Are South African Banks and lenders Hiding Their Bad Debt?
A technically obscure accounting IFRS 9 manoeuvre may be allowing lenders to mask the true scale of consumer financial fragility — and the NCR needs to investigate it
By Kaveer Beharee, Founder & CEO, Ubiquity AI
I heard this via the grapevine from someone present at a recent National Credit Regulator roundtable. An NCR-registered entity, major South African bank was directly confronted by another NCR entity with an allegation: the first lender operating in the mass-market had reported impossibly low delinquency rates, and was doing so by systematically extending loans to delinquent customers, charging them fees in the process, and using that transaction to avoid triggering mandatory impairment provisions under international accounting rules. I cannot verify the account of the meeting independently. What I can do is explain, in detail, why the allegation is technically coherent — and why, if true, it matters enormously.
The Setup: What Banks Are Required to Disclose
Since 2018, South African banks have been required to account for bad debt under IFRS 9 — the International Financial Reporting Standard that governs how financial institutions recognise and provision for loan losses. The framework is sophisticated. It requires banks to classify every loan into one of three stages:
Stage 1 is a performing loan with no significant deterioration in credit quality. A small provision is held — a 12-month expected credit loss.
Stage 2 is a loan where the borrower’s credit risk has increased significantly since the loan was originally made. A larger provision is required — a full lifetime expected credit loss.
Stage 3 is where a loan is credit-impaired: typically 90 or more days in arrears, or where there is objective evidence that the borrower cannot repay. Here, the bank must provision for the full lifetime expected loss, often 30–60% of the outstanding balance or more, depending on collateral.
The difference between Stage 1 and Stage 3 provisioning on a R1 million unsecured loan could be the difference between a R15,000 provision and a R500,000 one. That is a direct, immediate hit to the income statement and to regulatory capital.
The 90-day delinquency threshold is central to the entire system. It is the most commonly used, most clearly defined trigger for Stage 3 classification. Which means it is also the most obvious target for gaming.
The Alleged Mechanic: Extend, Charge, Reset
Here is what is alleged to be happening:
A customer is 60 to 75 days in arrears — approaching the 90-day cliff. Under normal circumstances, the clock keeps ticking, the loan moves to Stage 3, and the lender is required to book a substantial impairment provision.
Instead, the lender contacts the customer and offers an extension. The loan term is lengthened. A new repayment schedule is issued. Crucially, an administration fee — and in some cases an extension fee — is charged and collected.
From the lender’s internal records, this now looks different. The customer has engaged. A new contractual arrangement exists. The lender can argue that the delinquency clock has reset, because a new agreement has been entered into. The loan avoids Stage 3 classification. No full lifetime expected credit loss provision is booked.
The fees serve a dual purpose. First, they generate immediate income, partially offsetting the cost of the exercise. Second, they create paper evidence of a genuine commercial transaction — making the arrangement look less like a forbearance and more like a product modification. This matters enormously when auditors come asking questions.
What this means in practice: the bank’s published loan impairment numbers look healthier than the underlying borrower population justifies. The bank’s income statement is protected. Its regulatory capital ratios remain elevated. And the financially distressed customer — who cannot actually afford the original loan — has now been charged additional fees to remain on a treadmill they cannot escape.
Why IFRS 9 Creates This Vulnerability
The architects of IFRS 9 were aware of the risk of “extend and pretend” — it was a significant failure mode in the Global Financial Crisis. The standard contains several provisions designed to prevent it. But IFRS 9 have a structural weakness in precisely this scenario.
IFRS 9 paragraph 5.4.3 requires that historical delinquency information be considered even after a modification. The modification does not automatically erase the credit history. Its application is highly judgmental - and judgment is exercised by the bank, not the regulator.
IFRS 9 also contains what is known as the “90-day backstop” — a presumption that any loan more than 90 days past due has defaulted and must move to Stage 3. But paragraph B5.5.37 allows banks to rebut this presumption if they have “reasonable and supportable information” demonstrating that a different default definition is more appropriate. Many South African banks have established internal default definitions that already differ from the 90-day backstop. The room for discretion is therefore considerable.
There is a further wrinkle. IFRS 9 requires that a “Significant Increase in Credit Risk” (SICR) trigger movement even to Stage 2. Systematically extending loans to delinquent customers and not triggering at least Stage 2 classification would be extremely difficult to defend under a proper audit. But the SICR assessment is model-driven, and the models are proprietary to the bank. The bank can calibrate its model to treat “customer entered into a new fee-paying arrangement” as evidence of reduced credit risk — a judgment call that is aggressive and extremely difficult to disprove without granular internal data.
Finally, even within Stage 2, banks have substantial latitude in their Expected Credit Loss models. The Probability of Default curves, the Loss Given Default assumptions, and the forward-looking macroeconomic overlays are all set by the bank. A bank could classify a loan correctly as Stage 2 but apply sufficiently optimistic assumptions to produce a small provision. The staging looks right. The numbers are still wrong.
What the Numbers Would Show — If Anyone Is Looking
The systemic signature of this practice is detectable, if you have access to the right data.
At a portfolio level, a bank engaged in systematic delinquency deferral would show some combination of the following: Stage 3 ratios that appear low relative to the macroeconomic environment and relative to peer banks; unusually high rates of loan modifications concentrated in the 60-to-90-day delinquency bucket; fee income lines that have grown faster than the loan book; ECL provisions that have remained surprisingly flat even as household financial stress indicators — arrears at credit bureaus, defaults on other obligations — have deteriorated.
Some of this data is visible, in published financial results and Pillar 3 regulatory disclosures. A forensic comparison of Stage 3 ratios, modification volumes, and fee income trends across the major banks should reveal if this practice exists. The NCR, Prudential Authority and financial journalists covering bank all have access to the published data necessary to begin this analysis today.
Why This Matters Beyond the Numbers
This is not a dry accounting issue. There are real people at the end of every extended loan.
South Africa’s consumer credit market is under extraordinary stress. Real wages have been stagnant for years. The cost of living has risen sharply. South Africa still has the world’s highest unemployment rate. Millions of South Africans are trapped by debt - both loans by established lenders, and then much more expensive short-term and pay-day loans.
When a bank extends a loan to a borrower who cannot service the existing terms, and charges them a fee for the privilege, that borrower’s debt burden grows. Their total repayment period extends. Their interest cost increases. And the fundamental problem — that they cannot afford the loan — remains entirely unresolved. Borrowers are being kept in a formal credit relationship that is slowly worsening their position, while the bank avoids booking the loss.
The National Credit Act is clear that reckless lending — granting credit to consumers who cannot afford it — is prohibited. The NCR role is to enforce regulations and protect consumers. Offering an extension to a borrower who is 70 days in arrears, for the primary purpose of resetting a provisioning clock, is not a genuine credit assessment — it is an accounting manoeuvre that looks exploitative.
A Direct Call to the National Credit Regulator
The NCR has jurisdiction over credit provider conduct and operations. While the prudential framework — capital adequacy, IFRS 9 provisioning — sits with the Prudential Authority at the South African Reserve Bank, consumer protection and lending practices falls within the NCR’s mandate.
The allegation made at the recent NCR roundtable deserves formal investigation, not dismissal.
Specifically, the NCR should:
Obtain transaction-level data on loan modifications and extensions from retail credit providers, including volume by delinquency bucket, fee types charged, and subsequent performance of modified accounts.
Cross-reference modification activity against credit bureau data to identify borrowers who are simultaneously being extended by one lender and defaulting on obligations to others.
Investigate any evidence of systematic reckless extension — extending credit to borrowers who demonstrably cannot afford it,
Formally escalate any evidence of material misrepresentation in IFRS 9 staging to the Prudential Authority for supervisory action.
A Direct Call to Financial Journalists
Published data is already in the public domain. Annual reports. Pillar 3 disclosures. Interim results. Analyst presentations. A comparison of the following across the major retail lenders would be a significant piece of financial journalism:
Stage 3 ratios as a percentage of gross advances, trended over the past four years.
The ratio of loan modification volumes to total book size. Fee income from credit-related modifications and extensions, trended over the same period.
ECL provisions as a percentage of Stage 2 and Stage 3 balances — the “coverage ratio” — and how it has changed. The gap, if any, between credit bureau-reported delinquency trends and bank-reported Stage 3 ratios.
These are all legitimate questions to ask of any public company. If these loan extensions exist, it exposes a gap that regulators, auditors, and boards of directors must account for.
Credit risk officers, internal auditors, collection teams, and impairment model experts inside these institutions know exactly what the models are doing. While there is very little incentive for whistleblowers in South Africa, I suspect (if this practice is happening) some are extremely troubled with the knowledge that consumers are being harmed and shareholders are being misled.
What’s at stake?
If the allegation is correct — if South African banks are systematically using fee-charging extensions to defer delinquency classification and suppress IFRS 9 provisions — then the published financial results of those banks are materially misstated. Investors are making decisions based on loan book quality that does not reflect economic reality. Th Prudential Authority, NCR and shareholders are assessing capital adequacy on the basis of provisioning that understates actual credit risk. And the next credit cycle turn will produce losses that appear sudden and surprising, but were always there, just hidden.
We have seen this depressing film before. The Global Financial Crisis in 2008/9 was, in significant part, a story about loan books that looked healthy until they did not. The extend-and-pretend practices of that era — many of which IFRS 9 was specifically designed to prevent — are not ancient history.
The allegation made at that NCR meeting may be unfounded. I am not in a position to know. But it is technically credible, the incentive structure that would produce it is clearly present, and the harm it would cause — to consumers, to investors, to financial stability.
Someone in a room full of credit regulators allegedly said it out loud. That should be the beginning of an investigation, not the end of a conversation.
This piece is based on a secondhand account of a regulatory meeting and on publicly available information about IFRS 9 accounting standards and South African consumer credit markets. The author is not a licensed financial advisor or attorney. Nothing in this piece constitutes investment advice. Readers with information relevant to these allegations are encouraged to contact the National Credit Regulator directly.
National Credit Regulator: www.ncr.org.za | Complaints: 0860 627 627
Prudential Authority (SARB): www.resbank.co.za
Kaveer Beharee is the Founder and CEO of Ubiquity AI, a predictive AI company solving the $7.4B global collections inefficiency. He was recognised as a Young Enterprise Initiative Laureate in France (2022) within the VivaTechnology ecosystem, and previously advised Parliament of South Africa on the National Credit Amendment Act.
Follow on LinkedIn: linkedin.com/in/kaveer | Twitter/X: @UbiquityAI