The magic deregulation tree

The regulators have hit their target. The Government's Growth and Competitiveness Strategy asked the FCA and PRA to cut the burden of the Senior Managers and Certification Regime by half. The FCA's CP26/37, published last week, delivers a 53% cut, worth £229m a year, and the PRA's CP14/26 a 47% cut worth £11.5m.

Impressive. But, on closer inspection the claimed saving is questionable. When introducing the regime the regulators said it would cost (at most) £231m a year to run. They now say it costs £459m, and that is before £153m of costs firms reported which the FCA could not match to any rule. They propose to save £240m.

So the saving is bigger than the regime was ever supposed to cost. Even after inflation, the regime costs half as much again to twice what was forecast. The two consultation papers do not explain why. Nor does anyone ask the question that really matters for growth: what firms will do differently as a result of the changes.

This is what a burden-reduction target does. Give a regulator a percentage to hit and it will find a baseline, cut what counts against it and declare success. The actual cost savings may be mostly imaginary in practice. And the harder question, what regulation would help firms invest and grow, will not be asked, because no target was set for it.

What a target does to a cost benefit analysis

In 2016 the FCA's impact assessment put the running cost of the banking regime at £26.75m a year. In 2018 the CBA for the extension put it at £140.8m to £200m for solo firms and about £4m for insurers. Together, £171m to £231m a year, or perhaps £230m to £320m in today's money. The 2018 analysis moderated firms' figures down because firms might overstate costs to get a lighter regime, and concluded the regime would pay for itself if it cut harm by 9% to 17%.

CP26/37 now puts the running cost at £434.5m a year, plus £153.2m of overheads it cannot tie to any rule. On the FCA's own figures the regime costs about twice what it said it would, or nearly three times with the overheads. The FCA does give a wide range: at the 25th percentile the regime costs £233m, which is what 2018 forecast; at the 75th, £935m. But it chose the median for its headline, and the median is what this piece compares.

The 2016 banking assessment gives a total only, so the 2016 and 2018 element rows cover solo firms and insurers. The FCA baseline covers FCA rules at all firms; the PRA baseline covers PRA rules at banks and insurers. The PRA treats Conduct Rules training as an FCA matter.

The banking figures can be traced further, because the FCA published the consultants' working in CP14/13. Europe Economics modelled the ongoing cost of certification for banks as negligible, with continuing fitness and propriety checks at £176,000 a year for a large bank and £12,000 for a small one, and Conduct Rules at £32,000 for a large bank and negligible for a small one.

So both original analyses assumed that the two things which now cost the most would cost almost nothing to run. Certification was negligible in 2014 because the regulators expected it to ride on firms' existing appraisal cycles; firms built a separate process instead. Conduct Rules training was a one-off in 2014 and was discounted as an ongoing cost in 2018. Twelve years later those two elements are £318m of the £434.5m baseline. Either firms built processes the regulators never expected, or the regulators never understood what they were asking for. The CP does not say which.

Three reviews led to these reforms: the PRA's 2020 evaluation, the regulators' 2023 discussion paper and the Treasury's 2026 consultation response. None measured what the regime cost. The Treasury committed to the 50% target in July 2025. Its impact assessment for the Bill, in March 2026, took the 50% as an assumption and produced an annual saving of £70m; the Regulatory Policy Committee passed it, noting that the origin of some assumptions was not clear. The regulators' survey of what 100% was ran from April to July 2026, and the FCA's £229m followed in October, three times the Treasury's figure for the same policy. The baseline was built after the target was set.

It shows. The unit costs are medians from 260 self-reported responses, a 10% response rate, which the FCA concedes include ordinary HR cost; no respondent data, no external benchmark. The saving is gross: every removed activity is counted as saved in full, on the reasoning that a firm which keeps doing the work has decided it pays for itself. The FCA calls the result the value of requirements avoided, not a prediction of what firms will save; the headline then uses it as if it were money in firms' pockets. The lost benefit is not measured. The 2018 case for the regime turned on a 9% to 17% cut in harm, measured by redress, complaint handling and fines; the 2026 case for removing half of it estimates no change in harm and says safeguards remain. And the 53% is measured against a baseline from which the £153m of overheads has been removed. Put them back, allowing that some would fall too, and the saving is above 39% and short of 50%.

The PRA's CP14/26 is the same exercise on a smaller scale: PRA rules at about 1,200 banks and insurers, 96 responses from the same survey, a baseline of £24.8m and a saving of £11.5m, or 47%. Its largest line, £7.4m of other costs the survey did not break down, is simply halved. Three points short of 50%, the PRA says the unquantified benefits of faster appointments may be at least as large as the quantified savings and declares the commitment met. The money is on the FCA side: 139,000 certified people against the PRA's 7,700, and £71m a year of Conduct Rules training the PRA leaves out as an FCA matter.

What a target does to policy

If the number is what matters, the reforms chosen will be the ones that save the most on paper, not the ones that change what firms do.

Removing the Certification Regime saves £133m on paper, the biggest line by far, and replacing Conduct Rules training with a notice saves £56.8m. What do they change? The certificate, the annual date and the training module go. The duty to employ competent people does not. Chapter 4 of the CP says so: SYSC 3.1.6R and 5.1.1R still apply, the Training and Competence rules still apply, and the Consumer Duty expects firms to make sure staff can deliver good outcomes. A firm that stops assessing the fitness of its traders and advisers, or sends an email in place of training, will have to explain to a supervisor how it satisfied itself that they were competent. My guess is most firms will keep the annual cycle, call it something else and the cost savings will be limited. The £153m of overheads the FCA left out of the baseline tells the same story: firms built those assurance frameworks because supervisors and enforcement ask for evidence, and nothing in the CP says they will stop asking. Compliance cost is driven less by what the Handbook says than by what a supervisor or enforcement investigator will accept. The risk is that the CP shrinks the Handbook while leaving the evidential expectation where it was.

The reduction in roles required to be approved as Senior Manager Functions raises similar questions. Removing the Other Overall Responsibility and Local Responsibility functions takes about 1,400 people out of the approved population. However, it changes their obligations less than might appear from the rulebook. They are no longer subject to the Senior Manager Conduct Rules and the Duty of Responsibility. But they must still comply with the Individual Conduct Rules, and the regulators have always read the duty to act with due skill, care and diligence as requiring a manager to run their area properly, whether or not they are an SMF. A manager who runs a business line without an SMF label still answers to the FCA if it goes wrong, and the steps a careful manager takes to protect themselves are the same. Yes, firms will save the costs and delays associated with applications for approval for these roles. That is not nothing, and the reduced delay in appointing people to these roles is certainly of value. But the actual cost saving for firms will likely be marginal. Firms will still need to be able to evidence to the regulator on request how they satisfied themselves as to the suitability of individuals for these roles. That will require process and paperwork.

Removing regulatory references for certified staff will deliver cost savings, because a firm cannot compel a reference from a previous employer once the rule has gone. The Fair and Effective Markets Review recommended mandatory references in 2015 because voluntary references were not stopping individuals with poor conduct records moving between firms. They were the industry's main tool for stopping 'rolling bad apples' moving from firm to firm. The CBA counts their removal as a saving and assigns no value to the increased risk of harm. It will look like a good trade until the next trading scandal.

The Directory goes too, on the ground that it confused people. It was introduced in part because the industry asked for it: certified staff had once been approved by the regulator, and once they were not, firms could no longer track them through the system. That need has not gone away. It does not count towards the target, so it has not been weighed.

Meanwhile the reforms that might actually help a firm are the ones the CBA cannot count. Moving seven SMF roles to notification saves £0.28m, but the CP admits the real value is in not waiting for approval. Extending temporary cover to 18 weeks saves nothing on paper, but it is of significant importance to firms. Faster approvals, the thing firms complained about most, are an operational matter and appear nowhere in the CBA. The changes that would help firms most contribute almost nothing to the target. The changes that contribute most to the target will help firms very little.

What the industry actually asked for

Step back from the numbers and ask a different question. Was SM&CR the right thing to reform at all?

The regime was not broken. A PRA survey of banks and insurers in 2020 found that 95% of firms thought it had improved behaviour, 86% thought certification worked and 79% thought the whole thing proportionate. Its conclusion was that no major change was needed. The firms that responded to the 2025 consultations said the same: the regime was valuable and they did not want it weakened. What they complained about was process. Forms, references, the annual cycle, the time it took to get someone approved. The changes now proposed by the PRA and the FCA address many of those concerns. But is this really worth the effort? Will it really move the dial at all on regulating for economic growth?

I question whether process was ever the real complaint that lay behind the calls for government and regulators to look at reform of SM&CR. In my experience of advising senior people under this regime, what worries them most is not the paperwork. It is the prospect of being investigated, years after the event, for a judgement that was reasonable at the time and turned out badly, and of having to prove with documents they did not know they would need that they took reasonable steps. The regime asks executives to carry personal liability for the ordinary risk of running a business, and leaves them to guess what will count as enough. And the guessing is done against a standard applied in hindsight. An enforcement case or a redress exercise starts from the outcome and works back to the decision, with full knowledge of how things turned out and none of the pressure under which decisions were made.

Nobody puts that in a consultation response. A chief executive does not write to the regulator that he is afraid of being investigated. The trade body writes "competitiveness" and "talent" instead. But the symptom is in every review. The PRA found in 2020 that half the firms it surveyed reported more risk aversion, that some blamed unclear expectations about documentation, and that some firms said the pool of non-executive candidates had shrunk. It felt the need to restate that the standard was reasonable steps, not perfection. The 2023 discussion paper asked, as a standing question, how to stop qualified people being deterred from senior roles. The Treasury's 2026 response records that international candidates, particularly from the United States, are put off by the accountability framework itself, and that this obstacle will survive the reforms.

So what do the reforms do about this central issue? Nothing. The Duty of Responsibility is untouched. The Senior Manager Conduct Rules are untouched. A notified Senior Manager answers for exactly what an approved one did. A manager who loses the SMF18 label keeps much the same exposure under the Conduct Rules. The reforms strip process from a regime the industry had largely accepted and leave in place the one feature it feared.

What might have been targeted instead? A usable standard for reasonable steps. The Handbook already lists the factors the FCA will consider; what it does not say is what will suffice, which is the only question a manager has. A safe harbour for a documented judgement that went wrong. A time limit on investigations of individuals, so that nobody carries an open file for six years. A statement of what evidence a supervisor will and will not expect, so that firms can stop building defensive records. None of these counts against a baseline. All of them go to whether a capable person takes the job.

The magic deregulation tree

We will not get economic growth from a magic deregulation tree. The UK's problem is not too much regulation. It is bad regulation: rules that are vague about what they want, supervisors who want evidence of everything, enforcement that judges decisions by their outcomes, and a cost benefit process that produces whatever number the policy needs. SM&CR was reformed because it could be counted, and the part of it that deters business could not. A regime that costs twice its estimate, and is then cut by a number nobody can check, while the thing the industry feared is left exactly as it was, is bad regulation at both ends.

This is not a criticism of the FCA or PRA staff. They were given a percentage and they delivered it. The fault is in the instruction. Asking a regulator to reduce a measured burden by a fixed amount rewards changes to the measure.

The useful response to CP26/37 is to tell the FCA, in answer to question 27, that the numbers cannot be checked and should not be relied on. Arguing the £229m up or down concedes that it has some meaning. The message for the Treasury should be that if it wants growth from better regulation it should set a target for what deters investment. If it sets targets for burden reduction, it will just get what can be counted.

Consultation closes 22 January 2027.

Sources: PRA CP14/26, Chapter 6; RPC opinion on the Financial Services and Markets Bill impact assessment, June 2026, measure (d); CP26/37, Annex 2; FCA, Extending the SM&CR: Cost Benefit Analysis, PS18/14, Tables 6, 8 and 10; FCA Regulator Assessment, Accountability regime for banking sector, 2016; FCA CP14/13, Annex 2 (Figure 15) and Annex 10, Europe Economics, Tables 1 to 3 and 3.3; PRA, Evaluation of the SM&CR, December 2020; FCA and PRA, DP23/3, March 2023; HM Treasury, Reforming the SM&CR: Consultation Response, April 2026.FCA CP18/19, Introducing the Directory, July 2018; FCA PS19/7, Finalising the Directory, March 2019