The Examiner We Never Hired

 The House of Lords has voted to constrain the regulators’ power to commission skilled person reviews. The debate was about how often the power is used. The more important question is why the UK relies on it at all.

The vote

On 9 September the House of Lords defeated the Government on Amendment 64 to the Financial Services and Markets Bill. The amendment constrains the power of the FCA and the PRA to require a report by a skilled person under section 166 of FSMA. It was moved from the Liberal Democrat and Conservative benches by Baroness Bowles of Berkhamsted and Lord Altrincham, carried by 226 votes to 162, and is now clause 25 of the Bill before the Commons.[1]

The movers argued that section 166 had drifted from its purpose. In their view it is used routinely where it was meant to be exceptional, and it lets regulators delegate supervision to expensive professional firms at the expense of the firms being supervised. For the Government, Lord Pitt-Watson accepted that reviews should be proportionate but rejected a statutory threshold. He made three points: the regulators already weigh the firm’s circumstances, the cost, and the availability of alternative supervisory tools; a further test could delay intervention; and FCA usage has been broadly stable, with 31 reviews commissioned in 2025/26, the second-lowest number since 2016.

Both sides argued about frequency and proportionality. Neither asked who ought to do the work a skilled person review does, or who ought to pay for it. Those are the more important questions, and the answers begin with history.

Two traditions

Most banking supervision grew out of examination: officials visiting a bank, reading its files, and testing its books. The Bank of England took a different path, and it did so deliberately. After the secondary banking crisis of 1973–74, its remedy was better returns (i.e. the regular data submitted by the banks) combined with regular discussion with banks’ senior management. In 1975 George Blunden, then the Bank official responsible for banking supervision and later its Deputy Governor, described the aim as “a relaxed two-way exchange not for an inquisitorial examination”. Such discussion, he argued, was “more conducive to the maintenance of good banking practices than the technique adopted in many other countries of sending in teams of inspectors to examine banks’ books.”[2] This was a claim about influence as much as information: that conversation with the people running a bank would shape its conduct better than inspection could. It left the Bank with one structural weakness. Its analysis, and consequently the quality of the conversation on which it placed so much store, could be no better than the returns the banks submitted.

Johnson Matthey Bankers (JMB) exposed that weakness in 1984, and the Bank admitted it. By June of that year, JMB’s two largest exposures stood at 76% and 39% of its capital; however, its returns showed 38% and 34%, with the larger apparently falling. The returns were not silent – the Bank acknowledged they gave some clues elsewhere – but its identification of the problem was “seriously hindered by misreporting of the large exposures”. The Bank’s subsequent review of JMB’s collapse named two features of its supervision: reliance on the accuracy of banks’ returns and the encouragement given to management to bring concerns forward early. Of course, these were the two features on which Blunden’s approach rested, and the Bank concluded that neither had proved justified in JMB’s case.

Most of the misreporting was caused by deficient systems rather than any deliberate attempt to mislead. What was missing was what examiners reading a bank’s books would have supplied: any detailed analysis of the quality of a bank’s loans, and any means of assessing whether its controls worked. The Bank’s approach meant it did not directly monitor these itself. For both, it relied heavily on the bank’s external auditors.[3]

The Bank had already paid for a targeted examination during the crisis itself. After JMB’s own auditors had looked at the two largest loans, it commissioned a separate examination of the loan book from Price Waterhouse. The Banking Act 1987 turned that workaround into a power. Section 39 allowed the Bank to require reports from accountants, and in practice these took two forms: a periodic check on the accuracy of a return selected by the Bank, and a review of systems and controls, both paid for by the bank. The work normally went to the firm’s own auditors, for a sensible reason. They already knew the books, so the arrangement was economical.

This system was further formalised after the failure of Barings Bank in 1995. The subsequent report concluded that s39 reports had become too infrequent, contributing to the Bank’s failure to spot the fraud that led to the collapse. From now on, the reports should be annual, and any “holidays” needed to be separately justified.

When FSMA replaced the Banking Act in December 2001, section 39 became section 166 and was applied to all regulated firms, not just banks. The systems and controls review survived, but was now commissioned case by case, no longer presumed to be annual. The routine returns check did not.

Two clocks

The need for the returns check did not disappear. The clearest evidence is the PRA’s 2019 enforcement case Final Notice to three Citigroup entities, which carried a penalty of £43.9m for failings in the controls behind their regulatory reporting.[4]

Citi, in effect, used two clocks. One measured the liquidity coverage ratio it reported to the PRA. The other was the internal liquidity measure it used to run the business. The PRA could see only the first clock. Between October 2015 and June 2016, the reported figure was misstated by up to 47%. The discrepancy came to light because the firm compared the two in a monthly internal check, and in late 2016 it told the PRA. The PRA had caught some smaller data-entry errors in the returns. The firm found the big problem using an instrument only it could see.

The PRA then required a s166 skilled person review, which the firm contracted. Over five months, this examined four sets of returns in detail. It gave an adverse opinion on the liquidity returns and qualified opinions on the others, recording 106 findings. Cumulatively, errors in the capital returns had understated risk-weighted assets by $15.4bn, which reduced the entity’s capital (CET1) ratio from 11.8% to 10.3%. The review’s scope covered both of the old section 39 functions: the accuracy of the returns and the systems and controls behind them.

The Final Notice, like the Bank’s account of JMB, describes failing systems and controls, not deception. Thirty-four years apart, the failure was the same: returns were wrong because the machinery producing them was weak, and they went to a supervisor with no independent means of testing them.

Citi was not an isolated case. From 2019 the PRA ran a programme of skilled person reviews of regulatory reporting at larger firms, which included an assessment of the accuracy of the returns themselves. Its 2021 findings described significant deficiencies at several firms, and the PRA never published the programme’s scale.[5] The need never went away, but the routine did, thereby removing any standing capability to meet it.

Close substitutes

In my work with other regulators, I have been able to see, at close quarters, systems with strong examination capabilities. There, examination is carried out by the supervisor’s own staff; it is a standing function, and it builds cumulative knowledge. It is therefore not the same as the Risk Specialist reviews the PRA conducts, and it is further still from a section 166 review.

From the position of a UK regulator, however, the last two could be seen to some extent as substitutes for an examination capability – although not perfect ones. In many areas the preference would be to use the regulator’s own resources. For one thing, it builds the regulator's human capital. But internal resources are always oversubscribed, and some skill sets are needed so infrequently that it is uneconomic to retain them in a relatively small team. In place of such teams, the regulator could reasonably rely on a properly scoped s166 skilled person report. The PRA’s supervisory statement on skilled persons acknowledges this, saying that, in deciding whether to use a skilled person, it will consider “whether the resources required are available within the PRA to conduct the review itself.”[6] When Lord Pitt-Watson told the House that the regulators already weigh the availability of alternative supervisory tools, he was describing a test of capacity.

The section 39 logic of economy has also survived in form. The FCA’s Handbook gives appointing the firm’s own auditor as its example of a cost-effective appointment.[7] In my experience, the auditor was the usual appointment, but this was less the default for the FCA.

The accounting

These two substitute options reach the regulator through different routes. A review by the regulator’s own Risk Specialist Team is funded from the annual fees charged to all regulated firms. The firm pays nothing at the point of delivery, and the cost appears in the regulator’s budget, which is published and scrutinised. By contrast, a section 166 review is invoiced to the relevant individual firm, and the cost never appears in the regulator’s budget.

That creates a visibility asymmetry. The costs the firm can see when it is billed for the s166, and which are typically only incurred by them and not their competitors, attract complaints, while the cost it cannot see (from a review team visit) attracts far less, usually the time burden on staff. For a regulator under budget pressure, the route that stays off its own accounts will always be attractive. The effect is supervision financed off the regulator’s balance sheet.

Some in the industry have also noticed the asymmetry. Sir Nicholas Lyons, Chair of Phoenix Group and former Lord Mayor of the City of London, told the Lords Financial Services Regulation Committee that the regulators should be required to bear half the cost of skilled person reviews. He accepted that firms would pay it back through higher annual fees. The benefit, on his account, would be discipline: once the regulators’ share appears in their own budgets, more care would go into how skilled persons are chosen and how reviews are scoped. Much would depend on whether the regulators’ budgets were held flat, and the resulting cost pressure would most likely mean fewer s166 reports but more by the in-house risk specialists who are the alternative to s166 in the first place. The committee’s report did not take up the proposal. Fifteen months later, the Lords chose to constrain the power instead and left the question of who pays unaddressed.[8]

The published figures give a sense of scale. The PRA’s operating costs in 2025/26 were £350m for an average of 1,521 staff, a fully loaded cost of about £230,000 per head, or roughly £1,050 per working day.[9] The PRA’s own published charge-out rate for a technical specialist, used when it recovers staff costs from firms, is £225 an hour, or about £1,690 a day.[10] In 2024/25, the PRA’s reviews cost £22.2m in total, equivalent to nearly 100 staff-years; its entire Supervisory Risk Specialists division has 233 staff. The FCA’s £47.6m of completed reviews equals just over 6% of its annual funding requirement, and none of it appears in the budget Parliament examines.[11]

One caveat matters. Not every skilled person review is an examination. As an example, some oversee customer remediation, which would always be undertaken by an external team. The regulators classify reviews by purpose but do not publish that breakdown, so the share that could be brought in-house cannot be measured. The absence of that figure deserves notice in its own right, because it is the one number that would settle the argument.

The objection

The strongest case for s166 is that the polluter pays. A levy-funded examination function would make well-run firms pay for the examination of the badly run, whereas the s166 invoice places the cost on the firm that caused the concern and gives it an incentive to put things right. The PRA’s own policy reflects this, listing among its cost considerations “whether the work to be carried out by the skilled person is work that should otherwise reasonably have been carried out by a firm, or by persons instructed by a firm on its own initiative”.[12] Some of what a skilled person does, on this view, is work the firm should have commissioned for itself. The policy therefore contains both sides of the argument, and also explains why a larger in-house function has never been built. Most firms are never subject to a section 166 review, and they would lose from a higher levy.

There are designs that meet this objection. The OCC charges an hourly fee for special examinations that it initiates and adjusts the rate explicitly to track its costs.[13] Delaware charges each institution the actual cost of its examination, including salaries and benefits. New York apportions its supervisory budget across the industry according to the supervisory hours each type of institution consumes, which is effectively a levy weighted by how much supervision a firm needs.[14] And FSMA fee rules already contain a Special Project Fee, which bills named firms for regulator staff time at published hourly rates. The billing machinery exists, and it is already used for tasks such as risk model reviews. It has simply never been applied to examination.

The choice

Cost is only part of the case. When the examiner is usually the firm’s auditor, the review is conducted by the firm’s own paid adviser, sometimes on systems it has already reviewed. The regulators manage that tension, but it is built into the model. The knowledge problem runs deeper still. What a skilled person learns largely stays with the skilled person, while what an examination team learns stays with the regulator and informs the next review and the next firm. A section 166 review only buys a report, but an examination function builds capability. A regulator’s own independent analytical capability is not something it can go on hiring by the day indefinitely.

Clause 25 of the Bill, which the House of Lords has now added, puts a valve on one pipe and leaves the other untouched. Taking a step back, three positions are open to Parliament, and each can be defended. The first is to keep section 166 as it is. The second is to constrain it and accept less supervisory intrusion. The third is to constrain it and build an examination capability inside the regulators, with its costs recovered from the firms examined at the regulators’ own rates. The position that cannot be defended is constraining section 166 without saying which of the other three follows.

In 1985 the Bank wrote that, for the quality of a bank’s loans and the effectiveness of its controls, it relied heavily on the bank’s external auditors. Four decades and two statutes later, it still does. Section 39 began as a deliberate choice of tool, made for good reasons after a near-failure; its successor, section 166, has survived as a budget convenience. The Commons now has an opportunity to make that choice explicit and own it.

Endnotes

[1] Hansard, HL Deb, 9 September 2026, col. 762, Division 2.

[2] George Blunden, “The supervision of the UK banking system”, Bank of England Quarterly Bulletin 15, no. 2 (June 1975), 188–194, at 191. A shortened version of a talk given on 17 March 1975 to a seminar organised by the Institute of European Finance, University College of North Wales, Bangor.

[3] Bank of England, Report and Accounts for the year ended 28 February 1985, annexe, “The Bank of England and Johnson Matthey Bankers Limited”, pp. 35–37.

[4] Prudential Regulation Authority, “Final Notice: Citigroup Global Markets Limited, Citibank N.A. London Branch and Citibank Europe Plc UK Branch”, 26 November 2019, paras 1.7–1.8, 2.11–2.17.

[5] David Bailey and Rebecca Jackson, “Thematic findings on the reliability of regulatory reporting”, PRA letter, 10 September 2021.

[6] Prudential Regulation Authority, SS7/14, “Reports by skilled persons”, November 2024 update, para 2.9.

[7] FCA Handbook, SUP 5.4.9G.

[8] Sir Nicholas Lyons, written evidence to the House of Lords Financial Services Regulation Committee, “FCA and PRA’s secondary competitiveness and growth objective”, SCG0067, 6 February 2025; House of Lords Financial Services Regulation Committee, Growing pains: clarity and culture change required, 2nd Report of Session 2024–25, HL Paper 133, 13 June 2025.

[9] Prudential Regulation Authority, Annual Report 2025/26, June 2026.

[10] PRA Fees Amendment Instrument 2025, ch. 5; FCA Handbook, FEES 3 Annex 9.

[11] Financial Conduct Authority, “Skilled person reports 2025/26”; FCA, PS25/8.

[12] SS7/14, para 2.8(ii).

[13] Office of the Comptroller of the Currency, “Fees and Assessments Structure, Calendar Year 2026”.

[14] Delaware Office of the State Bank Commissioner, “Fees for Examination and Supervisory Assessment”; New York Financial Services Law, s. 206.