More than just ripples in SWIFT’s pond?
Cross-border payments have long been the bane of treasury: too slow, too expensive and too opaque. Nimble FinTech service layers get all the hype, but the key competition is about incumbency and the fundamentals of correspondent banking. By Mark Parsley.
Ask a corporate treasurer about the pain points in payments and you’d better prepare for a rant: how long a payment takes for the beneficiary to be paid; the predictability of that time; impossibility of tracking payment status; inconsistency of data requirements by different banks; poor quality of remittance data sent with payments; costs and predictability of costs of making a payment; the time and difficulty of dealing with rejections, stopping payments and performing payment repairs.
All of these problems conflict directly with treasurers’ need to automate processes, prevent fraud, increase cash and fee visibility and optimise liquidity management. They arise partly from the archaic structure of the correspondent banking system. This turns a simple payment between two parties into a game of pass the parcel for six: payer, payer’s bank, payer’s bank’s correspondent, beneficiary bank’s correspondent, beneficiary bank, beneficiary.
These parties do not themselves act consistently: some have straight-through processing (STP); some do not; some sit on the payment longer than others; the fees charged along the way differ bank to bank, country to country. The addition of KYC/AML, OFAC and other sanctions and an inconsistent global regulatory playing field has simply thrown sand into what was already a misfiring engine.
The contrast between this process and the digital world most of us experience as individuals is extreme – as it is in other areas of treasury. And it has not gone unnoticed. Existing players, such as SWIFT, Visa and Western Union have launched new or improved B2B payments solutions based on the existing system of correspondent banks, and these compete with new players like PayCommerce and, though it is more for P2P and SMEs, TransferWise. Others claim to have abandoned the old model and developed completely new systems based on distributed ledger technology alone, Ripple being the largest and best-established. (None actually uses a true Bitcoin-style blockchain and so all rely more or less on existing payment systems, whether ACH or correspondent banking.) So which is best for corporate treasury?
Building on the past
One way to address payments’ pain points is to try to mitigate them within the current infrastructure. This is the approach taken by SWIFT with their Global Payments Innovation (GPI). This is simply the traditional SWIFT messaging and correspondent banking system plus what SWIFT calls a new set of ‘business rules’ captured in a set of multilateral service level agreements (SLAs) between participating banks.
In other words, SWIFT’s member banks – at least those committing to GPI – agree to change their behaviour, to homogenise their offerings, and to provide “same day use of funds, transparency and predictability of fees, end-to-end payments tracking and transfer of rich payment information.” Any banks that sign up to GPI agree to cut down on the opaque charging and delays from which they may previously have benefitted and they also have to ensure that all their partner banks in the transaction chain also agree.
To improve transparency, SWIFT has built an ‘observer’ system which allows GPI banks to monitor the SLA compliance of their partners across the system as well as a payment tracker, on which payments’ progress can be viewed in real time. This will be white-labelled by banks for their clients.
Clearly a key challenge for SWIFT in implementing GPI is persuading banks to sign up to something that at least in the short term appears to hurt them commercially. The demands of the SLA force them to do an existing job more quickly and transparently, and without employing the tricks of the trade from which they previously extracted additional profits. These previous wrinkles include percentage rather than flat fees, ‘lifting fees’, additional correspondent banking fees by intermediaries, float and FX spreads that should simply be a fixed spread over a real-time market rate. Why should banks give these up?
SWIFT’s answer, according to Wim Raymaekers, Programme Director, SWIFT GPI, is that “GPI is about providing a better service by enhancing the customers’ experience of making cross-border payments. As a result, banks will be able to retain and attract new customers and ultimately show leadership in global payments innovation. They will also achieve significant cost savings as a result of the network and claim management efficiencies they can achieve in correspondent banking. Moving forward with the availability of additional SLAs and innovations, additional savings will subsequently result from enhanced compliance practices, optimised intraday liquidity flows and increased straight-through-processing rates.”
These arguments seem to have been persuasive so far. According to SWIFT’s website, “over 110 leading transaction banks from Europe, Asia Pacific, Africa and the Americas are already signed up [to GPI] and more are expected to join.” These banks represent the overwhelming majority of cross-border payments on the SWIFT network.
A step in the right direction
The first iteration of GPI is not the endgame – and SWIFT itself acknowledges that. “The second phase goes one step further, enhancing the digital transformation of cross-border payments by enabling banks to offer new services such as, for example, the facility to immediately stop and recall a payment, no matter where it is in the correspondent banking chain.”
Same-day settlement is better than the three to five days it could currently take, and it is a significant step towards real-time (which is still not a priority for most treasuries, according to a recent SWIFT/EuroFinance white paper). Similarly, with fee transparency and predictability, GPI delivers post-transaction clarity, and this can, over time enable treasurers to predict costs better. But treasurers’ ultimate wish is for accurate upfront cost information.
And the payment tracker, critical as far as any corporate treasurer is concerned – is clearly just a step – welcome though it is – along the road to a more modern system. It functions as a central payments database, hosted at SWIFT, updated via MT199 or API with data consumption via GUI. So each bank in the chain updates the SWIFT database via MT199 producing a set of messages that can be viewed by the payer via a GUI that gives details of when each bank did what. By itself this does not provide full fee transparency or predictability and it still relies upon old technology at the banks tracking payments through their systems quickly enough to produce meaningful data.
SWIFT’s claim that GPI will feature “transfer of rich payment information” is somewhat undercut by the fact that its own documentation confirms that unaltered remittance information is limited to the 140-character standard in its messaging. While this could be used to deliver a URL to a set of documents, it’s far from delivering full remittance information via the bank. Again though, the improvements in information delivery are significant and in many cases treasurers do not need to key information via the banks.
Also, phase two of the SWIFT roadmap for 2018 expects to deliver extended payment data.
But most important, no solution of this kind addresses any of the fundamental problems inherent in the traditional infrastructure (the MT messaging standard was designed for X.25 networks in the 1970s). In particular, it makes no change to the basic counterparty structure of a traditional payment which creates the cost, delay and opacity in the first place. If you believe that new consumer demands and the digital-first companies that service them are already stretching the capabilities of the old system, and that developments like the Internet of Things (IoT) will make this worse, then the GPI may well only be a bridge to some new system of the future.
SWIFT understands this. As Raymaekers says, “In addition, and more broadly, SWIFT is also engaging the FinTech community to assess available technology against customer requirements. In September we are hosting an ‘Industry Challenge’ competition for FinTechs to develop overlay services leveraging the SWIFT GPI platform. The FinTech winners will be awarded 100,000 EUR each to work with banks and SWIFT on collaborative innovation concepts that solve additional industry challenges in cross-border payments on top of SWIFT GPI.”
In addition, in January this year SWIFT’s exploration of the blockchain resulted in the launch of a proof of concept (PoC) – scoped in collaboration with leading correspondent banks – to determine if distributed ledger technology (DLT) could help banks reconcile their nostro databases in real time. Thirty SWIFT GPI member banks are participating in this PoC, set to show early results at the next Sibos meeting. Wells Fargo, Bank of New York Mellon, ANZ, BNP Paribas, DBS Bank, and RBC Royal Bank are among those participating.
As SWIFT explains on its website, “under the current correspondent banking model, banks need to monitor the funds in their overseas accounts via debit and credit updates and end-of-day statements. The maintenance and operational work involved represents a significant portion of the cost of making cross-border payments. This PoC will test whether distributed ledgers may be able to help banks reconcile those nostro accounts more efficiently and in real time, lowering costs and operational risk.”
SWIFT will deploy open-source Hyperledger technology, and combine it with key SWIFT assets to bring it in line with the financial industry’s requirements. Using a private blockchain in a closed user group environment with specific user profiles and strong data controls; user privileges and data access will be strictly governed.
The fact that SWIFT is looking at blockchain/DLT is a quiet acknowledgement that the competition is doing the same, and may well be stealing a march.
Baby steps to the future
For example, in an attempt to remove some steps in the payment chain, and to benefit from blockchain technology, Visa is working with Chain to develop Visa B2B Connect, to give financial institutions a simple, fast and secure way to process business-to-business payments globally via a new near real-time transaction system designed for the exchange of high-value international payments between participating banks on behalf of their corporate clients.
Chain, Inc. is a technology company that partners leading organisations to build, deploy, and operate blockchain networks and is author of the Chain Protocol, which powers the Chain Core blockchain platform. Its strategic partners include Capital One, Citigroup, Fiserv, Nasdaq, Orange, and Visa.
As Visa and Chain describe it: “Visa is working with Chain to build Visa B2B Connect using Chain Core, an enterprise blockchain infrastructure that facilitates financial transactions on scalable, private blockchain networks. Building on this technology, Visa B2B Connect will facilitate a consistent process to manage settlement through Visa’s standard practices.”
In other words, this platform will cut out the majority of the correspondent banking steps in the standard payment by allowing banks and corporates to make a payment direct to one another using the Visa infrastructure as a central clearing point. A bank can pay into B2B Connect, and Visa will pass the payment onto the final counterparty, removing the need for multiple intermediary banks.
The advantages, according to the developers, are familiar from the claims of GPI: predictability and transparency – banks and their corporate clients receive near real-time notification and finality of payment; security – signed and cryptographically linked transactions are designed to ensure an immutable system of record; and all parties in the network are known participants on a permissioned private blockchain architecture that is operated by Visa.
Are these claims true? We will know when Visa rolls out the pilot and time will tell if it emerges as a real competitor to SWIFT on SWIFT’s home ground.
As an aside, Mastercard added blockchain APIs to its developer site to “facilitate new commerce opportunities for the digital transfer of value”, following Visa’s announcement that the B2B Connect pilot would start this year. MasterCard has, however acquired VocaLink, the operator of UK ACH clearing which is a partner in many immediate payments schemes worldwide, including the US, Singapore, the UK and Sweden. This is a shrewd move as VocaLink understands both ACH and card payments and has experience of corporates. While the components don’t yet fit together, watch this space.
But the biggest threat to SWIFT’s supremacy and the largest DLT-utilising payments initiative is still Ripple. Ripple claims to be a real-time gross settlement system (RTGS), currency exchange and remittance network that uses a sub-set of blockchain technology called Interledger Protocol (ILP).
So what happens in a Ripple transaction? (You can find out on YouTube.) Ripple still uses the idea of correspondent banking. So in a transaction between two institutions there still sits an intermediary bank. These are the key stages:
The originating bank sends out a request to the beneficiary bank and correspondent bank to obtain their processing fees and, if required, FX rates.
Pre-transaction validation: this includes compliance screening and account verification checks. Since all parties have this information, they can pre-validate the transaction to ensure STP.
The originating bank accepts the best quote for which they can meet the compliance requirements. The beneficiary bank can then lock the quote. At this point Ripple initiates a hold on the funds in the banks’ ledgers.
The ILP ledgers generate cryptographic signatures to verify that funds are committed to the transaction. The funds are simultaneously released across all the parties’ ledgers ensuring no settlement risk.
Upon completion Ripple provides a confirmation message to all parties.
All of this happens within one or two seconds – faster, cheaper and less complicated than the current process with end-to-end visibility and rich information exchange.
The benefits for treasurers are clear: settlement risk is almost completely eliminated (risk remains if banks are not operating at the same time); Ripple’s auction system assures best FX execution; counterparties in the transaction chain can be limited to those who match pre-specified compliance criteria; Ripple uses standard ISO and MT messaging (which of course means it is subject to some of the same limitations as SWIFT); there is no loss of data in the transactions – meaning higher auto-reconciliation rates and all fees and costs are known upfront.
Also, corporates using Ripple are only indirectly exposing themselves to a FinTech start-up because their banks, not they themselves, are the counterparties in the network. Treasurers are still interacting with trusted bank partners.
Is blockchain the answer?
For Ripple to become a significant alternative to the existing infrastructure, the blockchain has to be proven suitable for the purpose. Plenty of people say that it is not. So, as explained above, SWIFT is exploring a blockchain PoC but in announcing that initiative it also says that “whilst existing DLTs are not currently mature enough for cross-border payments this technology, bolstered by some additional features from SWIFT, may be interesting for the associated account reconciliation,” says Wim Raymaekers, Head of Banking Market and SWIFT GPI at SWIFT. “This PoC gives us the opportunity to test DLT and determine if it can be applied to this particular use case.” It will initially show early results at the Sibos meeting in Toronto.
He is backed up by Stephen Grainger, Head of North America at SWIFT, who says, “there is a recognition in the industry that perhaps distributed ledger technology at this point isn’t ready for the wholesale application to manage all correspondent banking”.
The bankers on SWIFT’s GPI Vision Group also explain GPI’s reliance on traditional processes by claiming that neither banks nor their regulators are ready to risk the international payments system on an untested technology. So Tony Brady, managing director and head of global product management for BNY Mellon Treasury Services, who is a member of the SWIFT GPI Vision Group, says: “Our early view is that while blockchain and distributed ledger have a fair amount of promise, it’s a little early to try to tackle cross-border payments, particularly high-value cross-border payments where we’re putting millions of dollars at risk.”
It’s not just SWIFT, which has perhaps a vested interest in downplaying the new technology. As explored in the blockchain article (page 30), the Canadian central bank has come to the same conclusion and in May, the Bank of England, in its paper, “A blueprint for a new RTGS service for the United Kingdom”, also said “the Bank has decided not to build the renewed RTGS service on Distributed Ledger Technology, in light of its findings that the technology is not yet sufficiently mature to provide the exceptionally high levels of robustness required for RTGS settlement … Further work is required to address privacy and system scalability in particular, and these and other topics suggested by this initial work will drive the Bank’s future research programme on this technology.”
Ripple rebuts the criticism of its use of the blockchain by saying that ILP gets around the issues raised because it simply connects existing bank ledgers rather than holding the ledger itself. In effect, banks connect their core systems to the Ripple network – analogous to how they currently connect their core systems to the SWIFT network and the ILP is used to co-ordinate the payments. This addresses the fears banks have expressed about the blockchain around privacy, scaleability, regulatory approval and the issue of having to get all parties to a blockchain to agree to validate a transaction.
However, even Marcus Treacher, Global Head of Strategic Accounts, Ripple and formerly HSBC’s Global Head of Payments Innovation and a member of the Global Board of SWIFT from 2010 to 2016, admits that “a single blockchain is not viable, the future is in different currency blocks with Ripple an interconnection.” The fact that he has joined the new company, along with Marjan Delatinne, who had been leading customer engagement for SWIFT’s GPI, suggests that Ripple believes the blockchain can be part of a new global payments system in some way.
For treasurers the battle looks like a win-win. If Ripple does what it says, treasurers will benefit. If it fails but in the process forces SWIFT to improve its core offerings, then treasurers benefit. And GPI, as Raymaekers says, “is real! More than 40 global transaction banks have begun actively using or implementing the SWIFT GPI service, with another 50 in the implementation pipeline. Hundreds of thousands of GPI payments have already been sent across more than 85 country corridors.” Treasurers are already winning.