The next UK bank-regulation argument is not whether ring-fencing survives. It is what banks are allowed to run on both sides of it.
The Prudential Regulation Authority said on 18 May that it will consult this summer on reforms to the shared-services rules for ring-fenced banks. HM Treasury published the wider review on the same day, keeping the core policy of separating retail banking from investment banking while looking for a less rigid operating model. For the largest UK banking groups, that puts technology, data-processing and back-office infrastructure at the centre of the regime.
Ring-fencing has always been a structural answer to a specific political and financial memory. After the financial crisis, the UK required large groups to split core retail banking from investment banking so deposits and basic banking services would not be dragged directly into trading-market stress. The regime applies where a bank has more than £35 billion of core deposits and material investment-banking activity. The separation is meant to protect depositors, maintain access to essential services and preserve financial stability.
The reform now being prepared is narrower than the deregulation label suggests. The PRA is not proposing to let capital, liquidity or risk flow freely across the ring fence. It is looking at how groups with ring-fenced entities use shared operational resources: data-processing services, information technology and back-office functions. In other words, the test is whether two legally separated banking entities can use more common operating machinery without weakening the supervisory purpose of the separation.
That matters because the cost of duplication is increasingly a technology question. A modern universal bank does not simply duplicate branch counters or treasury teams. It duplicates data architecture, cyber controls, resilience testing, vendor governance, reporting processes and recovery playbooks. The more AI-enabled compliance, fraud detection and customer operations become embedded in the stack, the harder it is to draw a neat boundary between the operating system of a retail bank and the operating system of the wider group.
HM Treasury’s review frames the change as part of a growth package, not as an abandonment of the post-crisis settlement. The government says ring-fencing still supports financial stability but can be made more flexible, proportionate and responsive to markets and the wider regulatory framework. It will bring forward changes through the forthcoming Financial Services and Markets Bill and later secondary legislation. It is also preparing a consultation on a New Growth Allowance and related reforms that could let ring-fenced banks provide a broader range of products and services to businesses.
The number attached to that growth case is large. Treasury said the reforms could enable banks to provide up to £80 billion in additional support to businesses, including better hedging tools and access to programmes connected to the British Business Bank and National Wealth Fund. That is the policy pitch: keep the depositor-protection architecture, but reduce the frictions that make UK banks less useful to firms as they scale.
For supervisors, the key question is operational substitutability. If a ring-fenced bank and a non-ring-fenced affiliate use the same technology provider, internal platform or service company, the PRA has to be able to see what fails, who controls remediation, and whether retail services remain available during stress. Shared services reduce waste only if they do not turn a legal separation into a dependency map that is too tangled to resolve.
The PRA’s rationale is that the toolkit has changed since the regime began in 2019. The authority points in particular to the UK’s bank resolution regime, which has developed alongside ring-fencing and gives supervisors more options when a large banking group gets into difficulty. That is why the shared-services consultation will sit alongside other recent changes: the Financial Policy Committee’s capital review, the Strong and Simple framework for smaller banks, and higher thresholds for MREL and Resolvability Assessment Framework reporting.
This is the pattern in UK prudential policy now. The authorities are trying to loosen some structural or administrative constraints while insisting that the safety case is being carried elsewhere: resolution, operational resilience, supervisory reporting, capital buffers and targeted thresholds. The FPC’s December 2025 capital assessment lowered the system-wide Tier 1 benchmark from 14% to around 13% of risk-weighted assets, but it did so inside a broader argument about usable buffers and a resilient post-crisis banking system. Ring-fencing reform follows the same logic.
There is a risk in that logic. Moving safeguards from visible structural separation into supervisory tooling can be rational, but it also makes the regime harder for outsiders to inspect. A duplicated service company is easy to criticise as waste. A shared technology platform with clean service-level agreements, tested exit plans and enforceable recovery duties is harder to judge from the outside. The more the UK relies on operational evidence rather than bright-line bans, the more the PRA’s consultation will need to define what counts as acceptable independence in practice.
That is where the summer consultation should be judged. The useful version of reform will not merely say that shared IT and back-office services are cheaper. It will spell out how boards document control, how critical services are mapped, how cyber and data incidents are contained across the group, and how a ring-fenced bank continues to function if a non-ring-fenced affiliate is in trouble. The wrong version would treat shared services as an efficiency gain and leave the resilience proof to supervisory judgement after the fact.
For the UK’s biggest banks, the policy direction is now clear. Ring-fencing is staying, but the operating perimeter is being reopened. The wall around retail deposits is no longer only a legal boundary. It is becoming an engineering, outsourcing and recovery-evidence problem.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.