After months of testing, Swift has published its first conclusions on its blockchain proof of concept (PoC), and presented a live demo of the technology to Sibos delegates. While results so far are “encouraging”, Swift is not yet entirely convinced it’s the best technology for the purpose.

The PoC was launched back in April to test whether distributed ledger technology (DLT) – also referred to as blockchain – can help banks reconcile their nostro accounts more efficiently and in real time, while lowering costs and operational risk. (A nostro account is one that a bank holds in a foreign currency in another bank: it is difficult to monitor real-time nostro positions, due to a lack of intraday reporting coverage.)

33 banks are currently participating in the sandbox, which is part of Swift’s global payments innovation (gpi) initiative.

Swift’s interim report – which surveys the initial founder group of six banks only – provides an overview of the PoC to date, including technical objectives, potential business benefits, as well as key challenges that still need to be addressed to achieve industry-wide adoption.


Initial conclusions

Swift’s findings are centred on four areas: functionality, technology, bank feedback and cost.

In terms of functionality, it finds that the Swift-developed DLT application is fit for purpose.  “DLT provides real-time visibility to both the account owner and its servicer on the available and forecasted liquidity on the nostro account and supports payment reconciliation and investigations by providing an enriched data model based on ISO 20022,” Swift says in a release.

On the technology side, however, Swift says it is “still early days”: while the DLT sandbox demonstrates significant technology progress, it will take time for the newer generation of blockchain to be mature enough for industry-wide adoption.

“I think we have to be realistic,” Damien Vanderveken, head of research and development at SwiftLab told GTR on the sidelines of the live demo at Sibos, one of the world’s largest finance events, held in Toronto this week.

He went on to comment on the fact that Hyperledger Fabric v1 – the blockchain framework on which the PoC is based – is still in its infancy. “The Hyperledger Fabric v1 technology was released during the summer, so it’s two months old. It’s still going to take some time before it’s ready for production usage.”

In fact, blockchain may not even be the best solution: Swift has not ruled out that real-time nostro visibility can be addressed using another kind of technology.

“Is blockchain the right technology to solve that business problem? What we have shown so far is quite encouraging: we believe it is the right fit and could solve the problem. But could we do it in another way? Yes, we could solve the problem by using another technology, and that’s one of the things we will confirm by the end of the year – how we are going to move it forward,” said Vanderveken. “Only then will we confirm whether we will do it with blockchain or actually, for example, to minimise the integration effort for the industry, if it may be better to do it differently.”

Even if Swift were to pick blockchain to address current inefficiencies, the choice of specific platform that it uses – Hyperledger Fabric – is not set in stone. “We selected Fabric as one of the blockchain technologies that we consider is suitable for the financial industry,” said Vanderveken at the demo. But he added that Fabric is “not the only one”, and that Swift has not bound itself to a final choice when it comes to platform type. “There are other platforms out there, and we are still very much open to experimenting with those as well.”

Returning to the results: bank feedback on the PoC has been that Swift needs to develop unique value propositions in response to the different levels of sophistication, automation and previous investments of banks. “Differences in the incremental value for each financial institution will depend on its existing liquidity management capabilities,” explained Vanderveken.

Swift has concluded that it’s “crucial” that integration with legacy back-office applications and co-existence with existing processes are taken into account.

Finally, in terms of costs, although Swift has not quantified the investment required, there is no doubt that the cost will be significant. “Banks have said Swift has a role to play to make sure that the adoption is there by putting together a solution which minimises the integration efforts,” said Vanderveken. “Swift needs to bring a solution which minimises investment costs and allows co-existence with existing infrastructure and solutions.”

The findings will drive Swift’s next steps in the project, and will be compared and aggregated with the feedback that they receive from the remaining 27 banks in the group, which are currently undergoing testing. The PoC will conclude in November 2017, with the final results available in December.

Also speaking at the live demo, Matt Shepherd, part of Wells Fargo’s innovation technology team supporting its DLT programme, told delegates: “For us, right now, we do believe that this is a compelling case and we will continue to test with Swift.” He added that the banks look forward to understanding “how this may be integrated with gpi or other platforms”.

Wells Fargo has been involved with the PoC from the start, and, in fact, tested it bilaterally with partner bank ANZ before handing it over to Swift to take it forward on a multilateral basis.


Blockchain for payments

Despite the generally positive findings thus far, Swift remains reticent on using blockchain to make payments. “It will take time before it is mature and scalable enough for mission critical applications,” Vanderveken told GTR.

The question of blockchain for payments is particularly relevant in the context of Swift – the system which has supported cross-border payments for decades – as it is now seeing increasing competition from providers offering payments solutions based on blockchain, while it has chosen instead to focus on its gpi service.

For example, more than 100 financial institutions have to date joined Ripple’s enterprise blockchain network, RippleNet, to settle global payments leveraging blockchain technology. (Although many banks at Sibos complained to GTR that Ripple’s progress has thus far been frustratingly slow.) Just earlier this week, IBM and a number of banks launched a new blockchain-powered solution to clear and settle cross-border payments, using cryptocurrency as a bridge between fiat currencies.

Nevertheless, Vanderveken defended Swift’s current payments offering under gpi.

“The approach that we have for payments today is through the gpi initiative, it’s to build on the existing rails, because they are extensively used today. We complement what we are building with gpi with blockchain,” he explained. Although gpi payments are not settled in real time, Swift guarantees settlement within a single day.

Shepherd from Wells Fargo agreed in his address at the Sibos demo: “At this time, Wells Fargo does not believe that DLT can support front-end large-value payments,” he said.

Swift announced this week that its gpi service is growing at a rapid rate, surpassing 2 million payments in September. More than 120 banks, representing over 75% of all Swift payments, are now signed up to the service. 24 of these banks are currently live and actively using gpi, and Swift expects this number to grow to 40 by the end of the year.