Someone is wrong on the internet about the SWIFT network

There's a chart that has been circulating for a while now on social media that shows payments traffic on SWIFT, a key global financial messaging network. Below is a version from the Economist, but I've seen other versions too.

Source: The Economist

When banks make cross-border payments between each other, say euros to dollars, they need to use a communications network to coordinate the debiting and crediting of accounts, and SWIFT is the dominant network for doing so. Think of it as WhatsApp for banks.

Here's the problem. The main conclusion that pundits are taking away from the chart is the wrong one. Most of them seem to think that the chart illustrates an erosion in the euro's global popularity (i.e. de-euroization) and a simultaneous move towards the dollar for global trade. The Economist, which entitles its chart "Dollarisation," is also guilty.

Today I'm going to show you why that's the wrong conclusion; there is no SWIFT-related de-euroization. The reason for going through this effort isn't just because it's fun to dunk on wrong folks. It can also teach us some interesting things about the massive bits of unsung payments infrastructure that underlie our global economy, including not only SWIFT but also Europe's T2 and the U.S.'s Fedwire, two of the world's busiest financial utilities.

Let's dig in. The problem with trying to analyze charts of SWIFT messages across various currency jurisdictions is the data isn't necessarily comparable. As I said at the outset, commercial banks around the world use SWIFT to coordinate cross-border payments with other banks, and that is what people are hoping to measure with the SWIFT chart at top. But muddying the waters is the fact that in the EU, banks also use SWIFT for domestic payments. Here's how:

The most important bit of payments infrastructure in both the U.S. and EU are their respective central bank's large-value payments (or settlement) systems. When commercial banks make crucial domestic payments with each other, typically on behalf of their customers, these payments are settled in real-time using each commercial banks' respective account at their central bank, in the U.S.'s case the Federal Reserve, and in Europe's case the European Central bank, or ECB. The ECB's mechanism for settling payments is known as T2 (and previous to that, Target2.) The Fed uses Fedwire.

To coordinate this "dance of databases," the central bank and participating commercial banks need to communicate clearly and rapidly with each other, and that's where financial messaging networks come in. Fedwire doesn't use SWIFT for this. It comes fitted-out with its own proprietary messaging network for member banks. But the ECB has chosen a different setup. Up until 2023 the ECB had outsourced all messaging to SWIFT, a bank-owned cooperative based in Belgium.

Now you may be able to see why comparing the amount of euro payments made using SWIFT messages to dollar payments made using SWIFT is an apples to oranges comparison. Both data sets include cross-border payments, but the EU dataset also includes a large amount of domestic payments. The U.S. dataset doesn't.

This means that the variations in the amount of euro payments messages that get captured in the chart at top may not reflect dramatic geopolitical shifts like "de-euroization, but may be linked to more banal things like changes in local EU payments habits. And indeed, I'm going to show why domestic and not international factors explain the 2023 drop in the euro share of SWIFT messages. 

In 2023, the ECB replaced its Target2 settlement system with a new system called T2. Two key upgrades were introduced with T2 that ultimately affected SWIFT message flows. 

The first of the upgrades was a new language for constructing messages, with the ISO 20022 messaging standard replacing the legacy MT messaging format. (I wrote about ISO 20022 in an article entitled The Standard About to Revolutionize Payments.) 

This change in payments lingo has had a big effect on the sum of SWIFT data displayed in the chart at top. Both the ECB and SWIFT provide explanations for this, but here is my shorter summary. Prior to the 2023 changeover, a type of euro payment known as a liquidity transfer was regularly captured in the SWIFT chart. A liquidity transfer occurs when a European commercial bank, which often has several accounts at the ECB, must rebalance between its accounts when one of them is running low. These within-bank liquidity transfer messages aren't terribly interesting and have nothing to do with global payments, but were included in the SWIFT dataset nonetheless up until 2023, thus fudging the results. 

With the arrival of ISO 20022, messages related to euro liquidity transfers are now conveyed using a new type of message. Thus the big decline in the euro's share of SWIFT messages in 2023 — liquidity transfers have effectively dropped out of the chart. This is a good thing, though, since the omission of these relatively unimportant within-bank transfers means we're getting a cleaner and more accurate signal.

The second upgrade introduced in 2023 was the opportunity for European commercial banks to choose among multiple messaging networks for accessing T2. Under T2's predecessor, Target2, banks only had one access choice: the SWIFT network. With T2, European banks can also use SIAnet, owned by the Nexi Group. (I wrote about this upgrade here, in which I described T2's switch from an older Y-copy topology to a network agnostic V-shaped topology.)

In that older post, I suggested that adding additional access points was a healthy step for Europe, since it meant more resilience should one network suffer an outage. And in fact, Europe is already reaping the benefits. When SWIFT failed for several hours on July 18, 2024, the ECB issued the following alert:

"T2 is operating normally. However, due to an ongoing SWIFT issue, some incoming messages do not reach T2 immediately. Similarly, some T2 outgoing messages might not reach the receiver immediately... There is no impact on traffic sent or received via NEXI. Participants may continue sending new instructions and queries to CLM/RTGS/CRDM. Updated information will be provided at the latest by 16:30."

Whereas an outage of SWIFT in 2021 or 2022 would have seriously slowed down Europe's financial activity, the addition of Nexi's SIAnet to the mix in 2023 limited the damage caused by the 2024 SWIFT outage. By contrast, the UK's central bank, the Bank of England, remains entirely reliant on SWIFT for messaging, and so the 2024 outage caused more disruption for Brits than Europeans, according to the Financial Times.

Unfortunately, I can't find any data on how many European banks have actually chosen to shift their T2 messaging needs over to Nexi. But I'd imagine that it isn't negligible, given that Nexi's SIANet is already being used by banks to access other key bits of Europe's payments architecture including STEP2, a pan-European automated clearinghouse. And so some non-negligible portion of the drop in the euro's share of SWIFT messages in the top chart is due to a shift away from SWIFT.

If the SWIFT chart at top doesn't mean what people think it means, what is the euro's status as a global trading currency? A 2024 article from the ECB clears this up. In short, the euro's international role hasn't eroded over the last few years. The de-euroization memes are all wrong.

The irony of all of this is that rather than reflecting a decline in Europe's status, the SWIFT chart illustrates the opposite. A bunch of healthy advances are driving the euro's share of SWIFT payments down, including a more accurate classification of financial messaging data thanks to a better messaging language, combined with a much needed de-SWIFTication of European messaging flows. It's not as juicy as euro critics make it out to be.