FirstRand: The Highest-Return Bank in Africa Learns What a Foreign Regulator Costs

I. Seven Days Before the Verdict

On the morning of 10 September 2026, FirstRand Limited will publish audited results for the year ended 30 June 2026.²⁵ It is an ordinary date on an ordinary corporate calendar, and it is the single most consequential piece of paper the group has produced in a decade — because it is the first set of numbers in which the cost of a decision made in a Johannesburg boardroom nine years ago will land in full, in one line, in rand.

The Arithmetic That Made FirstRand Famous

For most of the last twenty-five years, FirstRand was the answer to a simple question: which bank in Africa compounds capital best? The answer showed up in the arithmetic. In the year to June 2025, the group earned normalised earnings of R41.8 billion, up 10%, on a return on equity of 20.2% — against a stated cost of equity of 14.65%.¹ That gap between what the business earns on shareholder capital and what shareholder capital costs is the entire game in banking. FirstRand had run it wide for a very long time.

Then came the letter from London.

On 30 March 2026, the UK's Financial Conduct Authority confirmed the final design of an industry-wide motor finance redress scheme, covering agreements written between 6 April 2007 and 1 November 2024. The regulator put the cost to consumers at £7.5 billion in compensation and the total bill to firms at £9.1 billion once administration is included, across roughly 12.1 million eligible agreements.⁹ Eight days later, FirstRand told the market what its share looked like. The group would raise its provision by a further £510 million, taking the cumulative charge for the UK motor commission matter to £750 million — about R17.7 billion.⁷

Set that number against the business it belonged to. FirstRand's UK car-finance operation, MotoNovo, had generated roughly £275 million of profit over the preceding decade.⁷ The provision was not a bad year. It was more than two and a half times everything the business had ever made, and it exceeded 40% of the group's 2025 earnings.⁷

Leaving Britain

FirstRand's response was not to fight. It was to leave. The group said that "the UK as a consumer finance jurisdiction will not deliver the returns that the group requires," that owning and operating a UK consumer finance entity was "not within the group's risk appetite," and that it would work with the Aldermore board and the regulators "to facilitate an orderly ownership transition."⁷ It also called the FCA's scheme "deeply flawed in its construction."⁸

Here is the detail that tells you what public markets actually thought of the UK strategy: the share price went up. FirstRand closed 7.14% higher at R93.21 on the day after the exit announcement, even as management guided that full-year normalised earnings would contract by between 4% and 9% once the provision was absorbed.⁸ Investors were not pricing a loss. They were pricing a release — the removal of an open-ended liability attached to a business that had, on the group's own disclosure, never earned its keep.

By then FirstRand had already surrendered something more symbolic. In February 2026, Capitec passed it to become South Africa's most valuable banking group, at R542.4 billion against FirstRand's R534.7 billion.²² By late May, Standard Bank had passed both, at R517 billion versus Capitec's R511 billion and FirstRand's R503 billion — a period over which FirstRand's market value had drifted down from R508 billion at the start of the year while its two largest domestic rivals climbed.²³ The bank that spent two decades as the JSE's compounding machine spent the first half of 2026 as its most argued-over financial stock.

So the story arriving on 10 September is not really about a provision. It is about whether the thing that made FirstRand exceptional — a domestic franchise that earns extraordinary returns on a difficult continent — is intact, and whether the group's judgement about where to deploy the cash that franchise throws off can be trusted after a nine-year experiment ended in a fire sale.

To answer that, you have to start with three young men, R10,000, and a business that was not a bank at all.

↑ Back to Top

II. Traders, Not Bankers: The Origin of an Owner-Manager Culture

In 1977, Johannesburg was not a place where you started a financial institution. Apartheid South Africa was a sanctioned economy, capital controls were tightening, and the banking system was a comfortable oligopoly of institutions with British parentage and very little appetite for risk. Into that, three young men — Laurie Dippenaar, GT Ferreira and Pat Goss — put up R10,000 between them and formed something called Rand Consolidated Investments.¹²

RCI was not a bank. It had no licence, no branches and no deposits. It was a lending and financing business, which in practice meant it did the deals the big institutions found too small, too structured or too strange. That is a critical detail, because it set the cultural DNA that FirstRand still markets today. The founders were not trained as bankers administering someone else's balance sheet. They were principals, risking their own money, and the whole organisational philosophy that followed — small teams, high autonomy, owner-manager accountability, pay linked to the profit you personally generated — grew out of the fact that in the beginning there was nothing else.

The Licence Swap That Made RMB

Goss left. Paul Harris joined. And then came the transaction that gave the business its name and its licence: the trio merged their operation with Johann Rupert's Rand Merchant Bank, taking the RMB name in the process.¹² It is worth pausing on how unusual that was. A merchant banking licence in 1980s South Africa was a scarce, protected asset. The founders effectively swapped their independence for regulatory standing — and kept control of the culture.

For the next decade RMB compounded quietly, then bought its way up the food chain. It acquired an interest in Momentum Life, the insurer. And in 1998 came the transaction that created the modern group: Anglo American disposed of its interests in First National Bank and Southern Life, and those assets were merged with RMB and Momentum. Momentum changed its name to FirstRand Limited and listed on the JSE.¹²

Read that structure carefully, because it explains a great deal about how FirstRand behaves. FirstRand was not a bank that bought an insurer or an insurer that bought a bank. It was a merchant bank's founders reverse-taking-over a much larger commercial bank and a life company simultaneously, using a listed insurance vehicle as the acquisition currency. The people who ended up running the group came from the smallest and most entrepreneurial of the three businesses. That is why FirstRand's operating model — a portfolio of separately branded franchises, each with its own management team and P&L, sitting under a small central holding company — reads more like a private-equity structure than a universal bank's org chart.

The 1838 Half of the Family

FNB itself brought the opposite inheritance: age. Its lineage runs back to the Eastern Province Bank, founded in Grahamstown in 1838 to finance the wool trade in the Eastern Cape export boom.¹³ It passed through the Oriental Bank Corporation in 1874, the Bank of Africa in 1879 and the National Bank, before being absorbed into Barclays Bank (Dominion, Colonial and Overseas) in 1925.¹³ It stayed Barclays until the anti-apartheid disinvestment campaigns forced the British parent to sell down, and in 1987 the business was renamed First National Bank of Southern Africa — wholly South African-owned for the first time since 1925.¹³

So the group that emerged in 1998 was a genuine hybrid: a 160-year-old deposit franchise with the country's deepest retail branch and payments footprint, welded to a young, aggressive, deal-driven merchant bank, run by people whose instinct was to allocate capital rather than administer it.

What the Structure Bought — and What It Cost

Two things followed from that combination, and both matter for the modern investment case.

The first is that FirstRand became structurally good at what it now calls financial resource management — the deliberate allocation of capital, funding, liquidity and risk capacity across competing internal uses, priced against a hurdle. Group CEO Mary Vilakazi described it in March 2026 as "a clear differentiator for FirstRand versus its peers," crediting it with balance sheet efficiency, margin uplift and capital strength.⁴ That is a management claim, not a proven fact, and the rest of this story tests it. But the claim has a real institutional origin: a firm founded by principals tends to ask what a rand of equity earns before it asks what a business unit wants.

The second is a specific vulnerability. An owner-manager model rewards the people running each franchise for growing their own book. It is superb at compounding a business that already works. It is much less obviously suited to the question of whether the group should be in a particular country at all — because nobody inside a franchise is incentivised to argue for its own liquidation. Hold that thought. It becomes the central tension of the UK chapter.

But first, the machine that made all of this affordable.

↑ Back to Top

III. The FNB Machine: What a 41% Return on Equity Actually Buys

There is a number in FirstRand's June 2025 disclosure that is easy to skim past and impossible to overstate. FNB's deposit base crossed R1 trillion during the year — R1.007 trillion in core deposits, up 8%, the largest deposit franchise in South Africa.¹

A trillion rand of customer deposits is not a marketing statistic. It is the raw material of the entire group. In banking, the cheapest funding in the world is a current account: money a customer leaves with you because that is where their salary lands and their debit orders run, on which you pay little or nothing. The gap between what a bank pays for deposits and what it earns lending them out, plus the fees it charges for moving money around, is the whole business. Whoever holds the transactional relationship holds the economics.

FNB holds it. In the year to June 2025 the franchise delivered normalised earnings of R23.6 billion — 56% of group normalised earnings of R41.8 billion — on a return on equity of 37.4%.¹ In the six months to December 2025 that ROE improved again, to 41.0%, with FNB contributing R13.1 billion, or 57% of the group total.³ To put 41% in context: the group's own cost of equity is 14.65%.¹ FNB is earning roughly two and a half times what its capital costs. Very few retail banks anywhere in the world do that on a sustained basis, and the reason FirstRand could afford a decade-long foreign experiment is that this franchise kept paying for it.

Three Engines Under One Account

Where does the return come from? Three places, and it is worth separating them because they carry different levels of durability.

The first is deposit gathering and payments. FNB served 12.37 million active customers at June 2025, roughly 10 million of them in South Africa, and its transactional volumes tell you where behaviour has moved: digital transactions up 11%, app volumes up 14%, card issuing up 6% — and ATM volumes down 5%.¹ Cash is draining out of the system, and FNB has been on the right side of that drain.

The second is a deliberate, and genuinely interesting, pricing decision. FNB cut its fees on low-value real-time payments early, ahead of the market. Real-time clearing volumes then rose 28%, and payments-related non-interest revenue in retail rose 16%.¹ This is a rare case of a bank voluntarily destroying a price point and being repaid in volume. It tells you something specific about the franchise's competitive position: FNB judged it could afford to make instant payments nearly free because the relationship, not the transaction fee, was the asset. That is a defensible read of the evidence — the volume response was real — but note the mechanism. It works only while FNB is the account the salary lands in.

The third is the layer on top: value-added services sold into the transactional base. FNB Connect, the group's mobile virtual network operator, alone carried R22 billion of volumes in FY2025, up from R18.6 billion, with roughly three million customers using this family of services and total retail revenue from them growing 15% to more than R2.9 billion.¹ Insurance income rose 9%, with pre-tax insurance profit up 8%.¹ This is the "integrated financial services" thesis in its concrete form: a bank using the account relationship as a distribution channel for airtime, insurance, rewards and investments.

The Jordaan Inheritance

The digital foundation under all of this was not built recently. It was built under Michael Jordaan, who ran FNB for a decade before announcing his departure in May 2013 and who had earlier headed eBucks.com, the rewards and internet-banking venture.¹⁴ Under Jordaan, FNB was voted the most innovative bank in the world, and the group explicitly credited him with a leadership position in digital banking channels.¹⁴ He was succeeded by Jacques Celliers, then 41, who had joined FNB in ecommerce development and helped launch eBucks.¹⁴ The lineage matters: FNB's app-first posture is the compounding output of a fifteen-year head start, not a recent pivot.

Testing the Fortress

Now the falsification, because a 41% ROE invites the assumption that this is a permanent structural advantage, and the record does not support the strong version of that claim.

Start with FY2020. In the year to June 2020, FirstRand's normalised earnings fell 38% to R17.3 billion, the credit impairment charge more than doubled to R24.4 billion, the credit loss ratio spiked to 191 basis points, ROE fell to 12.9% — below the cost of equity — and the board declared no final dividend, citing the Prudential Authority's guidance to preserve capital.¹⁷ Then-CEO Alan Pullinger called the pandemic "a once in a generation event."¹⁷ It was. But the point stands: this franchise's returns are not weatherproof. A single severe credit cycle took ROE from the low twenties to below the hurdle in twelve months and stopped the dividend outright.

Second, the growth underneath the headline return is modest. FNB's advances grew 5% in FY2025, with retail advances up only 3% and a deliberate tilt to commercial, where the book grew 11% — because household affordability was under pressure and the bank was cutting risk in retail.¹ Its non-performing loans sat at 7.64% of advances and its credit loss ratio at 1.80%.¹ That is a franchise harvesting a mature customer base in a low-growth economy, not one taking share by volume.

Third, and most concretely: FNB's ROE went up while its balance sheet went sideways, and management said why. The improvement to 41% in the December 2025 half was attributed to capital optimisation initiatives, improving retail credit performance and non-interest revenue growth — not to a step-change in the underlying franchise.³ Capital optimisation raises ROE by shrinking the denominator. It is real value creation, and it is also finite.

The honest reading: FNB's advantage is real, mechanically explicable and evidenced by peer-beating returns across a full cycle — but the 41% print flatters it, and the cyclicality is documented in the company's own 2020 accounts. What would confirm or break the durable version of the claim is not the ROE line. It is whether the transactional and deposit franchise keeps growing customers and payment volumes while Capitec and Tyme attack the base beneath it.

The other two South African franchises make that concentration a little less uncomfortable.

↑ Back to Top

IV. RMB and WesBank: The Rest of the Portfolio

If FNB is the annuity, RMB is the option. And in 2025 the option paid.

Rand Merchant Bank — the direct descendant of the founders' original business — delivered normalised earnings of R10.7 billion in FY2025, 26% of the group, on pre-tax profit growth of 11% and a return on equity of 20.7%.¹ The composition is where it gets interesting. South African pre-tax profit rose 18%, while the broader Africa portfolio fell 2%.¹ In the six months to December 2025, RMB accelerated to 13% earnings growth, the fastest of any franchise, contributing R5.4 billion.³

Investment banks are supposed to be the volatile bit. RMB has been the opposite: its credit loss ratio on core lending advances was 0.21% in FY2025, down from 0.31%, against non-performing loans of just 1.35% of the core book.¹ That is a corporate lending book with almost no visible stress, in an economy that grew barely at all. The plausible explanation is client selection — RMB lends to the large South African corporates and multinationals that survive whatever the domestic economy does — rather than any structural cleverness. It is a good business precisely because it is narrow.

The HSBC Handover

The most revealing recent RMB move is an acquisition almost nobody outside South Africa noticed. On 26 September 2024, FirstRand announced a proposed transaction with HSBC to take over selected banking assets, liabilities and employees of HSBC's South Africa branch, transferring the corporate and multinational portfolio to RMB; regulatory approval followed on 10 June 2025, with completion expected in the 2026 financial year.¹ By the December 2025 interim, management confirmed the transaction would complete in the second half and that its costs were already inside guidance, with only a marginal earnings contribution expected.³

Think about what that trade represents. A global bank retreating from a mid-sized emerging market sold its client book to the local incumbent. FirstRand bought distribution and relationships, not assets — the kind of low-risk, in-footprint, capability-adjacent deal that is the opposite of the UK adventure. Whether it earns its cost of capital is not yet knowable; management has guided only to a marginal contribution. But as a signal of where the group now wants to spend, it is unambiguous.

The Same Product, Two Very Different Returns

WesBank is the third and smallest leg: vehicle and asset finance, R2.4 billion of normalised earnings in FY2025, 6% of the group, on pre-tax profit growth of 15% and an ROE of 21.8%.¹ Retail vehicle finance pre-tax profit rose 22% on 10% advances growth; the corporate and commercial side fell 2% as costs rose while it integrated with FNB Commercial.¹ Its credit loss ratio improved to 1.14%, which management attributed to a deliberate focus on better-quality risk and on FNB main-banked customers.¹

That last phrase is the strategic point. WesBank's edge is not that it underwrites car loans better than everyone else. It is that it can originate them inside FNB's existing customer base, where the bank already sees the salary, the current account and the spending pattern. Lending to someone whose bank account you run is a fundamentally different risk proposition to lending to a stranger who walked into a dealership. Cost-to-income fell to 47.5%, the best of the three South African franchises.¹

Hold that thought too, because it is the cleanest possible statement of what went wrong in Britain. The identical product — used-car finance — earned 21.8% in South Africa inside a proprietary customer base and, as the next sections show, never earned its cost of capital in the UK, where it was originated by third-party dealers whose incentives FirstRand did not control.

The Broader Africa Question

There is one more thing the segment disclosure reveals, and it cuts against the "diversified across Africa" framing. FNB's broader Africa pre-tax profit fell 12% in the December 2025 half, hit by constrained client activity, higher funding costs and credit provisions in Botswana on the back of macroeconomic pressure and liquidity problems, plus higher costs in Ghana from implementing a new core banking platform.³ Management's own outlook flagged Botswana as remaining under pressure in the short to medium term and Mozambique as facing significant fiscal challenges.³

The rest-of-Africa portfolio is not, on this evidence, a growth engine offsetting South African maturity. It is a collection of small, individually cyclical, individually policy-exposed businesses that in aggregate contributed R3.7 billion of FNB's R33.6 billion pre-tax profit in FY2025 and shrank in the most recent half.¹³ Investors should size it accordingly.

Which brings us to the deployment that was supposed to be the answer to all of this concentration.

↑ Back to Top

V. The Owner-Manager Model, Tested Against Its Own Deal Record

Every bank says it is disciplined about capital. The useful question is what the discipline produced when it was tested — and FirstRand's record here is genuinely mixed in a way that the group's own narrative tends to smooth over.

Start with the wins, because they are real. In 2011, FirstRand walked away from acquiring Nigeria's Sterling Bank, a deal reported at roughly US$400 million, because the parties could not agree on price.¹⁸ In July 2013, its plan to acquire a 75% stake in Merchant Bank Ghana for about R750 million also collapsed without agreement.¹⁸ At the time the group framed this as "consistently executing its strategy to grow its franchise on the African continent, matched with a highly disciplined approach to protecting shareholder returns."¹⁸

Two abandoned deals in two years, in the two most-hyped African banking markets of the era, at the exact moment when every South African financial institution was being told by its advisers that the continental land grab was existential. In hindsight, not buying a Nigerian bank in 2011 looks like one of the better decisions FirstRand's capital committee ever made. Nigeria's subsequent decade of naira devaluation and banking stress punished almost everyone who did.

Then FirstRand Paid

So the "we walk away on price" characterisation is supported for the 2011–2013 African window. What it does not survive is being generalised.

Because in 2017 FirstRand did not walk away. It paid.

On 6 November 2017, the group announced a cash offer of 313 pence per share for Aldermore, valuing the UK specialist lender at approximately £1.1 billion — R20 billion — a 22% premium to the undisturbed price and 1.80 times net tangible book value.⁶ Read that multiple next to the deals it had refused. FirstRand declined to buy Nigerian and Ghanaian banks on price, and then paid nearly two times tangible book for a British challenger bank founded eight years earlier by a private equity firm.⁶

The rationale, in then-CEO Johan Burger's words, was that the transaction was "the latest step in FirstRand's strategy of protecting and building shareholder value by achieving a more diversified revenue profile" and that it would "provide the platform to fulfil our growth objectives in the UK," while allowing the group "to allocate more financial resources to our operations in Africa."⁶ Burger added that FirstRand was "very comfortable that the financial impact of this transaction is supportive of FirstRand's previous guidance to shareholders on growth, returns, capital position and dividend policy."⁶

Every element of that thesis is worth testing against what happened, and it is a useful exercise because it shows the difference between a strategy being wrong and a strategy being unlucky.

Grading the 2017 Thesis, Line by Line

The diversification logic was internally coherent. FirstRand's UK business at the time — MotoNovo, acquired in 2006 and generating about 4% of group earnings — was, in the group's own words, "undiversified from a product and market perspective."⁶ It made used-car loans, in one country, funded off FirstRand's own South African balance sheet through the London branch. Aldermore brought a £9.6 billion balance sheet, a deposit franchise, and positions in SME, invoice, asset and mortgage finance.⁶ Once integrated, MotoNovo's new business would be funded by Aldermore's deposits rather than FirstRand's, freeing South African balance sheet capacity for African growth.⁶

That funding logic actually worked. By June 2025 the UK operations were 92% funded by customer deposits, with deposits of £17.0 billion against advances of £16.9 billion.¹ The structural problem the deal was meant to solve was solved.

The returns thesis is where it fails, and it failed before any regulator got involved. In FY2025, before a single pound of motor commission provision, the UK operations' underlying return on equity was 10.7%, down from 12.0%.¹ On a normalised basis, after the provision, it was 7.7%.¹ In the six months to December 2025 it was 9.2%.³ Against a group cost of equity of 14.65% and a domestic franchise earning 37–41%, the UK business was destroying economic value in every one of those periods on the group's own numbers — not because of the scandal, but because specialist lending in a mature, competitive, high-capital-requirement market simply does not earn South African returns.

Management knew. FirstRand's FY2025 disclosure listed "improving its return profile" as Aldermore's key focus area, naming capital stack optimisation, risk-reward optimisation for enhanced margin, and "unlocking additional returns from the motor business" as the levers.¹ Eight years after acquisition, the primary strategic objective was still to get the returns up to something acceptable.

There was one genuine capital-management win. In FY2025, Aldermore Group paid a dividend of £125 million to FirstRand — explicitly noted as the first since the 2018 acquisition — taking its CET1 ratio from 16.1% to 14.9% against a medium-term target of 13–14%.¹ Seven years of retained earnings finally came home. Aldermore also obtained a Moody's Baa2 long-term issuer rating with stable outlook in January 2025 and established a euro medium-term note programme in March 2025.¹

The Verdict on the Record

So how should an investor score the capital allocation record? Not as "disciplined," and not as "reckless." The evidence supports a narrower claim: FirstRand's capital committee has been genuinely willing to abandon deals on price in emerging markets it understood, and it overpaid for, and then persistently under-earned on, a developed-market platform it did not. The disconfirming evidence here is not an old failure in an unrelated business — it is the second-largest earnings contributor of the last five years, under the current governance regime, still returning below the hurdle when the current CEO took over.

That is the version of the record to carry into the next chapter, which is where the merely disappointing became something else entirely.

↑ Back to Top

VI. The Machinery of Mis-Selling: How a Car Loan Became a £750 Million Problem

To understand what happened to FirstRand in Britain, you need to understand a piece of plumbing so mundane that for fifteen years nobody thought to look at it.

When a customer in the UK bought a used car on finance, the loan was almost never arranged by the lender. It was arranged by the dealer, acting as a credit broker. The dealer had a panel of lenders, and the lender paid the dealer a commission for the introduction. In many arrangements the dealer had discretion over the interest rate charged — a discretionary commission arrangement, or DCA — and the dealer's commission rose with the rate. The customer, standing in a showroom, was told what their monthly payment would be. They were not told that the person negotiating the rate on their behalf was paid more if the rate was higher.

FirstRand's own disclosure describes the mechanics without euphemism: a dealer could adjust the price of the car or alter the term of the agreement so the monthly repayment met both the customer's and the lender's affordability criteria, and could bundle in additional parts or services.¹ It was a system optimised for closing the sale.

Regulator, Ombudsman, Courts, Claims Industry

The unravelling followed a familiar British sequence: regulator, ombudsman, courts, claims industry, redress scheme.

In January 2024 the Financial Ombudsman Service ruled in favour of consumers in two DCA complaints, finding that discretionary commissions created an inherent conflict between broker and consumer and that lenders' disclosures were inadequate — while ruling for lenders on a non-DCA complaint.¹ The same month, the FCA paused complaints handling for DCA cases while it worked out what to do.¹ Claims management companies and claimant law firms, often backed by litigation funders, poured into the gap and filed heavily in the county courts, where the industry defended with what FirstRand described as a large majority win-rate on the underlying facts.¹

Then, in April 2024, the FCA told firms in scope they must at all times maintain adequate financial resources and signalled that its process might result in an industry-wide redress scheme. Under accounting standards, that made a provision necessary. FirstRand raised £127.4 million — R3.0 billion — for the year to June 2024, covering discretionary commissions from April 2007 to January 2021.¹

In October 2024 the Court of Appeal detonated the sector. Ruling for customers in three cases — including the Wrench and Johnson matters against FirstRand — it extended the issue far beyond discretionary commissions to all commissions paid by lenders to dealers where the customer's informed consent was not explicitly obtained.¹ The industry changed its disclosure practices immediately. FirstRand received permission to appeal to the Supreme Court on 11 December 2024 and, believing it had a strong case, chose not to raise an additional provision at its December 2024 interims.¹

Winning the Law, Losing the Money

The Supreme Court heard the appeal from 1 to 3 April 2025 and delivered judgment on 1 August 2025. FirstRand characterised the outcome as "a significant success," and on the law it largely was: the court found that motor dealers do not owe customers a fiduciary duty when arranging finance, which removed the foundation for bribery-based claims and superseded the Court of Appeal's findings of dishonesty against FirstRand Bank's London branch.¹

But in the Johnson case, the court found an unfair relationship under section 140A of the Consumer Credit Act 1974 on the specific facts — and the facts were unhelpful. The commission FirstRand paid amounted to 25% of the advance and 55% of the total charge for credit; there was a commercial tie giving FirstRand a contractual right of first refusal that was not disclosed; and Mr Johnson was commercially unsophisticated.¹ The court also emphasised that courts have wide discretion in awarding remedies.¹

FirstRand's defence of scale is worth quoting because it is both true and insufficient: less than 3% of total pre-2021 commissions matched the Johnson outcome, and Mr Johnson had in fact received the lowest interest rate available from FirstRand's London branch through that dealer.¹ The bank was arguing that its worst case was an outlier. That argument works in a courtroom, case by case. It does not work against a mass redress scheme, which is precisely why the FCA chose one.

On 3 August 2025 the FCA said it would consult on a scheme, potentially extending beyond discretionary commissions to fixed commissions and to inadequate disclosure generally, with interest at base rate plus 1% and most individuals likely receiving under £950 per agreement.¹ FirstRand recalculated, extending its provision to cover all commission arrangements from January 2007 to October 2024, and reported a total pre-tax provision of £240 million — R5.8 billion — at 30 June 2025, with a R2.7 billion income statement hit for the year.¹ It noted, with what reads in retrospect as considerable optimism, that this was "a best estimate of what can be raised as a conservatively struck accounting provision."¹

There is a governance detail here that deserves flagging as an accounting judgement, not a footnote. Between December 2024 and August 2025, FirstRand's provision reflected management's confidence in its own legal case. Management was substantially vindicated on the law and comprehensively wrong on the money. The final number was more than three times the June 2025 provision.

For the year to June 2025, the motor commission matter cost the group R2.19 billion of normalised earnings, following R2.42 billion the year before.¹ Excluding it entirely, group normalised earnings would have been R44.0 billion at a 21.2% ROE rather than R41.8 billion at 20.2%.¹ FirstRand presented both, which is appropriate disclosure — and it also began habituating investors to a version of its earnings with the bad part removed.

The final rules were still nine months away.

↑ Back to Top

VII. The Exit: April 2026 and the Anatomy of a Reversal

The FCA published its policy statement on 30 March 2026. The design was more punitive than FirstRand had modelled: a hybrid compensation calculation, higher interest assumptions, and a scheme opening on 30 June 2026 for agreements written from 1 April 2014 and on 31 August 2026 for earlier ones.⁹ The regulator had trimmed the eligible population to 12.1 million agreements from 14.2 million at consultation, and cut the projected redress to £7.5 billion from £8.2 billion — but the total cost to firms still landed at £9.1 billion.⁷⁹ Around 90,000 consumers in cases most like Johnson would receive full commission plus interest.⁹

Eight days later, FirstRand did something South African banks almost never do. It quit a developed market.

The £510 million top-up took cumulative provisions to £750 million.⁷ Management's language was pointed: it called the revised redress model "disproportionate and unfair," singling out the hybrid calculation and higher interest assumptions, and described the scheme as "deeply flawed in its construction."⁷⁸ It also said MotoNovo would require further recapitalisation given the provision, and that this would "severely constrain" the financial resources available to grow motor finance and would weigh on the group's excess capital position.⁷

And then the sentence that ended the strategy: the business case for FirstRand to own and operate a UK consumer finance entity was "not within the group's risk appetite."⁷

Nine Years, Two Opposite Sentences

Sit with the gap between that and Burger's 2017 framing. In 2017 the UK was the platform to fulfil growth objectives, a diversification of the revenue profile, comfortably supportive of guidance on growth, returns, capital and dividends.⁶ In 2026 it was a jurisdiction that would not deliver required returns, in a business requiring recapitalisation, being sold.⁷ Aldermore itself was described as "a resilient and sustainable business serving an important need in the UK market" — the problem was not the asset, it was the country and the regulatory risk attached to owning consumer credit inside it.⁷

How Much of This Was Foreseeable?

Now, the fairness point, because a neutral reading requires it. FirstRand did not mis-sell in a vacuum; the entire UK motor finance industry operated the same model, which is why the FCA built an industry-wide scheme covering 12.1 million agreements rather than pursuing one lender.⁹ Nor was the retrospective standard predictable in 2017: the Court of Appeal's October 2024 extension to all commissions was widely regarded as a surprise, and the Supreme Court subsequently overturned most of it.¹ A South African board underwriting a UK acquisition in 2017 could reasonably not have modelled a 2024 appellate ruling reinterpreting commission disclosure back to 2007.

What was foreseeable — and this is the part that belongs on the group's ledger rather than the regulator's — is that FirstRand was buying into a business model whose economics depended on paying intermediaries it did not control, in a jurisdiction with a documented history of retrospective consumer-redress schemes. Britain had already run exactly this playbook on payment protection insurance. Conduct risk in UK retail lending was not an unknown unknown in 2017. It was the single most expensive known feature of the market.

The market's verdict was the share price rising 7.14% on a day the group guided to a full-year earnings contraction of 4% to 9%.⁸ That is investors saying, in the only language they have, that the option value of exiting exceeded the earnings being given up. It is also, implicitly, a judgement on the original allocation: shareholders were pleased to be rid of an asset the company had paid 1.8 times tangible book for and which, on the group's own disclosure, generated roughly £275 million of cumulative profit against £750 million of provisions.⁶⁷

An Auction With an Open Liability

The exit is now a live process rather than a completed transaction. By 24 August 2026, private equity firms were expected to bid, with CVC Capital Partners and Lloyds Banking Group among the potential buyers and Shawbrook also weighing an offer; bidders could be given the option of making separate offers for Aldermore's banking operations and for MotoNovo.¹⁰ No price has been disclosed, and none should be assumed. A bank being sold with an open redress liability, in a process the seller publicly initiated in protest, is not a strong negotiating position.

Two forward-looking questions matter more than the headline sale price.

The first is whether £750 million is the end. The FCA's scheme runs on take-up rates and case-level determinations, and FirstRand's June 2025 disclosure explicitly noted its provision did not cover certain items — potential dealer liability given brokers' own regulated responsibilities, and other recoverable amounts assessable only once the scheme was defined.¹ Those could cut either way. Investors should treat the number as an estimate under an accounting standard, not a settled bill.

The second is capital. At the December 2025 interim, before the final rules, management stated that its capital position meant that even if a further provision were required, under any of its expected scenarios it would still pay a final dividend within its cover range — on normalised earnings before any UK motor commission provision.³ That is a carefully constructed sentence. It preserves the dividend by measuring cover against a profit figure that excludes the charge. It is defensible, since the provision is genuinely non-operational and non-recurring. It is also exactly the kind of framing a sceptical investor should notice, because the cash leaving the group is entirely real.

The group entered this period with a CET1 ratio of 14.0% at June 2025, rising to 14.4% by December 2025.¹³ That is a comfortable buffer, and it is why a £750 million hit is an earnings event rather than a solvency event. The balance sheet passed the test. The capital allocation process did not.

Which raises the harder question: with the UK gone, what does FirstRand actually own, and who is coming for it?

↑ Back to Top

VIII. The War at Home: Capitec, Tyme, and the Battle for the Main-Banked Customer

The most dangerous competitor FirstRand has ever faced does not have a merchant bank, an insurance arm, a UK subsidiary or a hundred and eighty years of history. It has a savings account, a card, and 26 million people.

Capitec's numbers for the year to February 2026 were the sort that force a re-rating. Headline earnings rose 23% to R16.8 billion, headline earnings per share climbed 23% to R146.06, operating profit before tax grew 25% to R22.18 billion, and the dividend rose 23% to R79.80.¹⁹ Net interest income rose 19% to R24.1 billion and loan disbursements jumped 34% to R98.3 billion, with business banking disbursements up 48%.¹⁹ Active clients exceeded 26 million; personal banking clients rose 7% to 25.2 million; and — the number that should concern FNB most — fully banked clients rose 12% to 9.9 million, or 39% of the active base, up from 37%.¹⁹

"Fully banked" is Capitec's term for the customer whose salary lands with them. It is the same asset FNB's entire economic model rests on. By 30 June 2026, that figure had reached 10 million.²⁰

Capitec Is Running FNB's Playbook

The strategic threat is not the count. It is the direction of travel and the runway. Capitec CEO Graham Lee described the group's market shares as "excitingly low" — 5% of personal consumer credit, 13% of savings and notice deposits, 2% of active life cover policies, 3% of business credit, 2% of business deposits.²⁰ That is a bank with the country's largest customer base and single-digit share of most profit pools, explicitly planning to move up-market into home loans, vehicle finance, insurance and business banking.²⁰

Read that against FNB's playbook and the collision is exact. FNB's model is: acquire the transactional relationship, then sell insurance, telecoms, investments and credit into it. Capitec now has more transactional relationships than FNB and is executing the identical cross-sell, from a lower cost base, with a brand built on being cheaper. Capitec even runs its own MVNO, Capitec Connect, at 1.4% subscriber share — the direct analogue of FNB Connect.²⁰

Tyme Comes From Underneath

The second attacker comes from beneath. TymeBank launched in February 2019 and became, on its own account, the first digital bank in Africa to reach profitability. In December 2024 it raised a US$250 million Series D — US$150 million from Brazil's Nubank, US$50 million from M&G's Catalyst and US$50 million from existing shareholders — at a US$1.5 billion valuation, roughly R26.7 billion, with 15 million customers across the group and 10 million in South Africa.²¹ Its then-CEO Karl Westvig stated the goal plainly: to be a top-three retail bank in South Africa within three years.²¹

Tyme's model is kiosks in supermarkets, near-zero fees, five-minute onboarding, and a cost structure with no branch network. It is not, today, taking FNB's affluent main-banked customers. It is taking the entry-level ones — the customers who become affluent main-banked customers in a decade.

Scoring the Domestic Battlefield

So what does the evidence actually say about FNB's competitive position? Something more nuanced than either bull or bear framing.

On the bear side: FirstRand lost its position as South Africa's most valuable bank to Capitec in February 2026 and then slipped to third behind Standard Bank by May.²²²³ FNB's South African active customer base grew 1% in FY2025, to 10.01 million, with retail up 1% and commercial up 6%.¹ Against Capitec's 7% personal banking client growth, that is share loss in the only currency that matters over a decade.¹⁹

On the bull side, and it is a serious one: FNB is not competing for the same rand. Its ROE of 41% comes disproportionately from higher-income segments, commercial banking and cross-sold products.³ Its retail non-interest revenue growth in FY2025 was explicitly driven by "the higher-income client segments."¹ Capitec's advance into that territory — private banking, business banking, insurance — remains mostly ambition backed by low share, and it is the hardest part of the market to enter, because affluent customers switch banks reluctantly and value service breadth over price.

Note also what the incumbency actually consists of. FNB operated 630 branches and 4,771 ATMs in South Africa at June 2025, plus 118 CashPlus agents domestically and 4,971 across broader Africa, and processed 1.11 billion card acquiring transactions.¹ The card-acquiring business — the terminals in shops — is a genuine two-sided position that pure digital banks find slow to replicate.

The calibrated conclusion: the history does not reject the claim that FNB has a durable franchise, but it narrows it substantially. The defensible version is that FNB holds a strong, high-margin position among affluent and commercial customers, with real switching costs and a payments infrastructure advantage — while losing the mass-market entry point to lower-cost competitors, and therefore losing the option on the next generation. The specific things to watch are FNB's South African active customer count and its main-banked mix, against Capitec's fully-banked count. If FNB's customer number keeps growing at 1% while Capitec's fully-banked base compounds at low double digits, the terminal value of the franchise is being eroded regardless of what this year's ROE prints.

There is one more competitive front worth a sentence: the macro. On 28 May 2026 the South African Reserve Bank raised the repo rate 25 basis points to 7%, its first hike since 2023, on a four-to-two vote, citing inflation risks from the Middle East crisis; it revised 2026 inflation up to 4.4% and cut 2026 GDP growth to 1.2%.²⁴ For a bank with a trillion rand of deposits, higher rates are a near-term earnings tailwind through the endowment effect. For a bank whose retail customers are already showing 7.64% non-performing loans, they are a medium-term credit headwind.¹

Both effects will land in the same set of accounts. Which is a reasonable moment to examine how FirstRand's board has behaved when shareholders pushed back.

↑ Back to Top

IX. The Revolt: What FirstRand's Shareholders Did to the Board in 2021

In December 2021, at FirstRand's annual general meeting, a majority of voting shareholders formally told the board that its executive pay decisions were unacceptable.

The numbers are unambiguous. On the non-binding advisory vote on the remuneration implementation report, 48.75% voted for and 51.25% voted against.¹¹ On the remuneration policy itself, 74.11% voted for and 25.89% against — narrowly failing the 75% threshold set by the JSE listings requirements.¹¹ Both votes failed. For a blue-chip South African financial institution, a majority vote against the implementation report is close to the strongest signal the shareholder base has available short of voting against directors.

The Covid-19 Instrument

What triggered it is a genuinely interesting corporate governance case, and it deserves to be told fairly, because it is not a simple story of executives grabbing money.

Covid-19 destroyed FirstRand's long-term incentive awards. The 2017, 2018 and 2019 LTI grants all failed their performance conditions — affecting more than 4,000 employees.¹¹ The remuneration committee's concern was retention: because FirstRand allocated LTIs to a far larger employee cohort than peers, a total wipeout created an outlier risk that key staff would leave. In response, in 2020 Remco introduced a one-off "Covid-19 instrument" for employees deemed critical to the business, struck at half the original grant value of the 2018 and 2019 awards, ignoring the 2017 failure entirely, and funded by the failed awards so that there was no additional cost to shareholders.¹¹

Remco also made an ex gratia payment to retirees whose 2017 and 2018 LTIs had failed.¹¹

Shareholders were unpersuaded. The company itself identified the two triggers as the treatment of retirees and the lack of support for the Covid-19 award.¹¹ The structural objection is easy to state: performance conditions that can be replaced with a discretionary instrument when they fail are not performance conditions. Shareholders had absorbed a 38% earnings decline and a cancelled final dividend in FY2020;¹⁷ management had absorbed a failed LTI, and then received a replacement worth half of it.

The board's defence was specific and partly evidenced. Remco argued the award achieved its objectives: 90% of the 480 recipients were retained over three years; group earnings exceeded 2019 levels a year earlier than expected; and shareholders recovered 100% of their position while management received 50%.¹¹ The company also noted that excluding one large dissenting shareholder, 94.9% of voting shareholders supported the policy.¹¹

That last statistic is the one to handle carefully. It is true, and it is also the kind of disclosure that reframes a defeat as a near-victory by removing the largest objector. A neutral reading: the policy vote failure was driven by concentrated dissent; the implementation vote failure, at 51.25% against, was not.¹¹

What did the board actually do? It engaged, repeatedly — a pre-AGM ESG roadshow with major shareholders in October 2021, a teleconference on 31 January 2022 inviting dissenting shareholders to give feedback, and a Remco subcommittee roadshow in June and July 2022.¹¹ It stated the Covid-19 award "will also not be repeated," while continuing to defend the retiree payment as the right decision.¹¹ It also acknowledged that ongoing non-support could keep affecting the implementation vote, since the award vested over three years to 2023.¹¹

What It Says, and What It Doesn't

So how should this weigh in an assessment of management credibility today? Carefully, and with limits.

The episode is now five years old, the instrument was one-off and has run off, the board committed publicly not to repeat it, and the executive team has since turned over almost completely — Alan Pullinger handed the group to Mary Vilakazi in April 2024, and FNB has had two CEOs since.¹⁵¹⁶ Scale matters too: the award was funded by forfeited grants, not incremental cost.¹¹ Judged by frequency, remediation and economic consequence, this does not support a general claim that FirstRand's board is captured by management.

What it does support is a narrower and still relevant point: when performance metrics delivered an outcome the board found operationally inconvenient, the board used discretion to soften it, and defended that discretion even after a majority of voting shareholders rejected it. That is a disposition, not an incident. It belongs in the same file as the decision to keep the UK provision at £240 million in June 2025 on the strength of management's own scenario modelling, and the choice to frame FY2026 dividend cover against pre-provision earnings.¹³ None of those is improper. All three are the same instinct: management's own judgement, applied where the rules would otherwise bite.

The governance overlay for the current period is thinner but worth noting. FirstRand's remuneration policy and implementation reports continue to be tabled annually for separate non-binding advisory votes, with Remco required to invite dissenting shareholders to engage if either is approved by less than 75%.¹¹ The next AGM is scheduled for 4 December 2026 — the first opportunity for shareholders to vote after both the UK exit and the FY2026 results.²⁵

Which brings the story to the people who will have to answer for all of it.

↑ Back to Top

X. Vilakazi's Group: New Leadership, a Restructure, and a Narrative Under Strain

Mary Vilakazi did not come up through a trading floor.

She spent eight years at PwC auditing financial services entities, becoming one of the firm's youngest partners in 2005, ran an advisory firm, and served as chief financial officer and then deputy chief executive of MMI Holdings before joining FirstRand in 2018 as chief operating officer, where she also held executive responsibility for diversification into insurance and investment management.¹⁵ She was named a World Economic Forum Young Global Leader in 2016.¹⁵ On 4 October 2023 FirstRand announced she would succeed Alan Pullinger, whose tenure with the group spanned 26 years, and she took the role on 1 April 2024 — the first woman to lead FirstRand.¹⁵

The background is not incidental. An auditor-turned-COO running a group whose founding culture was principal risk-taking is a specific kind of appointment. It suggests a board that wanted rigour and integration rather than another dealmaker — and it arrived precisely as the group's biggest problem turned out to be a conduct and controls failure in a foreign subsidiary.

Two Years, Good Numbers, One Reversal

Her first two years produced good operating numbers and a strategy reversal. On the FY2025 results in September 2025, she framed the year as demonstrating "the benefits of a diversified portfolio with deep franchise value, the continued disciplined allocation of financial resources, focused cost management and successful execution on growth strategies," noting that a 16% increase in operational earnings enabled the group "to absorb a further accounting provision for the UK motor commission matter and still deliver strong normalised earnings growth of 10%."² She added that the operational momentum in FNB, RMB, WesBank and Aldermore set the group up "for another strong performance in the coming year."²

Six months later, on 5 March 2026, the framing had shifted in a small but detectable way. Announcing 11% interim earnings growth and a 21.1% ROE, Vilakazi credited "the group's diversified portfolio of leading client franchises — FNB, RMB and WesBank."⁴ Aldermore had dropped out of the list of franchises supporting the performance and appeared later, separately, as having "delivered a mixed operational performance."⁴

That is not a contradiction, and it should not be presented as one — the UK genuinely did have a mixed half, with pre-tax profit down 13% and higher impairments on a worsening forward-looking macro view.³ But it is a visible narrowing of the story five weeks before the exit announcement, and it tells you the internal conclusion preceded the public one.

More importantly, the interim disclosure was explicit that the June 2025 provision had not been updated in the period, because the group was awaiting the FCA's final rules.³ Only £2.2 million of legal and professional costs, and R333 million of related group costs, were booked.³⁵ Management stated plainly that it would update shareholders following the FCA's announcement.³ Whatever else can be said, the group did not hide the pending exposure; it flagged the trigger and named the date it would speak again. That is guidance discipline, and it counts.

The Restructure Nobody Outside South Africa Noticed

On 30 March 2026 — the same day the FCA published its final rules — FirstRand announced the biggest reorganisation of its South African banking operations in years. Lytania Johnson, a 25-year group veteran who had run FNB's personal banking sub-segment, was appointed CEO of FNB effective 1 April, succeeding Harry Kellan, who took early retirement at the end of the year.¹⁶ Kellan had been FNB CEO since April 2024, after 22 years with the group including ten as group CFO.¹⁶

The restructure collapsed the old multi-layered segment model into three units: a Retail and Business Banking segment under Johnson, who also took the RBB CEO role; a standalone private banking and wealth management division under Sizwe Nxedlana, who had led it since 2023; and a commercial and corporate bank under Muneer Ismail, absorbing the enterprise and public sector sub-segments.¹⁶ Gert Kruger became group chief operating officer, with Emma Mer moving into his former role as group chief risk officer.¹⁶ Kellan described the reconfiguration as "the next step required for the business to be even more agile and responsive to customer needs," and Johnson said the simplified structure would "accelerate the bank's ability to serve its customers better."¹⁶ Vilakazi's own framing was operational: "FNB is in a good place, delivering 10% growth in pre-tax profits, with overall ROE increasing to 41%."¹⁶

Two readings are available and both are plausible. The generous one: with the UK being exited, the group simplified the domestic structure to sharpen accountability against Capitec, and elevated an insider who knows the retail customer. The sceptical one: a bank losing entry-level customers reorganised its retail hierarchy and changed its FNB CEO after less than two years, in the same week its foreign strategy collapsed. The market treated the leadership change as one of the factors behind the share's underperformance in the period.²³

The honest position is that Johnson's tenure is five months old and there is no execution record to judge. What can be assessed is the group's guidance behaviour, and there the record is reasonably clean. In September 2025 management guided to mid-to-high single-digit net interest income growth, non-interest revenue in the low-to-mid teens, a credit loss ratio at the bottom of the through-the-cycle range, and costs rising above inflation.¹ In March 2026 it reaffirmed all of it as unchanged, with staff increases settled at 5%, an additional 2–3% of expense growth from systems investment and headcount, and an explicit warning of a higher effective tax rate.³ It also flagged that expected private equity realisations were "not guaranteed to close before June 2026" — a caveat most managements would have left out.³

The results on 10 September 2026 will be the first real test of whether that discipline holds when the number is bad.

↑ Back to Top

XI. Myth Versus Reality: What FirstRand Actually Is

Every widely-held stock accumulates a consensus story. FirstRand has four, and they are worth taking apart individually, because in each case the truth is narrower than the myth.

Myth one: FirstRand is a diversified pan-African and international banking group.

Reality: it is a South African retail and commercial bank with a good domestic investment bank attached, plus a set of small foreign operations that have not moved the needle. In FY2025, FNB, WesBank and RMB together produced 88% of pre-provision segment earnings, and the broader Africa portfolio inside FNB contributed R3.7 billion of R33.6 billion in pre-tax profit — before shrinking 12% in the following half.¹³ The UK, the largest diversification ever attempted, is being sold. After completion, FirstRand will be more concentrated in South Africa than at any point since the Aldermore acquisition. Investors should underwrite it as a South African bank, because that is what it is.

Myth two: the group's financial resource management is a structural, peer-beating advantage.

Reality: the evidence is genuinely good, but the claim needs bounding. The mechanics are real and visible — a group ROE of 20.2% in FY2025 rising to 21.1% at the December 2025 interim, a cost-to-income ratio improving to 50.8%, economic profit of R11.6 billion growing 12%, and net income after cost of capital as the group's stated primary performance measure rather than earnings growth.¹²⁴ Very few banks anywhere run their internal capital allocation off an economic-profit measure and publish it. That is a real discipline.

But the same discipline presided over an eight-year holding of a UK business that never earned the group's cost of equity, and whose stated key focus area in its final full year of ownership was still "improving its return profile."¹ FRM optimises capital within businesses the group has decided to be in. It did not, on this record, force a timely exit from one it should not have entered. The revised claim that survives: FirstRand is unusually good at squeezing returns from an existing balance sheet, and unexceptional at deciding which balance sheets to own.

Myth three: the UK provision is a one-off, so look through it.

Reality: mostly true, with two caveats that matter. It is genuinely non-operational, and the group's underlying performance excluding it was strong — R44.0 billion of normalised earnings at a 21.2% ROE in FY2025.¹ But the cash is real, MotoNovo requires further recapitalisation, the recapitalisation will constrain the group's excess capital position, and the final number remains an accounting estimate that explicitly excludes certain potential offsets and liabilities.¹⁷ "Look through it" is a reasonable analytical stance. "It has no consequences" is not.

Myth four: FNB's 41% ROE proves an unassailable moat.

Reality: the ROE is partly a denominator story, improved by capital optimisation initiatives management named explicitly, and the underlying franchise grew South African customers 1% in FY2025 while its principal competitor grew personal banking clients 7% and fully banked clients 12%.¹³¹⁹ A high return on a slowly growing base is a superb cash generator and a mediocre compounder. The distinction determines what an investor is actually buying.

The Blind Spot: Conduct Risk Is Now the Biggest Loss Line

There is a fifth item that is less myth than blind spot, and it is worth a short aside: the group's non-financial risk profile is now demonstrably its largest source of loss. The UK provisions did not come from credit losses, market moves or a funding crisis — the three things bank risk management is built around. They came from how a product was sold by third parties, reinterpreted retrospectively by courts and a regulator. FirstRand's credit loss ratio of 85 basis points in FY2025 sat comfortably at the bottom of its through-the-cycle range, its liquidity coverage ratio was 129% and its net stable funding ratio 120%.¹ The bank was, by every traditional measure, safe. It still lost more than 40% of a year's earnings to conduct risk.⁷

That is the genuine second-layer lesson, and it generalises. As FNB pushes deeper into insurance, telecommunications, rewards and cross-sold credit inside a single relationship — with gross written premiums on group licences of R8.5 billion in FY2025, up 13% — it accumulates exactly the kind of bundled-selling, disclosure-dependent exposure that generated the UK bill.¹ South Africa's conduct regulator has not run a mass redress scheme on the British model. That is a fact about the past, not a guarantee about the future.

One final piece of consensus deserves correction. FirstRand is frequently described as the highest-quality bank in Africa on the strength of its returns. On returns, that is defensible. On market judgement, the last five years have delivered a lost UK strategy, a failed pair of remuneration votes, the loss of the most-valuable-bank title to a competitor with a fraction of its product range, and a domestic customer base growing at 1%.^11192223 The franchise is excellent. The record of the institution around it is more ordinary than the ROE implies.

Which is the right frame for the final question: what does the case for and against actually rest on?

↑ Back to Top

XII. The Bull and Bear Case: Powers, Forces, and the Three Things to Watch

Put FirstRand through Hamilton Helmer's 7 Powers and the results are unusually lopsided — strong on a couple, absent on most.

Switching costs are the group's genuine power, and they are concentrated in FNB. A main-banked relationship carries the salary deposit, the debit orders, the card, the rewards balance and, increasingly, the insurance and the phone contract. Moving it is a day of administrative work with real risk of a missed payment. This is why FNB can grow non-interest revenue 6% while advances grow 5% and why it could cut real-time payment fees and be repaid in volume.¹ It is a real, evidenced power.

Scale economies exist but are contested. FNB's cost-to-income ratio of 50.3% is respectable for a bank with 630 South African branches and 4,771 ATMs, and RMB's 48.2% is good for an investment bank.¹ But scale in retail banking has an ugly property: fixed distribution assets that were an advantage become a liability when a competitor arrives without them. Capitec and Tyme are attacking with structurally lower cost bases. FirstRand's scale is a moat against Standard Bank, Absa and Nedbank. It is a disadvantage against digital-native entrants.

Cornered resource is arguable in one place only: FNB's card-acquiring position, at 1.11 billion transactions in FY2025, and the merchant terminal estate underneath it.¹ Two-sided payment networks are slow to replicate.

Counter-positioning is where FirstRand sits on the wrong side. Capitec's low-cost, single-product-origin model and Tyme's kiosk-and-app model are counter-positioning plays against exactly FirstRand's structure — and incumbents cannot respond without cannibalising their own fee income. FNB's fee cut on real-time payments was, in essence, a partial and voluntary surrender to that dynamic.

Branding, network economies and process power do not carry material weight here. FNB has genuine brand equity, but South African banking brands do not command price premiums; they compete on fees and rates.

Five Forces: An Oligopoly Attacked From Below

Through Porter's five forces, the picture is a mature oligopoly under attack from below. Rivalry is intense and rising, with five credible domestic competitors and Capitec explicitly targeting adjacent profit pools where its share is low single digits.²⁰ Buyer power is rising as switching becomes easier and price transparency improves. Supplier power is effectively regulatory: the SARB sets the funding cost through the repo rate, most recently at 7%.²⁴ The threat of substitutes is real and structural — mobile money, retailer credit, embedded finance and MVNO-linked wallets all disintermediate a piece of the account relationship. The threat of new entrants is no longer theoretical: Tyme reached 10 million South African customers within six years of launch and attracted Nubank as an investor at a US$1.5 billion valuation.²¹ The one favourable force is regulation, which makes a full banking licence and a trillion-rand deposit franchise expensive to replicate.

The bull case, stated at its strongest: a franchise earning 41% on equity in its largest division, with the country's biggest deposit base, is being handed back capital as a loss-making foreign business is sold, in an economy where structural reform in energy and logistics is slowly improving the operating backdrop.¹³ The group has grown its dividend ahead of earnings — 12% in FY2025 and 18% at the December 2025 interim — while raising CET1.¹³ The HSBC South African portfolio adds corporate relationships to RMB at low risk.¹ Post-exit, FirstRand becomes a simpler, higher-return, more capital-generative business with a cleaner story.

The bear case, stated at its strongest: this is a low-growth domestic bank in a 1.2%-GDP economy, losing entry-level customers to a competitor now worth more than it is, whose management just wrote off the entirety of a decade-long international strategy at a cost exceeding two and a half times that strategy's cumulative profits, and whose retail non-performing loans sit at 7.64%.^1192324 Its ROE is being supported by capital optimisation with a finite runway, its rest-of-Africa portfolio is shrinking, and it now faces a rate-hiking central bank with an already-stretched consumer.¹³²⁴

The activist's angle, which nobody has publicly pressed but which is available: FirstRand's structure invites the challenge that the holding company adds cost without adding judgement. It ran a multi-brand portfolio, made an expensive foreign acquisition its domestic franchises did not need, took eight years to conclude that acquisition would not earn its hurdle, and has just spent a restructure collapsing the segment layers it had itself built.¹⁶ An activist would ask why a group generating 20%+ ROE domestically needed a UK platform at all, what the board's process was for the 2017 approval, and whether the disclosure convention of presenting earnings before the provision is a reporting choice or a framing choice.¹

The Three Numbers That Decide It

What to actually watch. Three metrics, and only three.

One: FNB's South African active customer count and its main-banked mix. This is the terminal value of the entire group. Everything else — the cross-sell, the payment volumes, the insurance premiums, the deposit cost — sits downstream of whose salary lands where. At 1% growth in FY2025 against Capitec's double-digit growth in fully banked clients, this is the number that decides whether FirstRand is a compounder or a cash cow.¹¹⁹

Two: group ROE against the 14.65% cost of equity, and the composition of any change in it. Not the headline. The composition. If ROE holds above 20% because the franchise is growing, the bull case is intact; if it holds because capital keeps being optimised out of the denominator while advances grow at 5%, the bull case is being borrowed from the future.¹³

Three: the final settled cost of the UK exit, including the sale proceeds, any further provisioning under the FCA scheme, and the capital released. The provision is an accounting estimate under a scheme that only opened for the first tranche of claims on 30 June 2026 and for older agreements on 31 August 2026.⁹ Until the sale closes and take-up rates are known, the number is not final.

The audited FY2026 results land on 10 September 2026.²⁵ They will show a group whose South African franchises, on the December half-year run rate, were performing about as well as they ever have — and whose reported earnings will nonetheless have gone backwards, by management's own April guidance, somewhere between 4% and 9%.⁸

That single set of accounts contains the whole argument. A bank can be operationally excellent and still lose a decade of foreign profits in one line. The interesting question for the next ten years is not whether FirstRand can run a 20% ROE in South Africa — the evidence says it can, through a pandemic and a rate cycle. It is whether an institution whose founding culture was built on backing its own judgement has learned anything durable about the limits of that judgement, or has simply been relieved of the evidence.

References

1. Analysis of Financial Results for the year ended 30 June 2025 — FirstRand Limited, 2025-09-11 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. Media release: FirstRand results for the year to 30 June 2025 — FirstRand Limited, 2025-09-11 ↩↩↩

3. Analysis of Financial Results for the six months ended 31 December 2025 — FirstRand Limited, 2026-03-05 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

4. Media release: FirstRand results for the six months to 31 December 2025 — FirstRand Limited, 2026-03-05 ↩↩↩↩

5. FirstRand Limited — Unaudited results and ordinary cash dividend declaration for the six months ended 31 December 2025 (SENS) — Sharenet, 2026-03-05 ↩

6. FirstRand announces terms of offer for UK specialist lender Aldermore — FirstRand Limited, 2017-11-06 ↩↩↩↩↩↩↩↩↩

7. UK motor finance claims force FirstRand to weigh exit — Business Day, 2026-04-07 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩

8. FirstRand's share price surges after announcing plans to exit UK bank Aldermore — Business Report, 2026-04-08 ↩↩↩↩↩

9. FCA confirms motor finance redress scheme — Financial Conduct Authority, 2026-03-30 ↩↩↩↩↩↩

10. Bids for Aldermore banking operations and MotoNovo expected — Motor Trader, 2026-08-24 ↩

11. Remuneration report for the year ended 30 June 2022 — FirstRand Limited, 2022 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

12. Our story — Rand Merchant Bank ↩↩↩

13. South Africa's Oldest Bank — An epic photographic journey — The Heritage Portal ↩↩↩

14. FNB's CEO Michael Jordaan passes the baton after 10-year run — Bizcommunity, 2013-05-22 ↩↩↩

15. Here's what you need to know about FirstRand's first female CEO Mary Vilakazi — The Citizen, 2023-10-05 ↩↩↩↩

16. Lytania Johnson promoted to FNB CEO as Harry Kellan exits — ITWeb, 2026-03-30 ↩↩↩↩↩↩↩↩

17. FirstRand Limited — Provisional audited results for the year ended 30 June 2020 (SENS) — Sharenet, 2020-09-10 ↩↩↩

18. FirstRand remains committed to growing African presence — Bizcommunity, 2013-09-11 ↩↩↩

19. Capitec's earnings climb as client base exceeds 26-million — Business Day, 2026-04-22 ↩↩↩↩↩↩↩↩

20. 'Excitingly low' market shares offer runway for Capitec's growth — CEO — The Citizen, 2026-08-06 ↩↩↩↩↩

21. TymeBank becomes a unicorn — Tyme Group, 2024-12-17 ↩↩↩

22. Capitec Bank overtakes FirstRand as it hits record valuation — Business Day, 2026-02-18 ↩↩↩

23. Standard Bank overtakes Capitec and FirstRand to become Africa's most valuable bank at R517 billion — Billionaires.Africa, 2026-05-25 ↩↩↩↩↩

24. South Africa Lifts Key Policy Rate as Predicted — Trading Economics, 2026-05-28 ↩↩↩↩

25. Key investor dates — FirstRand Limited ↩↩↩