Last week, only 15 banks reported €13.5bn of transactions on Monday — down from 30 banks and €47bn on the previous day. As a result, the ECB was compelled to use its contingency calculation method for the first time ever.
The ECB blamed a technical fault for the drop, and said that it had been remedied. The next day, the results seemed to indicate this was true. Some 29 banks reported €43bn of data.
It seemed that all was well until Friday’s data landed this Monday. Only 20 banks had reported transactions, and the volume had fallen over €12bn to €28.8bn. The ECB was able to calculate the rate normally, but only just. Its own rules indicate that if fewer than 20 banks report data, it must use the contingency method, so it scraped into the normal category, but by the skin of its teeth.
Some claim these sudden drops may hide a perhaps more serious threat to €STR’s methodology. Because €STR was designed to be based on actual transactions a stable and reliable flow of transaction data is absolutely crucial to its robustness. That flow might be under threat from the ECB itself. The latest round of TLTRO — some €1.3tr — means that banks are flush with cash. With their liquidity needs taken care of by the central bank, fewer are turning to the interbank market for funding.
That is certainly true in CP markets, where bank issuance in the first half of 2020 was 14% down year on year. It’s equally clear from the compression of the Euribor-OIS spread. But in the overnight market, from which €STR is calculated, transaction volumes have actually remained broadly healthy with the notable exceptions of August 10 and August 14. Even then, they are far higher than the volumes on which Eonia was based in its waning months.
But will this remain the case? TLTRO is far from over. Bank liquidity provision is unlikely to disappear, and could well begin to encroach on €STR volumes.
However, the drops of August 10 and 14 look more like aberrations than part of a declining trend. The ECB has not admitted to a second technical fault relating to the August 14 data. Accordingly, we are forced to conclude that the interbank overnight lending market suffered an actual abrupt drop in volume and the absence of nine reporting banks in the same week as a technical fault led to an artificial drop in volume and the absence of 14 reporting banks. This is, to say the least, highly unusual.
It points to a worrying degree of fragility in the €STR methodology. While the ‘thumb-in-the-air’ method used for the rates €STR is displacing has its problems, it must be said that it at least had the merit of flexibility in the face of unique circumstances.
In the event of €STR volumes continuing to be prone to these sudden drops, it may become necessary for the ECB to re-examine its models for when a contingency method is called for and how it is calculated.
And if €STR is to be the beacon of stability and transparency that replaces the murky old-world risk-free rates, then it is surely important for the ECB to be as transparent as possible so that market participants can understand what is causing the sporadic absence of a third of the reporting banks and their transactions.