The positions we are willing to sign.

The ESRB will deliver the 2027 adverse scenario as an incremental hardening of the 2025 template, not a methodological reboot. The cumulative GDP path lands in the −6% region across the three-year horizon — broadly in line with 2025 — with the geopolitical overlay tightened on the energy-price and sovereign-spread legs, and a modest additional depreciation assumed on CRE. Policy-rate assumptions widen the range on both ends but do not deliver a new shock topology.

The climate module remains thematic, not Pillar 1. Expect a dedicated climate overlay block in the methodological note, consistent with the 2023 and 2025 treatments, and possibly a one-off sectoral deep-dive in the published results. It does not bite the CET1 number in 2027. This matters: most of the work we see on climate is currently being scoped as if it will.

We concede the counter-position exists: BaFin has signalled that it wants climate integrated into Pillar 1 measurement, and has used the 2025 ECB thematic review as a staging ground for that argument. We expect BaFin to lose the argument inside the ESRB Working Group on Stress Testing for the 2027 cycle. We could be wrong; if we are, we will update this position under the date of the ESRB publication.

A bank that arrives in the exercise planning to rights-issue its way through adverse has already lost the methodology argument. The 2027 EBA methodology, like its predecessors, constrains both the quality and the quantity of management actions that can be recognised inside the CET1 depletion calculation. Dividend policy is capped. DTA recognition is capped. Rights issues outside a committed backstop are excluded. The bank is left with run-off economics and a small operational-cost lever. That is not comfort.

What matters is that the unconstrained run — the one the JST reads first — shows a credible CET1 path. If a bank needs management actions to pass the exercise, it fails the exercise in the way that matters (SREP read-across), even if the headline ratio is above the fully-loaded 5.5% line. Supervisors do not score that generously.

The right place for the mitigation narrative is the ICAAP, not the EBA templates. ICAAP is where you can set out your full recovery-plan linkage, your conditional liquidity posture, and the forward-looking capital-build thesis on a basis that is not artificially constrained. The EBA exercise is an engineering test of your unconstrained engine; ICAAP is where you defend the business model. Treat them accordingly.

The mental model most risk functions run — “we will influence the P2G once the SREP decision letter lands” — is inverted. By the time the JST is writing the 2028 decision letter, the P2G number has already been formed, in substance, from the stress-test output, the ICAAP benchmark, the bank-specific risk profile and the supervisor's prior-cycle view. The decision letter ratifies it.

The practical influence window on the P2G number is therefore the 18 months before submission — roughly Q3 2026 through Q1 2028 — not the four months of formal dialogue after. Specifically: the T-0 balance-sheet bridge into 2027 is built in Q3 2026; the methodological choices that drive bottom-up projections are locked in Q4 2026; the ICAAP internal adverse is set in Q2 2027. After the submission in Q1 2028, the available levers are limited to narrative corrections and calibration sensitivities. Those matter at the margin; they don't move the bucket.

Operationally, this means Risk functions should treat the 2026 calendar year as the high-leverage year. The 2027 year is execution; the 2028 conversation is cosmetics.

Climate remains thematic in the 2027 EBA EU-wide stress test. No Pillar 1 capital add-on is delivered through the exercise, no sectoral RWA multiplier is applied inside the engine, and the climate output lives in a separate disclosure block alongside the main CET1 depletion run. We hold this position with high conviction.

This is the controversial position. BaFin has spent 2025 signalling that it wants the opposite outcome, and a reading of the ECB thematic-review tone and the EBA’s own climate roadmap suggests direction of travel toward integration. Our view is that the machinery is not ready — the transition-risk PD curves are not harmonised across cohorts, the physical-risk geography is not in the templates, and the Basel framework itself has not endorsed Pillar 1 treatment. The ESRB will take the incremental path.

We are on the other side of the BaFin position. If the 2027 methodological note includes a Pillar-1-bite climate block, we will write a public correction under the date of publication and update our downstream advice accordingly. For the avoidance of doubt: we will state, in the correction, that we were wrong.

Stress testing is a senior-judgment exercise, not an hours-billing exercise. The hard calls — the T-0 bridge, the scenario translation under the new methodology, the defence of management-action choices in front of a JST — are three-to-five-person conversations that need partners in the room, not 30-person pyramids with analysts doing pivot-table gymnastics. Boutique wins on that structurally. Price and speed follow.

There is one workstream where Big-4 wins and we concede it openly: 20-country thematic co-ordination for cross-border groups — the kind of exercise where the ECB runs a thematic review and needs a unified approach delivered in Germany, Spain, Italy, France, the Nordics, Poland and the Benelux, plus the US and UK subsidiaries, all at once, on a synchronised calendar. That is a physical-footprint problem, and we do not pretend to solve it.

Where Big-4 win on footprint, Ezelman declines the mandate rather than staff it thin. Our preferred configuration is a carve-out: we do the senior-judgment workstreams and the Big-4 do the country-footprint co-ordination. Where that is not on offer, we recommend the client to the Big-4 directly. We keep the boundary clean because staffing a footprint mandate thin is how a boutique fails a senior client.