Telegraph to Tether: Challenges in the Global Payments System and the Struggle between Private and Public Interests

Abstract

Cross-border payments systems are fundamental to the smooth operation of the global economy, but they have attracted relatively little public or academic interest. To support trade and commerce, the system needs to be fast, efficient, inclusive and transparent but these characteristics are challenged by long transaction chains, fragmentation and complexity, legacy technology and high compliance costs. The geopolitical importance was highlighted by the exclusion of Russia from Society for Worldwide Interbank Financial Telecommunications (SWIFT) in 2022. This chapter examines the changing technology and the origins of two key institutions (SWIFT and Clearing House Interbank Payments System in New York, or CHIPS) with special focus on the private ownership of this public good. It draws on new archive evidence from central banks, commercial banks and multilateral institutions to offer fresh insights into how the global payments system has evolved.

Research was funded by the ‘Global Correspondent Banking 1870–2000’ project which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant Agreement No 883758). I’m grateful for research assistance from Thibaud Giddey.

In March 2022, the governments of the US and Europe decided on various sanctions in response to the Russian invasion of Ukraine, including asking for Russian banks to be excluded from the global payment system Society for Worldwide Interbank Financial Telecommunications (SWIFT).¹ That this sanction relied on cooperation from a private sector institution drew public attention to an otherwise obscure but fundamental aspect of globalisation: namely, the way funds move across borders. While most of the public was unaware of how their foreign goods and services were paid for, governments and multilateral agencies were already investigating the vulnerabilities in the underlying plumbing of global payments. In order to support global trade and commerce, the system needs to be fast, efficient, inclusive and transparent but these characteristics have been challenged by long transaction chains, fragmentation and complexity, legacy technology, weak competition and limited operating hours.² Improving the way that cross-border payments are made has been a key focus of the G20 and the Financial Stability Board since 2015.³

A key concern is that the correspondent banking system has been contracting over the past ten years.⁴ Figure 1 shows the decline in active pairs of correspondent banks since 2011 while the value and volume of SWIFT messages has increased (including a surge in value and fall in volume at the start of the pandemic in March 2020).

Fig. 1

Value, volume of message; number of correspondent relations (Index Jan 2011=100)

The fall in the number of active bilateral bank relationships on the SWIFT messaging service is particularly stark for some regions compared to others, likely due to the increased costs of checking that counterparties conform to rules to stop money laundering and the finance of terrorism (see Fig. 2). But some regions may be excluded from the benefits of efficient cross-border payments. Or, customers might have to divert their payments through less reliable routes.

Fig. 2

Index of number of active correspondents (2011=100)

The time taken to settle payments also varies widely from 12 minutes for payments to the United States to over 22 hours for some countries in North Africa and South Asia using SWIFT’s new streamlined service.⁵ Understanding how the global payments system developed can deliver important insights into how it might be reformed. How did this obscure but essential system develop over time? Why is it primarily in private ownership rather than public control? What are its weaknesses and how might digital currencies disrupt the system?

The challenges posed by time and distance are as old as international trade itself. For centuries traders went to their banks to borrow funds to cover them for the time it took for their goods to travel over distances before they were paid by their customers.⁶ Access to credit shrank time and distance, but settlement ultimately required a debit in the account of the buyer’s bank in one country and a credit to the seller’s bank in another, and therefore relied on a communication system, such as letters posted between banks to adjust their accounts. The speed and efficiency of this correspondence was hugely enhanced by the spread of undersea and coastal telegraph in the second half of the nineteenth century.⁷ These cables, heroically unwound across vast ocean floors, are still the fundamental framework of telecommunications today. Most internet traffic still relies on undersea cables that closely follow the networks laid before 1914.

Late nineteenth-century globalisation was an era of intense flows of people, goods and capital. With telegraph cables in place, the network of banks expanded quickly so that traders could transact with buyers anywhere in the world through banks in London or other regional centres. The importance of communication is evident in the name for this relationship formed between banks for the purpose of settling accounts, which became known as correspondent banking. Instructions between banks to credit/debit customer accounts were sent in complex code to shorten the messages and thereby reduce costs. Processing was labour intensive and costly requiring complex and error-prone coding and decoding of instructions and confirmations. The Bank of England, for example, used a 5-letter code for all verbal telegrams, which consisted of 582 code words and 388 pages of English phrases in 1939.⁸ Nevertheless, cables remained the standard messaging system well into the twentieth century when telex began to be adopted. Developed in the 1920s, Telex machines were more automated and reduced spending on staff, albeit with higher upfront costs to purchase the machine and slightly higher message fees. Coding was no longer necessary since fees were charged by time rather than by length of the message itself. Although the platforms and technology changed, the correspondent banking system of mirrored account debits and credits remains the fundamental framework of global payments despite the dramatic transformation of global trade and finance over the past 150 years. A simplified diagram is shown in Fig. 3.

Fig. 3

Global payment system. (Note: Payment may or may not go through a national payments system)

After the telegraph and Telex, the most important payments innovation was the computer, which helped a large volume of transactions to be settled among a complex network of correspondent banks, but introducing this technology was challenging. In the 1970s, two important private sector initiatives introduced computer technology for cross-border payments; Clearing House Interbank Payments System in New York (CHIPS) formed in 1970, and the Society for Worldwide Interbank Financial Telecommunications (SWIFT) activated in 1977 in Europe. CHIPS is a clearing and settlement system for global USD payments while SWIFT is a messaging system to communicate payments instructions between banks. These initiatives emerged in an era of intensifying cross-border banking and set the norm for cross-currency payments infrastructures to be controlled and governed in the private sector.⁹

Origins of Clearing House Interbank Payments System (CHIPS) and Settlement Risk

The most recent era of globalisation is usually dated from the end of the Cold War in 1989, when there was a burst of ICT innovation and the volume of financial flows swept ahead of the growth of international trade or production, but the basis for the great financialisation of the international economy was set in previous decades. The innovation of the Eurodollar market from the end of the 1950s led to exponential growth in international dollar balances in the later part of the 1960s.¹⁰ The opportunities presented by this innovation, the acceleration of US multinational companies (MNCs) moving abroad and the prospect of European integration all encouraged banks into a new phase of internationalisation.¹¹ American banks surged into the City of London to take advantage of the dollar money markets there, outside the jurisdiction of their home regulators.¹² As British and European banks confronted this competition for their staff and customers, they sought to maintain their market share by expanding abroad. The float of the pound sterling exchange rate in June 1972, followed by the US dollar in March 1973 created new foreign exchange trading business for banks, but also new market risk from less predictable currency rates.

CHIPS was prompted by the urgent need to expand capacity once capital controls became less effective and the volume of international payments increased.¹³ Until April 1970, international payments in New York were settled by cheques issued by correspondent banks—a paper-based system. Foreign banks did not keep large balances with their New York correspondents while they moved money through their accounts and daily turnover was many times the capital of the banks themselves, which created liquidity and operational risks.¹⁴ First, the process involved several steps, each of which introduced the risk of clerical error. Foreign banks cabled or telexed their New York correspondents to debit their account to make payments on behalf of customers. This required careful checking of the messages and balances available and then printing, typing and physically delivering official bank cheques to the Clearing House in New York. Using bank cheques (not corporate cheques drawn directly on the customer) meant that settlement was an obligation of a bank, not of the individual customer, so the New York correspondent had to make sure of the creditworthiness of its counterparty, ensure the telex was authentic, that either the foreign bank had sufficient funds in its account or that the correspondent New York bank was willing to extend credit for the amount of the payment, and that the cheque was filled out correctly and properly delivered to the Clearing House by one of an army of bank messengers. Bankers tended to wait until the last of the five daily clearings to submit most of their cheques to try to ensure that funds had already arrived to cover them, and this caused congestion at the end of the day.¹⁵

By the late 1960s, the volume of transactions made this system excessively cumbersome. Large numbers of staff had to be employed to undertake the clerical tasks, mainly young and with minimal training. In 1974 Robert J. Crowley, Vice President at the Federal Reserve Bank of New York (FRBNY), reported that ‘one New York bank had its payments staff drop from an average of 22 years in the bank to an average of 8 months in a two-year span’.¹⁶ This staff constraint made back-office operations more prone to clerical errors.¹⁷ Cost was another issue: Crowley observed that ‘Two banks told me they estimated their processing cost per official check to be between $5 and $7’.¹⁸ The size of the business challenged the back-office resources even of large banks.

CHIPS allowed banks to make coded entries through computer terminals during the day, with a final netting out at the end of the day at the hub computer in the New York Clearing House. Each member had 30 minutes to meet any overall deficit, with final payment through accounts with the FRBNY. The value of transactions through the system surged but the perils of time and distance had not been overcome. Several issues quickly emerged and some persisted until the mid-1980s. Access was restricted to banks in New York who were members of the Clearing House and so excluded branches and subsidiaries of foreign banks as well as most US banks directly. Banks that were not members of the Clearing House had to access the system indirectly through a member. Secondly, Clearing House funds were traditionally only freely available to payees on the business day after payment to allow time for clerical work to reconcile payments among the New York banks. Thus, dollars deposited at correspondent banks in exchange for foreign currency were available for use the following day, but the foreign exchange could be used immediately. This left banks exposed to counterparty credit and liquidity risk. A further implication was that Friday funds were not available until Monday, so there was an incentive to book on Friday and hold over of the funds to put into money markets over the weekend, creating more congestion. Thirdly, turnover during the day for each bank was many times greater than the deposits they held for their respondent banks. These ‘Daylight Overdrafts’ could become extremely large while the interbank deposits covered only a tiny proportion of turnover. More fundamentally for an international system, the final clearing in New York took place hours after European and Asian banks had closed so the system supported billions of dollars of unsecured credit between banks over the course of the day until payments were finally cleared and settled. As Crowley noted, ‘it is a business in which trust and confidence is essential’.¹⁹ This trust was soon tested.

In July 1974 the collapse of the Herstatt Bank in Germany almost brought down the USD payments system when it exposed the risks in correspondent banking.²⁰ The fraudulent Herstatt Bank was closed by the German authorities while its New York correspondent, Chase Manhattan Bank, was still waiting for inflows to release funds to Herstatt’s counterparties in New York. Trust evaporated and CHIPS payments froze. Banks only agreed to resume making payments under pressure from the Federal Reserve, and even then, they insisted on waiting for funds to arrive before paying out to customers, causing a lag that could last up to several days in the world’s main dollar settlement system. The episode was so striking that cross-border settlement risk posed by the potential collapse of one counterparty is still known as Herstatt Risk.

When the Herstatt crisis struck, public authorities were drawn in to get the market going again and to pressure this private sector institution to reform. The Fed encouraged the CHIPS banks to introduce more safeguards over the next ten years including minimum size of participant banks, letters of comfort from parent banks for subsidiaries, speeding up settlement and limits on intraday credits. As an indicator of the erosion of trust, Fig. 4 shows that it took 12 years for the average size of transactions through the system to recover to the level of 1974—and that was nominal value during a period of rapid inflation and global trade growth.

Fig. 4

Average size of payment, in thousands of USD

Part of this decline in the average size of transactions may have come from greater netting of payments before they went through CHIPS, but netting was not widely practised until the 1990s as will be shown below. In a rare survey in 1989, intraday credit in CHIPS was measured at almost $54 billion.²¹ In 2000 CHIPS finally moved from end of day (deferred net settlement) to intra-day final settlement by requiring participants to deposit funds upfront at a special account at the FRBNY. The liquidity of the system was further enhanced in mid-2008 by the ability of banks to add Supplemental Funding to clear payments. The Fed’s injection of bank liquidity in the wake of the crisis of September–October 2008 increased the resources for banks to provide Supplemental Funding to keep the system functioning, although payments left unresolved by 5 pm did increase in the short term.²²

Beyond CHIPS, settlement risk continued to challenge the resilience of global payments into the 2000s. While central banks remained concerned about systemic risks arising from cross-currency payments, they did not seek to replace private sector initiatives and restricted themselves mainly to encourage and inform and latterly to provide some form of oversight that fell short of supervision. A key method they promoted to reduce bank exposure was through schemes that net payments on a continuous basis so that a smaller value needs final settlement at the end of the day. After a range of private sector netting proposals emerged, the G10 central bank governors set up an ad hoc Committee on Interbank Netting Schemes in 1989 to review the issues and then established a more permanent Committee on Payment and Settlement Systems (CPSS) in 1990. It quickly aimed to set the terms for central bank oversight.²³ While recognising each nation’s sovereignty over its own national payments system, the report encouraged communication among central banks for cross-border payments to promote systemic stability and harmonisation. Nevertheless, ‘[T]he primary responsibility for ensuring that netting and settlement systems have adequate credit, liquidity, and operational safeguards rests with the [private sector] participants’.²⁴ Central banks did not want to take responsibility. During the 1990s a series of ripples in the market caused by the failures of counterparties (such as Bank of Credit and Commerce International (BCCI) in 1991) drew central bankers back into discussion about whether the private sector was best placed to reinforce the framework’s scaffolding. In 1996 the CPSS ‘call[ed] upon individual banks and industry groups alike to improve current practices and devise safe mechanisms for addressing settlement risk’.²⁵ The market response was mixed; a survey two years later found that ‘over 60% of the banks in the surveys are still underestimating their exposures’.²⁶ At this point the CPSS acknowledged that the 1990s strategy ‘founded on the belief that the private sector, with the active support of the public sector, had the power to contain the risk that first came into focus at the time of the 1974 failure of Bankhaus Herstatt’ had not produced robust solutions. This time they sought to push banks further by threatening to increase regulatory capital for foreign exchange exposure.²⁷ In 1997 seven banks launched CLS (continuous linked settlement) in London as a cross-border netting scheme for foreign exchange transactions on a payment-versus-payment basis so that payments were not released unless there were funds available to clear them. It took a further seven years to become operational, initially for a limited group of 39 banks and seven currencies.²⁸ Thirty-eight years after Herstatt collapsed, this private sector solution did finally reduce counterparty risk. In September 2008, CLS was able to clear almost all of Lehman Brothers’ trades via Citibank, its primary cash dealer.²⁹ The private sector system proved its resilience, albeit with the support from central banks to restore global liquidity in the weeks that followed.

Origins of SWIFT and Messaging

When CHIPS launched in 1970, European banks had already spent a year planning for a standardised messaging system to underpin international payments. The aim was to save time translating from one system to another and hopefully prompt fewer errors while making global payments more efficient. But unlike the case of CHIPS, which was based in a single national clearing house, agreement at a multilateral level proved challenging. SWIFT is a non-profit cooperative among member banks, headquartered in Brussels, and is the dominant global messaging standard. But there were moments when central banks challenged whether it should be a private sector initiative. As in the case of settlement and clearing, there is a public interest in the stability and resilience of the messaging system that instructs correspondent banking transactions. Archive evidence shows that in 1969–1971 technicians from the central banks of major industrialised economies tried to take part in the design of SWIFT and to secure a place in its governance, but they were rebuffed by the commercial banks leading the planning. Commercial banks had better technical knowledge about the new-fangled computer systems and collectively had the capital to invest in the technology required. Most central banks were behind the curve technologically and in the end merely joined as members alongside the world’s commercial banks.

For Scott and Zachariadis, the ‘galvanising’ event that hastened SWIFT was the launch of First National City Bank’s own proprietary messaging standard, MARRTI (Machine Readable Telegraphic Input), developed in 1973 and implemented in 1974/1975.³⁰ FNCB tried to use its market power to enforce the use of its standard, giving fresh impetus to the European-led, cooperative framework of SWIFT. But new archival evidence shows that the initiative rested with European bankers who were anticipating the integration of European banking systems. During the late 1960s groups of banks from the key markets in Western Europe formed clubs and consortia to confront competition from larger American banks in their home markets and to prepare for European financial and monetary integration. These started out as loose agreements, but sometimes led to distinctly incorporated institutions that often included a US bank. For the most part, these were non-competition agreements, but they also offered the opportunity for more practical cooperation.³¹ The international payments system became one of the areas of cooperation.

In 1969 three plans were underway in Europe to develop cross-border inter-bank communication networks.³² Societe Financiere Européenne (SFE) was a consortium of Barclays Bank, Algemene, Banca del Lavoro, Bank of America, Banque Nationale de Paris (BNP) and Dresdner Bank. Members of SFE met at the end of 1969 to set up a Steering Group representing all the countries that made up the consortium plus Switzerland, Sweden and Belgium. In January 1970 Barclays and BNP wrote a feasibility report for an International Inter-Bank Message Switching System and it was circulated to other banks both within and beyond the SFE. Meanwhile another set of bankers were brought together by the Bilderberg Group chaired by Prince Bernhard of the Netherlands. A third system was developed separately by Societe Generale de Banque, which was at an advanced stage and ready to become operational to link Brussels, London and New York in 1971.³³ It was clear that competing systems made no sense given the importance of network externalities to achieve the gains from a messaging system, so the European banking consortia eventually fell in behind the SFE proposal. It was more inclusive than SGB’s scheme albeit less well advanced.

Evidence from the archives of central banks, commercial banks and the Bank for International Settlements (BIS) shows that central bankers did worry about the dominance of the private sector in the design of cross-border payments. At the end of January 1969 the BIS convened its first meeting of experts on the use of computers in central banking.³⁴ At this point there was a range of take up of new technology among central banks; the Swiss National Bank had no computer and did not see that its operations would require one for the next five years,³⁵ Canada rented computer time rather than having its own and the Bank of England used computing for a range of administrative purposes including registration of government debt, clearing cheques and giro payments with commercial banks.³⁶ The Banca d’Italia had the widest use for all large government receipts and payments as well as its own research purposes.³⁷ A key obstacle was the difficulty in attracting and retaining skilled computing staff due to the relatively low salaries on offer at central banks. Given the mixed levels of engagement and expertise in computerisation, it is perhaps not surprising that the initiative for computerised cross-border messaging came from the private sector. Nevertheless, this central bank computing committee became a main conduit between MSP/SWIFT and the BIS and G10 central banks as they came to consider the computerisation of international payments.

John J. Clarke, Vice President and Special Legal Adviser at Federal Reserve Bank of New York (FRBNY) was particularly concerned about the absence of central bank influence over the Message Switching Project (MSP). He wanted G10 central bank representatives meeting at the UN Commission on International Trade Law (UNCITRAL) in December 1970 to seek a consensus that ‘central banks interest themselves in proposals now being pursued by sectors of the commercial banking communities in several countries looking toward an international payments mechanism that would be computerized on a multilateral basis’.³⁸ Clarke worried about the proliferation of incompatible standards and believed it was ‘natural that the central banks should be the leaders’ even if such a system was a long way off ‘and may take a form now undreamed of’.³⁹ Despite a lack of support from most central bank participants in the UNCITRAL meeting, Clarke tabled his paper, but he faced resistance from, for example, Bank of England representatives.⁴⁰ As the central bank governing the world’s largest foreign exchange market at the time, the Bank of England was hugely influential.

In New York, Clarke’s position had hardened. At the start of 1971, three years before the Herstatt Bank collapse, he argued that ‘one of the most serious problems facing international banking today is that of promoting and maintaining efficiency in effecting international transfers of funds’ and suggested that the central banks at Basle set up ‘possibly with commercial bank participation’ an organisation to establish ‘a computerized multilateral payments mechanism, harmonized standards for negotiable instruments’.⁴¹ He noted that European commercial banks were seeking to create their own message switching system for the international flow of payment orders ‘as well as a clearing house type operations’ without any central bank representation.⁴² Clarke believed, ‘The central banks, it seems, have the prestige and the resources to be of assistance to the international banking community in regard to this matter of common interest and concern’.⁴³ It is not clear what level of support Clarke had at the FRBNY for his views, which were firmly couched in personal terms, but he was not a lone voice.

The Banque de France contributed a paper from January 1971 on ‘The Use of Computers in International Transfers of Funds’ between correspondent banks. Echoing Clarke’s views, the paper emphasised the problem of incompatible standards emerging.⁴⁴ Given the momentum of proposals it was ‘of some importance for the central banks to take up a position in this matter’. The report concluded that ‘in view of the development of private projects, we believe that the central banks should rapidly come to an agreement to set up a permanent body to study the automated transmission of data concerning international payments’ and recommended that the G10 central bank governors take a position on whether to pursue this or not.⁴⁵ The National Bank of Belgium also pushed its peers to give more consideration to automated international payments.⁴⁶ Like others, they pointed out the risks of incompatible networks and the benefits of standardising codes etc. but they also emphasised the special needs of the EEC as it moved towards monetary union. Clearly there were several straws in the wind when the topic was due to be discussed at the May 1971 BIS Meeting of Experts on Computing.⁴⁷ At this point, BIS staff also sought views on whether the BIS itself could ‘establish a multilateral mechanism for effecting international payments, transfers and clearings on a computerised basis in an off-line or on-line mode, both for central and commercial banks’.⁴⁸ There was no general enthusiasm among central banks for the BIS to expand its role in this way.

Meanwhile commercial banks continued their planning, setting up national working groups of local banks to broaden the scope of engagement.⁴⁹ At this point, the scheme was to be as ‘widespread as possible with the stress on clean payments; the study of a clearing should not be excluded’.⁵⁰ A clearing system was still an ambition of the group and this drew in central bankers at the highest level.

At the meeting of Governors at the BIS in September 1971, the central banks of the G10 agreed formally that the Computing Experts Meeting should ‘study the possibility of using a communications system for effecting international payments similar to the one which was being studied at that time by the Central banks of the EEC’ and that the Meeting should contact commercial banks that were ‘considering establishing alternative networks with the aim of discouraging the development of a series of possibly incompatible telecommunications facilities’.⁵¹ This proved too late for the MSP discussions, which by December 1971 had already moved to feasibility studies. Rather than creating their own system, central banks tried to make a direct contribution to the design of the commercial banks’ scheme.

A month later, the chair of the BIS Meeting of Experts wrote to the commercial banks’ MSP Steering Committee to ask about the possibility of central banks and the BIS joining the project. The Steering Committee considered this at its meeting in Rome in March 1972, but the outcome was not what central bankers had hoped for. The banks decided only to invite central banks to participate as individual members ‘in the same position as the participating commercial banks’. Each Steering Committee member agreed to contact their national central bank to discuss the project, and to invite them to join through their national working group, but central banks were excluded from the Steering Committee itself. The banks also proposed to take advice from their national central banks on how the BIS should be involved rather than respond directly.⁵² The preference by the Steering Committee at this stage was for the BIS to join merely as part of the Swiss national working group.⁵³

A BIS Meeting of Experts was quickly convened in April 1972 in Basle to discuss this outcome. Central bankers were urged to pay the fee and join the MSP ‘in order to obtain the full documentation presented to the existing membership’.⁵⁴ For most central banks, the MSP scheme itself offered few practical benefits because their own correspondent traffic was relatively small but they needed to know what was going on and try to introduce their own views as market supervisors. The computing experts still hoped that the BIS could represent central banks on the Steering Committee, separate from the national committees. By this time the clearing facility, which had been part of the original vision, was dropped ‘largely because of the troublesome political and jurisdictional problems it was likely to generate’.⁵⁵ Technically and constitutionally, though, the banks were designing the system so that ‘a clearing facility could be grafted on without drastic changes to the basic structure’.⁵⁶

Shelving the payments element reduced the impetus for central bank involvement. As agreed, the chair of the Experts Group requested that the BIS should join the MSP as a ‘separate entity’ in the Steering Committee because of its status as an international organisation but this request was turned down.⁵⁷ The MSP Steering Committee’s view was that ‘it is the only object of the project to forward messages relating to international payments by means of a message switching system. As such, the system will not affect actual banking activities and it was therefore that the members of the Steering Committee could not see why a special position would be taken by the central banks’.⁵⁸ Since payments were not being settled or cleared through the system, the MSP Steering Committee saw no special role for central banks.

Opinion among central bankers was mixed. The Bank of England viewed the scheme mainly as a communications channel between banks, with no processing or settlement features, at least so far.⁵⁹ Participant banks would still be subject to each country’s exchange control for the payments entered through the network so that the scheme was not ‘fundamentally different from the present system’. The impact on central banks themselves would not be great because of the low level of their own correspondent business. Most other G10 central banks ‘had reservations as to the further evolution of the proposed network, with a number of participants feeling that it was as a mechanism for effecting clearing and settlement transactions that the MSP would in fact be of most value to the network users’ and therefore was a likely outcome.⁶⁰ The MSP was also likely to be expensive in terms of hardware and subscriptions compared to direct telex costs.⁶¹ Nevertheless, by July the Dutch, Italian, UK, Swedish, German and Canadian central banks had joined their national working groups and paid the $3200 fee to get access to the MSP documentation. The Federal Reserve’s position in their national working group was complicated because it was led by the American Bankers’ Association, which had difficult relations with the Fed.⁶² The French and Swiss were waiting for the outcome of the July Experts Meeting and the National Bank of Belgium had not paid ‘but professed to be fully informed’.⁶³ The drawbacks to joining were mainly the cost and ambiguity in the proposed by-laws. On the positive side, there were several advantages that central banks could get from joining⁶⁴:

• protecting the interests of small banks and non-commercial credit institutions

• establishing a link with national clearing systems

• ensuring compatibility of standards

• possible facilities for collecting relevant statistics

• central banks’ responsibilities as monetary authorities

By September all G10 central banks had joined except for the Swiss National Bank (for constitutional reasons) the FRBNY (because of issues with the ABA), the Bank of Japan and the National Bank of Belgium. Given that the organisation was due to be based in Brussels, the position of the National Bank of Belgium was intriguingly negative. They agreed that it was useful to maintain informal contact with SWIFT developments but believed that central banks would not be able to influence it from inside or use it to get access to data to fulfil their supervisory responsibilities.⁶⁵ In the end, however, the Governor of the National Bank of Belgium did not oppose the general view that in principle the G10 central banks should join SWIFT.⁶⁶ Still, the potential for SWIFT to pose both challenges and opportunities for central bank supervisors meant central bank governors continued to worry about whether they should be more active participants in the design of the system.

In early 1974, the BIS Group of Computer Experts advised the Governors of the BIS of their concerns about the security and reliability of the newly founded SWIFT cooperative. As a result, the G10 central bank governors approved a proposal to set up a Working Party on Electronic Systems for International Payments that included representatives from the Bank of Italy, the FRBNY and the Bank of England to determine whether central banks should take further action in regard to SWIFT before it became operational.⁶⁷ After three years of deliberations, as SWIFT was due to be launched in 1977, they concluded it was unlikely that helpful regulatory information could be collected by central banks from the proposed system. By this time, it was clear that SWIFT was set to remain merely a messaging system rather than morph into a settlement or payments system per se. As a result, the Working Party recommended that central banks did not have a direct interest in SWIFT. Instead, the BIS and central banks should merely join the system and ‘monitor developments’. They continued to have concerns over the decades that followed but the chance to be in on the design of the system had been rejected.

These sometimes tortuous discussions demonstrate the gulf in expertise between the commercial banks and central banks but also highlight that the private sector ownership and control was contested by several central banks.⁶⁸ Moreover, the pace of developments among commercial banks was faster than the ability of the public authorities to arrive at a consensus for action. With the Bank of England, in particular, keen to leave it to the commercial banks there was little basis for the FRBNY, Banque de France or the Bank of Belgium representatives to hold sway.

SWIFT had a rocky start with computer glitches, high costs (to join, to buy terminals, to pay fees), and an evolving rulebook that caused uncertainty and legal ambiguity. Once it started in 1977, however, SWIFT quickly gathered volume and members but it was a very European organisation with only 8% of shares owned by US banks in 1979. Figure 5 shows the number of users and the number of countries they came from. Opening to securities firms and national clearers in 1987 and to fund managers in 1992 increased the scope of members and the benefits of network externalities. In 1998, 20 years after it started operating, the G10 central banks began to engage in ‘cooperative oversight’ of SWIFT because so many critical payments systems depend on it. As noted on the SWIFT website, in the absence of formal regulation, ‘the main instrument for oversight of SWIFT is moral suasion’.⁶⁹ The details of transactions through SWIFT remain confidential.

Fig. 5

SWIFT connections (firms and countries), 1977–2011. (Source: Scott & Zachariadis, 2014: 113)

Conclusions and Prospects

Technological change in the late nineteenth century made the network of global payments more efficient and facilitated globalisation. The framework of bilateral or nested correspondent bank relationships based on the telegraph that developed in this era persists as the fundamental method of final settlement for a significant proportion of global payments. In the nineteenth and early twentieth centuries, commercial banks developed this network themselves in response to customer needs and commercial interests. By the late twentieth century, when computers began to transform international communications, commercial banks cooperated to apply this new technology to enhance the existing network of correspondent bank relationships. Public sector actors such as central banks and the United Nations sought oversight of these developments but the system remained in private sector control even as the volume of cross-border transactions surged and the potential for systemic crisis was exposed. By the 1990s, international institutions like the BIS were concerned about the stability of the cross-currency settlement system but continued to encourage private sector solutions. As the correspondent banking system entered a new phase in the 2010s, with a contraction in the network, public sector interest has re-emerged in the form of an ambitious programme of research commissioned by the G20. Geopolitics has also drawn in public sector influence. In March 2012 the EU passed specific law to prohibit messaging services being provided to Iran, which required SWIFT to disconnect Iranian banks (until 2016). In 2022 this was followed by similar regulations prohibiting SWIFT’s services to Russian and Belarusian entities. The global payments system thus becomes part of geopolitical action.

Political and technological frictions have prompted new thinking about how finally to replace the 150-year-old system of correspondent banking through new digital technologies. Like the telegraph and the computer in the past, digital currencies and block-chain may be the next disruptive technology for the global payments system. The use of smart contracts in distributed ledgers could make cross-border payments much faster and final than the current system, but there are challenges also in this solution.

The prospects for cryptocurrencies have not been rosy since 2020, exemplified by the collapse in value of the flagship Bitcoin in 2022. A new generation of crypto-currencies with more transparent foundations sparked considerable optimism, but in 2022 the stability of so-called stable coins has also been questioned. Tether was subject to Securities and Exchange Commission fines for misleading the public about the extent of its US dollar backing in October 2021 and then dropped below its $1 peg in May 2022. Another stable coin, Terra/Luna, which operated with an algorithmic base, collapsed entirely in the same month. Nevertheless, the block-chain technology that emerged with cryptocurrencies might offer a cheaper and faster means of payment, especially through smart contracts.⁷⁰ The medium could be stable coins (which are anonymous and make supervision difficult) or central bank digital currencies (CBDC).

Central banks have trialled a bewildering range of regional experiments in different forms of cross-border CBDC since 2019 but obstacles to governance, harmonisation and supervision as well as technical challenges remain.⁷¹ The new inflationary environment and higher interest rates may also complicate these initiatives. Nevertheless, this may be the moment for public agencies to take over the governance and security of the global payments system from the private sector. On the other hand, history shows that the pace of change has not been fast in global payments and that public authorities have not led innovation. Correspondent banking relationships persisted over the past 150 years through two waves of globalisation, dramatic information and communications innovations and many geopolitical disruptions. They may yet survive a bit longer.