CENTRAL BANK DIGITAL CURRENCIES FOR CROSS-BORDER PAYMENTS A Review of Current Experiments and Ideas NOVEMBER 2021 FINANCE, COMPETITIVENESS & INNOVATION GLOBAL PRACTICE Payment Systems Development Group © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This volume is a product of the staff of the World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. RIGHTS AND PERMISSIONS The material in this publication is subject to copyright. Because the World Bank encourages dissemination of their knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution is given. CENTRAL BANK DIGITAL CURRENCIES FOR CROSS-BORDER PAYMENTS A Review of Current Experiments and Ideas ACKNOWLEDGMENTS This note was drafted by Biagio Bossone (Senior Payments Bank), and Holti Banka (Financial Sector Specialist, World Advisor, World Bank), Oya Ardic (Senior Financial Sector Bank). Harish Natarajan (Lead, Payments and Market Infra- Specialist, World Bank) with contributions from Ahmed structure, World Bank) provided overall guidance and co- Faragallah (Senior Financial Sector Specialist, World Bank), ordination. This note is part of a three-piece World Bank Sheirin Iravantchi (Financial Sector Specialist, World Bank), package on CBDC. The other two include: Central Bank Maria Chiara Malaguti (Senior Legal Advisor, World Bank), Digital Currency: A Payments Perspective (flagship report); Gynedi Srinivas (Senior Financial Sector Specialist, World Central Bank Digital Currency Background Technical Note. ii TABLE OF CONTENTS List of Abbreviations iv I. Context and Background 1 II. CBDCs for Cross-Border Payments 4 A. The Boc-Boe-Mas Models 4 B. The Project Inthanon-Lionrock 9 C. The R3 Models 11 D. Project Aber 13 E. Project Stella 16 F. The Architecture Of CBDCs For Cross-Border Payments: Drawing Lessons From Africa 18 III. How Do Cross-Border CBDCs Address Existing Challenges? 21 IV. Use of Cross-Border CBDCs: Legal Issues 24 A. Legal Obstacles To CBDC For Cross-Border Payments 24 V. Conclusion: “It Takes Two to Tango” 27 Annex 1. Changes And Initiatives In Cross-Border Payments 29 Annex 2. Non-CBDC Cross-Border Payment Models 30 Endnotes 33 List of Charts and Table Chart 1 Model 3a 5 Chart 2 Model 3b 6 Chart 3 Model 3c 7 Chart 4 Project Japser-Ubin—Cross-Border Transaction Approaches 9 Chart 5 Inthanon-Lionrock—Proposed Model 10 Chart 6 Central Bank-Issued Intermediate Cryptocurrency Model 12 Chart 7 Cross-Registered Intermediate Cryptocurrency Model 13 Chart 8 Project Abel—The Model 14 Chart 9 Aber Protocol—Issue Workflow 15 Chart 10 Aber Protocol—Transfer Workflow 15 Chart 11 Project Stella—Cross-Border Funds Transfers 18 Chart 12 G20 Roadmap To Enhance Cross-Border Payments 23 Table 1 How Cross-Border CBDCs Address Existing Challenges 21 iii ACRONYMS AND ABBREVIATIONS AMF Arab Monetary Fund AML/CFT Anti-Money Laundering and Combating the Financing of Terrorism ATM Automated Teller Machine AWPS Africa Wide Payments System BGP Byzantine Generals Problem BOC Bank of Canada BOE Bank of England BOT Bank of Thailand CAD Canadian Dollar CBDC Central Bank Digital Currency CPMI Committee on Payments and Market Infrastructures CRDR Cross-Registered DR CS Credit Swiss DEA4 DLT Distribute Ledger Technology DR Depository Receipt FATF Financial Action Task Force FBA Federated Byzantine Agreement FIGI Financial Inclusion Global Initiative FSB Financial Stability Board FX Foreign Exchange GCC Gulf Coordination Council GPI Global Payments Innovation HKD Hong Kong Dollar KHMA Hong Kong Monetary Authority HTLC Hashed Time Lock Contracts IIN Interbank Information Network IOI Indication of Interest ITU International Telecommunication Union KYC Know Your Customer LVTS Large Value Transfer System MAS Monetary Authority of Singapore MSME Micro Small Medium Enterprise OMS Omnibus Accounts OTC Over the Counter iv PoC Proof of Concept POS Point of Sale PvP Payment versus Payment RBC Royal Bank of Canada RTGS Real-time gross settlement RPW Remittances Prices Worldwide SADC Southern Africa Development Community SCP Stellar Consensus Protocol SDEX Stellar Decentralized Exchange SDR Special Drawing Right SGD Singapore Dollar SPV Special Purpose Vehicle THB Thai Baht TTP Trusted Third-Party UAE United Arab Emirates USC Utility Settlement Coin U-W-CBDC Universal W-CBDC WBG World Bank Group W-CBDC Wholesale CBDC XLM Stellar Lumen cryptocurrency XRP Ripple cryptocurrency I. CONTEXT AND BACKGROUND Over the years, the demand for seamless and inexpensive ity to adopt straight-through processing procedures, the cross-border payments has grown in parallel with growth high use of liquidity and the large involvement of manual in international e-commerce, remittances and tourism.2,3 operations, which result in increased cost for end-to-end Yet, cross-border payments have not kept pace with the payments processing. Finally, different regulations and intensive modernization that has characterized domestic standards set by central banks for domestic large-value payment services worldwide. Cross-border payments con- payment and settlement systems that are used for the pro- tinue to be largely based on the old correspondent banking cessing of cross-border payments create barriers for all but model, which has not quite benefited from the same flow the largest banks to join multiple systems and increase the of innovations as domestic payments have over the recent need for, and the number, of intermediaries required to decades.4, 5 This is mainly because managing change in the complete cross-border payments.6 As a result, the cost of cross-border payment and settlement space is considerably services to end-users increase. more challenging than doing so in the area of domestic pay- ments and settlements, due to the inevitable presence and Other significant challenges affect cross-border payments complex interactions of multiple jurisdictions that feature to/from emerging market economies and developing different policy and regulatory requirements, use dissimilar countries. Since the global financial crisis, banks have been standards and operating procedures, and difficulties in the reducing the number of their correspondent networks and, organization of the necessary collective action. over the past decade, cross-border correspondent bank relationships have declined by about one fifth. One of the When effecting cross-border payments, some key chal- largest drivers of this phenomenon appears to be banks’ lenges affect end-users, commercial banks and central reconsideration of their business strategy.7 Another key banks. Cross-border payments require intermediaries, and driver relates to risk considerations. As correspondent banks existing intermediaries benefit from high barriers to entry. conduct business globally, they must comply with the rel- In many cases, barriers stem from high fixed and sunk evant laws and regulations in all jurisdictions in which they costs required to interface with users, comply with regu- operate. These include anti-money laundering and com- lation, build trust in services, and operate large back-of- bating the financing of terrorism (AML/CFT) regulations, fices. In addition, size matters for these institutions: there tax transparency codes, and economic and trade sanctions. are scale economies in liquidity and risk management, net- In recent years, stringent enforcement of AML/CFT regu- work externalities are prevalent in messaging, and access lations, tax transparency requirements and economic and to multiple counterparties facilitates transactions. Restric- trade sanctions has resulted in high-profile actions and tions on operating hours and payment processing cut- penalties across the banking industry.8 Faced with higher off times make end-users experience uncertainty on the regulatory expectations, several banks have chosen to scale status of payment transactions, while lack of transparency down or stop providing correspondent services, concen- does not give them clarity on the fees that are charged trating the business in larger global transaction banks,9 the for their execution. Commercial banks, on their side, suffer so called “de-risking”.10 De-risking may threaten progress from the fragmented settlement infrastructure, the inabil- that has been achieved on financial inclusion, and also 1 2  •  Central Bank Digital Currencies for Cross-Border Payments has the potential to reverse some of the progress made had a sound reputation for managing currency issuance and in reducing remittance prices and fees, if banks close or circulation, for implementing monetary and exchange rate restrict access for money transfer operators.11 Furthermore, policies effectively, and for being in charge of national finan- the loss of corresponding banking relationships may have cial stability. precarious consequences on monetary transmission chan- nels, as small banks and payment services providers that This report discusses the use of CBDCs for cross-border are unable to bear the increased compliance costs by cor- payments.14 The report reviews the models that have been respondent banks are pushed out of the market. Finally, developed for this purpose to date and discusses critical and almost paradoxically, de-risking can frustrate AML/CFT legal issues that arise in the context of cross-border use of objectives and may not be an effective way to fight finan- CBDC. While no CBDC project has an explicit focus on pay- cial crime and terrorism financing. By pushing higher risk ments beyond the jurisdiction of the issuing central bank,15 transactions out of the regulated system into more opaque, a number of central banks are working on cross-border informal channels, they become harder to monitor (the so payment models in parallel to their CBDC efforts,16 and called “re-risking”). international cooperation among central banks on CBDC, including for cross-border payments, is intensifying.17 The Against this background, consideration has been given currently ongoing cooperative efforts, some of which will recently to whether and how new technology solutions be reviewed in this report, focus on wholesale types of applied to finance (FinTech) could reshape the cross-bor- CBDC. Their analysis, and more broadly the analysis of how der payments landscape.12 New technologies may reduce CBDC can improve the safety and efficiency of cross-bor- service shortcomings and alter market structure by favoring der payment service delivery, are within the spirit of the market platforms over intermediaries, remodeling business WBG’s recent work in the area of cross-border payments plans and firm boundaries, and encouraging entry. In fact, (Box 1). the emergence of distributed ledger technologies (DLT) as well as the entry into the market of providers like Transfer- The report is organized as follows. Section II specifically Wise, and the global payments innovation (GPI) initiative by discusses the models developed jointly by the Bank of Can- SWIFT are revolutionizing cross-border payments, enabling ada, Bank of England, and Monetary Authority of Singa- transactions to be executed within minutes (see Annex 1). In pore, those developed jointly by the Bank of Thailand and addition, the role of BigTechs and global Stablecoin arrange- Hong Kong Monetary Authority, those proposed under the ments as providers of cross-border payment services might R3 initiative, the model developed by the Saudi Arabian become prominent, due to the large network effects their Monetary Authority Central Bank of the United Arab Emir- solutions could bring that would lower transaction costs, ates and the joint undertaking by the European Central widen access, and open the possibility of complementary Bank and Bank of Japan, and draws from the experience services offered on social networking and e-commerce in Africa with cross-border payment systems lessons that platforms of global scale. Obviously, these same solutions could be used for “regional” CBDC arrangements. Section could also raise significant issues of monetary sovereignty III evaluates how cross-border CBDCs address challenges and financial stability, creating policy challenges that would of the existing correspondent banking arrangement. Sec- require the utmost attention.13 tion IV discusses the legal issues involved in cross-border use of CBDCs, and Section V concludes the report with An alternative avenue to modernize delivery of cross-bor- some general remarks. Annex 1 describes recent import- der payment services is being increasingly explored in the ant changes and initiatives in the realm of cross-border context of central banks issuing their own digital currency. payments and Annex 2 illustrates experimental models of A central bank digital currency (CBDC) could well incorpo- cross-border digital payment systems that are not based rate options and features specifically designed to execute on CBDC and discusses the outlook for cross-border pay- cross-border payments, with a view to reducing the ineffi- ments looking forward. ciencies and rents discussed above by shortening the pay- ments value chain. CBDC would have many of the same A caveat is in order as regards the scope of this report. implications on cross-border payment services and market The report is intended only to show how CBDC-based solu- structure as a hub & spoke network, but possibly with some tions can facilitate cross-border payments. To this end, the unique qualities: by virtue of being a central bank liability, report reviews and evaluates models that are being con- public trust in CBDC would be greater if the central bank sidered by the international central banking and payments Central Bank Digital Currencies for Cross-Border Payments  •  3 BOX 1  SUMMARY OF THE KEY WORK DONE RECENTLY BY THE WORLD BANK GROUP ON CROSS-BORDER PAYMENTS In the context of the work done by the Financial Sta- existing sub-regional systems- as part of the Digi- bility Board (FSB) and Committee on Payments and tal Economy for Africa (DE4A). The AWPS is seen as Market Infrastructure (CPMI), the WBG has high- critical to reduce the cost of intra-Africa transactions lighted the specific challenges faced by micro, small to less than 1% of transaction value. We had sup- and medium enterprises (MSMEs) and individuals in ported the establishment of one of the sub-regional emerging market and developing countries and has systems in the SADC region back in the early 2000’s also provided insights from the Remittances Prices and are now contributing to expanding the offering Worldwide (RPW) database: of this sub-regional system to a cross-border retail payment system. • The RPW results are contributing to bring attention to the topic of foreign exchange fees, role of compe- • The WBG is also working in the Pacific Region to tition and innovation. develop domestic infrastructure in individual island countries which will all be also interconnected in the • The WBG has highlighted the need to segment next stage. the cross-border payments landscape into: (i) large corporates; (ii) smaller local corporates, MSMEs, • The World Bank’s work on Digital ID (ID4D Initia- cross-border ecommerce and freelancers; and (iii) tive) and contributions to FATF’s new guidance on international remittances. Digital ID has relevance for cross-border payments as well. AML/CFT related compliance processes are • The WBG has also shared its experience as leader one of the key frictions in the cross-border pay- of the global effort on remittances price reduc- ment processes. Increased coverage and usage of tion, which could be useful for the global effort on Digital ID is seen as a key solution to this issue. The improving cross-border payments. World Bank’s ID4D initiative is seen as a key source More broadly, the World Bank has pioneered and con- of knowledge and research on Digital ID and fur- tinues to be involved in some relevant work on regional ther our deep country experience in implementing integration of payments and market infrastructures. ID systems would be of relevance to the work on Some examples are: cross-border payments. • Informing the Arab Monetary Fund efforts to estab- • The International Finance Corporation made a few lish the Arab Regional Payments System—which investments in entities working on cross-border seeks to bring real-time cross border payments payments—notably Earthport (divested in 2019) reducing the time to seconds from the current 2.2 and Currencycloud, which provided a platform for days and the cost from $33 to around $5. WB served cross-border payments by leveraging accounts that on the advisory committee guiding this implemen- Earthport and Currencycloud opened in multiple tation, it went live in late 2020. countries. Something like a Western Union but for more larger amounts but much smaller than pay- • Supporting the development of the Africa Wide ments of large corporates. Payments System (AWPS)—a virtual integration of community and discusses some of advantages they fea- Moreover, the report is part of a World Bank three-piece ture and challenges they raise. The report does not assess package on CBDC, supplementing a Technical Background or rank the models discussed on the observance of given Report on CBDC as well as the flagship report on Central standards or resolution of given challenges, nor does it Bank Digital Currency: The Payments Perspective. make recommendations as to which model(s) would or should be preferable. II. CBDCs FOR CROSS-BORDER PAYMENTS THE BOC-BOE-MAS MODELS of the W-CBDC model and its variants will remain indicated as Model 3a, 3b e 3c, respectively. Central banks and the payments industry are considering ways to improve cross-border systems. The Bank of Can- The descriptions below of Model 3 and its variants build on ada (BOC), Bank of England (BOE), and Monetary Authority the following basic structure. Two countries (Country A and of Singapore (MAS), in consultation with payment indus- Country B) are considered, each with its own central bank try stakeholders,18 have joined efforts to review the exist- (Central Bank A and Central Bank B) and one or more com- ing challenges and frictions that arise when undertaking mercial banks (A1, A2, etc. in Country A, and B1, B2, etc. in cross-border payments and issued a report that explores Country B). The scenario is one where Bank A1 needs to make proposals for new and more efficient models for processing a payment across the border to Bank B1, and that Bank B1 cross-border transactions—from whose contents this sec- needs ultimately to receive currency B. Bank A1 holds a set- tion draws. Specifically, the report identifies the following tlement account with Central Bank A and, similarly, Bank B1 challenges from end-users, commercial banks and central holds a settlement account with Central Bank B. Both coun- banks prospective: i) lack of transparency regarding pay- tries have their existing real-time gross settlement (RTGS) ment status, visibility and certainty of outcome; ii) limited platforms in place (RTGS platform A and RTGS platform B, availability of cross-border payment services; iii) time taken respectively) for interbank payments and settlements within for payment processing; iv) high costs associated with the their jurisdictions. In both countries a new platform is created correspondent banking model; and v) challenges associated for the issuance, exchange, redemption and cancellation of with legacy payments infrastructure across networks, central W-CBDCs (referred to below as “W-CBDC platforms”). Thus, banks and commercial banks. the W-CBDC platforms would operate in parallel with the existing RTGS platforms. W-CBDC platforms are assumed to The above report proposes three models. The first two mod- be based on DLT. Finally, the conversion of W-CBDCs denom- els (which will not be discussed here) are based on enhanc- inated in different currencies could take place through a new ing existing domestic interbank payment systems using W-CBDC-specific foreign exchange (FX) market. traditional technology.19 Without changing the underlying correspondent banking model, these two models could Some general considerations apply to Model 3 and its vari- overcome some, but not all, of the identified challenges and ants. The three variants assume a common W-CBDC infra- frictions. The third model builds on the experience from the structure as well as common solutions to address service Bank of Canada and Monetary Authority of Singapore with availability, payment visibility, and harmonized and data- research on tokenized forms of central bank liabilities for rich messaging standards. Also, to enable interoperability of domestic use cases and considers three variations based on the various systems interlinked with the common W-CBDC issuing wholesale central bank digital currencies (W-CBDCs) infrastructure, consensus will be required of the participat- as tokenized, limited-access form of central bank liabilities ing jurisdictions around a governance framework, common used for wholesale interbank payment and settlement trans- standards, oversight requirements, etc. The onus will there- actions. In order not to confuse readers that may want to fore be on the founding jurisdictions to agree on common consult the original source, in the following the numeration solutions. Furthermore, the model implies an enhanced role 4 Central Bank Digital Currency: The Payments Perspective  •  5 for the central banks to process the issuance, tracking and network in Canada using and Hashed Time Lock Contracts redemption of W-CBDCs, both within and beyond their (HTLC) to support the atomicity in transferring two assets jurisdictions. The governance structure for operating the across two separate ledgers.20,21 common W-CBDC platform will need to be evaluated and defined appropriately. Model 3 does not address the issue Model 3a of access for non-banks or smaller banks to the common Model 3a is based on currency-specific W-CBDCs where W-CBDC payment platform. Access policies will remain a these W-CBDCs can be transferred and exchanged only decision for each national central bank. Overall, the cost of within their home jurisdictions and cannot be transferred implementing any of the Model 3 variants in countries with outside their home jurisdictions. In this model, each cen- less developed financial market infrastructure may prove tral bank provides single-currency accounts (wallets) for prohibitive. This is likely to reduce the likelihood that Model W-CBDC, that’s is, denominated only in their own currency. 3 will be used by these countries to alleviate the challenges This would require commercial banks to open multiple and frictions identified of existing cross-border payment account (wallets) with multiple central banks if they wish to methods. Finally, a parallel market for the exchange of hold multiple currencies. W-CBDCs within each jurisdiction might emerge in addi- tion to the existing currency exchanges. Further analysis Description of Model 3a. Each central bank issues its own would be required to evaluate the implications of such crit- W-CBDC, against its country’s local currency, to the partici- ical development. pating banks in its own jurisdiction. The two central banks enter into an agreement whereby participating banks from The BOC and MAS have further collaborated to explore each jurisdiction maintain a W-CBDC account (wallet) with the technical architecture for two of the proposed models the central bank of the other jurisdiction denominated in (Model 3a and 3b). Under the Jasper-Ubin project, they the currency of that jurisdiction. These are single-currency have built a proof of concept (PoC) to understand the tech- accounts (wallets) and thus may hold only one digital cur- nical challenges in implementing these models. The project rency. Alternatively, a bank operation in a third country assumes that DLT-based domestic RTGS systems sit on dif- could hold W-CBDC accounts (wallets) with the two central ferent platforms in each country—the R3’s Corda platform banks and offer correspondent services to banks in the two in Canada and JP Morgan’s Quorum platform in Singapore. countries. As Chart 1 shows, Bank A1 maintains a W-CB- The project successfully implemented and demonstrated DC-A account (wallet) in Country A and similarly Bank B1 the ability to perform “atomic” transactions between a maintains a W-CBDC-B account (wallet) in Country B. On Quorum-based network in Singapore and a Corda-based the other hand, Bank C1 in Country C holds W-CBDC-A and CHART 1 MODEL 3A Country A Country B RTGS Platform RTGS Platform (Traditional)—A W-CBDC Platform—A W-CBDC Platform—B (Traditional)—B Central Bank A Central Bank A Central Bank B Central Bank B W-CBDC-A Currency B 5 1 2 Platform Currency A connectivity W-CBDC-B 4 and Bank A1 Bank A1 interoperability Bank B1 Bank B1 (W-CBDC-A account) (W-CBDC-A account) 3.i 3.ii Or W-CBDC-B 3.1 W-CBDC-A W-CBDC-A Bank C1 Bank B1 Bank C1 (W-CBDC-A (W-CBDC-A (W-CBDC-A account) account) account) 6  •  Central Bank Digital Currencies for Cross-Border Payments W-CBDC-B accounts (wallets) in Countries A and B, respec- from the use of correspondent banks. Finally, the develop- tively, and offers correspondent services to A1 and B1. ment of a Model 3a platform for cross-border payments could deliver some relief from some challenges, in particular Example of funds transfer. If Bank A1 wants to remit funds to as regards interoperability between members, transparency Bank B1, it can transfer W-CBDC-A to Bank B1’s account (wal- for users, and 24-7 availability. However, such reliefs would let) at Central Bank A. Bank B1 would then have to exchange be due to the new platform, not the use of W-CBDC per se. W-CBDC-A into W-CBDC-B. Alternatively, Bank A1 can effect an atomic, synchronized transfer of W-CBDC-A from itself to Model 3b Bank C1 and W-CBDC-B from Bank C1 to Bank B1. Model 3b is similar to Model 3a but based on currency-spe- cific W-CBDCs that can be transferred and exchanged Considerations. This solution retains a dependency on beyond their home jurisdictions. In this model, commercial intermediary (i.e., correspondent) banks for cross-border banks can hold multiple W-CBDC accounts (wallets) with payments and settlement. In fact, it is nothing more or their home central bank (e.g., a bank based in Canada can less than a tokenized form of the currently existing model. hold W-CBDC in Canadian dollars as well as pounds sterling Importantly, for the model to substantially address the chal- and Singapore dollars in a wallet with the Bank of Canada). lenges of the status quo, access to settlement accounts This would require each central bank to support multiple should be broadened to enable entities to hold W-CBDC W-CBDC tokens. wallets in each RTGS system (which remains a decision for each participating central bank to take): without broader Description of Model 3b. Central Banks A and B enter into access, the solution would end up resembling the current an agreement whereby participating banks in both coun- correspondent banking model, where banks with liquid- tries may hold and exchange the W-CBDCs issued by both ity in multiple digital currencies will offer services to those central banks with each other. This means, for instance, that without. Also, as in the traditional model, in Model 3a corre- W-CBDC-A can be held by banks in Country B and W-CB- spondent banks would need to ensure adequate funding of DC-B can be held by banks in Country A. Each participating W-CBDC accounts to be able to honor payment obligations. bank maintains multi-currency W-CBDC accounts (or wal- This requires the monitoring of positions and appropriate lets) with the central bank of its own jurisdiction. As Chart 2 balance sheet management similar to current practices: shows, Bank A1 maintains W-CBDC-A and W-CBDC-B in one trapped liquidity would remain a significant issue for banks or more account (wallets) with Central Bank A, and likewise with networks of nostro/vostro accounts. Furthermore, for Bank B1 with Central Bank B. usage of W-CBDCs does not remove the credit risk arising CHART 2 MODEL 3B Country A Country A RTGS Platform RTGS Platform (Traditional)—A W-CBDC Platform—A W-CBDC Platform—B (Traditional)—B Platform connectivity Central Bank A Central Bank A and Central Bank B Central Bank B interoperability W-CBDC-A 2 W-CBDC-B Currency B 5 Currency A 1 4 Bank A1 Bank A1 Bank B1 Bank B1 3.i W-CBDC-A W-CBDC-A W-CBDC-A account account Fx Fx Conversion Conversion 3.ii W-CBDC-B W-CBDC-B W-CBDC-B account account Central Bank Digital Currencies for Cross-Border Payments  •  7 Example of funds transfer. If Bank A1 wants to remit funds lenges might impact the willingness of a jurisdiction to join to Bank B1, it can transfer W-CBDC-B form its account (wal- the W-CBDC scheme, limiting the uptake of the solution and let) held at Central Bank A to the W-CBDC-B account (wallet) thus its overall success. In a scenario where this model were that Bank B1 holds at Central Bank A. Alternatively, Bank A1 widely adopted, there might be an increase in the com- can transfer W-CBDC-B from its account (wallet) at Central plexity of the system based on it because every participant Bank B and further transfer W-CBDC-B to the W-CBDC-B would need to hold multiple W-CBDC accounts (wallets) in account (wallet) that Bank B1 holds at Central Bank B. multiple currencies, and the technical challenges would have to be considered of synchronizing transactions across two or Considerations. In this model, holding a W-CBDC beyond more W-CBDC platforms. the home jurisdiction offers the possibility of greater effi- ciency via peer-to-peer exchange, with reduced reliance Model 3c on the correspondent banking model. However, the model requires the opening or holding of multicurrency accounts Model 3c is based on a universal W-CBDC that is backed in each RTGS—a significant departure from the status quo. by a basket of currencies and accepted by all participating Also, central banks would have to consider the impact that jurisdictions.22 Unlike models 3a and 3b do, model 3c does the creation of W-CBDCs would have on money supply and not involve the use of multiple currency-specific W-CBDCs; monetary policy when W-CBDCs are circulated in other juris- rather, it involves a single universal W-CBDC. dictions. In addition, the impact should be evaluated for par- ticipating central banks holding on their balance sheet other Description of Model 3c. Several participating jurisdictions, central banks’ W-CBDC intra-daily and potentially overnight. through either their respective central banks or a global The model would enable a settlement account holder in multilateral institution, agree to create a “universal” whole- Country A to hold a digital account (wallet) in Country B sale CBDC (U-W-CBDC) (Chart 3). The U-W-CBDC would without having to go through the on-boarding process for be backed by a basket of currencies issued by the partici- the RTGS system of Country B; participating central banks pating central banks and would be issued via an exchange would thus have to agree on a defined set of eligibility cri- specifically created to allow for its issuance and redemp- teria for this new platform. Furthermore, Bank A’s holding tion. The conversion of a jurisdiction’s currency into the of W-CBDC-B would have to be collateralized via reserves U-W-CBDC would create an exchange rate between that held at Central Bank A; an exchange rate risk would there- currency and the U-W-CBDC. A framework for how this fore emerge, which would need to be carefully managed would be managed should collectively be determined by by central banks and participants. This raises fundamental the participating central banks. questions for central banks about control over the money supply, exposure to exchange rate risk, and the relationship Example of funds transfer. There would be nothing peculiar between CBDCs and reserves. The scale of these policy chal- in the transfer modalities under this model, since transfers CHART 3 MODEL 3C Country A Country A RTGS Platform RTGS Platform (Traditional)—A W-CBDC Platform—A W-CBDC Platform—B (Traditional)—B Currency A Currency B ‘Universal’ Central Bank A Central Bank A 2 W-CBDC 6 Central Bank B exchange Central Bank B U-W-CBDC U-W-CBDC Currency B U-W-CBDC 3 7 Currency A 1 U-W-CBDC 5 Platform connectivity Bank A1 Bank B1 Bank A1 and interoperability (U-W-CBDC Bank B1 account) (U-W-CBDC account) 4 U-W-CBDC 8  •  Central Bank Digital Currencies for Cross-Border Payments would take place like in any real-time large-value payment ities. The second is to consider further the policy implica- and settlement system. Banks (and in general all entities tions of some of the more radical changes outlined in the permitted to participate in the system) would use the U-W- report, especially on monetary policy, broader access to CBDC to settle peer-to-peer cross-border transactions. central bank money, and the role of the RTGS operator in the future state. Finally, while the report focuses on change Considerations. This model seems most comprehensively to driven through revolution in the central payment infrastruc- address the identified challenges and frictions of the exist- tures, further thinking should be done on how policymakers ing arrangements. By introducing a universally accepted and and industry could work together on private-sector inno- traded W-CBDC, it offers a solution that could be more easily vation to address, in the shorter term, the challenges and implemented in many jurisdictions as it lacks many of the frictions faced by users of existing cross-border payment policy challenges outlined in Model 3a and Model 3b. Yet, services. the model raises important policy issues that might limit its feasibility. Under this model, central banks would need to Finally, particular consideration should be given to the manage and monitor the supply of funds in cash, domestic potential impact of U-W-CBDC on monetary policy and RTGS and U-W-CBDC, and there will need to be frameworks financial stability of participating jurisdictions. The critical to ensure adequate collateralization of U-W-CBDC with cen- concerns with regard to these areas, which are raised by tral bank reserves in the face of a potentially volatile intra- the prospect of using domestically CBDC issued by foreign day exchange rate. The creation of a U-W-CBDC exchange countries, would only be amplified if the prospect were that introduces a single point of failure in the model that is not of a single CBDC becoming universally available. The risk of present in the other variants of Model 3. This exchange currency substitution, in particular as a source of problem would facilitate the trading and use of the U-W-CBDC for for the domestic transmission mechanism of monetary pol- purposes other than transactions and, as a result, the U-W- icy and the stability of national financial institutions and sys- CBDC could take on the properties of a financial asset rather tems, would become material especially in jurisdictions with than those of a simple medium of exchange—speculation unsound macroeconomic policies and poorly credible policy and hoarding in particular could impact the price and hence institutions, which reflect in weak national currencies.23 the utility of such a token as a medium of exchange. The model involves a huge change and might present signifi- Further developments: The Jasper-Ubin Project cant frictions in onboarding a new currency to the basket of currencies backing the U-W-CBDC: the resulting complexity More recently, building on the 3 models just discussed, the might hamper its adoption. BoC and MAS explored how their originally CBDC projects could be developed so as to offer cross-border payment The BOC-BOE-MAS report offers only a starting point for solutions.24 In 2016, the two central banks had embarked on further analysis of the above models. Several aspects of the Project Jasper and Project Ubin, respectively, to investigate models would still require further and deeper consideration, the use of DLT for the clearing and settlement of payments including the legal basis and risks associated with each and securities. Specifically, the two projects envisioned a model, and in particular the legislative changes required to tokenized form of W-CBDC issued on blockchains by the recognize W-CBDCs as legal tender for interbank payments central bank for use by commercial banks. The two central and settlements; the cross-jurisdictional governance frame- banks then joined forces to understand how the Jasper and work required to ensure harmonized standards; the impact Ubin prototype networks, developed on different blockchain on monetary policy; and the eligibility criteria for financial platforms, could interoperate, allowing for cross-border pay- institutions and payment system participants to become ments to be settled on central bank digital currencies. Essen- direct participants in the CBDC platforms. tially, the model allows for cross-border, cross-currency, and cross-platform atomic transactions to take place through an In particular, the report identifies three areas of future intermediate account and without the need for transacting research for policymakers and industry. The first is to con- parties to hold funds with a third party. In this model, the duct further research and experimentation to better evalu- intermediate escrow account is used and operated autono- ate the different models, in particular the hypothesis that a mously as a smart contract with predefined rules, such that holistic approach to infrastructure change can deliver more no action proceeds if any preceding action fails, thus ensur- far-reaching benefits than incremental improvements to the ing the end-to-end consistency of each transaction. In the current model. This could include the creation of a technical context of cross-border payments, where the transaction solution aimed at assessing the delivery of future capabil- consists of two parts, one in a home country and one in Central Bank Digital Currencies for Cross-Border Payments  •  9 CHART 4 PROJECT JAPSER-UBIN—CROSS-BORDER TRANSACTION APPROACHES CROSS BORDER PAYMENTS Widened access to a 2.  Multiple currency support 3.  1. Intermediaries approach network (direct access) within a network (direct access • also known as asset swap • also known as direct • also known as asset transfer via intermediary access • allows for multiple currencies • needs intermediary for • does not involve an within the same network foreign exchange and intermediary • still need intermediary transfer (which could be the central banks) for transfer a foreign country, the Hashed-Timelock Contracts (HTLC)24 work in relevant areas, explore business cases and con- protocol is used to manage both parts of the transaction in nections to other platforms, and encourage participation a Payment Vs Payment (PvP) mode eliminating principal risk. of banks and other relevant parties in cross-border funds transfer trials.27 The BoC-MAS project proposes three broad conceptual design options for cross-border payments. (Chart 4). The The model first option involves using intermediaries, and the second The THB-HKD corridor network acts as a bridge to connect and third involve granting transacting parties access to the the domestic W-CBDC payment networks of the two coun- central bank’s liabilities. Access to the central bank’s liabili- tries. All the participating banks have their own nodes on ties can be achieved through two different designs. The first both the local payment network and corridor network, while design achieves direct access by granting transacting par- the two central banks have their own nodes in the local pay- ties direct access to accounts or wallets on the network. i.e., ment network and a separate node in the corridor network allowing a financial institution to hold the currency issued (called the “operator node”), whose control is shared by the by the foreign central bank; the second design allows the BOT and HKMA. In principle, a foreign currency liquidity pro- local currency to flow into foreign currency networks where vider node should exist in the corridor network in order to it can be transacted directly. This latter arrangement can be facilitate the provision of foreign currency liquidity. viewed as a multi-currency settlement system. Description of the model. In the proposed model, each central bank issues its own W-CBDC (Chart 5). The domestic THE PROJECT INTHANON-LIONROCK settlement networks (i.e., Inthanon network and LionRock Project Inthanon-LionRock is a joint initiative by the Bank of network) are separated from cross-border transactions. Thailand (BOT) and Hong Kong Monetary Authority (HKMA), Non-resident banks are not allowed to access the domestic initiated in May 2019, to explore the application of CBDC network and to hold foreign W-CBDC. Participants in the to cross-border payments.26 A Thai Baht (THB)–Hong Kong corridor network are the banks participating in the Inthanon dollar (HKD) cross-border corridor network prototype was and LionRock networks, respectively. The corridor network developed, allowing participating banks in Hong Kong and provides cross-border settlement services. These services Thailand to conduct funds transfers and foreign exchange include liquidity management processes for both local and (FX) transactions on a peer-to-peer basis, which helps foreign currencies, through a queueing mechanism, grid- reduce settlement layers. Leveraging on smart contracts, lock resolution procedures, and liquidity provision. In parallel the cross-border funds transfer process was enhanced to a with the corridor network, each central bank plays a role in real-time and atomic PvP modality. Project Inthanon-Lion- its respective domestic settlement network to facilitate the Rock was completed in December 2019 and a DLT-based conversion of W-CBDC into a special vehicle called Depos- PoC prototype was developed successfully together with ten itory Receipt (DR) denominated in domestic currency and participating banks from both places. The key findings of vice versa. To settle transactions in the corridor network, the the project were presented in January 2020, and the two DR is used for transferring value amongst all participants.28 In authorities agreed to proceed with further joint research the corridor network, participating banks may hold DR-THB 10  •  Central Bank Digital Currencies for Cross-Border Payments CHART 5 INTHANON-LIONROCK—PROPOSED MODEL Inthanon network Corridor network LionRock network THB (W-CBDC) domestic DR-on-W-CBDCs for cross-border payment on and FX-PvP HKD (W-CBDC) domestic BOT Liquidity provider Liquidity provider HKMA (Central Bank node) Operator node (Central Bank node) DR-THB DR-HKD DR-on-W-CBDCs for cross-border payment and FX-PvP TH Bank 2 TH Bank 2 HK Bank 2 HK Bank 2 node node node node TH Bank 1 node HK Bank 1 node THB (W-CBDC) HKB (W-CBDC) TH Bank 1 HK Bank 1 node node DR-THB DR-THB DR-HKB DR-HKB W-CBDC THB W-CBDC THB W-CBDC THB and DR-HKD for cross-border funds transfer and FX PvP FX transaction, there are three different ways of doing in the transactions, which are performed on a peer-to-peer basis Inthanon-LionRock model (Board Rate, Request for Quote, with finality.29 Corridor network parties include the corri- and Off-corridor Arrangement) and are discussed below. dor operator node, the participating bank nodes, and the All settle in an atomic PvP fashion, and smart contracts are foreign currency liquidity providers. The corridor operator developed to track the settlement process of the trade. node is a joint BOT-HKMA body, which is responsible to issue and destroy DR-THB and DR-HKD (in response to DR Board rate conversion request by participating banks), provide gridlock This option allows banks to seek the best FX bid-offer rate resolution services, ensure regulatory compliance. Partici- from other participants in the corridor network. A bank pating bank nodes in the corridor network initiate and set- choosing to conduct an FX transaction via the board rate tle cross-border payments and manage their own liquidity method can obtain HKD/THB rates published by market in both local and foreign currencies. The foreign currency maker banks. The system takes the best board rate avail- liquidity providers provide foreign currency liquidity when able into the FX transaction. The bank, as a market taker, will deadlock occurs. get the best rate amongst rates submitted by market maker banks. To publish the rates, the market maker banks input Example of funds transfer. Cross-border transfers of funds the HKD/THB quote into the system.31 For the market taker may involve a sending bank submitting instructions to trans- looking for a board rate, the system will automatically match fer funds in either local or foreign currency to a receiving the market taker’s board rate request with the best available bank. A transfer of funds can thus happen in one of the fol- board rate and book the FX transaction. Once matched, the lowing three types of transaction: transfer of local funds from market maker’s available amount will reduce accordingly. a local bank to a foreign bank; transfer of foreign currency funds from a local bank to another local bank; and transfer of Request for quote foreign currency funds from a local bank to a foreign bank.30 The request for quote provides banks with the option of In all cases, a sending bank submits a cross-border payment selecting specific counterparties within the corridor network. instruction and the transaction is settled simultaneously (if The bank, as a market taker, asks for a quote directly from a there are sufficient funds) or is placed in the queue (if there market maker bank in the corridor network. To request for are insufficient funds) and settled when liquidity is suffi- a quote, the market taker bank inputs the settlement details cient. The most relevant cases are those involving FX trans- which are the required amount, currency, preferred coun- actions. When a bank in the corridor network wants to do an terparty, and option of quoting an FX rate from one or mul- Central Bank Digital Currencies for Cross-Border Payments  •  11 tiple market makers. Once the market makers respond, the Further operational considerations market taker then reviews and confirms which quoted rate An issue to be addressed concerns the expansion of the net- it wants to execute. work to include other foreign currencies or central banks. The proposed corridor network model allows for flexible Off-corridor arrangement expansion. Yet, integration issues from the technical and This option provides an alternative way of FX dealing out- operational perspectives will have to be explored for con- side the corridor network between participating banks and necting the corridor network to other systems. There will non-participating (off-corridor) banks. Once the FX rate be governance issues of how to set up the multi-currency has been agreed upon, the transaction can be settled via a DR issuing node’s entity in terms of governing laws, data participating representative bank of the non-participating privacy, and node’s location. In addition, roles and respon- banks in the corridor network. The representative bank and sibilities will need to be deliberated in relation to the devel- the counterparty bank both input the transaction details in opment of new functionalities (e.g., for liquidity provision in the system to settle the transaction. The two transactions the system through cross-currency repo operations). entered by the two transacting parties must be matched using the same reference number and transaction details. Further technical considerations The matched transaction then proceeds to settlement. If the Performance, scalability, security and operational resilience transaction details do not match, the deal is rejected. are key concerns for facilitating real-time payments and onboarding new participating members. Under the PoC Considerations. The PoC has shown that, with the use of testing scenario, it was found that if a bank holding a cer- CBDC and the connection of the Inthanon, LionRock and tain amount of cash tokens needed to make payments to corridor networks, cross-border payments can be com- different banks at the same time, the payment transactions pleted within seconds without intermediaries or settlement would occur in a sequential order whereby one transaction layers (assuming that the sending bank has sufficient liquid- chain must be completed before the other(s) can be exe- ity in the corridor network). The PoC has shown additional cuted. Consideration will have to be given to parallel com- important benefits in terms of enhanced liquidity efficiency, puting and optimization processes for the transfer of cash better regulatory compliance and improved reporting abil- tokens without order dependency. Security is another area ity, and scalability. These benefits, however, are the effects of which needs to be addressed. DLT-related controls must be the technology solutions adopted, not necessarily of CBDC implemented to mitigate risk. For example, private keys, use. The PoC has also indicated that the model requires which are generated and stored in DLT nodes to identify supplementary analysis from a number of important angles nodes and sign transactions, should be integrated with the (legal and regulatory, operational, and technical), in prepara- hardware security module and managed properly to avoid tion of moving the project to next stage of developing a full being compromised. Finally, high availability is related to the production-grade system in a sandbox environment. capability for disaster recovery, which requires the presence of a procedure to handle different degrees of system com- Further regulatory considerations ponent failures. Future work may include exploration of new Particular attention will have to be placed on the model’s high-availability deployment configuration options for the legal basis in the presence of different national regulatory nodes and higher capability to monitor running flows in the frameworks. An ongoing periodic revision of existing reg- system. ulations will be necessary to make sure that the proposed model design is consistent with the ever-changing regula- tory environment comprising Thai, Hong Kong and global THE R3 MODELS32 regulations (any change in the regulatory environment will raise compliance issues in the proposed model) and con- Recently, R3 conducted a study intended to stimulate sideration will have to be given to the possible harmoniza- ideas for new, different approaches to improve wholesale tion of the different regulatory frameworks. Also, the model cross-border settlement using DLT.33 Three of the proposed should integrate AML/CFT standards as well as mechanisms models assume active involvement of central banks: in the to deal with legal claims to central banks and data integrity first model (Option 1) central banks issue their own CBDC; and privacy concerns. Another legal dimension to consider is and in the second model (Options 2), the central banks settlement finality: prior to rolling out the PoC into produc- coordinate to issue a dual-registered digital currency for a tion, settlement finality should be clearly defined in the legal specific currency pair; the third model (Option 3) involves and operational sense. credit lines. Only the first two models are discussed below, 12  •  Central Bank Digital Currencies for Cross-Border Payments since their further development was deemed to be particu- Example of funds transfer. Using the example from the R3 larly useful. Conversely, the third model significantly departs study, assume Canada and Singapore issue, respectively a from the traditional role of the central bank and has the Canadian-dollar CBDC (CAD) and a Singapore-dollar CBDC potential to dramatically expand the central bank’s balance (SGD). Royal Bank of Canada (RBC) needs to pay Credit Swiss sheet and influence the money supply; thus, it will not be (CS) in SGD (Chart 6). Both CAD and SGD are intermediate discussed in the following. Similarly, the other model pro- cryptocurrencies in the language of the R3 study (W-CB- posed by the R3 study (based on a trusted third party) will CDs). As RBC seeks to buy SGD in exchange for CAD, which not be considered in this report but will only be briefly refer- RBC holds on its account (wallet) with BOC, exchange pro- enced below. The R3 study evaluated each model based on viders or brokers match RBC’s indication of interest (IOI) and its monetary supply implications, impact on liquidity man- return a quote. Once RBC selects and accepts a quote, the agement for commercial banks, settlement risk, credit risk, FX transaction is executed and the transfer of SGD to CS is and complexity for central banks. then effected. Transaction settlements take place through omnibus accounts (OMS) held by RBC, CS and the dealers Underlying the two models illustrated below is the concept (brokers) at BOC and MAS.36 Equivalently, RBC could pay CS of “intermediate cryptocurrency,” introduced by the study. using CAD and CS would convert CAD on SGD through the An intermediate cryptocurrency is a W-CDBC adopted dealers (broker) network. worldwide, which can thus flow freely across borders. The ledger where the intermediate cryptocurrency is issued Considerations. This option places a heavy burden on cen- provides open access such that foreign financial institutions tral banks as it requires their ledgers to be opened to a much can hold accounts on it. Cross-border transactions would broader range of financial institutions than is currently the therefore take the form “Currency A ⟶ Intermediate cryp- case. The legal and policy frameworks of most jurisdictions tocurrency ⟶ Currency B,” which would allow for greater would be incompatible with this model and would require efficiency since the intermediate cryptocurrency can be significant adaptations. transferred directly from the sender’s account to the receiv- er’s account. DLT and smart contracts support a decentral- Option 2: Cross-Registered Intermediate ized PvP solution for settling FX transactions. Cryptocurrency This option proposes the use of a “depository receipt” that Option 1: Central Bank-Issued Intermediate can be circulated in multiple currency jurisdictions. A depos- Cryptocurrency34 itory receipt is issued by a central bank in the form of dig- This option assumes that intermediate cryptocurrencies ital tokens and confers on its holder the title to receive (a are all issued on an interoperable, common ledger.35 With claim on) an equivalent net balance of central bank-issued a common ledger, the need for inter-ledger transactions is currency payable on redemption by the central bank. A effectively eliminated. The more currencies are issued on cross-registered depository receipt (CRDR) consists of cen- an interoperable ledger, the less likely a given intermediate tral bank-issued digital tokens that are backed by collateral cryptocurrency needs to be converted into off-ledger assets. held at the central banks of issue.37,38 A CRDR that is issued CHART 6 CENTRAL BANK-ISSUED INTERMEDIATE CRYPTOCURRENCY MODEL RBC account BOC OMS linked Dealer network OMS linked MAS CS account CAD account or exchange account SGD → FX 101 ← FX 101 Sell CAD/Buy SGD Sell SGD/Buy RBC CS +SGD, –CAD + CAD, –SGD Central Bank Digital Currencies for Cross-Border Payments  •  13 CHART 7 CROSS-REGISTERED INTERMEDIATE CRYPTOCURRENCY MODEL Retire redeemed CRDR SGD Issue CRDR MAS BOC CAD CAD SGD Redeem CRDR Redeem CRDR CDCR CAD SGD RBC –CAD, + SGD to Bank A1 by Central Bank A against its own currency A can the FX risk is transferred to central banks. The hedging be redeemed at Central Bank B against its own currency B at cost on the central banks might in turn be priced in as a a predetermined exchange rate.39 By providing an alterna- premium on the CRDR or as haircut on the collateral paid tive—widely accepted—settlement asset, Option 2 is useful against CRDR issues. Second, the model might be difficult to in those cases where transactions involve currency pairs that scale up as it requires each pair of central banks to commit are not heavily traded, for which the market is therefore thin to honoring two CRDRs. Third, the model would introduce and a counterparty to an FX transaction cannot be found arbitrage opportunities for CRDR holders as the foreign easily. CRDRs may eliminate the need for currency exchange exchange market dynamics might cause actual exchange and hence the need for finding a counterparty to the cur- rates to diverge from the predetermined rates applied on rency exchange in a thin market. CRDRs. Fourth, as exchange activities using CRDR are con- ducted in a decentralized way, they might be anonymous. Example of funds transfer. Taking the same example of Option 1, RBC needs to pay CS (Chart 7). RBC can make the The R3 study proposes also a model that is based on a payment using CRDR issued by BOC, which CS can redeem trusted third-party (TTP) issuer central bank. Since the TTP at MAS against SGD; in turn, MAS redeems CRDR at BOC in model (in its various variants) does not envisage the adop- exchange for SGD. As an alternative, RBC can redeem CRDR tion of CBCD solutions, it will not be discussed here and holdings against SGD at MAS and pay CS in SGD. interested readers are referred to the R3 study for illustration of the model.40 Considerations. A major advantage of the CRDR as settle- ment asset for bilateral trading is that the instrument can PROJECT ABER flow freely across the borders of the two currency jurisdic- tions involved in the exchange. In the best case, only one In early 2019, the Saudi Arabian Monetary Authority and ledger would be involved for a single trade if the CRDR is the Central Bank of the United Arab Emirates announced used as settlement asset. One challenge arising from using in a joint statement the launch of their pilot Project Aber CRDR is that money supply in the issuing and receiving (after the Arabic name that stands for “one who crosses jurisdictions might be subject to erratic fluctuations due boundaries”). The vision of the project is to create a CBDC to the cash-in-and-out preferences of traders, since issuing instrument that can be used for settlement of cross-border CRDRs denominated in a currency shrinks the supply of the payment obligations between commercial banks in the two currency while the opposite happens when CRDR holders countries as well as domestically. The initiative aims to imple- redeem their holdings. In other words, the imbalances of ment a proof of concept for, studying, understanding, and cross-border transactions are reflected as instability of the evaluating the feasibility of issuing wholesale CBDC, with a money supply. There are other risk implications as well. First, view to reducing transfer times and costs between banks, in the holder of CRDR is exempted from the foreign exchange addition to experiment the direct use and actual application risk because the exchange rate is predetermined; however, of technologies such as the distributed ledgers. Initially, the 14  •  Central Bank Digital Currencies for Cross-Border Payments joint venture for digital currency will be restricted to bank- CHART 8 PROJECT ABEL—THE MODEL ing institutions and will not be open for public usage. A few selected commercial banks (three Saudi and three UAE) were selected to participate in the development of the currency. In November 2020, the two central banks issued a report on 1 Pledge the status of the project, which documents the solutions, collateral results, and main lessons learned through the pilot stage.41 The model 6 2 Destroy Generate currency currency The Abel model is based on a DLT called Hyperledger Fab- ric (HLF).42 HLF is a permissioned blockchain technology, with a pluggable architecture that allows “plug and play” of important components such as consensus and member- ship service. This technology allows for two types of peers: endorsers and committers. Execution of smart contract or 5 3 Redeem Full bank’s chain code is limited to endorsing nodes, while committers currency for account only maintain the ledger. cash The HLF technology allows for the system to be as decen- 4 tralized as possible. The purpose is to enable commercial Exchange with counterparty banks to settle with each other even in cases where the cen- tral bank is unavailable or disconnected from the network. The rationale behind this is for the system to offer a higher level of architectural resilience than traditional centralized rimary channel: All the banks (commercial as well as • P systems, which depend on the availability of centralized ser- central) participate in this channel (ledger). The endorse- vices, and thus to avoid a single point of failure. Given that ment policy on this channel requires at least 5 (out of the two national currencies (i.e., the Saudi Arabia Riyal and the total 8) participants to endorse transactions (with at the UAE Dirhams) are pegged to the USD, the digital cur- least 2 of them from each jurisdiction). CBDC is issued rency will be pegged as well and the exchange rate or con- by central banks through a special “issue” transaction on version rate for fiat currency to CBDC will be fixed. Thus, the the primary channel. same currency will be used for both domestic and cross-bor- ilateral channels: these are the peer-to-peer channels • B der transactions and will be transferrable and redeemable (ledgers) between each pair of commercial banks. The both domestically and cross-border. Moreover, each central central bank of each peer also participates in the chan- bank’s issuance can be used/redeemed in the other juris- nel. The endorsement policy on this channel requires diction; thus, each central bank will be able to see the total only the two commercial banks to endorse a transaction. amount generated and issued by the other central bank and, This allows a payment to be confirmed even when one therefore, it will have full visibility to all CBDC issued in the or both central banks are unavailable. network. Since CBDC will remain a liability of the issuing bank, regardless of the jurisdiction of redemption, the sys- rivate channel: A private channel (ledger) between a • P tem will support settlement between central banks when commercial bank and its central bank. This is used for redeeming cross-border issued CBDC. making private requests such as CBDC issue and redeem requests. The endorsement policy on this channel While all CBDC issued goes through this cycle, step 4 can be requires both the bank and the central bank to endorse repeated as many times as needed as CBDC is exchanged a transaction. between participants. In step 6, issued CBDC is destroyed by the central bank, as part of a redemption request in which Description of the model. The Aber model involves a life- converted back to cash and deposited back with the com- cycle process of CBDC, from issuance to destruction, which mercial bank. Aber uses three types of channels for transfer- involves the following steps (Chart 8): ring funds: commercial bank pledges cash collateral in an account • A held by the central bank in its jurisdiction Central Bank Digital Currencies for Cross-Border Payments  •  15 • T he central bank converts cash collateral to generate Example of CBDC issue. As bank A makes an issuance CBDC request, the process can be described as follows (Chart 9): he central bank funds the newly created currency in the • T sing the bilateral channel (ledger), bank A and bank B • U commercial bank’s account on the ledger generate a pseudonym each, against which CBDC will be issued. Bank B also issues a consent agreeing to full own- he commercial bank transfers the new currency to an • T ership by bank A account belonging to the counterparty on the ledger sing the private channel (ledger), bank A requests the • U he counterparty redeems the currency for cash collat- • T central bank to issue shares (slices) of CBDC on its bilat- eral via the central bank in its jurisdiction eral channel with bank B he central bank that has facilitated redemption returns • T he central bank issues the shares (slices) of CBDC in the • T the redeemed CBDC to the issuing central bank primary channel (ledger) The issuing central bank destroys the created CBDC. • ank A requests allocation of the CBDC shares to its bilat- • B eral channel (ledger) with bank B. CHART 9 ABER PROTOCOL—ISSUE WORKFLOW A Ledger: (A,B) B Ledger: (A,CB) CB Primary RTGS 1: Consent request 2: Consent Resp 3: Slice request 3.1: Debit/RTGS A/C Debit 4: Issue slice 5: Add slice CHART 10 ABER PROTOCOL—TRANSFER WORKFLOW A Ledger: (A,B) B CB Primary 1: Payment 1 1: Payment 2 1: Payment N 2: Initial settlement 3: Settlement request 4: Settlement response 5: Revoke consent 6: Settlement end 16  •  Central Bank Digital Currencies for Cross-Border Payments Example of funds transfer. Aber uses a distinctive trans- there could be a differential opportunity cost imposed for action flow that goes through the steps of endorsement, banks holding CBDC where a higher interest is paid. Another ordering and validation. When bank A wants to make a pay- issue concerns finality, since both parties involved in a trans- ment to bank B, the process works as follows (Chart 10): action are legally accountable for the transaction, the two central banks should give unambiguous direction to deter- anks A invokes the chain code (smart contract in the • B mine when finality would occur in the system. Yet another HLF language) on the bilateral channel (ledger) between issue is about CBDC redemption, in light of the dual-issued bank A and bank B to activate a transfer of CBDC to B. nature of the currency: since the funding of redemptions will In case of insufficient balance, payments get processed take place at the Nostro account that each central bank holds in an on-chain queue. Bilateral netting is implemented; with the other, a bilateral agreement should be required thus, before adding a payment to a queue, netting between the two central banks on how overdrafts would be opportunities with payments queued on the other side handled and on how each central bank would be expected are explored to replenish funds when acquiring the other currency. he settlement cycle gets triggered. This could be after • T every transaction, and would be based on time, number of payments or movement in positions/balances, follow- PROJECT STELLA ing the last settlement Project Stella was born as a joint research undertaking by ank A issues consent with B based on the new shares • B the European Central Bank (ECB) and the Bank of Japan (slices) of CBDC to be transferred (BOJ). Launched in December 2016, the project aimed to contribute to the ongoing debate with experimental work onsents previously issued by bank A on those slices of • C and conceptual studies exploring DLT’s opportunities and CBDC need to be invalidated by bank B. This operation is challenges for financial market infrastructures. In 2018, the performed on the primary channel (ledger). two central banks built on the insights gained from the ear- lier stages of the project to explore innovative solutions for Considerations. A single digital currency moving across a sin- cross-border payments, i.e., payments between currency gle network for settlement of cross-border payments would areas.45 Stella Phase 3, thus, studied whether cross-border be especially useful in a region like the GCC [Gulf Coordi- payments could be improved, especially in terms of safety, nation Council] states, where there is substantial intra-re- by using new technologies. In particular, it studied a led- gional trade and movement of citizens and residents. Also, ger-agnostic protocol that synchronizes payments across in the Aber system, the movement of funds would occur in different types of ledgers and assessed the safety and effi- real time, with no need for commercial banks to hold corre- ciency implications of a variety of payment methods that spondent bank Nostro account in each country. This would could be used in the cross-ledger payment. These models address the inefficiencies in the existing correspondent show distinctive characteristics, which can be summarized banking-based payment systems, which often results in as follows: i) whether individual payments are settled on-led- delays and require commercial banks to maintain substan- ger or recorded off-ledger, ii) whether funds are locked or tial Nostro account balances with correspondent banks. This escrowed, iii) whether payments are enforced when the issue, characterized as “trapped liquidity,”43 currently causes predefined condition for the payment is fulfilled, and (iv) significant opportunity and compliance costs for commer- whether specific ledger functionalities are required to con- cial banks. Also, by requiring that only counterparties to the duct transfers, such as the functionalities to enforce condi- payment transactions need to be online for payments to be tional transfers and process signed claims. settled,44 the Aber system offers a higher level of operational resilience than traditional centralized systems, which depend Description of the models. The different payment methods on the availability of the central bank and to avoid single are the following: point of failure. Finally, in point of visibility, cross-border pay- ments will be auditable by both central banks. On the other rustline: This model involves an arrangement between • T hand, the project still needs to address some critical issue. the payer and the payee outside the ledger where the One concerns the impact that in a context of a dual-issued payer promises to make a payment if the payee fulfils a digital currency could derive from differences in monetary predefined condition. At the same time, the total of pay- policy between the two jurisdictions, in particular differences ments which has not been settled must not exceed the in interest rates. As CBDC would be backed by pledged com- predetermined maximum amount that the payer can pay mercial bank funds held with the respective central banks, without settlement on the ledger. Central Bank Digital Currencies for Cross-Border Payments  •  17 n-ledger holds/escrow: This model uses the HTLC tech- • O All individual payments along the payment chain depend nology (referred to earlier), which allows for conditional on the fulfilment of the following condition by the payee transfers that are recorded on the ledger and enforced by (Receiver or Connector(s)): presentation of the preimage for the ledger if the payee fulfils a predefined condition. a cryptographic hash value before a predetermined time (timeout). The hash value is used to define the payment hird party escrow: This model is conceptually similar • T condition (in the context of a smart contract), while the to the on-ledger escrow but relies on a third party which corresponding hash preimage marks the fulfilment of that is trusted by the payer and the payee rather than on the condition. Before a cross-ledger payment is initiated, the ledger to enforce the conditional transfers. preimage and its cryptographic hash value are produced by • S imple payment channel: This model involves an the Receiver. The hash value must then be shared with the arrangement between the payer and payee using Sender together with other terms of the payment, which escrowed funds in a shared temporary account on the may include payment amount, payment currency, payment ledger. Both parties promise to exchange signed claims timeout and Receiver information, using external means of off-ledger, which represents their entitlement to a spe- communication (e.g., email). After the initial bilateral Send- cific portion of escrowed funds, if the payee fulfils a pre- er-Receiver communication, each individual payment of the defined condition. Only the final net position of multiple cross-ledger payment chain goes through two main phases: bilateral payments is actually settled on the ledger. irst, the payer (the Sender or Connector(s)) prepares • F onditional payment channel: This model uses HTLC • C the payment to the payee (the Receiver or Connector(s)) and is similar to the simple payment channel in the sense according to the specific payment method used. that both parties exchange signed claims off-ledger, but in addition has an enforcement mechanism by the ledger econd, there are three possible scenarios. If the hash • S for the transfers based on whether the payee fulfils a pre- preimage is presented by the payee (the Receiver or the defined condition. Connector(s)) before the timeout and is verified as cor- rect, the condition for the payment is fulfilled and the Example of a funds transfer. Here only a general illustra- payment to the payee is executed (fulfilment scenario). tion is reported of how the Interledger Protocol—a led- Alternatively, if the timeout expires without the correct ger-agnostic protocol—allows the sender to make payments preimage being presented, the payment is aborted (tim- across different types of ledgers, The building blocks of the eout scenario), or if the payee does not accept the pay- protocol are Participants, Ledgers and Payment Methods, ment, the payment could be aborted even before the whereby Participants are entities that have accounts on one expiration of the timeout (reject scenario). or more ledgers and participate in the cross-ledger payment, ledgers indicate any system used to track transfers of value Payment processes based on the protocol require infor- between, and balances on, accounts, and payment methods mation exchanges between participants as well as with (as above) are bilateral agreements between participants on other relevant entities. These include the initial informa- specific methods to make payments and settle obligations tion exchanged between the Sender and the Receiver on the ledger. The choice of payment method depends on which does not vary along the payment chain and does not participants’ preferences and ledger functionalities. Partici- depend on the ledger and the payment method used, as pants can assume three roles within the Interledger Protocol: well as other information which varies among each payment Sender, Receiver, or Connector. Connectors are entities with in the payment chain (e.g., fees and timeouts for individual accounts on two or more ledgers, which act as liquidity pro- payment conditions). viders that relay payments across ledgers and play a critical role for the successful execution of cross-ledger payments. Considerations. The Stella report concluded that, from a tech- A liquidity provider enables a cross-ledger payment by nical perspective, the safety of today’s cross-border payments exchanging an incoming payment from its account on one could potentially be improved by using payment methods ledger for an outgoing payment to its account on another that synchronize payments and lock funds along the pay- ledger. When Connectors relay a payment across ledgers ment chain, noting in addition that further reflections would denominated in different currencies, they conduct currency be needed on legal and compliance issues and the maturity conversion. Where a single Connector cannot link the pay- of the technology. Specifically, in relation to safety, the on-led- ment between the Sender and the Receiver, or cannot do so ger escrows, third- party escrows, and conditional payment in an efficient way, multiple Connectors can be composed channels (all of which have enforcement mechanisms) can into a payment chain (Chart 1 1). ensure that each transacting party who completely satisfies 18  •  Central Bank Digital Currencies for Cross-Border Payments CHART 11 PROJECT STELLA—CROSS-BORDER FUNDS TRANSFERS 1. Prepare 2. Prepare 3. Prepare Execution of payments along a payment chain Sender Connector 1 Connector 2 Receiver 6. Execute 5. Execute 4. Execute (fulfilled) (fulfilled) (fulfilled) 1. Prepare 2. Prepare Cross-ledger payments aborted due to timeout Sender Connector 1 Connector 2 Receiver 4. Abort 3. Abort (timed out) (timed out) Chart 11 shows both a successful payment scenario and a rejection scenario. Both scenario feature two Connectors and illustrate the order of payment preparations and executions. In the successful payment scenario, the Sender and Connectors 1 and 2 prepare the payment to Connector 1, to Connector 2, and then to the Receiver, respectively, in this order. Then, the Receiver and both Connectors fulfil the payment conditions by presenting the hash preimage before the timeout. Depending on the arrangement between the Sender and Receiver, pos- session of the preimage by the Sender could be regarded as evidence of the receipt of payment by the Receiver. In the rejection scenario, the payment is first prepared but then aborted due to timeout. The Sender and Connector 1 prepare the payment to Connector 1 and Connector 2, respectively, in this order. Then, if Connector 2 remains unresponsive and does not present the hash preimage to Connector 1 before timeout, payments to Connector 2 and Connector 1 are aborted, in this order. its responsibilities in the transaction process is not exposed THE ARCHITECTURE OF CBDCs FOR CROSS- to the risk of incurring a loss on the principal amount being BORDER PAYMENTS: DRAWING LESSONS FROM transferred. On the other hand, with regard to liquidity effi- AFRICA ciency, the trustline method appears to be superior to the others since it is the only post-funded payment method. The design of the architecture of CBDCs for cross-border payments use can draw lessons from existing regional pay- Interestingly, the study identified the so called “free option ment systems in Africa.47 These systems comprise those in problem,” a risk aspect that could materialize even when all the Southern Africa Development Community, West Afri- of the preconditions used in the assessment are met. This can Economic and Monetary Union, West African Monetary is the exchange rate risk that participants are exposed to Zone, Economic and Monetary Community of Central Africa, when relaying a cross-ledger, cross-currency payment. In East African Community, Common Market for Eastern and the preparation of such a payment, participants enter into Southern Africa, and BUNA (formerly known as the Arab a commitment to deliver a fixed amount denominated in Regional Payment System, led by the Arab Monetary Fund one currency in exchange for a fixed amount denominated (AMF) and launched in December 2020 for the Middle east- in another currency. This could potentially be exploited ern and northern African countries that are members of by malicious actors, without the safety of payments being the AMF). As it is to be expected, each model has advan- undermined.46 This “free option problem” currently remains tages and disadvantages, and model variants typically aim at open and is being actively discussed within the Interledger adapting the model to the realities of the countries involved, community. However, it should also be noted that it does while at the same time trying to preserve the advantages not pose direct risk to the safety of the payments. Moreover, and minimize the disadvantages. Yet, some general ele- this may not be a significant issue if the participants value ments and their basic features could be considered when reputational risk above the potential gains from exploiting designing CBDC models for cross-border payments. the Connector’s binding commitment. Central Bank Digital Currencies for Cross-Border Payments  •  19 The deepest form of payment system integration across bor- cation of low intra-regional trade and overall economic inte- ders is a fully centralized architecture. This would involve a gration. It should as well be noted that bilateral agreements common technical-operational facility to process payment require less upfront investments relative to a central plat- transactions to which central banks, bank and non-bank form, and thus may be more affordable, although they may PSPs connect directly. This model is mostly identified with not be as scalable as multilateral network to include multiple regional integration projects that have evolved into a mon- payment corridors and currencies. etary union.48 There is little distinction between domestic and intra-region cross-border payments (in the region’s cur- Finally, as the experience from Africa suggests, a key aspect rency), and both types of payments can be processed in the that would need to be explored in great detail is the choice same payment system seamlessly. of currency. Any CBDC model for cross-border use would need to choose the CBDC(s) that would be used for set- An alternative general architecture for the integration of pay- tlement. This would bear implications for the underlying ment systems across borders is the “hub-spoke” arrange- liquidity and FX risk management. In Africa, where partic- ment. The “hub” is a central platform for payment exchange ipants in cross-border payment arrangements do not share and processing, the “nodes” are the points of delivery and a single currency, a global reserve currency like the US dollar the “spokes” are the communication routes between the or the Euro is normally chosen as the settlement currency, nodes and the hub. In practice, this means that bank and mainly because at least one of these two currencies is widely non-bank PSPs do not connect directly to the central plat- available. The latter is a very important aspect for liquid- form but rather do so through their central bank or domestic ity risk management in a cross-border payment system in payments system, and all cross-border traffic go through the which the central bank acting as settlement agent is not the hub and then out again. issuer of neither of these two currencies. Because of this rea- son, this central bank is unable and/or unwilling to act as a A third general architecture would consist of a fully decen- source of liquidity in the system. However, the fact that the tralized architecture. There would be no central platform two currencies are widely available could mean that some to which participants would connect to, either directly or large commercial banks operating may be willing to act as indirectly. Typically, such a model would rest on bilateral or liquidity providers. This option, however, would not be avail- multilateral interlinking platforms that would link domestic able in the absence of CBDCs functioning as global reserve systems together, with harmonized solutions for messaging currencies.49 and communication and for scheme management, respec- tively, and with settlement taking place on a bilateral basis, Regarding FX risk and currency conversion costs (e.g., fees), for each country pair. As an alternative to bilateral links, a these are typically borne by the originating and beneficiary decentralized architecture could also make use of DLT. Use end-users in a transaction rather than by the bank or non- of DLT has been piloted for some retail payment uses cases, bank PSPs that participate in the cross-border system. Two like cross-border remittances. To date, experience with the of the CBDC models discussed above, in fact, transfers the application of DLT for cross-border large-value and corpo- FX risk on the participating central banks.50 In the other rate payments is limited. models described, each bank or non-bank PSP could set its own exchange rates and related rules for its clientele. Alter- Evidence shows that in Africa most cross-border payment natively, a rulebook could be conceived of, which would systems are highly centralized. On one hand, this reflects the include specific rules covering this aspect (e.g., a maximum greater efficiency and reduced costs and risks of centraliz- spread over the prevailing exchange rates quoted by the ing the payment clearing and settlement functions and with central banks).51 a rulebook that promotes robust risk management. On the other, greater usage also reflects the fact that the countries Another element should be kept in mind when reflecting that participate in these regional systems are more highly on currency choices for cross-border CBDC arrangements. integrated in terms of their intra-regional trade of goods In many corridors, countries mainly exchange USD or other and services and their financial system. Centralized solutions dominant and widely accepted currencies. This is because have also required very high commitment of stakeholders to each trading country uses such currencies both to build materialize, particularly of central banks. In turn, more lim- reserves and for trading internationally. Upon using domestic ited usage of regional payment systems that have a partly currencies for cross-border transfers (the same would hold or fully decentralized architecture may be reflecting limited for domestic CBDCs), there will always be the possibility of benefits of these systems, although it may also be an indi- such currencies accumulating depending on the balance of 20  •  Central Bank Digital Currencies for Cross-Border Payments trade and remittances, and hence the risk of currency fluc- Finally, FX controls applied in some countries are an issue tuations and of piling up a stock of currencies that have no that could potentially deter participation of those coun- value outside their counties of issue. This would be avoided tries in regional arrangements that use multiple CBDCs. by cross-border arrangements where CBDC were backed by This aspect would need to be analyzed on a case-by-case a basket of reserve currencies. However, the pegging to a basis, depending on the specific type of controls that are currency basket would constrain the supply of CBDC, much applied, to determine whether it is technically possible and as any fixed exchange rate regime would do, and change its financially viable for the cross-border CBDC arrangement nature into something analogues to a stablecoin. Moreover, to develop functionalities that would allow participation the practical complexities of such arrangements should not of bank and non-bank PSPs of countries where FX controls be neglected, such as fixing the right weights of the basket, apply. For countries adopting a “hub-and-spoke” architec- mitigating the fluctuations in CBDC valuation (ad hence the ture, the respective central bank through which bank and risk to users) due to the exchange rate volatility of the bas- non-bank PSPs send CBDC payments could ensure adher- ket currencies, or managing the reserves investment portfo- ence to transaction limits in FX currencies. lio fund optimally—all elements that would complicate the CBDC peg. H OW DO CROSS-BORDER CBDCs III.  ADDRESS EXISTING CHALLENGES? A general perception is that cross-border payments lag affect a number of different stakeholders on the supply side behind domestic ones. A recent assessment by the Financial (bank and non-bank PSPs, payment system operators and Stability Board (FSB) of the existing payment arrangements technical service providers) and the demand side (end users has identified four types of important challenges, namely: composed of individuals, businesses and government agen- cost, speed, access and transparency.52 These challenges cies), and affect each of them in different ways. TABLE 1 How Cross-Border CBDCs Address Existing Challenges CORRESPONDENT BANKING CROSS-BORDER CBDC53 CHALLENGE 1: High Cost Supply Side •  High operational cost Lower cost due to scale and network economies •  (ceteris paribus) High barrier to entry and unwillingness to do business with less profitable customers •  Direct participation available to all applicants •  •  Smaller banks and non-bank PSPs may need to rely on other banks in foreign compliant with access requirements jurisdictions, with accompanying liquidity and credit risk. Others may not be able Direct participation avoids the need for indivi- •  to find correspondents or bank partners dual participants to rely on others (although this •  Multinational PSPs offering services in various countries and currencies need depends on the scalability of the model used) liquidity access in several currencies and face related FX risk Individual PSPs manage their own liquidity needs •  (in the relevant countries and currencies) and the associated FX risk Demand Side Individuals and MSMEs are impacted by high transaction fees in relation to sma- •  PSPs’ lower access cost to a cross-border CBDC •  ller value payments infrastructure, and transparency of PSP’s access •  High costs for maintaining an account or for individual transfers may discourage cost structure, should discourage PSPs from adop- use of the regulated financial system for cross-border transfers, exacerbating ting unfair pricing practices with service users54 financial exclusion and driving some payment flows underground. In other cases, Lower costs should encourage PSPs to (at least •  individual users may be discouraged from making cross-border payments at all partly) pass them on to individual and MSME customers CHALLENGE 2: Low Speed Supply Side Speed is impacted by the dependence on several correspondents/ •  A single infrastructure would eliminate speed limi- •  providers, cut-off times, asynchronous opening times or regulatory checks. tations When processing speed is low, the cost for liquidity as well as FX •  settlement risk increases and liquidity management becomes more complex •  Lack of system interoperability slows transactions Non-harmonized messaging and processing standards furthermore reduce speed •  21 22  •  Central Bank Digital Currencies for Cross-Border Payments TABLE 1 continued Demand Side Low speed of cross-border payments brings delays and thus increases uncer- •  Higher speed would enhance certainty and •  tainty, liquidity and credit risk, impacting all customers. Moreover, it can facilitate negatively impact business and investments, in particular where payments are business and investment time-critical CHALLENGE 3: Limited Access Supply Side Challenges to access to payment systems and wholesale services can occur on •  Open, transparent and risk-based access criteria •  the supply side owing to technical and financial entry barriers, regulatory require- governing participation in a cross-border CBDC ments, or liquidity access limitations. PSPs may not be able to access directly local infrastructure, and their consistent application, and foreign payment systems and possible funding in foreign currencies. This may would facilitate direct or indirect participation make them dependent on other providers impacting their cross-border payments from PSPs, thereby removing all such challenges offerings 55 Demand Side Access limitations may exist for MSMEs and individuals possibly limiting financial •  Higher speed would enhance certainty and facili- •  inclusion and pushing customers toward inefficient or costly third-party services. tate business and investment When unregulated payments channels are used instead, this can exacerbate financial integrity risks CORRESPONDENT BANKING CROSS-BORDER CBDC53 CHALLENGE 4: Limited Transparency Supply Side Limited transparency can lead to uncertainty and missed service levels to custo- •  As a central bank’s standard-compliant infrastruc- •  mers ture, a cross-border CBDC would mitigate, if not •  Dependency on third parties can lead to difficulty in controlling the payments eliminate, such risks process and tracking the status of payments and resolving disputes Information gaps can create a lack of transparency for AML/CFT and other purpo- •  ses Demand Side Limited transparency concerns all stakeholders on the demand side due to the •  • As above uncertainty it causes. For corporates, lack of information about the speed, fees and FX rates of payments in process leads to uncertainties over the timing and amount of payments and can impact business service levels, and may lead to hedging and insurance costs to address the risks Limitations may exist for MSMEs and individuals possibly limiting financial inclu- •  sion and pushing customers toward inefficient or costly third-party services Based on the above description, the models discussed national standards. Also, the analysis (realistically) assumes suggest that cross-border CBDC can be designed in ways that central banks running a CBDC facility would adopt that address effectively the above challenges. Table 1 con- cost-recovery rules and pricing criteria that would take siders the specific supply and demand side challenges into account the volumes and values of the transactions that fall under each of the four types and points to how ordered by participants but would apply no surcharges the cross-border CBDC could addresses them. The anal- aimed to extract extra-profits from its operation. ysis reported in the table excludes references to BOC- BOE-MAS Model 3a (discussed above) since, as noted, the The models discussed, also, mark a significant progress along model closely resembles the existing correspondent bank- the G20 roadmap to enhance cross-border payments.56 The ing arrangement and features many of its same challenges. models address two of the five focus areas identified to Also, the analysis (realistically) assumes that central banks address the key challenges faced by cross-border payments, would run a CBDC facility as a financial market infrastruc- by contributing to five of the nineteen building blocks of ture or payment scheme and apply for it all relevant inter- the roadmap (Chart 12).57 In particular, by experimenting Central Bank Digital Currencies for Cross-Border Payments  •  23 with their respective models, the stakeholder communities as a commitment mechanism to drive change, and they are around the central banks undertaking the pilots are building creating the potential for new payment infrastructures and on the shared understanding of the targeted improvements arrangements for cross-border payments, which could offer in users’ experience with cross-border payments and acting solutions to existing challenges. CHART 12 G20 ROADMAP TO ENHANCE CROSS-BORDER PAYMENTS Focus Areas and Associated Building Blocks 1. Develop common cross-border payment vision and targets 2. Implement international guidance and principles 3. Define common features of cross- border payment service levels Public and private sector commitment 17. Consider the feasibility of 4. Align regulatory, supervisory new multilateral platforms and oversight frameworks A 5. Apply AML/CFT consistently and arrangements for cross-border payments and comprehensively New payment Regulatory, 18. Foster the soundness of 6. Review interaction between infrastructures supervisory global stablecoins data frameworks and and and oversight arrangements cross-border payments arrangements E B frameworks 19. Factors an international Enhance 7. Promote safe payment corridors dimension into CBDC cross-border 8. Foster KYC and identity designs payments information-sharing D C Existing Data and payment 14. Adopt a harmonised version market 9. Facilitate increased adoption infrastructures of ISO 20022 for message practices of PvP and formats 10. Improve (direct) access to arrangements 15. Harmonise API protocols for payment systems data exchange 11. Explore reciprocal liquidity arrangements arrangements 16. Establish unique identifiers 12. Extend and align operating hours with proxy registries 13. Pursue interlinking of payment systems 23 IV. USE OF CROSS-BORDER CBDCs: LEGAL ISSUES CBDC for cross-border payments is inherently exposed to A different consideration would hold for cross-border CBDC a plurality of legal systems. Anything that crosses borders that is backed by a basket of currencies as one of the models implies that it is potentially subject to (at least) two different described above (and under the limitations therein noted). legal systems. These systems do not necessarily have simi- In this case, what is transferred is not the currency of a coun- lar rules to govern same events or contingencies. And even try but an asset similar to the IMF’s SDR. As noted earlier, the when similarities exist, such as, for instance, when interna- SDR is essentially an artificial currency instrument, built from tional standards apply to both systems, the actual applica- a basket of a number of national currencies, which the IMF tion of the same standards in each country might vary and uses for accounting purposes.58 The SDR is not regarded as might even conflict with each other. Conflicting standards a currency or a claim against the IMF assets. Instead, it is a and, even worse, conflict of laws engender the risk that the prospective claim against the freely usable currencies that expected effects of a transaction do not materialize or that are issued by select IMF member states. Although the legal unexpected consequences occur. definition of CBDC would change in this case from that a fiat digital currency, once all participating jurisdictions rec- CBDC that is legally issued in a country should be fully rec- ognized its use and value by way of agreement, its validity ognized by the other countries. If, in the issuing country, might not be challenged in any of the relevant jurisdictions. the central bank is acknowledged to possess the authority to issue CBDC, any other country should accept this CBDC as the currency of the issuing country in the same way as it LEGAL OBSTACLES TO CBDC FOR CROSS-BORDER accepts foreign currencies under any other form. This means PAYMENTS that the legal issues associated to CBDC transfers should be the same as those associated to any other cross-border In the first place, CBDC models designed for cross-bor- transfers. A situation where a country would accept a for- der payments might pose different legal issues according eign currency under a specific form (say, cash) while it would to their individual specificities. An analysis would thus be refuse another currency under another form (hypothetically, required for each model, based on the legal order involved CBDC) would be inconceivable in legal terms especially if in each case, which is not within the scope of this report. the latter is understood to be a means of payment and not a commodity (precisely the case of CBDC). However, in light However, one common issue is that of “applicable law.” of the fact that the central bank permits non-residents, at When an established infrastructure is entirely situated within least in principle, to freely purchase their domestic CBDC, a jurisdiction under a specific legal order, that legal order thus placing both currencies (as means of payment) in com- shall apply to its operation as well as to its participants, petition with each, the circumstance cannot be ruled out including when these are foreign entities. The interlinking whereby the receiving country would impose limitations on of different infrastructures situated in different countries foreign CBDC use within its jurisdiction in order to safeguard would require of the authorities involved to adopt common its own currency. standards that would mitigate risk transfers from one infra- structure to the other. Yet, this would not change the fact 24 Central Bank Digital Currencies for Cross-Border Payments  •  25 that each infrastructure is subject to its own jurisdiction and Inter-jurisdictional differences in data collection and data its respective legal system. The situation is different when protection rules may affect CBDC use for cross-border pay- the infrastructure is cross-border. In this case, the issue of ments. Such rules require specific consideration, since digi- relevant applicable law does arise. tal payments involve transfer of data. Consequently, relevant differences between national legislations might impair the Usually, this issue is resolved by way of an express choice execution of cross-border payments. taken by the parties involved at the time of setting up the legal basis of the infrastructure. The agreement would seek The use of DLT raises legal issues. In addition to the fact to address most of the foreseeable issues and contingencies, that digital transfers must be fully recognized in all rele- leaving to the application of general principles the resolution vant jurisdictions, transfer of data needs to be adequately of eventual gaps arising from issues or contingencies that regulated and data need to be protected. This is both to are not covered under the legal basis. In this regard, where ensure effective transfers and to prevent any form of abuse parties have not elected a specific law, the principle that of data. Moreover, the decentralization of transfers through currently applies to conflict-of-laws cases is the applicable the progressive record into nodes challenges the relevance law. In the case of a payments system whose operations are of the applicable law. Indeed, in a horizontal and decentral- integrated and carried out for the purpose of executing pay- ized mechanism, each leg of a transfer may in principle be ments, the applicable law is usually the law of the jurisdiction considered as a separate transaction. In fact, this approach where the processing, clearing and settlement of payments is not new to law theory and also to the legal practice in the take place.59 This, at least, as long as the activities are central- realm of payments. For instance, international wire transfers ized. Different considerations would hold for decentralized have been often considered as a chain of separate transac- infrastructures such as those based on DLT (see below). tions from the standpoint of potential conflict-of-laws issues. However, this fragmentation can affect the legal soundness Payment systems or platforms are typically regulated by of cross-border payments, since, as noted, different legal agreements. To that extent, it is inconceivable that an infra- systems may govern the same events differently. structure operates without a detailed and articulated agree- ment having been established since the very outset, covering When public entities (such as central banks) are involved in most of the relevant issues and identifying the applicable cross-border arrangements, they can stipulate agreements law. The oversight authorities usually require for these agree- subject to international law. This might be the case for ments to be submitted for their approval or authorization domestic CBDCs, and it would be even more appropriate before being adopted. When multiple central banks are for cross-border CBDCs. If an infrastructure is planned for involved, they are expected to cooperate with each other the operation of a cross-border CBDC, it can be established and try to solve any emerging conflict. by way of international agreement and made subject to international law. An infrastructure that would be subject to Even in the presence of an agreement governing an infra- international law would be governed independently of the structure, however, each legal system has mandatory rules national laws of the participating countries. However, this that cannot be superseded by way of agreement or con- solution might have its own drawbacks, since central banks tract. Possible conflict between the legal systems involved offering services under international law, which compete cannot be fully resolved by an agreement or the selection of with services supplied by the private sector under national the applicable law. Rules that are considered of a mandatory law, would risk distorting competition. Moreover, in the event nature, irrespective of the applicable law, need to be applied. that international law applies, it is still common practice to Thus, an effort would be required of the participating juris- respect high standards, and especially when domestic rules dictions to adopt some key common rules and standards to derive from the application of international standards, cen- the extent possible. In particular, in the case of cross-border tral banks that would be found not to be fully compliant with CBDC, each participating country should recognize digital such standards would face reputational risk. transfers as enforceable for CBDC to be accepted within its own jurisdiction and to be protected from legal risk. In the Finally, differences across jurisdictions may weaken the legal same vein, participating countries should share the same basis of cross-border CBDC. Regulatory inconsistencies standards for verification and enforceability of the final exe- might emerge if, for instance, different criteria exist for the cution of transfers under all circumstances, and similarly for authorization of financial institutions as participants, or if cybersecurity and foreign exchange risks. different conditions apply for access of participants to cen- 26  •  Central Bank Digital Currencies for Cross-Border Payments tral bank systems and facilities or to open accounts at central ciency and even prejudice the achievement of achievement banks, or still if laws differ on data privacy, storing, sharing of common regulatory standards for the use of the same and management (which are of special relevance in the dig- CBDC across the jurisdictions involved and ultimately alter ital space). These differences may not only affect the legal the homogeneous quality of CBDC services to the public soundness of the infrastructure but may also weaken its effi- across the whole arrangement. V.  CONCLUSION: “IT TAKES TWO TO TANGO” This report has illustrated the CBDC-based models currently fall into three broad categories: legal and regulatory barriers, under consideration. The report was intended only to show technological incongruities, and monetary risks. how CBDC-based solutions can facilitate cross-border pay- • First, a divergence in regulatory strategies in differ- ments. It has not ranked the models discussed, and it has ent jurisdictions may impede CBDC’s ability to improve not made recommendations as to which model(s) should cross-border payments. For instance, the transfer of a be preferable. CBDC between two countries would require the currency to comply with the legal requirements (such as money The report suggests that CBDC-based solutions could lower laundering, terrorist financing, and others) of both, which costs and reduce significantly the number of intermediaries may vary dramatically. Achieving smoother cross-border involved in cross-border payments. Payments into a country transfers would require significant harmonization across using a CBDC would go from the payer’s account (or wal- various legal and regulatory domains. let) to the central bank of the receiving country and then directly to the payee or directly to payees’ account (or wal- • Second, technical variables—such as different blockchain let) if it were peer-to-peer, without having to go through / DLT standards and applications—may reduce the effi- a network of commercial banks. If both countries were to ciency of CBDCs across borders. Underlying blockchain issue an interoperable CBDC (or interoperable CBDCs), pay- / DLT systems would have to be interoperable, which ments would only need an exchange market to function may not be the case without harmonization in design across borders. and implementation, and considering this is a new tech- nology, it may take some time for standards to emerge. However, as the heading of this section suggests, any Similarly, the presence of legacy systems and infrastruc- cross-border arrangement necessarily involves two (or more) tures in various countries contributes to the technological partners and requires them all to agree on the rules and divergences at play. procedures needed to make the underlying exchange pro- • Third, the use of CBDCs for cross-border payments may cess possible. As the model description has shown, and the raise risks. It would not eliminate the exchange rate risk, legal analysis suggests, CBDCs could not possibly be used and costly processes would remain in place for cur- for cross-border payments without the central banks of the rency exchange. In other words, even if systems became jurisdictions concerned being intimately involved in the pro- interoperable, currency conversion and cross-currency cess of setting up and operating the interlinking or com- rates would still pose obstacles. Moreover, the cross-bor- mon infrastructure, and making its various (legal, technical, der use of CBDCs may put monetary sovereignty at risk. operational, financial, risk management) components mutu- The cross-border use of a CBDC denominated in a certain ally consistent or commonly shared: the cross-border use of (major) currency may have far-reaching implications for CBDCs is essentially a collective, cooperative undertaking.60 monetary and financial policy independence. This may reduce and obstruct the functionality of CBDCs denomi- There are several preconditions and steep obstacles along nated in other currencies as means of payment and store the way to achieving system interoperability and to improv- of value. Also, since CBDCs must be fully fungible and ing the flow of money across borders using CBDCs. These 27 28  •  Central Bank Digital Currencies for Cross-Border Payments convertible into and from fiat currency, the most effec- preventing CBDCs from producing powerful efficiency gains tive solution to ensure that CBDC improves cross-border in cross-border payments. This raises significant questions payment processes would be for issuing countries to about liquidity management across borders. For instance, cooperate and jointly devise harmonized, interoperable some CBDC systems might be transferring funds via Swift solutions. Finally, while principal risk would be obviated messages, while others may be using DLT-backed infrastruc- by using PvP modalities, the systems would still funda- ture reliant on tokens or stablecoins. mentally rely on the sender having adequate liquidity, which (as is evident from experience of financial markets) It is clear that some kind of unified corridor or harmoniza- is not a given. Thus, adequate liquidity support mecha- tion is required to achieve maximum efficiency gains. This nisms would be needed, which would introduce credit warrants further research, as many of risk and thereby necessitate developing linkages with collateral management systems that might still be based the benefits of CBDCs for cross-border use hinge on on traditional technology models. For such solutions to addressing this challenge effectively. In any case, it will be be truly scalable, putting in place additional market sup- important that the private sector be closely involved in the port mechanisms would likely be necessary. dissemination and management of any prospective CBDCs and that risks be mitigated as much as possible that usage In fact, central banks are in general skeptical about the abil- of CBDCs might destabilize the financial sector. ity of rules to apply across boundaries and are concerned about the potential incongruities that might arise between In conclusion. Further work is still needed in this area various CBDCs.61 The stringency and properties of AML/CFT before workable solutions for wholesale cross-border CBDCs requirements vary across countries, making it difficult for can be arrived at, and the issues to be addressed are still CBDCs to cross borders. This also applies to data protec- many and relevant. Once this is done, taking the next step, tion standards, which differ between jurisdictions and may that is, making CBDCs available to retailers for cross-border pose concerns around the anonymity and privacy of CBDC payments would require developing interfaces for users transactions. Another major concern is about digital iden- to interact with their banks and PSPs.63 Further, a range of tity management and how this might work in cross-border other market intermediaries (like payment gateways) will transactions. Ideally, a single verifying mechanism should need to emerge in order to integrate acceptance of CBDCs be in place not only to confirm the identity at each end of for ecommerce and other payment needs.64 This, however, the transaction and in general to support all processes that would also call for the development of some form of scheme require verification. This would be especially necessary to rules and brands. In the meantime, national authorities and fulfil KYC rules in multiple country contexts.62 industry stakeholders must remain committed to continue strengthening domestic systems, adopting international This applies not only to questions around legal and regu- standards, and enabling access to cross-border payment latory harmonization, but also to cross-country differences service providers, and improving to the extent possible the in technical and operational standards, which might be the efficiency and transparency of the existing cross-border main obstacles to the cross-border interoperability of CBDC payment arrangements. systems. These standards might diverge across economies, ANNEX 1 CHANGES AND INITIATIVES IN CROSS-BORDER PAYMENTS DISTRIBUTED LEDGER TECHNOLOGIES pound, US and Canadian dollars, and euro. TransferWise routes most payments not by transferring the sender’s DLT is a tool for recording ownership. A distributed ledger money directly to the recipient as it is in the case of corre- is a database of transactions that is spread across a network spondent banking, but by matching the amounts with other of many computers, rather than stored in a central location. TransferWise’s users sending the other way around. Trans- It is consensually shared and synchronized across multiple ferWise then uses these pools of funds to pay out transfers sites, institutions or geographies. The participant at each via local bank transfer. This process avoids currency conver- node of the network can access the recordings shared sion and transfers crossing borders. A small commission is across that network and can own an identical copy of it. Any charged per transaction and the inter-bank mid exchange changes or additions made to the ledger are reflected and rate is used, unlike traditional currency transfers where there copied to all participants in a matter of seconds or minutes. are buy and sell rates and the broker takes the difference Applied to cross-border payments through a so-called “hub between the two. & spokes” model, users can exchange fiat money into DLT- based tokens held in digital wallets, through ATM machines, SWIFT GPI POS terminals, online interfaces, or other means (i.e., the spokes). These tokens are then transferred across borders SWIFT GPI is a new facility that makes for quicker payments over a virtual currency’s secure network (i.e., the hub) to with full transparency on accompanying costs, while provid- the payee’s digital wallet. Finally, tokens are exchanged into ing on-the-spot information on the status of transactions. foreign fiat money, as desired, through the same means as With GPI, a payment is effective on the same day the pay- above (again, the spokes). The attributes of payment ser- ment process is initiated and a breakdown of all of the costs vices offered by hub & spoke networks may look attractive involved in the payment, including exchange-rate costs, are to the public, except for three important caveats: first, the provided to the party initiating the payment. The status of potentially erratic valuation of virtual currencies introduces the payment situation with respect to correspondent banks risks and could limit their adoption, at least for large value is available at all times. This traceability is possible due to a payments; second, the lack of trust in hub & spoke networks unique reference associated with each payment that is kept could erode their value; and third, the lack of interopera- and shared by the different banks involved in the operation. bility among networks could keep prices of hub & spoke The process finalizes with confirmation that the payment payments high. has been deposited in the beneficiary’s account. Information regarding the payment remains unchanged and homoge- TransferWise neous during the process, and details provided by the issuer are the same as those received by the recipient as a result TransferWise is a British online money transfer service of the commitment established between the banks taking founded in January 201 1 by Estonians Kristo Käärmann and part in SWIFT GPI. These technologies facilitate end-to-end Taavet Hinrikus and is based in London. The company sup- tracking of payments (just like tracking a courier) and offer ports more than 750 currency routes across the world and transparency and allow companies to optimize liquidity. provides multi-currency accounts expressed in UK sterling 29 ANNEX 2 NON-CBDC CROSS-BORDER PAYMENT MODELS65 This annex summarizes the main current cross-border bank positions more efficiently. Moreover, USC may also be payment models—alternative to corresponding banking— issued by transferring non-operating cash balances into a that are not based on CBDCs. The annex concludes with special purpose vehicle (SPV) that manages cash without a some considerations on the future outlook of cross-bor- profit-making objective. This has the benefit of reducing der payments. the regulatory collateral requirements of the participating financial institutions by moving non-profitable high liquidity flight risk deposits off their balance sheets while being able PRIVATE LABEL CRYPTOCURRENCIES to use the liquidity for transaction settlement purposes. In this model, consortiums of commercial banks or non- Similar to USC, JPM Coin is also an experimental digital coin bank entities issue digital tokens on an agreed upon digital built on the DLT, originated by J.P. Morgan Chase. JPM coin currency standard to execute cross-border settlement and is a digital token, redeemable in a 1:1 ratio to the US dollar. other exchanges of value in a closed system. The digital The purpose of JPM Coin is transferring value in cross-bor- token is typically backed by central bank issued fiat cur- der settlement. JPM Coin is permissioned, its users are exclu- rency in the form of collateralized liquidity or other asset sively institutional customers, and it employs the Quorum or liability deposits at or guaranteed by the central bank or ledger. Additionally, in 2017, J.P. Morgan launched the Quo- some other trusted entity. The two examples of these are rum-based Interbank Information Network (IIN), as a pilot the Utility Settlement Coin (USC),66 and the JP Morgan Coin program. The IIN allows member banks to exchange pay- (JPM Coin).67 Cross-border settlement using either USC or ment information to overcome the challenge of sharing such JPM coin leverages the blockchain to ensure faster and more information in cross-border settlement. As of April 2019, secure cross-border payments by straight-through process- there were 220 member banks across the world participat- ing and improvements to transparency. ing in the IIN.68 Both the IIN and JPM Coin aim to address the shortcomings of correspondent banking in terms of infor- USC is an experimental digital cash instrument built on DLT mation sharing and settlement. However, they may also be and originated by the Swiss global financial services com- disruptive to central clearing services offered by organiza- pany UBS in partnership with Clearmatics Technologies tions such as Payments Canada and sideline the traditional in a consortium of fourteen banks. s of 2019, the consor- wholesale payments space. tium raised $63 million in funding from the 14 shareholder banks.20In the USC model, large private banks and FinTech Looking at potential disruption to wholesale payment sys- firms create digital tokens (i.e., USCs) representing money tems is crucial. A business case for USC would be to provide from multiple countries that can be exchanged on a dis- a vehicle for settlement within derivatives markets. An exam- tributed ledger platform. The digital token is fully collater- ple would be the market for cross-currency and interest rate alized by the cash balances of participating banks which swaps. To this end, it is important to note that the gross are held on the books of the central bank. Unlike Bitcoin, notional value of outstanding contracts in the over-the- USC is created through liability securitization. The decen- counter (OTC) derivatives markets at the end of Q2 2018 was tralized nature of USC enables its members to settle inter- US$595 trillion while the gross market value of OTC deriva- 30 Central Bank Digital Currencies for Cross-Border Payments  •  31 tives in the same period was US$10 trillion. In the context of dators must come to a consensus on the transaction. While disruption to existing wholesale payments clearing and set- Ripple claims to be decentralized, the fact is that the 55 val- tlement infrastructures such as Canada’s Large Value Transfer idator nodes all belong to Ripple. This is expected to change System (LVTS), payments made as part of the trade life cycle as third-party validator nodes join the network. As this hap- events in broader securities and derivatives markets may no pens, each of the Ripple validator nodes will be removed longer require such infrastructure. Indeed, the instantaneous for every two third-party nodes that join. Under this setup, cross-border transfer of digital tokens exchangeable for cash Ripple will become de facto decentralized with time. assets eliminates the need for cumbersome processes of intraday collateral management required by existing whole- Any accepted type of currency or asset can be used to sale settlement systems. Moreover, as such tokens are not transact on the Ripple Network. Cross-border payments tied to the operating hours of the wholesale payments sys- using Ripple transact using XRP, a digital currency works as tems across jurisdictions, transactions are free to flow fric- a liquidity source whenever it is necessary besides acting tionlessly hours a day and seven days a week. as a bridge between two currencies. XRP, which is an open source blockchain operates on the interface of peer-to-peer servers. It is supposedly capable of settling a payment within STABLECOINS FOR CROSS-BORDER PAYMENT 4 seconds and handling 1,500 transactions every second. Each transaction costs $0.00001 and requires a small frac- Stablecoin refers to a class of digital currencies that are tion of XRP to be destroyed in the process (0.00001 XRP), relatively stable in terms of their price. Stablecoins offer meaning the total supply of XRP decreases over time and, in instantaneous processing and security of payments as many theory, maintains its value. cryptocurrencies do. They also offer stability with respect to their parity against fiat currencies. Two digital curren- By contrast, Stellar enables individuals (end-users) to trade cies that fall into this category are RippleNet’s native digital money directly with each other across jurisdictions, using currency Ripple (XRP),69 and Stellar network’s native cryp- entrusted intermediaries to handle FX and funds transfers. tocurrency, the Stellar Lumen (XLM).70 Both Ripple and Stel- Stellar is an open source, decentralized protocol for digital lar enable faster and more efficient cross-border payments currency to fiat money transfers, which allows cross-border relative to correspondent banking. However, they differ in transactions between any pair of currencies. The Stellar pro- that the former aims at improving cross-border settlement tocol is supported by the Stellar Development Foundation, a between international banks, whereas the latter aims at pro- non-for-profit organization. viding low-cost cross-border payment financial services to end-users and the unbanked population. Transaction confirmation time using Stellar has been observed to range from 1,000 transactions per second to Ripple is a real-time gross settlement system, currency approximately 10,000 transactions per seconds in a 2016 exchange and remittance network created by Ripple Labs Barclays Africa and Deloitte pilot. Stellar’s transaction fees Inc., a US-based technology company. Ripple enables mul- remained a fixed rate at 0.000001 XLM per transaction, tinational corporations to settle cross-border payments by thereby making XLM cost effective for retail cross-border transferring XRP through the Ripple network, resulting in transactions. on-demand liquidity. The three parts of the Ripple ecosys- tems are i) servers that maintain the ledger, ii) clients, and iii) For a transaction to be processed in a few seconds, the Stel- intermediaries. Unlike Bitcoin or Ethereum, Ripple does not lar network needs to reach consensus fast while ensuring run proof of work nor does it run a proof of stake consensus accuracy.72 The Stellar Consensus Protocol (SCP) works by mechanism. Instead, Ripple transactions rely on a Byzantine utilizing groups of trusted nodes that communicate among Generals Problem (BGP) consensus protocol, known as Rip- themselves to verify transactions. Consensus is achieved, ple gateways, to validate account balances and transactions on average, every two to five seconds between the trusted of the system.71 The Ripple settlement process involves the nodes. Unlike Ripple where decentralization is controlled creation of a transaction that is signed by the account owner by Ripple, decentralization in Stellar is expanded by trusted and submitted to the network. Badly formed transactions nodes deciding to extend trust to other nodes according to will be rejected immediately, otherwise, they are provision- a Federated Byzantine Agreement (FBA).73 The Stellar Lumen ally included on the ledger. The Ripple network has many can be traded to different currencies, which means that per- validating nodes which are used to validate and verify trans- forming a cross-border transaction using XLM is essentially actions. For a successful transaction to take place, the vali- using XLM as the bridge currency between two jurisdic- 32  •  Central Bank Digital Currencies for Cross-Border Payments tions. This currency bridge function is facilitated through address these frictions, solutions have emerged across dif- the Stellar Decentralized Exchange (SDEX), an exchange ferent sectors. Large corporates and FinTechs operating on a allowing trading between domestic fiat currencies and XLM. global scale entering the payments space are able to address As a decentralized exchange (DEX), SDEX is borderless and a number of these frictions through their reach in various therefore is not subject to jurisdictional controls or frictions jurisdictions. These firms are able to timely and transpar- and currency exchange is done by way of atomic swaps.74 ently facilitate cross-border payments by simply engaging Similar to Ripple, the price of XLM depends on active trading in transfer pricing activities between their operations across against other currencies including fiat and the widespread jurisdictions. In doing so, they also potentially also bene- adoption of the digital currency itself. fit from tax arbitrage. SWIFT has launched and is actively expanding the core functionality of its SWFT GPI platform, More recently, Facebook announced Libra (which has now discussed earlier, to facilitate the instant and always on trans- become Diem, see fn. 63), its own permissioned blockchain mission of payments across borders. In addition to these ini- digital currency. Note that Libra is merely one of many social tiatives, there has been an expansion of DLT-based payment media digital currency platforms with some like LBRY and protocols that appear to be yielding promising results. Steem being operational for a number of years now.75 Like Steem before it, Libra’s primary use would be in the area of As these protocols gain ground, a proliferation of digital cur- person-to-person, person-to-business, and to pay for goods rency might take place in the mid- to long-term. Whether it and services online. While Libra is still a relatively new digi- is going to be CBDC such as that envisioned under Project tal currency, given the vast global network of approximately Jasper-Ubin, or a private label cryptocurrency like Ripple or 2.6 billion (or one third of the global population) users in Stella, or perhaps a consortium of commercial banks that the Facebook ecosystem, Libra has a substantial advantage emerge with their digital currency (USC or JPM Coin), this over many other stablecoins and digital currencies. As a dig- emerging trend suggests that blockchain is here to stay. ital currency not bound by borders, Libra has the potential to disrupt existing correspondent banking models of retail Although DLT-based payment protocols and the internet cross-border settlement. layer for payment services remains nascent and unlikely to replace correspondent banking infrastructure on a large THE OUTLOOK OF CROSS-BORDER PAYMENTS scale in the short term, its low cost, transparency, and real- time processing speed provide encouraging signs to the Currently, cross-border payments are dominated by cor- sector in their role in the future of cross-border payments. respondent banking, characterized by slow, costly and For sure, competition is on the rise and will continue to nontransparent execution. In the digital age, consumers (cor- strengthen putting pressure on the incumbents. The emerg- porates and individuals) more and more expect cross-bor- ing DLT-based payment protocols are competing in the der transactions to be fast, frictionless, and affordable, at wholesale and retail payments space and have the potential least as much as domestic payments are becoming every- to erode established clearing and settlement services pro- where. But while there have been many attempts to improve vided by central banks and institutions. the cross-border payment systems, some of the core chal- lenges in correspondent banking remain unresolved. To ENDNOTES 1 Drafted by Biagio Bossone with contributions by Oya Ardic, Ahmed 10 Relevant sources on this topic include: Faragallah, Sheirin Iravantchi, Maria Chiara Malaguti and Gynedi Srini- – De-risking in the Financial Sector, Brief, The World Bank, 7 vas, and under the oversight and coordination of Harish Natarajan. October 2016, available at https://www.worldbank.org/en/topic/ 2 The “cross-border payments” referred to in this report fall within financialsector/brief/de-risking-in-the-financial-sector the realm of retail payment services and include those payments or – The Report on the G20 survey in de-risking activities in fund transfers and remittances that are sent by an individual, busi- the remittance market, available at https://documents. ness, financial institution or government agency in one jurisdiction worldbank.org/en/publication/documents-reports/docu- to a recipient in another jurisdiction. Cross-border payments may or mentdetail/679881467993185572/report-on-the-g20-survey- may not involve a currency conversion. in-de-risking-activities-in-the-remittance-market 3 For un update on the recent growth of cross-border payment until – The decline in access to correspondent banking services in the period prior to the start of the Covid-19 pandemic, see Enhanc- emerging markets : trends, impacts, and solutions - lessons ing Cross-border Payments—Stage 1 report to the G20: Technical learned from eight country case studies, The World Bank, 2018, background report, by the Financial Stability Board, 9 April 2020. available at https://documents.worldbank.org/en/publication/ 4 Defined by Financial Action Task Force (FATF), “correspondent documents-reports/documentdetail/552411525105603327/ banking” is the provision of banking services by one bank (the the-decline-in-access-to-correspondent-banking-services- “correspondent bank”) to another bank (the “respondent bank”). in-emerging-markets-trends-impacts-and-solutions-lessons- Large international banks typically act as correspondent banks for learned-from-eight-country-case-studies thousands of other banks around the world. Respondent banks – The decline in access to correspondent banking services in may be provided with a wide range of services, including cash emerging markets : trends, impacts, and solutions - lessons management (e.g., interest-bearing accounts in a variety of curren- learned from eight country case studies, The World Bank, 2018; cies), international wire transfers, cheque clearing, payable-through the FSB reports on correspondent banking and remittances, 16 accounts and foreign exchange services. Domestic banks often rely March 2018, available at https://www.fsb.org/2018/03/fsb-re- on correspondent banks to handle transactions conducted involv- ports-on-correspondent-banking-and-remittances/ ing foreign counterparts. This allows domestic banks to gain access to a wider scope of financial markets and to extend their services – Erbenová, M., L. Yan, N. Kyriakos-Saad, A. López-Mejía, G. Gasha, abroad without the hassle of opening overseas branches. E. Mathias, M. Norat, F. Fernando, and Y. Almeida (2016), “The Withdrawal of Correspondent Banking Relationships: A Case for 5 Alternative arrangements for cross-border payments, which are Policy Action,”, IMF Staff Discussion Note, SDN/16/06, June; and less relevant than correspondent banking, today include the single De-risking, Council of Europe, available at https://www.coe.int/ platform or in-house/intragroup transfer model, the interlinking en/web/moneyval/implementation/de-risking. between the national payment infrastructures of different countries, and the peer-to-peer model, with the first of the three play a 11 This is particularly important as remittances have overtaken private relatively large role. For a description of these arrangements, see capital flows, FDIs, and aid in terms of total cross-border capital Enhancing Cross-border Payments—Stage 1 report to the G20: inflows and, in several countries, they represent a substantial por- Technical background report, cit. tion of GDP. 6 For a comprehensive analysis of the costs associated with cross-bor- 12 One of the first contributions in this area is Fintech and Finan- der payment and settlement systems, see Cross-Border Interbank cial Services: Initial Considerations, IMF Staff Discussion Note, Payment and Settlements: Emerging Opportunities for Digital SDN/17/05, June 2017. Transformation, a report by the Bank of Canada, Bank of England 13 See Digital Money Across Borders: Macro-Financial Implications, and the Monetary Authority of Singapore, November 2018. On the International Monetary Fund, October 2020. retreat of correspondent banking relationships for the past decade, 14 For all the aspects that relate to the concept of CBDC and use of see Rice, T., P. von Goetz, and C. Boar, On the global retreat of CBDC within jurisdictions (for domestic payments), the report refers correspondent banks, in “International banking and financial market to Central bank digital currencies, joint report by the Committee on developments,” BIS Quarterly Review, March 2020, 33-52. Payments and Market Infrastructures and Markets Committee, Bank 7 About a third of banks reported that they exited relationships for International Settlements, March 2018. which were no longer profitable or cost-effective because of the 15 To date only the Bahamas Central Bank has officially launched retail required due diligence or other economic reasons See Correspon- CBDC in its jurisdiction, and a recent BIS survey of central banks dent banking data report, Financial Stability Board, 4 July 2017. indicates that about 80 percent are engaging in work related to 8 See Recent trends in correspondent banking relationships—further CBDCs, and 40 percent have progressed to experiments or proof considerations, Policy Papers. International Monetary Fund, 21 April of concept. Se Boar, C., H. Holden, and A. Wadsworth (2020),. 2017. “Impending Arrival—A Sequel to the Survey on Central Bank Digital 9 See Correspondent banking data report, cit. 33 34  •  Central Bank Digital Currencies for Cross-Border Payments Currency.” BIS Papers 107, Bank for International Settlements, Basel. 22. The qualification of “universal” refers to a single CBDC being used 16 For a technical discussion of the architecture options for three types by all participating jurisdictions. Such a solution might raise an of CBDCs (wholesale, retail, and cross-border), based on stylized analogy with the use of the IMF’s Special Drawing Right (SDR) as examples, see Reference Architecture and Use Cases Report, ITU-T CBDC that the central banks of member governments would make Focus Group Digital Currency including Digital Fiat Currency, Focus available to their commercial banks (and possibly other financial Group Technical Report, International Telecommunication Union, institutions) for cross-border payments activity. Yet, the SDR would July 2019. not quite be associable to CBDC. The SDR is an international reserve asset, created by the IMF in 1969 to supplement its member 17 In January 2020, six central banks with significant expertise in countries’ official reserves, and it serves as the unit of account of exploring digital currencies created a working group to share CBDC the IMF and some other international organizations, including the experiences and assess use cases, which will assess CBDC use cases, World Bank. Its value is based on a basket of five currencies (the US including cross-borders. See Central bank group to assess poten- dollar, the euro, the Japanese yen, the British pound sterling and, tial cases for central bank digital currencies, Press Release, Swiss from 2016, the Chinese renminbi). However, the SDR is neither National Bank, Press release, Zurich, 21 January 2020. a currency, nor a claim on the IMF; it cannot be used directly in 18. Stakeholders included a group of commercial banks led by HSBC. market transactions. It can only be exchanged for these five major Other commercial banks in the group were Oversea-Chinese currencies and can therefore play a role in providing liquidity Banking Corporation, Toronto-Dominion Bank and United Overseas and supplementing member countries’ official reserves. SDRs are Bank. KPMG Services Pte. Ltd helped facilitate a workshop between generally allocated across member states according to the IMF’s these participants to discuss views on this topic, and compiled quota formula and principally used by developing states in need of views from them to assist in the development of the report. See hard currency. Members can also buy and sell SDRs in a voluntary Cross-Border Interbank Payment and Settlements: Emerging market. The SDR mechanism is self-financing and levies charges on Opportunities for Digital Transformation, cit. allocations which are then used to pay interest on SDR holdings. For 19. Just for brief reference, Model 1 is the collection of current and further information on the SDR, see Special Drawing Right (SDR), planned industry initiatives, aimed at enhancing in-country RTGS International Monetary Fund, March 24, 2020 systems (to expand operating hours, enhance payment status 23. This issue is discussed in the IMF’s Digital Money Across Borders: notification, improve standards, and reduce frictions) along with Macro-Financial Implications, cit. other improvements such as the adoption of common messaging 24. See Jasper—Ubin Design Paper: Enabling Cross-Border High Value standards, payments tracking and status visibility, improved KYC Transfer Using Distributed Ledger Technologies, Accenture and JP features, initiatives to link domestic payments infrastructures, and Morgan, 2019. initiative to facilitate the establishment of common rules and stan- dards for cross-currency/ cross-border payments. Model 2 is based 25. https://interledger.org/rfcs/0022-hashed-timelock-agreements/ on an expanded role for in-country RTGS operators that act as 26. Prior to Project Inthanon-LionRock, the two central banks had “super-correspondents” for settling cross-border payments instead embarked on their separate journeys to investigate the prospect of relying on intermediary banks as correspondent banks. Central of W-CBDC in the local context. The HKMA led some local financial banks allow RTGS operators of different jurisdictions to open institutions in 2017 to commence Project LionRock to study the accounts in their (central bank’s) books in the currency of the given benefits and risks of W-CBDC. The project included a PoC study RTGS operator. The central bank money for each currency resides on token-based CBDC and debt securities issued into a single in home jurisdictions only while, domestically, RTGS operators have DLT system. In 2018, the BOT initiated Project Inthanon with local a mirror account reflecting their balances with other central banks. participating banks to explore the feasibility of DLT to enhance This enables RTGS operators to effectively hold multicurrency Thailand’s financial infrastructure, as well as to encourage collabo- clearing accounts for their member banks without the need for the rative learning amongst the involved parties. Upon completion of latter to open nostro/vostro accounts globally. When banks need their respective domestic projects, the next step brought together to fund their nostro accounts, the process is similar to the process the BOT and the HKMA to explore how W-CBDC could improve today (i.e., the commercial banks still take the risk and source the efficiency in cross-border payments. funds); the only difference here is that the nostro accounts are now 27. See Inthanon-LionRock: Leveraging Distributed Ledger Technology with the RTGS operator and not with a correspondent bank. The to Increase Efficiency in Cross-Border Payments, report by the Bank various RTGS operators are linked, potentially via a common shared of Thailand and the Hong Kong Monetary Authority, powered by platform. C.RDA. 20. Drawing from the field of computer science, the term “atomic” 28. This is how links between CSDs are established currently allowing refers to an operation that requires a set of distinct changes to take securities listed in a exchange in one jurisdiction to be traded in effect altogether for the operation to be completed; otherwise, another jurisdiction—for example American Depository Receipts the operation has no effect and is reversed. Applied to financial of Glaxo Smithkline listed in the London Stock Exchange or ADR of transactions, an atomic transfer is complete if a number of events HDFC Bank listed in Bombay Stock Exchange (India). take place altogether; if it this does not happen, the transfer is not 29. While the BOT-HKMA report does not define the DR, the concept executed. As an example, consider a 100-dollar funds transfer from draws on the financial literature which defines it as a negotiable account X to account Y: first, the balance of account X must be financial instrument issued by a bank to represent a foreign compa- checked, and 100 dollars are removed from the account and, sec- ny’s publicly traded securities. The DR typically trades on local stock ond, 100 dollars are added to account Y, and the balance of account exchanges and its purpose is to facilitate buying shares in foreign Y is checked as the end. If the entire operation is not completed as companies, so that shares do not have to leave the home coun- one whole “atomic commit,” then several problems could occur. If try. A more precise definition of the DR in the context of foreign the system fails in the middle of the operation, after removing the exchange markets and cross-border payment services is offered by money from X and before adding into Y, the 100 dollars have just the R3 study discussed below in sub-Section II.C. disappeared. Another issue is that if the balance of Y is checked before the 100 dollars is added, the wrong balance for Y will be 30. The type of transaction that has been excluded is the sending of reported. With an atomic transfer neither of these circumstances local funds from a local bank to another local bank, since this would would materialize. In the former case, the transfer would be rolled be executed in the domestic settlement network. back, and the money would be returned to X. In the latter case, the 31. The system requires three details to be filled out: (1) type of board check of the balance of account Y would not occur until the transfer rate, whether bid or ask, (2) available amount, and (3) the quoted were completed. rate. Inputted quotes can be updated at any time. 21. See Enabling Cross-Border High Value Transfer Using Distributed 32. The denomination of these models derives from the company—R3 Ledger Technologies, Jasper-Ubin Design Paper, report by the LLC—that has developed them. It must be noted that they differ Bank of Canada and Monetary Authority of Singapore, powered by from R3 Corda model underpinning the Jasper and Ubin projects. Accenture and JP Morgan, 2019. R3 LLC is an enterprise blockchain technology company, headquar- Central Bank Digital Currencies for Cross-Border Payments  •  35 tered in New York City. It leads an ecosystem of more than 300 payment by rejecting it within the timeout or by letting the time- firms working together to build distributed applications on top out expire. Thus, the colluding parties can take advantage of the of Corda (known as CorDapps) for usage across industries such as Connector’s binding commitment depending on the movement financial services, insurance, healthcare, trade finance, and digital of the exchange rate. That is, they only execute the payment if the assets. It was founded in 2014 by David E Rutter. R3’s blockchain exchange rate moves in a favorable direction for them, otherwise platform Corda records, manages and synchronizes financial the payment is aborted. agreements and standardizes data and business processes. R3 47. This section builds on the extensive analysis reported in a World made Corda open source in November 2016, allowing the global Bank’s still-in-progress study on Africa-Wide Payments Platform: developer community to contribute to it, build on top of it, and to Framework Document. drive its design and adoption. 48. A fully centralized model for settlements denominated in multiple 33. See Cross-Border Settlement Systems: Blockchain Models Involving currencies is nevertheless also possible, the prime example being Central Bank Money, R3 study by Zhao, X., H. Zhang, K. Rutter, C. CLS Bank International. Thompson, and C. Wan, 22 January 2018. 49. This would of course require that issuers of global reserve curren- 34. Notice that the term “cryptocurrency,” which is part of the original cies (US, Eurozone) would be issuing their own CBDC and accept to name of the model chosen by its authors (and for this reason make their CBDC available as settlement currency of regional CBDC retained in this subsection), might be misleading as it is usually arrangements. meant to indicate privately mined coins with no central issues. Here, 50. See Project Inthanon-Lionrock, and R3 Option 2. in fact, the model developed by the author refers to an “intermedi- ate cryptocurrency” as a liability jointly issued on a single ledger by 51. Inter-ledger protocols could be used for this purpose. These pro- the central banks participating in the arrangement. Project Aber, to tocols are used for payments across different networks. A protocol be discussed below, is a good example of such a model. connects ledgers from two different entities, such as banks. 35. Conceptually, this would be similar to the Universal W-CBDC dis- 52. See Enhancing Cross-border Payments—Stage 1 report to the G20: cussed earlier. Technical background report, cit. 36. The report does not further describe the underlying payments 53. All that follows in this table as regards cross-border CDBCs excludes infrastructure. references to BOC-BOE-MAS Model 3a (discussed above), since, as noted, the model closely resembles the existing correspondent 37. Existing platforms such as Stellar could provide part of the infra- banking arrangement and features many of the latter’s same chal- structure for this type of solution. lenges. 38. In a CBDC context, a settlement cycle would typically start with 54. This (realistically) assumes that central banks would run a CBDC participants’ pledging cash collateral into a special pooled account facility as public utility, possibly under cost-recovery rules and held by the central bank against the issuance of an equal amount pricing criteria that would take into account transactions volumes of a central bank-issued digital cash that can circulate through the and value by participants, but with no surcharges aimed to extract distributed ledger. Two banks can then send digital-cash payments extra-profits from its operation. to each other in real time to meet payment obligations they have agreed to settle on DLT. Alternatively, a bank may also cash out its 55. It is (realistically) assumed that central banks would run a CBDC digital-cash holdings by converting them back into cash at the facility as a financial market infrastructure or payment scheme and central bank. apply for it all relevant international standards. 39. At CRDR expiration, Central Bank B redeems it in its own currency B 56. See Overview of Saudi Arabia’s 2020 G20 Presidency—Realizing at Central Bank A at the same exchange rate. The foreign exchange Opportunities of the 21st Century for All, Riyadh, 1 December 2019, risk is thus born by Central Bank A. and Enhancing Cross-border Payments Stage 3 roadmap, report by the Financial Stability Board, 13 October 2020. 40. The TTP model is presented under four variants, differentiated according to the type of entity that issues the depository receipt, its 57. The models address to Committing to a joint public and private role in the clearing and settlement processes, and the risks involved sector vision to enhance cross-border payments (Focus A) and and who bears them. Thus, the report evaluates a private sector-is- Exploring the potential role of new payment infrastructures and sued intermediate cryptocurrency, a pre-funded escrow account arrangements (Focus E), by contributing to Developing a com- model, a Continuous Linked Settlement (CLS)-like model, and a mon cross-border payments vision and targets (Building block 1), deficit-fund ed account model. Developing a common cross-border payments vision and targets (Building block 3), Considering the feasibility of new multilateral 41. See Project Aber, Saudi Central Bank and Central Bank of the platforms and arrangements for cross-border payments (Building U.A.E. Joint Digital Currency and Distributed Ledger Project, Block 17), and Factoring an international dimension into CBDC available at https://www.centralbank.ae/sites/default/files/2020-12/ design (Building block 19). Aber%20Report%202020%20-%20EN.pdf 58. The World Bank Group and international agencies like the Bank 42. See Hyperledger Fabric documentation, available at https://hyper- of International Settlements, the African Development Bank, the ledger-fabric.readthedocs.io. Arab Monetary Fund, the Nordic Investment Bank, and others, may 43. See McKinsey, Global Payments 2016: Strong fundamentals despite acquire and use SDR in transactions by agreement with the IMF. uncertain times, available at https://www.mckinsey.com/~/media/ 59. This is not always the case, however; see, for instance, the case of McKinsey/Industries/Financial%20Services/Our%20Insights/A%20 the EBA Clearing’s payment systems. mixed%202015%20for%20the%20global%20payments%20 industry/Global-Payments-2016.ashx. 60. One type of CBDC cross-border payments would be an exception to this conclusion. It includes the payments and transfers of funds 44. That is, commercial banks can agree on payments bilaterally with- denominated in the currency of the issuing central bank juris- out involving a trusted third party. This ensures that a commercial dictions and effected in favor of residents of other jurisdictions bank can move its funds freely between the channels in which it (individuals, businesses, financial institutions, government agen- participates without having to depend on the central banks or any cies). These transactions would be made possible by the issuing other central party. central bank granting CBDC access to non-residents, with the only 45. See Project Stella: Synchronised cross-border payments, joint requirement that non-residents be allowed by their jurisdiction research project of the European Central Bank and the Bank of of residence to hold accounts or wallets denominated in foreign Japan, June 2019. currencies. Of course, such type of cross-border payments would 46. An example would be one where the Sender and the Receiver col- permit only payments denominated in the currency of the issuing lude after initiating a transaction and thereby effectively locking the central bank jurisdiction. Connector’s liquidity. In this instance, the Receiver has the option 61. See Retail CBDCs: The next payments frontier, report by OMFIF to either execute the payment (fulfil), or alternatively abort the and IBM, 2019. The report findings were informed by 23 central 36  •  Central Bank Digital Currencies for Cross-Border Payments banks, which participated in an OMFIF survey conducted between cannot cause the loyal generals to adopt a bad plan. For further July-September 2019. details refer to Marshall Pease, Robert Shostak and Leslie Lamport 62. Digital identity verification is essential to the operation of CBDCs, (1982) “The Byzantine Generals Problem”, ACM Transactions on particularly in cross-border transactions. Tradeable digital assets Programming Languages and Systems, Volume 4 Issue 3, July 1982, must be tied to a digital identity system, which in turn should be 382-401. See https://www.academia.edu/27660079/The_Byzan- tied to an automatic KYC and AML/CFT verification system. This is tine_Generals_Problem.pdf?source=swp_share. a foundational step to the potential use of CBDCs, and emerging 72. Consensus means that the entire network reaches an agreement developments in regulatory and compliance technology may bene- on the transacting value. It is a vital component of decentralized fit central banks’ experiments in the digital currency space. network. 63. An example is what Novi (an app of Facebook) proposes to do for 73. FBA comes to agreement on state updates using a unique slot Diem (erstwhile Libra) stablecoin. See Facebook-backed digital where update dependencies between nodes are inferred. Nodes coin Libra renamed Diem in quest for approval, by Anna Irrera and must agree on the slot update in each round of consensus. How- Tom Wilson, Reuters, 1 December 2020, available at https://www. ever, since the system is open to nodes joining and leaving the reuters.com/article/facebook-cryptocurrency-int/facebook-backed- network at will, a majority-based quorum consensus mechanism digital-coin-libra-renamed-diem-in-quest-for-approval-idUSKB- will not work. Instead, the FBA in the SCP employs quorum slices N28B574. that are subsets of quorums that are capable of convincing particu- 64. See Central Bank Digital Currency and the future: Visa publishes lar nodes of an agreement. See the SCP White Paper (https://www. new research, by Cuy Sheffield, Head of Crypto, Visa, 17 Decem- stellar.org/papers/stellar-consensus-protocol.pdf) and Blockonomi ber 2020, available at https://usa.visa.com/visa-everywhere/blog/ (https://blockonomi.com/stellar-consensus-protocol/) for more bdp/2020/12/17/central-bank-digital-1608165518834.html. The details. report explores the offline exchange of digital cash and how it 74. Atomic swaps, or atomic cross-chain trading, is the exchange could benefit consumers and economies worldwide. of one cryptocurrency for another cryptocurrency (or rarely, fiat 65. This annex draws on Li, Z., and S. Bewaji, How cross-border pay- currency), without the need to trust a third-party. A relatively new ments are evolving, Payments Canada, 26 November, 2019. piece of technology, atomic cross-chain trading looks to revolution- ize the way in which users transact with each other. For example, 66. See What is ‘Utility Settlement Coin’ really?, available from: https:// if Alice owned 5 Bitcoins but instead wanted 100 Litecoins, she ftalphaville.ft.com/2017/09/18/2193542/what-is-utility-settle- would have to go through an exchange, i.e. a third-party. However, ment-coin-really/. with atomic swaps, if Bob owned 100 Litecoins but instead wanted 67. See J.P. Morgan creates digital coin for payments, available from: 5 Bitcoins, then Bob and Alice could make a trade. In order to https://www.jpmorgan.com/global/news/digital-coin-payments. prevent, for example, Alice accepting Bob’s 100 Litecoins but then 68. These included four of the Canadian designated systemically failing to send over her 5 Bitcoins, atomic swaps utilizes what is important banks. known as hash time-locked contracts (HTLCs). For more on atomic 69. See Ripple, available on https://www.ripple.com/. swaps see https://www.cryptocompare.com/coins/guides/what-are- 70. See Stellar, available on https://www.stellar.org/lumens/. atomic-swaps/. 71. First proposed in 1982 by Marshall Pease, Robert Shostak and Leslie 75. Steem is the Steemit ecosystem issued stablecoin with a construc- Lamport, the BGP conceptually imagines that several divisions tion similar to a self-contained macroeconomy having what is akin of the Byzantine army are camped outside an enemy city, each to government securities with short to longer term note maturity/ division commanded by its own general. The generals can com- term structures. For an overview of the economics of Steem, refer municate with one another only by messenger. After observing the to https://steemit.com/steem/@spectrumecons/steem-explained- enemy, they must decide upon a common plan of action. However, by-an-economist. some of the generals may be traitors, trying to prevent the loyal generals from reaching agreement. The generals must therefore devise an algorithm to guarantee that i) all loyal generals decide upon the same plan of action and ii) a small number of traitors