Short-term credit (payday loans, salary advance products, and micro-loans below R8,000) carries a specific affordability challenge that payslip-only assessment cannot adequately address. The borrower base is disproportionately informally employed, month-to-month earners, or consumers managing multiple small obligations. A payslip confirms last month's gross salary from one source. It does not show the three debit orders to other lenders that started last month, the R2,400 going to Hollywoodbets each month, or the informal loan being repaid in cash.
Regulation 23A applies to short-term credit in full. The size of the loan does not reduce the verification obligation. What changes is the borrower profile and the data landscape. For a significant share of short-term credit applicants, the payslip is either unavailable or materially incomplete as a verification instrument.
The Reg 23A income verification requirement for short-term credit
Regulation 23A(4) specifies three acceptable forms of income verification: the consumer's latest three payslips, three months of bank statements, or income statements. These are alternatives, but not equivalent in information value. A payslip confirms gross earnings for one pay period from one employer. A three-month bank statement shows income from all sources, the consistency of those deposits month on month, the total debit order load, and the consumer's actual cash conduct.
The three-document hierarchy matters most when the consumer is self-employed or informally employed. South Africa's informal sector accounts for more than 30% of the working population. For these borrowers, payslips do not exist. Three months of bank statements or the latest financial statements are the prescribed alternatives. A short-term lender whose affordability process is built entirely around payslip capture is structurally unable to conduct a compliant assessment for this segment, and it is precisely this segment that dominates micro-lending origination volumes.
The Reg 23A affordability framework operationalises the full calculation: gross income verified, statutory deductions, existing monthly obligations enumerated, living expenses assessed against income-band norms, and discretionary income calculated. A payslip contributes to step one. Steps two through five require additional data that the payslip cannot provide.
What payslip-only assessment misses
A consumer earning R14,500 per month submits a payslip. The calculation looks clean: R14,500 gross, statutory deductions to R11,200 net, standard living expense estimate of R2,500, leaves R8,700 apparent discretionary income. Easy approval for a R4,000 short-term loan.
The bank statement for the same consumer tells a different story. There are debit orders to three micro-lenders that started in the last 60 days (R1,200, R850, and R1,100 per month), none of which are on bureau yet because monthly submission cycles lag by four to six weeks. There is R1,800 per month going to an online gambling platform. There is a monthly EFT of R900 to an individual, consistent with informal loan repayment, and R2,400 in ATM withdrawals that suggest additional cash-based obligations. Actual discretionary income after these obligations: approximately R2,200. The application that looked comfortable is substantially over-indebted.
This is the practical consequence of payslip-only assessment. The obligation picture is invisible. Behavioural credit scoring captures exactly these signals (the debit order enumeration, the gambling spend, the informal obligation patterns, the cash withdrawal behaviour) from the same bank statement that Reg 23A already requires the lender to obtain.
Why “all monthly obligations” means more than the bureau says
Regulation 23A's obligation enumeration standard is explicit: “all monthly debt repayment obligations of the consumer.” The word “all” is not qualified by “bureau-listed” or “registered.” Informal lending repayments paid as EFTs to individuals or unlicensed lenders represent real monthly obligations. Bureau data, which reflects only registered credit provider submissions on monthly cycles, structurally cannot capture these.
Loan stacking compounds this gap. A consumer who has taken two additional short-term loans in the past 30 days from lenders who submit on monthly cycles appears clean on bureau today. The bank statement shows both disbursement credits and the associated new debit orders. The obligation picture the bureau provides is incomplete by up to a month behind the consumer's actual financial position.
Section 81(3) of the NCA adds the verification responsibility: a credit provider cannot shelter behind a consumer's failure to disclose obligations if the credit provider could have taken reasonably practicable steps to verify. Obtaining and analysing a bank statement that would have shown the undisclosed obligations is precisely the “reasonably practicable step” the section contemplates. A lender who obtained a bank statement and did not systematically review it for obligation signals is in a worse position than one who acknowledged the limitation of bureau-only assessment and addressed it.
The NCT enforcement record on inadequate short-term lender assessment
NCR v Tsoelli (Pty) Ltd t/a Tswelopele Cash Loans [2023] ZANCT 26 is the benchmark enforcement precedent for short-term lenders. Tswelopele was a registered cash loans provider. The NCT sampled credit files and found systematic affordability assessment failures: income not verified, obligations not enumerated, no contemporaneous assessment records. The outcome was cancellation of registration.
The case is not instructive because the failures were unusual. It is instructive because they were common: the same failures that appear in the file-sampling of many small short-term lenders: stated income accepted without payslip or statement, flat expense norms applied regardless of consumer circumstance, no retrievable assessment record for older agreements. Tswelopele is the enforcement record for what happens when these are systemic rather than isolated.
The CDH November 2023 alert identified “disregarding information indicating financial stress” as a reckless lending trigger. For short-term lenders specifically: if the bank statement shows gambling spend, loan stacking, or returned debit orders, and the credit provider proceeds without addressing those signals, the assessment may not satisfy the substantive requirement that 2026 court judgments have now reinforced.
The August 2025 regulatory tightening: expense norms are coming
The draft Regulation 23A amendments published in August 2025 introduce mandatory minimum household expense norms by income band. Under the proposed sub-regulation 23A(10), a credit provider would be prohibited from using an expense figure below the relevant income-band minimum, even if the consumer declares lower actual expenses. The comment deadline was September 2025; promulgation is expected in 2026 with a transition period.
For short-term lenders who have historically used a flat low expense estimate (R1,500 or R2,000 across all applicants), this is a significant methodological change. An applicant on R8,000 per month has a materially different minimum expense profile than one on R25,000. The proposed norms table will require differentiated treatment. The credit provider who implements income-band-calibrated expense norms now, ahead of promulgation, is both more defensible under current enforcement standards and ready for the amended regulation on day one.
This is also where the ABSA v Goolam precedent becomes a near-term operational risk for short-term lenders with flat expense practices. That case found that using incorrect expense norms in an otherwise completed assessment was itself a ground for a reckless credit finding, separate from the question of whether any assessment was done at all.
A practical bank statement checklist for short-term lenders
A compliant, defensible affordability assessment for short-term credit requires six components from the bank statement:
- Income credits: Identify all income deposits: salary, irregular credits, secondary income. Classify by source. Note recurrence across the three months. Flag income that appears in only one or two months as irregular.
- Debit order payees and amounts: List all recurring debit orders by payee name and monthly amount. Separately categorise financial-services payees from utilities, insurance, and subscriptions. Note any debit orders that started in the most recent month, as these may represent recent loan stacking not yet on bureau.
- Cash withdrawal patterns: Total ATM withdrawals by month. Compare to income. Flag months where ATM withdrawals exceed 20% of gross income as potentially indicating cash-based obligations not visible in structured transactions.
- Gambling transactions: Total debits to gambling merchant category codes and known gambling platform descriptions (Betway, Hollywoodbets, Sportingbet, similar). Express as a percentage of gross income. High-frequency or high-value gambling directly reduces available discretionary income and is a behavioural stress indicator.
- Net disposable income calculation: Gross income minus statutory deductions, minus identified debit orders and recurring obligations (including informal payment estimates from cash patterns), minus minimum living expense norm for the income band, minus gambling spend treated as a recurring expense. The residual is the defensible discretionary income figure.
- Collection date recommendation: Where applicable, note the date range when salary deposits typically clear, to inform debit order collection timing. A debit order collection aligned with the consumer's salary receipt materially reduces the risk of a returned debit.
AffyScore provides this structured bank statement checklist as a scored, documented output: income extraction, obligation enumeration, gambling flag, loan stacking signal, discretionary income calculation, and collection date recommendation. The output is a single decision pack per application, generated from one PDF or open banking upload, structured to satisfy Reg 23A evidence requirements. One upload, complete audit trail.
The alternative is 30–45 minutes of manual review per application. At ten applications per day, that is seven hours of analyst time spent on documentation that should be automated. At fifty applications per day, it is not possible at all without cutting corners that produce exactly the assessment failures the NCT enforcement record describes.
