This is a follow up on a recent post that explored asymmetry in short-term money market rates.
I have become interested in better understanding the link between key central bank and interbank rates using the point of view of a microeconomist. The market for short-term liquidity in the euro area is excellent for this, as its core structure is easily dissectable: suppliers act in the role of banks with access to the ECB’s deposit facility rate, whilst consumers are the smaller financial institutions that deposit their funds at so-called “large” banks (for more on the rationale).
In particular, I am interested in learning if the market structure of interbank lending influences policy pass-through. This comes from the hypothesis that price-setting power of larger banks that have access to the DFR allows them to artificially widen the DFR-€STR spread, which is the profit headroom that these banks make.
Last time it was established that in the last policy cycle, the €STR was aggressively more responsive to changes in the DFR as rates were cut as opposed to increased. Now we explore if the market concentration of interbank liquidity had any role in the slower rise in the €STR.
Finding this out comes with limitations, especially pertaining to a lack of publicly available datasets featuring individual market shares in the unsecured short-term lending market. Nevertheless, some elementary conclusions can still be drawn. To this end, I enlist the help of Anthropic’s helpful research assistant Claude AI to guide me through converting simple linear regressions into Python code.
We begin by assessing if changes in the €STR are influenced by the market share over time of the top 5 players in the interbank lending market. Under our guiding assumption of price-setting ability, higher market concentration should correlate with a higher spread. A classic linear regression¹ is set up to gauge the impact of market share, including a consideration to account for the fact that yesterday’s spread is the strongest indicator of what today’s spread will be. In short, this estimates the correlation between changes in the spread with changes in the share of the top 5 players.
As it turns out, the largest banks’ market share has a negligible impact on the spread under this model. Lower market volume also does not correlate with stronger pricing influence given higher market share. Nothing changes significantly if changes to top 5 are smoothed.
Differentiating between the hiking and cutting phase of the 2023 bout of inflation tells another story: the results become extremely statistically significant as rates were cut², although the impact of concentration still remained minimal: a 10% increase in market share among the top 5 players corresponded to 0.003bp wider spread, as opposed to no link when interest rates increased. The positive correlation corroborates the presumption that market power tends to widen spreads (i.e. profits), but the hypothesis of price-setting power was unfounded.
A reinforced albeit small link between the spread and market share as rates were cut connects interestingly to the finding in the earlier post. As stated before: rates are more aggressive on the way down because the opportunity cost of large institutions holding deposits at DFR increases rapidly. Thus, as opportunities are scarce, financial corporations without access to the DFR have more power to lend at rates closer to the DFR. This causes a more aggressive convergence to a small spread.
The old findings provide a probable reason as to why the market concentration of big firms is more instrumental in influencing €STR pricing on the way down the mountain. This is when power matters more, even if the actual influence in the end is marginal.
These elementary results drawn from a short time series and even more basic correlation analyses open a window between macroeconomics and micro analysis. The results indicate that the market for short-term unsecured interbank lending in the eurozone is highly price competitive, with large banks not being able to influence pricing even as the five largest banks’ collective market share reached over 65% of transactions. Whilst the findings could be improved with more granular data, the micro conclusions already bear important repercussions for economic policy.
As for this blog, this moderately technical tour marks the end (for now) of our two-part venture into the market for unsecured overnight lending. The next post will be far more approachable from the get-go, although this does not constitute a promise on my part.
²This regression takes the same model highlighted above but applies it separately to the hiking, neutral and cutting periods of the 2022-2025 monetary policy cycle.
Economitron 
Leave a comment