Summary
While the EU estimated the cost of open banking API changes, permissions dashboards and payee verification, those figures may not be reliable. The estimates also missed several significant costs of implementing regulatory change, including legal interpretation, gap analysis, programme management, preparing the required compliance pack, and more.
The European Commission published an impact assessment (IA) to accompany its proposal for a new payment services regulation (PSR) and third payment services directive (PSD3). However, the EU's estimate of nearly €300m in implementation costs omits significant adjustment and administrative costs of regulatory change, including:
- legal interpretation;
- gap analysis;
- programme management;
- preparation of the compliance pack necessary to stay authorised,
and more.
This article outlines those hidden costs, and complements our analysis of how the EU estimated PSD3 and PSR implementation costs, and why those figures might be too low.
Implementation costs not estimated in the IA
There are a few reasons why we don't think the estimated figures in the IA reflect the likely costs of the items they have tried to price, but the IA also misses or leaves entirely unquantified a few significant line items:
- Familiarisation and legal interpretation: understanding what's in the payment services package and how it differs from PSD2 and the EMD.
- Gap analysis and impact mapping: taking the new rules as interpreted and understanding how the changes affect each firm and its products, contracts, policies, systems and controls.
- Programme and change management: making, organising, monitoring and testing the changes, through to board sign-off.
- Preparing a compliance pack: required for existing firms to obtain reauthorisation under the new regime.
- Training: informing staff not directly involved in implementation what changes are coming and what they mean for their day-to-day work.
We've built a PSD3 implementation tool to handle 1-3, with support for 4 and 5 planned for closer to the time they're needed. Get in touch to learn more.
Although the IA acknowledges "Legal interpretation and uncertainty" as the third largest cost item associated with firms' implementation of PSD2, "familiarisation with new rules" is priced only in the column for competent authorities. The next cell, on business costs, reads "None". That these items are missing doesn't seem to be because the Commission has judged them to be small; they're simply not mentioned.
What would the missing line items cost?
Without an official estimate of these costs for PSD3 and the PSR, we've borrowed a method from the UK Financial Conduct Authority (FCA). The FCA's cost-benefit analyses split implementation into the same buckets every time: familiarisation and gap analysis, training, change projects and IT, each of which get specific financial estimates, tailored to small, medium and large firms.1
Familiarisation and gap analysis
Here's how the FCA's 2024 consultation on changes to safeguarding of payments balances estimated the costs for familiarisation and legal review of a 151-page package, based on firm size,2 and how that might scale to the 600-page Level 1 texts for PSD3 and the PSR:
| Regulatory change | Small firm | Medium firm | Large firm |
|---|---|---|---|
| FCA safeguarding rules update (151 pages) | £1,790 | £9,900 | £30,100 |
| PSD3 and the PSR (~600 pages, before L2 texts) | £7,000 | £40,000 | £120,000 |
Read that estimate as the FCA's view of the time it takes to read the rules and check them against what you do now, not as a quote for your legal workstream. Those numbers don't cover external counsel, drafting or arguing about what an ambiguous article requires in practice. In our experience, it understates the real work by a wide margin. Even the FCA's conservative model doesn't treat this cost as zero.
When the EU published the Anti-Money Laundering Regulation (AMLR) in 2024, we saw firms spending six weeks or more just to compare the AMLR against the directive it was replacing and interpret those differences. This work ended up being done by the firm's legal team and left them with a spreadsheet. See what we've built to save your firm from this when implementing the PSR.
Everything else
Without a clear benchmark for the ratio of IT costs to programme costs, we've prepared a bottom-up estimate of the other implementation costs using the FCA's methodology applied to the Council's first-reading texts of 21 October 2026.
This sketch assumes a mid-sized EMI with a few hundred staff, offering e-money accounts and cards, passporting into a handful of member states, and operating as an ASPSP for open banking purposes working over a 21-month implementation window.
| Workstream | What it involves | FTE-months |
|---|---|---|
| Programme management | Planning, sequencing, RACI matrices, steering committees, chasing | 9–21 |
| Policies, procedures and controls | Fraud and transaction monitoring, complaints and refunds, safeguarding, outsourcing and agents, disclosures | 4–10 |
| Contracts and disclosures | Framework contracts, terms and conditions, agent and distributor agreements, pre-contract information | 2–6 |
| Product and operations design | Payee verification flows, refund handling, spending limits, dashboard journey design | 3–6 |
| Testing and assurance | User acceptance testing, second-line review, internal audit | 3–6 |
| Re-authorisation compliance pack | Initial capital, safeguarding, governance and ICT arrangements, incident handling, business continuity, group application history and a winding-up plan, plus regulator Q&A | 2–4 |
| Training | Designing and delivering (not staff time spent attending) | 1–2 |
| Governance | Board and committee papers, sign-off | 0.5–1 |
| Total | ~25–56 |
At a salary and payroll tax cost of around €8,000 to €11,000 per FTE-month, plus the 25% overhead the Better Regulation Toolbox applies, that's roughly €250,000 to €770,000 for one firm. That excludes the familiarisation floor above, any engineering, and external counsel. For comparison, the IA's estimated implementation cost for this firm is the €40,000 for payee verification (for firms yet to implement it in response to other requirements) and, for ASPSPs, an open banking API adjustment.
Smaller firms, with fewer products and simpler governance, could come in below this figure. A firm with agents in several countries, a card programme and a regulator already asking questions could spend into the millions. In any event, the work the IA leaves out is measured in person-years, not person-days; that work needs skilled professionals and falls on every firm.
Authorised push payment (APP) fraud refunds
The IA says a full refund right for APP fraud "would not require any major upfront implementation costs from PSPs". It says the same about its preferred, narrower option: "Thus, this option would not require any major upfront implementation costs from PSPs…".3
This is because the IA treats these refund obligations as a contingent liability and a transfer from a PSP to a fraud victim, not a cost of implementation or compliance. However, in practice, PSPs will make changes to reduce the risk of such frauds and the resulting refund obligations, and rightly consider this part of the cost of preparing for the PSR and PSD3.
Looking to the UK as an example, payment firms have had to reimburse in-scope APP fraud victims "in most cases" since 7 October 2024, split 50:50 between the sending and receiving firm, up to £85,000 per claim.4 The only real defence to this reimbursement obligation is the consumer standard of caution, which comes down to gross negligence by the customer.
The UK Payment Services Regulator called that "a high bar" and put the burden of proof on firms. Its guidance then raised the bar further:
- a warning only counts if it's "bespoke… consumer, scam, and transaction specific", not "boilerplate", and
- ignoring such a warning isn't automatically gross negligence for the consumer.5
In practice, about 3% of claims get rejected for the consumer's gross negligence.6 One firm told the regulators it hadn't declined a single claim, "which reflects the extremely high threshold set for PSPs to apply the gross negligence exception."7
PSPs in the UK have made a number of changes to help manage this risk since the APP fraud refund rules were announced, including:
- upgrading transaction monitoring systems and rules;
- buying third-party behavioural and biometric tools;
- redesigning and testing warnings by scam type;
- hiring specialist teams to talk customers out of a scam mid-payment;
- introducing payment holds (enabled by some later legislative changes); and
- restricting payments to crypto exchanges (which were seen to have been a common vector for APP fraud payments).7
When proposing these rules, the cost-benefit analysis flagged this preventative spend as a high-to-medium cost and then admitted "we cannot fully quantify the likely scale of the costs to PSPs".8 The independent evaluation published in 2026 puts it at £31m to £42m a year in prevention and detection, on top of the refunds themselves, in a payments market smaller than that in the EU/EEA.7 While this is the kind of spend that likely has positive returns on fraud prevention and consumer protection, it would be a mistake to ignore the costs entirely.
The Council's first-reading texts of 21 October 2026 point the same way, and in places they go further than the proposal estimated for in the IA. The better reading of the IA's "no major upfront implementation costs" line is "no costs we were asked to count". Firms that don't want to fund the refunds will fund the controls instead, and the UK's experience gives a rough view of what that looks like at scale.
See how PSD3changes can accelerate and simplify your implementation, from the comparison with PSD2 and the EMD through to delivery.
Read more in this series
Footnotes
-
FCA, Statement of Policy on Cost Benefit Analyses (July 2024): familiarisation and gap analysis (p. 75), training (p. 76), change projects (p. 79). ↩
-
These figures combine the familiarisation costs and legal costs provided in paragraphs 112 and 114 of CP24/20. ↩
-
SWD(2023) 231, options 1d and 1e. ↩
-
PSR, PS23/3 (June 2023), p. 6; cap from PS24/7 (Oct 2024), p. 3. ↩
-
PSR, Consumer standard of caution guidance (Dec 2023). ↩
-
PSR, APP scams reimbursement dashboard (updated 30 July 2026): 88% (£316m) reimbursed across 301,500 in-scope claims; "3% of claims were rejected due to the customer not taking enough care". ↩
-
Frontier Economics, independent evaluation of the APP scams reimbursement requirement, technical report (June 2026). The cost estimates (£44–56m a year of non-reimbursement costs, of which £31–42m is prevention and detection, based on 17 firms covering ~75% of Faster Payments volume) are shown in Figure 48 and described by Frontier as "indicative high-level approximations". ↩ ↩2 ↩3
-
PSR, PS23/3 Annex 4: cost benefit analysis, para 1.77, p. 19. ↩