Blue-tinted photograph of a person putting a coin into a pink piggy bank, with the FC:LAB logo

Full analysis

Dr Nicola Harding, Financial Crime Lab

On 7 October 2026, the UK's mandatory reimbursement scheme for authorised push payment (APP) fraud turned two. Since 7 October 2024, payment service providers using Faster Payments and CHAPS have had to reimburse eligible victims up to £85,000, with liability split equally between the sending and receiving firms (Grant Thornton, 2026). The anniversary comes with two headline numbers that appear to contradict each other. The Payment Systems Regulator's (PSR) independent evaluation estimates that the policy has reduced APP fraud losses by £73 million a year (Payment Systems Regulator, 2026). UK Finance reports that APP fraud losses rose by 19 per cent in 2025, to £576.4 million (UK Finance, 2026).

Both figures can be true, and the reason why tells us something important about the policy. In this piece I read APP reimbursement through the criminological idea of responsibilisation. My argument is that the scheme has succeeded in moving responsibility away from the victim, but it has concentrated that responsibility at the point in the fraud where prevention is hardest.

Responsibilisation, and who it is aimed at

Garland (1996) described responsibilisation as one of the main ways late modern states adapted to the normality of high crime rates and the limited capacity of criminal justice agencies. Instead of acting on crime directly through police and courts, the state works indirectly, activating crime control by private organisations, communities and individuals. He was clear that this is not simply the state withdrawing. The state keeps its powers but governs at a distance, setting the terms on which others take responsibility (Garland, 1996).

Fraud policy has long been an example of responsibilisation aimed at individuals. Awareness campaigns tell the public to stop, think and check, and the implicit corollary is that people who are defrauded failed to do so. Cross's (2015) interviews with 85 older people in Queensland, Australia, found a powerful victim-blaming discourse around online fraud, in which victims are cast as greedy and gullible and held responsible for their losses. Victims and non-victims alike share this discourse, and it isolates those who are harmed and discourages them from disclosing (Cross, 2015).

Mandatory reimbursement reverses the direction of responsibilisation. The scheme places the default cost of fraud on the firms that move the money, while keeping the state in the background as the rule-setter. It is still responsibilisation in Garland's sense, but its target is the institution rather than the individual.

The evidence that it works

The early evidence is encouraging. Frontier Economics' evaluation for the PSR found that the number of APP scams has fallen by nearly 35,000, that reimbursement across all claims has risen from 54 to 65 per cent, and that firms are reimbursing 97 per cent of claims within the scope of the policy. The biggest improvements came from firms with the highest levels of APP fraud before the policy began. That is what we would expect if liability creates an incentive to prevent (Payment Systems Regulator, 2026).

The evaluation also tested the industry's most persistent objection, that guaranteed reimbursement would create moral hazard by making customers careless or encouraging false claims. It found no evidence that either had happened at scale (Grant Thornton, 2026). Criminologically, this matters. The moral hazard argument rests on the same premise as the victim-blaming discourse Cross identified, namely that losses follow from victims' lack of care. The evidence so far does not support that premise.

The 50:50 split also creates an incentive to tackle money mule activity. By making receiving firms jointly liable, the policy gives them a direct financial reason to detect and act on accounts used to receive and move stolen funds. Mule detection becomes a cost-saving activity as well as a compliance obligation.

Why losses can fall and rise at the same time

The apparent contradiction between the headline figures is largely a matter of measurement. Frontier's estimate is a counterfactual, built on in-scope transactions and dated by when the scam payment was made. UK Finance's figures cover all APP fraud, including international and business payments that fall outside the scheme, and are dated by when the firm closed the claim. The PSR's own dashboard excludes out-of-scope transactions and those made before October 2024 (Payment Systems Regulator, 2026).

So the policy can be reducing in-scope losses below where they would otherwise have been while total recorded APP losses continue to climb. UK Finance recorded 248,070 APP cases in 2025, up 7 per cent and the highest number for which comparable data exist (UK Finance, 2026).

A 97 per cent reimbursement rate measures how well victims are compensated. It is a valuable measure of fairness, but it is not a measure of prevention, and the two should not be reported as though they were the same thing. Garland (1996) identified "redefining organizational success" as one of the ways agencies adapt when they cannot control crime directly, and this risk is present here.

Responsibility at the last link

The deeper limitation is where responsibility has been placed. UK Finance's latest data show that 66 per cent of APP cases began online, accounting for 32 per cent of losses, and that a further 17 per cent began through telecommunications, accounting for 28 per cent of losses (UK Finance, 2026). In crime script terms, the approach, grooming and persuasion take place on platforms and phone networks long before the payment. By the time a customer authorises the transfer, they have often been persuaded that it is legitimate, and the bank is intervening at the final and most difficult stage of the script.

Calabresi's (1970) law and economics principle of the cheapest cost avoider offers a useful way to frame the problem. Liability should rest with whichever party can prevent harm at the lowest cost (Calabresi, 1970). For some scams the bank may well be that party, particularly where payment data reveal anomalies or mule accounts. For a fake marketplace listing, a fraudulent advertisement or a spoofed call, the platform or network that hosted the first contact may be better placed and cheaper to act. A model that allocates the whole cost to payment firms creates strong incentives at the payment stage and weak incentives upstream.

This should not be overstated. techUK argues that the 66 per cent figure rests largely on victims' self-reports, which cannot be independently verified and are vulnerable to misattribution when a scam runs across several channels (techUK, 2026). That is a fair methodological point. It is an argument for better data on where scams originate, however, rather than a reason to leave upstream actors without incentives.

What the December consultation should address

The PSR will consult in December 2026 on changes to the scheme, with a decision expected in May 2027. The consultation will cover the consumer standard of caution, claims unresolved within 35 business days, the treatment of "returns from investment", and annual data requirements on the platforms and services fraudsters use. A report on platform data for 2024 and 2025 will be published alongside it (Grant Thornton, 2026).

Three issues deserve particular attention. First, the evaluation found that firms were applying the consumer standard of caution inconsistently (Grant Thornton, 2026). This exception is where responsibilisation of the victim can quietly return, so it needs clear and narrow criteria that resist victim-blaming assumptions. Second, the costs of the policy fall unevenly. Changes in annual reimbursement costs ranged from a £14 million reduction to an £18 million increase, and for one firm the increase was equivalent to 2.6 per cent of revenue. Smaller and growing firms reported the sharpest pressure. Third, the £85,000 cap may leave the highest-harm victims short. Of the £5.3 million above the cap that was not voluntarily reimbursed, £4.9 million would have been covered by the originally proposed £415,000 limit, and this may grow as slower-reported investment frauds come through (all Grant Thornton, 2026).

The platform data report may prove the most consequential element. The PSR's managing director, David Geale, said that "The evidence is clear" and that "APP reimbursement is working", while calling for "a step-change in the approach taken by tech firms and telcos" (Payment Systems Regulator, 2026). Reimbursement has shown that liability changes behaviour. The open question is whether that lesson will be applied to the places where scams begin.

References

Calabresi, G. (1970) The Costs of Accidents: A Legal and Economic Analysis. New Haven: Yale University Press. Available at: https://archive.org/details/costsofaccidents00cala_0

Cross, C. (2015) 'No laughing matter: Blaming the victim of online fraud', International Review of Victimology, 21(2), pp. 187-204. Available at: https://doi.org/10.1177/0269758015571471

Garland, D. (1996) 'The limits of the sovereign state: Strategies of crime control in contemporary society', British Journal of Criminology, 36(4), pp. 445-471. Available at: https:// doi.org/10.1093/oxfordjournals.bjc.a014105

Grant Thornton (2026) 'APP fraud reimbursement requirement: were industry's concerns justified?', 21 September. Available at: https://www.grantthornton.co.uk/insights/app-fraud- reimbursement-requirement-were-industrys-concerns-justified/

Payment Systems Regulator (2026) 'Payment fraud falls by £73m following PSR reimbursement scheme', 1 July. Available at: https://www.psr.org.uk/news-and-updates/latest-news/news/payment- fraud-falls-by-73m-following-psr-reimbursement-scheme/

techUK (2026) 'Fraud demands collaboration, not finger-pointing'. Available at: https:// www.techuk.org/resource/fraud-demands-collaboration-not-finger-pointing.html

UK Finance (2026) 'Fraud remains a national security threat as criminals steal almost £1.3 billion', press release for the Annual Fraud Report 2026, 15 June. Available at: https:// www.ukfinance.org.uk/news-and-insight/press-release/fraud-report-2026-press-release

Publication details

Published by The Financial Crime Lab on 2 Nov 2026.

Cite as: The Financial Crime Lab (2026). Who Pays and Who Prevents: Two Years of APP Reimbursement.

Advisory

Put this evidence to work in your organisation.

Book a consultation with Dr Nicola Harding to translate these findings into prevention strategy, product design and frontline practice.