, ACATS process: typically between three and six business days) preserve broker market power and fees.
- A shared ledger could:
  - Make transfer and reconciliation instantaneous, removing clearinghouse needs and reducing operating costs.
  - Lower switching costs, intensify competition among brokers, and exert downward pressure on brokerage fees.
  - Enable direct retail trading or purchases from issuers, further increasing competitive pressure.
- Empirical counterpoint:
  - Bergquist, McIntosh, and Startz (2024) find that even when mobile-phone accessible ledgers were introduced (Uganda agricultural ledger), most small farmers continued to use intermediaries rather than trade directly, suggesting persistent demand for intermediation despite direct access.

*FINTECH NOTES Tokenization and Financial Market Inefficiencies — INTERNATIONAL MONETARY FUND*

### introduction of the ledger continue to do so afterward. Nevertheless, the trading fees charged to the small

### FINTECH NOTES Tokenization and Financial Market Inefficiencies

### Conclusion
- Definition: Digital token ledgers are defined by sharedness (to allow ownership transfer), programmability (to execute contingent transactions), and trust (participants’ confidence in system integrity).
- Analytical framework: The note uses a technology-neutral, market-inefficiency–based conceptual framework to analyze how tokenization may affect financial markets; the analysis takes existing market structures and regulations as given and applies to all models of tokenization except where explicitly noted.
- Potential positive effects stemming from improved sharedness and programmability:
  - Mitigation and reduced costs associated with frictions throughout the asset lifecycle by reducing the need for certain intermediaries.
  - Automation of procedures.
  - Mitigation of counterparty risk through simultaneous settlement.
  - Enabling faster settlement.
  - Reducing search frictions.
  - Increased innovation and liquidity externalities that could benefit market functioning.
- Potential adverse effects:
  - Facilitation of the spread of shocks across financial institutions, increasing financial stability risks.
  - Augmentation of costs associated with operational risk events.
  - Exacerbation of retail investor internalities.
  - Effects on market power (including possible extraction of monopoly rents by ledger owners).
- Research and policy implications:
  - Further research is needed to quantify benefits and risks associated with tokenization’s impact on market inefficiencies.
  - Regulatory adaptation may be required to harness benefits and mitigate adverse effects; such regulatory changes are beyond the scope of this note and are the remit of standard-setting bodies and financial sector regulators and supervisors.

*Key excerpt on market power and ledger ownership:*
- If the ledger is privately owned, owner(s) could extract monopoly rents; more shared and dominant ledgers may increase market power of owner(s).
- Privileged access to data could be monetized (for example, data sales or direct credit provision).
- Market power monetization strategies may include charging fixed or variable fees (for example, per transaction) for firms or households active on the ledger.
- A private provider could establish free access initially and, once dominant, begin to charge fees; network effects, setup costs, and returns to scale can act as barriers to entry that prevent challengers from undercutting fees.
- Additional concern: a private ledger with significant market power may lack incentives to invest in infrastructure needed for rural or poorer populations to obtain access.

### Annex 1 — Ledger Design and Risks to Trust
- Ledger models and trust sources:
  - Permissionless DLT: open access; anyone can read, write, or validate transactions without central authority approval.
  - Permissioned DLT: participants must be certified by an entity or consortium before connecting to read, write, or validate.
  - Single-operator (centralized) systems: a central authority operates the ledger.
- Permissionless DLT risks to trust:
  - Consensus attacks: feasibility of a successful attack varies by consensus mechanism; under proof-of-work, a successful attack requires control of more than 50 percent of the computational power devoted to validating the network’s transactions.
  - Transaction sequencing and fee-based prioritization create opportunities for front-running and “sandwich” attacks.
  - Oracles (services that transmit off-ledger information to smart contracts) present vulnerabilities and account for a significant fraction of cyberattacks.
  - Imposing penalties for lack of regulatory compliance is more challenging because of decentralized governance.
- Permissioned DLT risks to trust:
  - Trust may fail if network authorities modify records or programming functions at their discretion (including potential manipulation of trading or prices).
  - Disincentives against manipulation include reputational costs and regulatory measures and penalties for noncompliance.
- Centralized ledger risks:
  - Faces similar risks as permissioned DLT but arguably to a greater degree because no consensus among permissioned authorities is required to modify the ledger or its programming functions.
  - Perceived trust risks depend on the credibility of the central authority (for example, a central bank’s reputation could make the probability of revocation low).
  - Centralized ledgers generally have multiple server backups, and permissioned distributed ledgers with multiple copies may achieve more security relative to a centralized ledger with no backups.
- Regulatory context and data governance:
  - Traditional trading platforms face regulatory requirements on data usage; examples cited include the General Data Protection Regulation (GDPR) in Europe and the California Consumer Privacy Act in the United States.
  - Regulatory bodies such as the Securities and Exchange Commission and the Commodity Futures Trading Commission impose requirements for reporting trades to ensure transparency in the United States.

### Annex 2 — Comparing Native and Non-Native Tokens
- Definitions:
  - Non-native tokens: representations of assets external to the ledger; the token’s unit of account is the off-ledger asset backing them. Examples: a token representing an ounce of gold, a house, a painting, or a US dollar, each backed by the respective asset.
  - Native tokens: tokens issued only on the ledger and backed by off-ledger assets that do not serve as their unit of account. Example: a stablecoin issued only on the ledger and denominated in US dollars but backed by off-ledger assets other than US dollars, such as government bonds or commercial paper.
- Price-decoupling risks:
  - If backing assets are also traded in off-ledger markets, token prices may decouple from off-ledger asset prices.
  - A stablecoin backed by off-ledger assets that are not the currency itself can lose its peg with the US dollar (or another pegged asset), preventing redemptions at par; such de-pegging is not classified as token decoupling.
- Token decoupling classification:
  - Assets are considered non-native only if the same asset (for example, money market fund shares) is issued both on- and off-ledger and the on-ledger assets are represented by units of the off-ledger issues held in escrow.
  - If the off-ledger asset and its on-ledger representation trade at different prices, this is considered a token decoupling.

*Tokenization and Financial Market Inefficiencies NOTE/2025/001*

---


_Source: https://www.imf.org/-/media/files/publications/ftn063/2025/english/ftnea2025001.pdf_