分布式账本技术下的DLT沙盒（英文版）.pdf

EBI Working Paper Series Dirk A Zetzsche Senior Scientist, ADA Chair in Financial Law (Inclusive Finance), University of Luxembourg. Electronic copy available at: 2 PilotR 2021 Contents I. INTRODUCTION . 3 II. DLT MARKET INFRASTRUCTURES AS A REGULATORY CHALLENGE . 4 1. Terminology: DLT and Market Infrastructure . 4 2. Opportunities . 8 3. Challenges . 10 III. THE PROPOSALS REGULATORY APPROACH . 12 1. Scope: DLT Trading and Settlement . 12 2. Exemptions and Requirements . 14 3. Procedure: Specific Permission and Exemptions . 18 4. Governance: The Business Plan Approach . 18 5. PilotR vs. the SFD . 20 6. EU Passport . 22 IV. PILOTR AS THE EUS REGULATORY SANDBOX? . 23 1. Learning as a Regulatory Objective . 23 2. The PilotR Learning Process . 23 3. PilotR vis-vis Other Regulatory Sandboxes . 24 V. POLICY CONSIDERATIONS. 25 1. Objectives . 26 2. Scope . 26 3. Supervision . 28 4. Technical Inconsistencies . 29 VI. CONCLUSION . 29 Electronic copy available at: 2020 Zetzsche those accounts constitute a copy of the master copy. Intermediaries with CSD access then function as a centralized ledger for other sub-level intermediaries: this situation has been likened to a hub-and-spoke system, with the CSD as the hub, and the intermediaries as the spokes. 15 Figure 1: CSD system for securities settlement Once a transaction takes place it must be reconciled on all ledgers concerned, at the level of sub-level custodians, that of top-level custodians with CSD access and, if the overall number of shares a top-level custodian holds changes due to the transaction, the CSD as well. This typically happens after transactions are settled individually on each ledger, for instance daily at the close of business. 14 See id., at 5. 15 See David Mills et al., Distributed Ledger Technology in Payments, Clearing, and Settlement, 2016-095 Fed. Rsrv. Bd., Finance and Economics Discussion Series 6-7 (2016). Electronic copy available at: 6 PilotR 2021 If financial securities are purchased (i.e. exchanged for fiat currency), clearing and settlement takes place through a security leg and a payment leg. The CSD first verifies that the seller holds a title in the securities being sold. Then, the CSD registers the name of the new owner (the buyer) in the ledger. Thereafter, the transaction price is taken from the buyers bank account, and added to that of the seller. If the buyer and seller do not have custodial accounts at the same institution, the ledger held in the name of each of the custodians directly or indirectly (through intermediaries with CSD access) at the CSD will be updated to reflect that the custodian of the seller now has a claim on the custodian of the buyer. This process typically takes 1-3 days. 16 In the meantime, both securities and money are often held as collateral and are rendered unavailable for other economic purposes. b. DLT Market Infrastructures DLT market infrastructures are aimed at replacing the centralized ledgers in todays market infrastructures with distributed ledgers. A distributed ledger is a database that is consensually shared and synchronized across networks that are spread across multiple sites, institutions, or geographies, allowing transactions to have multiple private or public “witnesses”. 17 The data sharing results in a sequential database distributed across a network of servers, which together function as a ledger. 18 Distributed ledgers are characterized by the absence or minimal presence of a central administration and no centralized data storage. They are, hence, distributed in the sense that the authorization for the recording of a given piece of information results from the software-driven consensus of many data-storage points (nodes). Coupled with cryptographic solutions, such features curtail the risk of data manipulation, thereby solving the problem of having to trust any third-party data-storage service provider. 19 Furthermore, since all nodes run the same software code and store, in principle, the same data, all nodes have simultaneous access to the data stored, resulting in an enhanced degree of transparency. Other ledger operators can observe if one ledger undertakes to rewrite the data, hence only if the majority of them agree (which usually happens by virtue of a 16 Morten Bech et al., On the Future of Securities Settlement, BIS QUARTERLY REVIEW 67, 68 (March 2020). 17 O. Wyman see also S. Davidson, P. De Filippi P. De Filippi Benos et al., The Economics of Distributed Ledger Technology for Securities Settlement, 4 LEDGER 121, 126-27 (2019). 26 Compliance becomes difficult if anyone can take part in validating transactions. Public ledgers are difficult to reconcile with privacy, insider trading, and market abuse legislation. Permissioned systems require less extensive consensus mechanisms because the nodes are pre-approved (trusted), which enables faster and cheaper processing of transactions. See, for all, Benos et al., supra note 25, at 127-28. 27 Optimistic estimates see the cost savings potential at $20 billion annually, while estimates assuming increasing efficiencies in traditional systems result in no savings at all. See Mariano Belinky et al., The FinTech 2.0 Paper: Rebooting Financial Services, SANTANDER INNOVENTURES / OLIVER WYMAN / ANTHEMIS (2015), at 15. 28 Benos et al., supra note 25, at 130. Electronic copy available at: 2020 Zetzsche in turn, less collateral would be necessary during the settlement period. 32 However, it has also been argued that most current settlement cycles are not long because of technological limitations but because of banks back-office processes, legal arrangements, and liquidity management practices, 33 considering the fact that parties sometimes need time to rearrange the liquid assets required for the transaction. 34 c. Enhanced Resilience The DLT-specific transparency and accounting trail may enable better risk management 35 and enforcement, but its full potential may only truly be seen in the context of cybersecurity because a distributed ledger enables other nodes to continue operations even if one or more are compromised. To clarify, if there are n nodes (instead of one concentrated ledger) and e describes the effort necessary to compromise any single server, all other conditions being equal (safety per server, etc.), the effort necessary to manipulate all of the linked servers would be n x e rather than 1 x e. 29 Benos et al., supra note 25, at 129-30. 30 For example, DTCC is currently able to reduce the volume of trades for final settlement through netting by approximately 97%. See Michael Mainelli Chapman et al., Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet?, BANK OF CANADA FINANCIAL SYSTEM REVIEW 130 (June 2017). 31 For example, by streamlining reconciliation and reducing the number of intermediaries. See Bech et al., supra note 16, at 75. 32 However, the cost of overnight collateral for settlement is small relative to the total cost of trade processing. See Mainelli Mainelli Benos et al., supra note 25, at 132. 41 See Benos et al., supra note 25, at 132 (who also note that DLT protocols exist that avoid this problem). 42 For example, the DLT supporting Bitcoin stores all data except the identity of the owners, which requires a private key. 43 See Zetzsche et al., supra note 36, at 1375. Electronic copy available at: 2020 Zetzsche Zetzsche et al., Decentralized Finance, 6(2) JOURNAL OF FINANCIAL REGULATION 172, 191 (2020). 52 See BIS, supra note 35, at 13. Electronic copy available at: 12 PilotR 2021 complexity. 53 This demands that regulators be at the top of their game with regard to old and new technologies alike. Simultaneously, regulators must learn to seize new opportunities. For example, a move to DLT, with immutable records of all historical transactions, may prove immensely valuable for compliance and law enforcement if regulated appropriately. 54 Drawing analogies with the 1980s to the 2000s period is sobering, when the popular adoption of computers sparked innovations such as securitization and credit default swaps. 55 Policymakers supported these developments, which eventually facilitated the global financial crisis of 2007-09. Today, the pace of innovation is probably even faster, and the financial system is more complex, with even more being expected of regulators. To come anywhere close to accomplishing their task, regulators must team up and learn together with the innovators. Start-ups often lack the necessary experience and budget for regulatory compliance, and their business cultures tend to deviate from financial industry orthodoxy. Yet, they must somehow join forces with regulators who are often trained the old-school way. III. The Proposals Regulatory Approach PilotR is designed to achieve nothing less than the squaring of the circle by ensuring mutual learning, openness to innovation, and risk controls all at the same time. 1. Scope: DLT Trading and Settlement PilotR retains control over risk by rigorously limiting the scope of the learning environment. 56 Such limits are composed of six elements. The first element is allowing only certain actors DLT multilateral trading facilities (DLT MTFs) and DLT securities settlement systems (DLT SSSs) to benefit from the regulatory safe space. A DLT MTF is in principle an MTF as defined in MiFID II 57 (subject to further conditions). Meanwhile, a DLT SSS without stating so seems to rely on the SSS definition in Article 2(1)(10) CSDR. 53 For example, to trade a DLT-based asset for conventional fiat currency would require an interface between the DLT and the bank accounts of each participant. With so- called settlement coins Delivery versus Payment (DvP) can take place directly on the DLT, but eventually the settlement coins must be exchanged for commercial bank or central bank money. See ESMA, supra note 44, at 8, 24, 16, 61; Saule T. Omarova, New Tech v. New Deal: Fintech as a Systemic Phenomenon, 36 YALE JOURNAL ON REGULATION 735, 777-78 (2019). 54 See ESMA, supra note 44, at 11, 35. 55 Brummer in theory any back-up storage device could be understood to constitute DLT. Accordingly, the DLT definition requires some rethinking for sure, which is already underway in the draft report by the European Parliament with regard to MiCAR, 59 but oddly not in the equivalent report for PilotR. 60 The fourth element here relates to permissible activities. An MTF must admit DLT transferable securities to trading, and an SSS settles transactions in DLT transferable securities against payment. 61 Trading is covered by PilotR to unlock opportunities for DLT-based disintermediation, 62 while the inclusion of settlement is aimed at merging 58 See EUROPEAN PARLIAMENT, COMMITTEE ON ECONOMIC AND MONETARY AFFAIRS, Draft Report 2020/0267(COD) (9 th March 2021), at 48. 59 See id., at 43-44. 60 See EUROPEAN PARLIAMENT, COMMITTEE ON ECONOMIC AND MONETARY AFFAIRS, Draft Report 2020/0265(COD) (25 th February 2021), at 6-7. 61 Article 2(4) PilotR. 62 See PilotR Proposal, supra note 9, at recitals 9, 15. Electronic copy available at: 14 PilotR 2021 the trading and post-trade environments 63 a necessity for the envisaged acceleration of settlement to real-time settlement (supra, at II.2.b). Beyond trading and settlement, FMIs also execute clearing and recording, as defined by the CPMI Principles. 64 These activites are not explicitly mentioned in PilotR, yet an MTF is allowed to also record, to settle transactions, and to provide safekeeping in relation to DLT transferable securities. We can only speculate as to why clearing and recording are not covered. One possible explanation is the absence of EU legislation from which exemptions are required for the clearing of shares and bonds: the clearing obligation pursuant to Article 4 EMIR and Article 29 MiFIR extends to derivatives only. 65 In the same vein, recording is largely governed by the rules of EU/EEA Member States. 66 To allow for the smooth functioning of PilotR, Member States regulators should consider whether PilotR should be supplemented by exemptions from national legislation. Looking at the fifth element, Articles 7(1) and 8(1) PilotR limit the benefits to legal entities already authorized as investment firms and market operators under MiFID or CSDs subject to the CSDR. This means that PilotR supports innovation by incumbent firms only, and that start-ups do not qualify. It also deviates from most regulatory sandboxes worldwide, which tend to allow (or even prefer) non-authorized entities. 67 Finally, the sixth element is that the activities covered by PilotR are limited in size in two ways. On the one hand, there is a limit per issuance. Only shares in issuers with less than EUR 200 million (tentative) market capitalization and bonds (including convertible bonds) with an issue size of less than EUR 500 million may be admitted to trading on a DLT MTF and recorded on a DLT SSS (Article 3(1) PilotR). While this excludes the shares of large issuers, it does not exclude their bonds, provided that each bond issue is below the limit. Smaller issuers can be completely DLT-based, with all their shares and bonds covered by PilotR. There is also a limit per authorized entity. The total value of DLT transferable securities recorded by a CSD in a DLT SSS or by a DLT MTF operator may not exceed EUR 2.5 billion (Article 3(3) PilotR). Within these limits, large incumbents may operate DLT market infrastructures as a side business to gain experience. However, in the absence of further legislative steps, they will not be able to turn their main business into a DLT market infrastructure. 2. Exemptions and Requirements PilotR offers a scheme of exemptions and requirements. The regulatory technique here is the same both for MTFs and SSSs: for MTFs, MiFID and MiFIR apply, while for SSSs, the CSDR applies, unless the 63 See PilotR Proposal, supra note 9, at recitals 9. 64 See BIS, supra note 13, at 7. 65 See ESMA, supra note 44, at 14, 49. 66 See id., at 16-17, 62-66. 67 See Dirk Zetzsche et al., Regulating a Revolution: From Regulatory Sandboxes to Smart Regulation, 23 FORDHAM JOURNAL OF CORPORATE one relates to delegation; one relates to participants in the SSS; one relates to cash settlement and delivery-vs-payment; one relates to access to the SSS; and one relates to settlement finality (cf. Article 5(2) to (6) and 5(8) PilotR). Figure 3: Exemptions available to DLT SSSs 68 Proposal for a Directive of the European Parliament and of the Council amending Directives 2006/43/EC, 2009/65/EC, 2009/138/EU, 2011/61/EU, EU/2013/36, 2014/65/EU, (EU) 2015/2366 and EU/2016/2341 Article 6(1)(4). Electronic copy available at: 16 PilotR 2021 Given the opportunity for disintermediation through the use of DLT, the opening for DLT MTFs to vertically integrate from trading to settlement is welcome. However, it is surprising that the reverse is not catered for, as there is no exemption available t