Target2 ‘fingerprint’ could flag liquidity problems – research

Deviations from regular payment patterns are a sign something might be wrong

Banks are creatures of habit, especially when it comes to large payments – meaning any departures from the norm can be a sign of operational foul-ups or impending liquidity problems, new research suggests.

Using data from the EU’s Target2 large-scale payment systems, Bundesbank researcher Marc Glowka constructed ‘fingerprints’ for the 210 most active participants, and showed they could be clustered into sets of similarly behaving institutions – making oversight easier. The research will appear in the Journal of Financial Market Infrastructures later this year.

Both individual customers and banks tend to have fairly predictable transaction patterns – varying from season to season, but in a repetitive way. When these behaviours change, this could mean something is wrong. At the customer level, previous research found, it can be a sign that liquidity shortages, operational breakdowns or other financial problems are forcing the customer to change its behaviour. For banks participating in large-scale payment systems, Glowka found in research published in JFMI in 2018, several consecutive 10-minute periods with activity well below the norm is a reliable sign of an operational problem. At a systemically important bank, this could signal a potential threat to system-wide liquidity and even financial stability.

Using cluster analysis, Glowka classified the Target2 users into 10 categories, based on the timing of their transactions over the day (in blocks of one hour), and the size of each transaction compared to total volume for the day. ‘Early Birds’ carried out up to 40% of their transactions in the first hour; ‘Long Sleepers’ brought in at least 20% between 10am and 11am, local time; ‘Late Payers’ executed the bulk of their transactions in the afternoon, and so on.

The categories emerged from the cluster analysis rather than being laid down in advance, and are therefore likely to represent actual underlying differences, rather than being a result of shoehorning institutions into arbitrary groups.

The categories are fairly robust over time, Glowka found. Very few participants – only 17 out of 210 – behaved in exactly the same way every day. But the majority kept to the same profile on at least 50% of working days in 2017. For the rest – 36 of the participants studied – there is no clear reason why they do not have a stable profile; more research needs to be done, Glowka admits, and the reason and timing of profile changes could reveal important information about institutions’ financial stability.

But for the banks that have stable or relatively stable payment behaviour, a deviation could, at least, be a prompt to investigate further.

Target2 data from the Netherlands showed that abnormal payment behaviour was a good indicator of liquidity stress before and during the financial crisis. Glowka suggests clustering analysis would make this indicator simpler to use and more powerful.

He also speculates that the profiles themselves could represent differences in financial strength. “The relationships between participants with the same and with different payment profiles as well as the reasons for the specific payment behaviour of participants are challenging questions for further analysis. For instance, are ‘Late Payers’ waiting for liquidity before they introduce their transactions?” he writes.