Fri, 14 Oct 2016
SWIFT (Society for Worldwide Interbank Financial Telecommunication) is nearly half a century old and was originally built to replace the antiquated telex machine. Telecommunication technology has changed drastically since then and it is unlikely that banks would want to build a bespoke telecommunication network if they were designing the system from scratch today. Cryptographic tools like SSL/TLS/HTTPS allow secure communications over ordinary telecommunication links. Of course, SWIFT is not just a telecommunication company: it also pioneered the standardisation of financial messaging formats like the famous
MT 105. However, over time, this role too has gradually been taken over by the global standard setting bodies (for example,
All this means that if SWIFT did not already exist, nobody would bother to create it today. But SWIFT does in fact exist, and until recently, there was no serious reason not to just let it be. If SWIFT were delivering security and piece of mind, why would anyone disturb it? The problem is that in recent months, the Bangladesh Bank SWIFT hacking and other breaches of SWIFT security in Ecuador, Vietnam and India have shattered the illusion that SWIFT provides unquestionable security. Suddenly, SWIFT is being viewed as a source of risk – a single point of failure. For example, last month, the Bank of England put out a Consultation Paper about the design of the next generation of the large value payment system -- the UK RTGS. Two of the proposals are:
“[I]ntroduce additional functionality to mitigate the impact of an outage in the core SWIFT infrastructure, should it ever occur ... to remove the current single point of failure.”
Use ISO 20022 messaging standards in the new RTGS infrastructure instead of the current SWIFT MT messaging standards. This is designed to increase interoperability, eliminate single points of failure, and enable richer payment data.
Then there is the blockchain, which has helped popularize the hitherto esoteric notion that critical systems must be designed for Byzantine fault tolerance. In other words, the system must function correctly even if a few participants are completely evil (and not just selfish). In a world where even the largest banks could get hacked by rogue nation states or terrorist organizations, it is reasonable to assume that at any point of time, some participants in the global financial network are evil. Even if the blockchain turns out to be a passing fad, the need for Byzantine fault tolerance is not going away anytime soon.
Where does all this leave SWIFT? It is by no means self evident that its half-centenary coming up in a few years’ time will be an occasion for much celebration.
CLS Bank is a much newer organization – less than 15 years old. Yet, it belongs to a different era in which the big global banks constituting the foreign exchange markets confronted national high value payment systems (Real Time Gross Settlement Systems or RTGS) designed to serve their respective domestic markets. The RTGS in each country tended to be open for a few hours each day corresponding to the trading hours in that country and the idea that an RTGS in one country could be interconnected with the RTGS of other countries did not occur to anybody at all. So CLS Bank emerged as a private sector solution that interconnected all the major RTGS by participating in each of them. During the short window of time during early morning in Europe when all the major RTGS are open, CLS Bank achieves a payment versus payment (PvP) settlement – European Mega Bank can pay euros to American Giga Bank and receive dollars in return, with CLS Bank ensuring that both payments happen simultaneously. There is no risk that the euros will flow out, but the dollars will be stuck or vice versa; the so called Herstatt risk is solved.
Over the last decade, payment systems have evolved and in some large countries like the US, the RTGS now closes for only a few hours. In the UK RTGS Consultation Paper, the only question that they are debating is whether the new RTGS should be open for 23 ½ hours a day or for 24 hours. Moreover, national RTGS are becoming more open to the idea of interfacing with another RTGS in a different country. Again the UK Consultation Paper proposes to create a synchronization functionality which “allows each RTGS system to confirm that the funds are earmarked in the system in which the linked transaction will take place, the two systems then simultaneously release the two transactions”.
Inter-RTGS synchronization would provide a settlement system with much lower risk than the CLS Bank solution. I remember in 2008, the principal Indian fixed income CCP (Central Counter Party) was accessing CLS Bank through a European settlement bank that needed a bailout from its home country governments. For a significant period of time, Indian entities settling through CLS Bank via this TBTF (Too Big To Fail) settlement bank actually faced a greater risk than bilateral settlement with Herstatt risk. Even in normal times, the CLS solution is too demanding in terms of timelines and liquidity needs to really solve Herstatt risk. The system functions only with the liberal use of so called In/Out Swaps that reintroduce Herstatt risk.
In fact, I think this is an area where the IMF has a legitimate role to play. Articles III and VIII give the IMF access to every currency in the world and it is also the issuer of its own quasi-currency, the SDR. It is possible for the IMF to run a global multi-currency RTGS allowing simultaneous exchange of any currency for any other currency on its own books virtually round the clock. Participants could move the money from IMF books to the respective central bank books at any time when the respective central bank’s RTGS is open. Alternatively, if it is desired to run all settlement only in central bank money, the IMF could run an SDR RTGS that allows synchronization with each national RTGS. European Mega Bank can then exchange euros for SDRs through a linked transfer between the IMF RTGS and the European TARGET2 RTGS. American Giga Bank can then exchange these SDRs for dollars through a linked transfer between the IMF RTGS and the US Fedwire RTGS. (Since the IMF is the issuer of SDRs, it is clear that SDR balances at the IMF count as central bank money).
The point is that CLS Bank was a second best solution to Herstatt risk that made sense at a time when the world was struggling with third best and fourth best solutions. Indeed CLS Bank is a solution by TBTF banks, for TBTF banks, and of TBTF banks: it makes little sense in today’s post crisis world. Advances in technology and changes in mindsets have made a first best solution feasible. I think that CLS Bank is now living on borrowed time, but the lobbying power of the TBTF banks cannot be underestimated.