TechCabal reported on 18 September 2026 that South Africa's estimated R350 billion ($21.5 billion) funding gap for micro, small and medium enterprises is "really a data problem". The framing deserves attention because the supply side looks healthier than the headline suggests. The same report, citing the OECD, says the number of small business funders has doubled from 148 in 2018 to more than 300 in 2025. Money is arriving. What lenders lack is a reliable way to tell one borrower from another.
What the numbers say
The OECD's Financing SMEs and Entrepreneurs 2026: An OECD Scoreboard, its 15th edition, is the source for the South African figures TechCabal cites. Per TechCabal's reading, 56% of South Africa's MSMEs are unregistered, only 7% used a formal financial-services provider's business loan to start up, and only about half have internet access. The same article reports that 85.6% of finance applicants have annual turnover below R1 million, and that 50.9% of owners in that band have poor or below-average credit scores.
The TechCabal piece attributes the R1 million and credit-score figures to a separate SA MSME Access to Finance Report, not to the OECD. The R350 billion gap figure also arrives without a named source. Treat that headline number as an estimate, not a measurement.
The pattern is familiar. A micro-enterprise with no registration, no formal loan history and thin bureau data is invisible to a scoring model. The lender is not being lazy. It is pricing an unknown.
The case for caution
The strongest argument against loosening data rules is that alternative-data lending can go wrong. Scoring people on payment patterns, purchase orders and invoices means collecting a lot of information about people who often cannot negotiate consent terms. Small traders could be profiled, mis-scored or locked into one platform. South Africa's Protection of Personal Information Act (POPIA) exists to prevent exactly that, and a regulator that insists on it is not obstructing innovation.
That concern is legitimate. It argues for well-designed data sharing, not for the current status quo, in which the cost of uncertainty falls on the smallest borrowers.
Where the public infrastructure gap sits
The hook is right to call this a digital public infrastructure question. Three layers matter, and South Africa is uneven across them.
Registration. Registration is cheap. The Companies and Intellectual Property Commission (CIPC) lists a private company registration fee of R125 and a name reservation at R50, and its BizPortal bundles company registration with SARS and UIF registration. Yet 56% of MSMEs remain unregistered. If fees were the barrier, that figure would look different. The likelier barriers are the perceived compliance burden that follows registration and the lack of any visible benefit. Nobody registers to be taxed if registration does not unlock credit. That is a design problem for the state: registration has to pay for itself in access to finance.
Payment rails. This layer is where the country has done best. According to ClearingPost's analysis of BankservAfrica data, the PayShap instant-payment service reached 905 million cumulative transactions by May 2026, up from 461 million at the end of December 2025, with roughly 89 million transactions a month in the latest five-month period. Every one of those payments is a data point a lender could use with consent. The same analysis notes that BankservAfrica has acknowledged fees and inconsistent user experience across banking apps as constraints on adoption. A rail that is expensive or clumsy leaves informal traders in cash, which produces no data at all.
Data-sharing. This layer is the weakest link. Payment histories, invoices and purchase orders sit in separate silos: banks, accounting software, payroll and tax systems. TechCabal's lenders want verifiable alternative data such as invoice discounting evidence, purchase-order financing records and digital payment patterns. Without a consent-based way to move that data between institutions, each lender must rebuild the picture from scratch, and small borrowers pay for that duplication in higher rates or outright refusal.
The internet-access gap cuts across all three layers. If only half of MSMEs are online, a purely digital data pipeline reaches only half of them. Policy should treat connectivity as a prerequisite, not a footnote. Meanwhile, close to 80% of MSMEs already use digital financial services, according to the TechCabal report. That suggests mobile and app-based payments are further along than web presence, and that lenders should build on payment data rather than waiting for websites.
A proportionate approach
Three principles follow, and none of them requires a new heavy-handed regime.
- Make consent portable. Let a business permit its bank or payment provider to share transaction history with a chosen lender, revocable at any time. POPIA already supplies the rights framework. What is missing is the technical standard and the incentive for incumbents to comply.
- Tie registration to something. Link registration in BizPortal to verifiable status that lenders can check, so that formalising has a payoff beyond compliance.
- Keep rail fees low. Fee levels on instant payments determine whether informal trade moves into systems that generate creditworthy data.
What policymakers should avoid is a blanket mandate that treats every small lender like a systemic bank. Heavy compliance requirements would push lenders back toward collateral-based lending, which the OECD's scoreboard notes remains prevalent across its sample, and would deepen the exclusion the data problem already creates.
South Africa does not have a capital shortage in small-business lending. It has a visibility shortage. Registration, payments and consent-based data sharing are cheaper to fix than a R350 billion gap, and every improvement compounds. The measure of success is simple: fewer borrowers who look identical to lenders because the system cannot see them.