Last reviewed · Version 1.0.0 · Evidence cutoff
The licensing-success register reads the survival curve from one side. This register reads it from the other. Across the same regulator population that produced 571 currently authorised EMIs, PIs and AISPs, 63 named entities have had their authorisation withdrawn, cancelled, revoked or otherwise removed from the active register and are documented here individually, each with a primary-source URL and a withdrawal-type classification.
This register is the counter-narrative to the licensing-success population. Where the success register answers the question “who got through?”, this register answers the question “who got stopped — or stopped themselves — and how was the stop classified by the regulator?”. Companion artefact: EMI / PI licensing-success forensic register, the verified-floor positive-control benchmark this register inverts.
Methodology
Every named entity sits in one of five withdrawal-type classes:
regulator-revocation— the authorisation was revoked or the registration was cancelled by the regulator under enforcement powers. The supporting source-document URL on each row points either to a Final Notice / Decision PDF or to a regulator press release naming the entity. Examples: every Bank of Lithuania revocation press release, every FCA E-Money register entry with status fieldEMD Revoked, the Finansinspektionen authorisation-withdrawal decisions.voluntary-cancellation— the firm requested cancellation; no enforcement action visible in the public record. The FCA register convention is that a status field ofCancelled - Authorised EMIorCancelled - Small EMI(without aRevokedqualifier) defaults to voluntary or administrative cancellation. Where the firm name appears in a verified Final Notice, the classification is upgraded toregulator-revocationregardless of the underlying status field.application-refused— the applicant never reached authorised state and the regulator refused authorisation. The FCA publishes Final / Decision Notices for refused PSD2 / EMD2 applications; most other EEA NCAs do not. The two named entities in this class — Awesome3 Limited (FCA Final Notice, 19 March 2020) and Yan International (FSA-era Final Notice, 29 August 2012) — are the verified UK-published rows; the population is structurally biased to NCAs that publish refusal data and is not a denominator for inferring overall refusal rates across the perimeter.lapsed-without-renewal— authorisation expired and the firm did not renew. EMD2 / PSD2 authorisations do not expire on a fixed date in EU law, so this class is rare and is preserved as a taxonomy anchor.regime-transition-non-completed— the firm failed to re-authorise under a new regime. Will populate as the PSD3 / PSR transitional period closes (currently expected to end in 2027 with an 18-month transitional period from OJ publication in 2026). A meaningful tranche of FCACancelled - Authorised EMIentries from 2021–2022 are arguably post-Brexit re-licensing exits, but the FCA register does not flag the cause and this register declines to infer it.
The five classes are exhaustive and mutually exclusive at the point of withdrawal. They do not encode reasons (AML failure, safeguarding failure, capital insufficiency); reasons are legal commentary that lives in the linked Final Notice itself, not in this register’s structured index.
Population landscape
The 63 named entities draw from four regulator populations: the FCA (UK) at 57 entries, the Bank of Lithuania (LT) at 4, Finansinspektionen (SE) at 1, and De Nederlandsche Bank (NL) at 1. The UK skew is structural — the FCA publishes the cleanest machine-readable cancelled / revoked cohort in the EEA + UK perimeter, downloadable as a CSV with named status fields, and the FCA also publishes Final and Decision Notices for refused PSD2 / EMD2 applications which most other NCAs do not. Per-NCA registers in BaFin (DE), ACPR (FR), CSSF (LU), MFSA (MT), CBI (IE), Banco de España (ES), Banca d’Italia (IT) and the remaining EEA NCAs were variously Cloudflare-protected, JS-rendered, or non-machine-readable in this run; they are documented at the methodology page as a coverage gap rather than as evidence of zero withdrawals.
This is not a complete withdrawal census across the EEA + UK. The EBA’s central Payment Institutions Register (EUCLID PIR) is the authoritative pan-EEA anchor but its bulk-extraction endpoint was not safely retrievable from the benchmark session, and absence of any specific entity from this register should not be interpreted as evidence that the entity remains authorised. Every named row is a positive observation; the population shape is constrained by what each regulator publishes in machine-findable form.
Jurisdictional patterns
In the verified subset, withdrawal-type distribution skews to the FCA’s bimodal pattern: 30 of 63 records are regulator-revocation (47.6%), 31 are voluntary-cancellation (49.2%), and 2 are application-refused (3.2%). Within the UK FCA cohort the pattern is sharper still — there is a verified Final Notice cluster spanning December 2025 through April 2026 against dormant Small Payment Institutions and one Authorised Payment Institution (FIRST MONEY SERVICES LTD 22 December 2025; iCorp Global Limited 22 January 2026; Stallion Money Limited 24 February 2026; Dania Money Transfer Ltd 5 March 2026; Taj Exchange Limited 16 April 2026) which the FCA has cancelled under Regulation 10 of the PSRs and the Money Laundering Regulations 2017 for failure to maintain HMRC MLR registration after dormancy in their initial 12-month window. This is the canonical late-2025 / 2026 cohort of regulator-driven cancellation against firms that were registered but never traded; the analytical signal is that the Final-Notice publication cadence for this failure mode runs to roughly one per month.
The Bank of Lithuania pattern is structurally different: each verified case is a regulator-revocation grounded in a public press release, with the entity named in the press release itself. The four documented cases (Foxpay 2024, Kevin EU 2025, Lock Trust 2025, PAYRNET 2023) span AML/CFT, safeguarding, capital-adequacy and outsourcing failures respectively. The Bank of Lithuania does not publish a discoverable register of voluntary cancellations in English, so the LT cohort here is regulator-revocation by source-availability rather than by population shape.
Finansinspektionen (SE) and DNB (NL) each contribute a single named regulator-revocation: Intergiro Intl AB (SE, authorisation withdrawal October 2025) and Suri-Change B.V. (NL, licence withdrawal October 2023, objection rejected May 2024). Both NCAs publish per-decision pages in English with reasoning narrative; neither is bulk-downloadable.
The implication for any reader using this register as a positive observation set: jurisdictional comparison is biased by NCA publishing convention, not by underlying withdrawal rate. A jurisdiction that publishes only enforcement actions will look like its withdrawal cohort is 100% regulator-revocation; a jurisdiction that publishes both will show the bimodal pattern visible in the FCA data. The right way to use this register is at the named-row level, not as a basis for inter-jurisdictional rate comparison.
Forward-look
The PSD3 / PSR provisional political agreement reached on 27 November 2025 starts an 18-month transitional period from Official Journal publication in 2026. Any EMI or PI authorised under PSD2 / EMD2 that does not complete re-authorisation under PSD3 / PSR by the end of the transitional period in 2027 will populate the regime-transition-non-completed class. Forward-looking applicants should plan for PSD3-readiness in addition to current PSD2 compliance; the most common avoidable failure mode in the historical cohort is the safeguarding model (single-bank concentration, out-of-MS, weak reconciliation), which is materially tightened under PSR.
Data-quality limitations
Three limitations apply to any quantitative use of this register. First, NCA-level publication-style bias: the UK skew above is a measurement artefact, not an underlying statement about UK authorisation outcomes. Second, the absence of a denominator: there is no public EU-wide refusal register, so authorisation success rates cannot be computed from this artefact alone. Third, the inference layer between FCA Cancelled and EMD Revoked register status and the 5-class taxonomy: where the FCA Final Notice URL is verified, this register classifies as regulator-revocation; where only the register status field is available, the literal status determines the class. Reasonable readers may classify some of the unverified Cancelled rows differently; the source URL on each row supports independent re-classification.
This is a structured-index page, not a legal commentary. For corrections to entity classifications or new Final Notice citations to upgrade entries from voluntary-cancellation to regulator-revocation, write to legal@finray.tech.