ECB paper links fragmented capital buffers to weaker corporate lending
Serguey Asael Shinder2 min readECB Working Paper No 3287, dated September 2026, asks whether the shape of the capital buffer framework — not its level — affects how banks lend. The authors are Markus Behn, Marco Forletta and Alessio Reghezza; the paper carries the standard disclaimer that the views are the authors' and not necessarily those of the ECB.
The contribution is an index. It measures, at bank level and quarterly frequency, how fragmented a bank's buffer environment is: how many buffers are simultaneously active, how geographically dispersed they are, and how often buffer rates changed in the preceding year. To keep the measure from simply tracking how strict a regime is, the authors net out how much capital is being demanded, how big the bank is, and where the credit cycle stands. What is left is meant to be shape, not severity.
Matched against the European corporate credit register, the results are quantified. Move the index up by one standard deviation and two things move with it: headroom above the requirement widens by roughly 36 basis points, and credit to non-financial companies grows about 50 basis points more slowly — measured between a bank and a firm that were already doing business. The authors put this forward as an indication rather than a proof: a bank unsure what its buffers will be next year keeps a margin and extends credit more carefully.
The framing in the non-technical summary is candid about the trade-off: the flexibility of the post-crisis toolkit lets authorities target specific risks precisely, and the same flexibility means a bank may face several buffers at once, set by different authorities and revised at different times.
What it means
The finding worth carrying is that uncertainty about a requirement has a cost separate from the requirement itself. The index deliberately strips out how much capital is demanded; what remains is how predictably it is demanded, and that residual still moves lending. A bank that cannot forecast next year's buffer holds a cushion against the forecast, not against the risk.
Note what the measured effect is attached to: lending growth within existing bank-firm relationships. That is the conservative place to look, because it holds the borrower fixed and removes the objection that fragmented banks simply face different customers.
Two honest limits. These are associations from an indicator the authors built, not an experiment, and they say so — "suggestive evidence" is their phrase, not a hedge we have added. And a working paper is research published by the institution, not a position held by it, which the disclaimer states outright. The policy inference — that harmonising the timing and geography of buffer decisions might be nearly free in resilience terms and not free in lending terms — is ours to draw, and the paper does not draw it.