The entire stress test — 3-year projections, adverse CET1 trajectories, Pillar 2 Guidance — rests on a single number: the starting point. The new RTS finally codifies how that number is built, and the parallel consultation on revised reporting templates reshapes how banks must evidence it. Get either wrong and every downstream projection is compromised.
The adverse scenario gets the headlines. The starting point decides the outcome. Every basis point of mis-anchoring propagates through the 3-year projection horizon and lands directly in the final CET1 depletion.
Complete asset, liability and off-balance sheet perimeter at T-0 — frozen for the duration of the exercise under the static balance sheet assumption.
COREP C01→C08 + FINREP F01→F18Starting stage 1 / 2 / 3 distribution sets the baseline for provision trajectory. Mis-allocation at T-0 compounds into €100m+ coverage gaps by year 3.
Must reconcile to FINREP F18.00Pre-stress net income, dividend assumptions, AT1 coupons, DTA and prudential filters all shape the CET1 start. Each element is now explicitly in scope of the RTS.
COREP CA1 lines 1–73Anonymised European G-SIB. The reconciliation grid, the NPE stage-allocation call, and the RWA-engine parameters that survived the supervisory challenge. Mandate-specific figures withheld; public-data calibration on request. One-pager written for CFOs, not modellers.
CFO Capital-Impact Case · coming soon Publishing shortly — currently under review.Prior exercises left the starting point largely to methodological notes and Q&A. The RTS moves it into binding technical standards — with explicit definitions, reconciliation duties and supervisory challenge rights.
The RTS defines line-by-line what belongs in the starting balance sheet. It closes historical grey zones on trading-book repo, off-balance commitments, held-for-sale portfolios and participations in non-consolidated subsidiaries.
Banks must now reconcile the starting point simultaneously against (i) audited IFRS financial statements, (ii) COREP regulatory returns and (iii) internal management accounts. Every discrepancy must be explained and signed-off.
The static balance sheet is no longer a simplifying convention — it becomes a binding constraint. Departures require explicit EBA approval and are limited to pre-defined categories of non-performing run-off and regulatory-driven disposals.
This is the first diagnostic every CRO should demand from their stress-test PMO. Each line of the starting balance sheet must tie to three independent sources. Tolerances are explicit.
| Starting Balance Sheet Item | COREP Reference | FINREP Reference | Tolerance | Severity if Broken |
|---|---|---|---|---|
| Total Risk-Weighted Exposure Amount | C02.00 r010 | — | 0 bps | CRITICAL |
| CET1 Capital (post filters & deductions) | C01.00 r020 | F01.03 r350 (bridge) | < 5 bps of RWA | CRITICAL |
| Loans & Advances at Amortised Cost | C07 / C08 by ptf | F04.03 r080 | < 0.1% of total assets | HIGH |
| Non-Performing Exposures (NPE) | C08.02 NPL r by ptf | F18.00 r070 | 0 tolerance | CRITICAL |
| IFRS 9 Stage 2 Allocation | — | F04.04 + F12.01 | < 2% of stage 2 volume | HIGH |
| Off-Balance Sheet Commitments | C07.00 / C08 OBS | F09.02 | < 1% by product | MEDIUM |
| Deferred Tax Assets (DTA) | C01.00 r340/r370 | F10.00 / F11.00 | 0 tolerance | HIGH |
| Expected Loss / Provisions (IRB deduction) | C01.00 r390 | F12.01 | < 2% of provisions | HIGH |
| AT1 & T2 Instruments | C01.00 r530/r700 | F01.03 capital bridge | 0 tolerance | MEDIUM |
CET1 reconciliation to FINREP is where 80% of starting-point findings are raised by the JSTs. A clean bridge between the accounting balance sheet and the regulatory CET1 — line by line, reviewer-ready — is now table stakes, not optional supplementary documentation.
The single largest source of starting-point challenge by JSTs is the allocation of the performing portfolio across IFRS 9 stages and the reconciliation of the NPE perimeter. Under stress, misallocation at T-0 manifests as a 20–40 bps CET1 drift by year 3.
If the ECB asked you today to bridge your FINREP F18.00 NPE stock to your COREP C08.02 NPL perimeter exposure by exposure, with the stage allocation and ECL coverage as of T-0 — how many days would it take you to produce a signed-off reconciliation? Anything above 5 business days indicates a structural data gap that will surface under RTS scrutiny.
Developed across three EBA stress tests and multiple SSM thematic reviews, the Ezelman ANCHOR™ framework is the practitioner's operating model for defending the starting point against supervisory challenge. It is designed to be reusable, auditable and boardroom-defensible.
Compile the complete reference balance sheet: loans, trading book, OBS, participations, DTAs, prudential filters — at legal-entity and consolidated level.
Harmonise segmentation between accounting, regulatory and management accounts. Eliminate definitional drift on exposure class, geography, and counterparty type.
Freeze the IFRS 9 stage allocation and NPE perimeter at T-0. Document the SICR thresholds, overlays, and curing rules in force on the reference date.
Reconcile the three views (IFRS × COREP × MI) line by line. Produce and sign off the tri-reconciliation bridge with explicit tolerance thresholds.
Assign single accountability by line item (1.5LoD). One named owner per starting-point domain, one named challenger in Risk, one signatory at CFO/CRO level.
Independent validation by 2LoD (Risk) and 3LoD (Internal Audit) before submission. Pre-emptive peer review vs. prior exercise starting points.
Build the supervisory defence pack before the JST asks. One-pagers per contentious item, challenger-ready, traceable to source systems.
Banks that anchor their starting point using ANCHOR™ materially reduce JST clarification requests on T-0, compress their data remediation cycles, and enter the stress test with an uncontested baseline — so every bps of depletion is attributable to the scenario, not to data quality issues. †
† Illustrative · based on the qualitative experience of ECB on-site and horizontal JST reviews observed across our mandate history. Not a measured dataset.
“Keep the balance sheet constant” is the most misunderstood instruction in the EBA exercise. The RTS is explicit about what constant means, what departures are allowed, and how run-off of matured exposures must be handled.
Maturing exposures are replaced with new exposures of identical risk profile (same exposure class, PD band, LGD band, maturity bucket) — unless the bank has a formal run-off strategy.
Total asset size is held constant. Banks cannot project growth-driven P&L, even where strategic plans include organic growth.
Departures from the static assumption are allowed only for NPE run-off and disposals committed and publicly disclosed before T-0 (e.g. binding sale agreements).
CRR3 standardised approach migrations (IRB → SA, permanent partial use) are recognised in the starting point and maintained through projection — not re-optimised under stress.
Dividend distributions are set to zero in the adverse scenario. AT1 coupon behaviour follows contractual MDA & ADI rules, strictly applied on projected figures.
The most frequent supervisory finding we see is implicit business-plan bleed-through: banks unconsciously apply their internal MTP growth assumptions to margin or cost projections even as they “freeze” the balance sheet. The RTS now explicitly prohibits this and requires evidence of static-assumption discipline in the projection engine outputs.
A mis-anchored starting point is not cosmetic — it materially shifts the 3-year adverse CET1 depletion. Based on our observations across three EBA exercises, the following sensitivities are typical.
| Starting-Point Driver | Typical Error | CET1 Impact by Year 3 | RWA Impact | Severity |
|---|---|---|---|---|
| NPE stock mis-classification | ± 5% of NPE volume | ± 15–30 bps | Up to +€1bn | CRITICAL |
| Stage 2 under-allocation | 1–3 pp of performing book | −10 to −25 bps | Indirect via ECL | HIGH |
| DTA recognition gap | ± 10% of DTA balance | ± 5–15 bps | Direct CET1 deduction | HIGH |
| OBS commitment scope gap | 3–7% of OBS | − 5–10 bps | + RWA via CCF | MEDIUM |
| Expected Loss − Provisions mismatch | €50–200m | ± 5–15 bps | Direct CET1 deduction | HIGH |
| Prudential filter mis-application | Instrument-level | 5–20 bps | Direct CET1 deduction | MEDIUM |
In aggregate, a poorly controlled starting point costs banks 30–80 bps of avoidable CET1 depletion by year 3 — directly eroding headroom to MDA and potentially triggering Pillar 2 Guidance uplift. Discipline at T-0 is one of the highest-ROI interventions available to any stress-test programme.
Based on our delivery experience across three EBA stress tests, this is the minimum viable programme to arrive at the reference date with a defensible starting point.
Run the ANCHOR™ diagnostic against the latest COREP/FINREP cycle. Identify tri-reconciliation gaps, NPE classification issues, stage 2 documentation weaknesses, and DTA recognition ambiguities.
Close the top-10 reconciliation gaps. Harmonise exposure segmentation between finance, risk, and MI. Establish a T-0 freeze protocol for ECL overlays and SICR thresholds.
Appoint line-item owners. Set up the tri-reconciliation committee with CFO, CRO and Head of Finance Reporting. Define sign-off protocol and escalation thresholds.
2LoD and 3LoD challenge of the starting point. Peer benchmarking against prior exercise T-0 and against public disclosures of peer G-SIBs.
One-pager per contentious item. Pre-empt likely JST clarification requests. CFO/CRO/Head of Risk signed dossier ready for submission.
Banks should use the consultation window to push for (i) clearer tolerance thresholds on low-risk bridges, (ii) recognition of banks with active Guideline-compliant NPE run-off plans, and (iii) a harmonised definition of “material” overlay to avoid inconsistent supervisory interpretation across Joint Supervisory Teams.
In parallel to the RTS on the starting point, the EBA has consulted on a revised set of stress-test reporting templates. Most banks treat the two as separate files. They are not. The new templates demand exposure-level granularity, tri-reconciliation lineage, and dynamic management-action logic — and every one of those demands traces back to the starting-point data you build at T−0. The reporting consultation is the operational test of whether your starting-point architecture can hold supervisory weight.
| Reporting Area | Old Template | New Template | Starting-Point Impact | Severity |
|---|---|---|---|---|
| Credit Risk — Losses | Portfolio-level aggregation, 15 segments | Exposure-level granularity, 45+ segments | T−0 reference balance sheet must be reconstructable at counterparty level | CRITICAL |
| NII Projections | Top-down NII with 5 repricing buckets | Bottom-up NII with 12 repricing buckets + behavioural modelling | Starting-point loan-level cash flows become mandatory — aggregate balances no longer sufficient | CRITICAL |
| Sovereign Exposures | Country-level aggregation | Instrument-level with maturity + accounting classification | HTCS/HTC split at T−0 must reconcile to FINREP — accounting classification drives CET1 path | HIGH |
| Capital Ratios | Point-in-time CET1/T1/TC | Dynamic capital with management action constraints | Static-balance-sheet assumption must be evidenced from the reference date forward | HIGH |
| Liquidity & Funding | LCR/NSFR snapshot | LCR/NSFR + cash flow survival analysis under stress | Starting-point cash flow ladders now in scope — new data required at T−0 | HIGH |
Exposure-level granularity across credit risk and NII is the single largest infrastructure challenge produced by the merged RTS + reporting consultation. Most banks’ stress testing platforms were built on portfolio-aggregate starting-point data. Retrofitting exposure-level T−0 data pipelines is a 12–18-month undertaking that must be completed concurrently with the starting-point reconciliation work described above — you cannot sequence them.
Four structural challenges emerge when the starting-point RTS and the reporting consultation are read together. Banks that treat them as two separate workstreams will fail the first supervisory dry run.
Counterparty- and instrument-level T−0 balances become the new minimum. This requires a single source of truth between FINREP, COREP, and the stress-test submission layer — all three anchored to the reference date with documented lineage. Portfolio-aggregate starting points will no longer survive JST challenge.
The reporting templates require dynamic capital paths under standardised management actions, while the RTS binds the balance sheet to static-replacement logic. Banks must document exactly where the two meet and where management action is permitted — and the boundary must be defensible at the board level.
Tri-reconciliation (FINREP ↔ COREP ↔ stress-test) is now a binding RTS output, not an internal QA step. At the same time, the reporting consultation compresses the submission window by ~40%. Banks that compress the tri-reconciliation into the last two weeks will produce visibly weak submissions that attract top-down adjustments.
The reporting consultation introduces Tier 3 supervisory benchmark comparison as a new QA step. Banks must be able to explain any deviation from EBA/ECB benchmark model outputs — and those explanations trace back to the starting-point choices (NPE staging, overlay refresh, FVOCI classification). Weak starting-point governance is visible directly in the Tier 3 output.
Can your bank deliver exposure-level starting-point pipelines, recalibrate projection engines, implement three-tier QA, and run a dry run — all within 18 months, while the starting-point RTS is finalised in parallel? For most banks the answer is no without starting infrastructure work now, ahead of final text.
The merged programme — starting-point RTS + reporting consultation — has one integrated timeline. Missing any milestone on either track delays the other.
Full programme delivery (RTS starting-point reconciliation + new reporting templates) requires 15–25 FTE across risk methodology, data engineering, IT and programme management over 18 months. Total investment for a G‑SIB: €5–12m including technology, external support, and internal resource costs. Under-investment creates execution risk that materialises as supervisory top-down adjustments.
Dates, paragraph numbers and document versions verified at publication. Where consultations are still open, we flag draft status in text. Corrections: research@ezelman.com.
Partner-led, senior-only, no audit conflict. If the question in this piece is live at your institution, the first conversation is with a partner — not a BD team.
Talk to a partner →