AnaCredit phases 1, 2 and 3 — and what replaced them

“AnaCredit phase 2” and “phase 3” are two of the most-searched terms in EU credit reporting — and the honest answer is that they do not exist as adopted law. AnaCredit went live in a single stage in 2018 and has stayed there. The further stages sketched in the ECB’s original planning — extending the dataset to more instrument types and more lender categories — were discussed and never adopted. What did happen instead is more consequential for anyone planning a reporting roadmap: the Eurosystem folded the future of granular credit reporting into the Integrated Reporting Framework, whose milestones are now public. This piece lays out the actual timeline, what Stage 1 covers today, what the envisaged later stages would have added, where IReF takes over, and — with worked scenarios — how to plan against it.

1. Quick recap — what AnaCredit is

AnaCredit (“analytical credit datasets”) is the ECB’s loan-by-loan credit register, established by Regulation (EU) 2016/867, adopted in May 2016. Euro-area credit institutions report granular data on credit exposures to legal entities, monthly, through their national central banks, which forward the data to the ECB’s central database. The operational detail sits not in the Regulation but in the AnaCredit Reporting Manual, published in three parts — Part I on general methodology, Part II on the datasets and attributes, Part III on case studies and scenarios — with the second edition issued in May 2019. For the scope test and the debtor threshold, see our AnaCredit for payment firms piece.

2. The actual implementation timeline

3. What AnaCredit phase 1 (Stage 1) covers — the live regime

A vocabulary note, because it drives the search terms: the ECB’s own documents speak of stages, while practitioners and vendors say phase 1. They mean the same thing — the single live collection that started with the September 2018 reference period. If a project plan refers to “AnaCredit phase 1”, it is describing the regime below.

• Reporting agents: euro-area resident credit institutions and resident foreign branches of credit institutions
• Instruments: loans, deposits other than reverse repos, overdrafts, credit-card credit, revolving credit, credit lines, factoring, financial leases and trade receivables — where the debtor is a legal entity
• Threshold: debtor-level commitment of €25,000 or more at the reporting agent
• Frequency: monthly, with some attributes updated quarterly
• Out of scope: natural-person debtors (consumer credit, residential mortgages to individuals), derivatives, off-balance-sheet items other than committed credit lines

The attribute-level detail — ten linked datasets and roughly 95 attributes, and which of them cause most rejections — is covered in our AnaCredit data model walkthrough.

4. AnaCredit phase 2 and phase 3 — what was envisaged, and what happened

Searches for phase 1, phase 2 and phase 3 map onto a staged build-out that appeared in the ECB’s early AnaCredit planning and was then overtaken. Read against what was actually adopted:

Two practical consequences follow. Household and mortgage lending remains outside AnaCredit — it is captured instead by national credit registers where those exist, and by securitisation-level datasets. And there is no adopted timeline for an AnaCredit scope extension, so speculative build-out against “phase 2” has nothing to build to.

5. What actually replaces the missing phases: IReF

The real successor is the Integrated Reporting Framework, the Eurosystem programme to consolidate euro-area banks’ statistical reporting into a single framework. Following the implementation plan published in June 2026, the ECB announced the roll-out milestones: a public consultation on the draft IReF Regulation in the second half of 2027; a one-year pilot phase starting in the second quarter of 2030, in which reporting agents are invited to test their ability to meet the new requirements; and first official IReF reporting in the second quarter of 2031, beginning with a one-year parallel phase during which existing statistical reporting within IReF scope continues alongside the new returns.

Scope matters here. IReF is aimed at the statistical requirements of deposit-taking corporations, drawing together collections that today sit in separate instruments — granular credit data alongside balance-sheet-items and interest-rate statistics (see our BSI and MIR reporting piece for what those look like today). The Eurosystem’s stated approach is that existing ECB regulations containing statistical requirements folded into IReF will be repealed or amended; the ECB has not published a discontinuation date for AnaCredit itself, and the parallel phase means both regimes run for a period. The implementation plan also flags that how individual national collection frameworks evolve is still being worked through — which is where most of the real cost for any single reporting agent will land.

For a reporting agent the sequencing is therefore: AnaCredit as-is through the rest of the decade, the H2 2027 consultation as the first point at which the future requirement becomes concrete enough to design against, and the Q2 2030 pilot as the first date on which you have to actually produce something new.

6. Why the timeline looked different per country

Regulation (EU) 2016/867 sets a floor, not a ceiling. National central banks were free to collect more, earlier, or with fewer derogations — and several did, which is why implementation experiences differ so much between member states. Two structural patterns:

• Integration with a pre-existing national credit register — Spain (CIRBE), France (Service Central des Risques), Belgium (via its central register infrastructure), Italy (Centrale dei Rischi). In these countries AnaCredit was implemented as an evolution of an existing national collection, and reporting agents saw a merged national+AnaCredit template.
• Fresh collection — member states without a comparable register built the pipeline new.

See our country-by-country AnaCredit comparison for how DNB, the Bundesbank, Banque de France and Banco de España each run the collection. The same divergence is the reason IReF’s national-implementation workstream matters: harmonising the ECB layer does not automatically harmonise what your own NCB asks for.

7. Derogations for small reporters

Article 16 of the Regulation lets national central banks grant derogations to small reporting agents — reduced reporting frequency or partial relief — provided the aggregate covered by derogations stays within defined limits. The practical consequence: whether your institution qualifies as a “small reporter” is a national determination, and it is revocable. Ask your national central bank, not the ECB, and treat the answer as a status to monitor rather than a permanent classification.

8. Three planning scenarios

Scenario 1 — the budget line for “AnaCredit phase 2”. A reporting agent’s multi-year plan carries a change budget for an AnaCredit phase 2 extension, on the assumption that derivatives and group-level reporting are coming. There is nothing to build: no amending regulation, no consultation, no draft. Redirecting that budget correctly means pointing it at the IReF track instead — a watch item until the H2 2027 consultation, a design exercise once the draft Regulation is public, and a delivery programme timed to the Q2 2030 pilot. The failure mode of getting this wrong is not under-investment; it is spending three years building for a requirement that never arrives while the one that does arrive gets a standing start.

Scenario 2 — the bank that built to the ECB floor. An institution in a member state with a long-established national credit register reads the AnaCredit Regulation and the ECB Manual, builds to that scope, and is then rejected repeatedly by its national central bank. The cause is that the NCB collects a superset: the merged national template carries attributes and counterparty coverage the ECB layer does not require. The remediation is to build to the national template as the binding specification and treat the AnaCredit dataset as a subset within it — the reverse of the instinct to start from the Regulation.

Scenario 3 — the derogation that lapses. A small reporting agent has a derogation permitting reduced-frequency reporting and has built a pipeline that only ever runs on that cadence, with several steps completed manually between runs. Loan growth takes it past the point where the NCB is willing to keep the derogation, and the institution is told to move to monthly reporting. The manual steps that were tolerable a few times a year are not tolerable monthly, and the lead time is short. The lesson is to keep the pipeline capable of monthly operation even while reporting less often — derogations are a frequency concession, not an architecture concession.

9. FAQ

10. What to do, today

• Plan against Stage 1 as a stable target — do not build speculative capacity for unadopted AnaCredit stages.
• Move any “phase 2” budget line onto the IReF track, with the H2 2027 consultation as the first design gate and the Q2 2030 pilot as the first delivery date.
• Treat your national central bank’s template, not the ECB Regulation, as the binding build specification.
• Confirm your derogation status in writing, and keep the pipeline capable of monthly operation regardless of the cadence you currently report on.
• If you lend to natural persons, track the national credit-register rules instead; AnaCredit does not reach them.

Related: AnaCredit for payment firms · The AnaCredit data model · AnaCredit across national central banks · CIRBE — the Spanish credit register · DAC8 XML reporting — the data build · What is a reporting agent · Building an AnaCredit submission