COREP wins the budget; FINREP loses the programme. The IFRS 9 stage-migration tables, the FINREP-to-COREP reconciliation and the ESG overlay are where 2026–2027 supervisory attention is actually landing. Here is what needs to be rebuilt, in which order, and why the data-quality regime changes the answer.
Three companion pieces cover what actually changes in each of the three reporting frameworks that matter most to supervisors. Read them in order.
35+ template changes under CRR3, Output Floor, FRTB, CVA. The headline rebuild — and the one already in most banks' 2026 plans.
IFRS 9 stage-migration, reconciliation to COREP, ESG overlay and the new data-quality regime. The quiet half.
~95 attributes per loan, 2016/867 ECB Regulation. How supervisors use it to triangulate COREP and FINREP, and what it means for data lineage.
Every CRR3 reporting programme we see treats COREP as the prudential artefact and FINREP as the accounting annex. That framing is wrong. FINREP is where three things collide that COREP alone cannot resolve.
Stage migrations, ECL coverage, forborne exposures, POCI assets — none of it is in COREP. Templates F 18 (non-performing exposures) and F 19 (forbearance) are where the JST first sees whether your provisioning curve is consistent with your book. A COREP-only rebuild leaves this data flow untouched while supervisors increasingly lean on it.
COREP RWA, FINREP exposure, and Pillar 3 disclosure must triangulate. If the three don't tie, the JST opens an inquiry — typically framed as a data-quality finding, occasionally as a disclosure failing under CRR Part Eight. Most banks have bridge logic; few have bridge discipline.
The EBA Pillar 3 ESG templates (climate-related transition and physical risk tables under the EBA/ITS/2022/01 regime) pull their anchor numbers from FINREP exposure classes. If FINREP taxonomies are wrong — NACE code mapping, collateral location data — the ESG return inherits the problem in public disclosure format.
The EBA's work on reporting data quality (mapped into the Supervisory Data Strategy) means FINREP validations are no longer just XBRL checks. Supervisors triangulate plausibility, period-over-period stability and cross-framework consistency. FINREP is the dataset that fails those checks first.
Templates F 18 and F 19 carry the data that supervisors use to test your IFRS 9 engine against the macro and against your own COREP credit numbers. Three flows matter most.
| Flow | Template | What the JST is reading | Common failure mode | Severity |
|---|---|---|---|---|
| Stage 1 → 2 migration (SICR trigger) | F 18.00, F 18.01 | Does the rate of stage-2 transfer move with your PD model and with cycle conditions? Is it flat, implying overrides? | Stable stage-2 coverage ratio across quarters despite macro deterioration — suggests management override | CRITICAL |
| Stage 2 → 3 (default recognition) | F 18.00, F 19.00 | Is default recognition aligned with the CRR Art. 178 definition of default as implemented at your institution (EBA/GL/2016/07)? | FINREP default flag diverges from COREP default flag — typically a data-feed issue, not a policy issue | CRITICAL |
| Forbearance classification | F 19.00, F 19.01, F 19.02 | Are forborne exposures captured at grant, held for the minimum probation period, and reclassified on a rules-based trigger? | Forbearance flag set by manual workflow, not systematically applied — inconsistencies across geographies | HIGH |
| Cure rate & return-to-performing | F 19.02 | Are cured exposures being reclassified to performing on schedule, or is there a curing backlog? | Curing pipeline not reconciled — exposures stuck in forborne-non-performing long after they should have cured | MEDIUM |
| POCI (purchased/originated credit-impaired) | F 18.00 | Are POCI assets correctly segregated? IFRS 9 requires different accounting — lifetime ECL from inception | POCI population drifts into Stage 3 treatment — wrong ECL measurement, wrong provisioning run-off | MEDIUM |
Stage-2 coverage ratio, quarter-over-quarter, against the macro scenario you declared in ICAAP. If the two curves are decoupled, you will be asked to reconcile them on paper at the next JST deep-dive. Banks that cannot reconstruct the stage-2 migration at portfolio level fail this interaction — the issue is almost always a data-lineage failing, not a methodological one.
Three returns, three different governance owners, three different validation calendars. Supervisors treat them as one dataset. That gap — between how banks produce the returns and how supervisors read them — is where data-quality findings come from.
FINREP F 01.01 gross carrying amount must reconcile to COREP exposure value pre-CRM at portfolio level. The bridge items are well known — accounting consolidation scope vs. regulatory scope, netting, on/off-balance treatment — but many banks have no documented bridge owned by a single function. The bridge exists in a reporting analyst's Excel workbook.
FINREP NPE in F 18 must equal the ratio disclosed in Pillar 3 Table EU CR1. If the two differ — even by small amounts driven by rounding or cut-off — the JST asks why. Under CRR3 the Pillar 3 format is more prescriptive; room for quiet divergence shrinks.
When the Output Floor template (C 06.02) reports "higher of IRB or floor × SA" per risk type, supervisors cross-reference the SA leg against the FINREP exposure class. A portfolio that is large in FINREP but small in C 06.02 raises a question — is there a scope issue, or an SA RWA calculation gap? This is a new reconciliation only because C 06.02 itself is new.
IFRS 9 provisions reported in FINREP must reconcile to the IRB shortfall/excess calculation in COREP and the disclosure in Pillar 3 Template EU CR9 (A-IRB approach). A persistent unexplained gap in this bridge is a classic trigger for an on-site inspection on the provisioning methodology.
The EBA Pillar 3 ESG templates pull anchor numbers from FINREP. If the accounting data is wrong, the ESG disclosure inherits the error — in public disclosure format, with reputational as well as supervisory consequences.
Template 1 (banking book exposures by sector) requires NACE codes down to a granular level. Most banks carry NACE at the obligor level in FINREP F 06; the code quality varies widely by geography and legacy system. A material error at the obligor level propagates into the ESG disclosure and the supervisory climate stress test.
Physical-risk templates require location data — typically NUTS-3 or postcode — for real estate collateral. FINREP does not mandate that field. Most banks are building it as a FINREP extension; the ones that haven't are reporting at country level with caveats that supervisors are now pushing back on.
Residential real-estate collateral requires EPC-style energy-efficiency data (Template 2, EBA/ITS/2022/01). Data coverage is the honest answer — banks typically disclose partial coverage with a remediation plan; supervisors accept the gap if the plan is credible. The supervisory question is not "why is coverage low" but "what is your plan and milestone to improve it".
Scope 1 / 2 / 3 financed emissions by obligor require external data or internal estimation (typically PCAF methodology). Where estimated, FINREP exposure is the base. An error in the FINREP classification — e.g., misassignment to the wrong sector — propagates into the emissions disclosure as a metric error, not a data-gap disclosure.
BCBS 239 was adopted in 2013 (BCBS d239). A decade of thematic reviews followed. The ECB's 2023 follow-up letter to G-SIB CEOs made clear that the implementation window has closed: supervisory patience on risk-data aggregation has ended (ECB 2023 — "Effectiveness of risk-data aggregation and reporting at significant institutions"). For most G-SIBs, the next on-site inspection on data quality will be judged against a FINREP-shaped dataset.
| BCBS 239 principle | Where FINREP evidences it | Typical supervisory test |
|---|---|---|
| Accuracy & integrity | F 01.01 ↔ general ledger reconciliation; FINREP ↔ COREP bridge | Reproduce a month-end FINREP submission from source ledgers with no manual adjustments > materiality |
| Completeness | Consolidation scope coverage in F 00.01; subsidiary rollup | Validate that FINREP perimeter equals regulatory consolidation scope under CRR Title II |
| Timeliness | Ability to produce FINREP pro-forma within days, not weeks | Ad-hoc request: provide a FINREP F 18 table as of a non-quarter-end date — on five business days |
| Adaptability | Ability to produce stressed FINREP projections for ICAAP / stress-test | Produce F 01.01 & F 18 under the adverse EBA scenario — with traceable methodology |
If an ECB OSI team arrived next quarter and tested risk-data aggregation against FINREP and the COREP bridge, could we pass the accuracy and timeliness tests above — without the heroics of one named reporting lead? If the answer is "only with heroics", the data-quality finding is already written; the inspection date is just waiting to be scheduled.
Most of the critical-path work on FINREP is not about new templates; it is about the bridge and the data-quality evidence layer. None of it is glamorous. All of it is supervisory hygiene.
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 →