The builders are here. The bank for them is not.
South Africa does not lack entrepreneurs. It lacks a financial system designed to back them. The country's small, medium and micro enterprises are the real economy, they generate close to 40% of GDP and around 60% of employment, and yet the formal credit system treats them as a residual risk to be managed down rather than a future to be financed.
The exclusion is stark. Only about 5% of formal MSMEs in South Africa can access credit1. The National Credit Regulator estimates that close to 780,000 formal firms and two million informal firms sit on the wrong side of the financing gap2. That gap is variously sized at R350-billion and, on the International Finance Corporation's broader measure, around R500-billion (roughly US$30-billion)2. Crucially, the businesses being turned away are the ones that build: more than 85% of firms seeking funding have annual turnover below R1-million, and that same cohort accounts for over 80% of all SMME jobs2.
This matters now because the structure of the economy is changing faster than the structure of its finance. The next generation of enterprise (digital, services-led, green, export-oriented, often asset-light) does not look like the collateral-heavy, fixed-asset borrower the credit system was built around. A firm whose value lives in software, contracts, brand and cash flow is close to invisible to a lender that underwrites against property and plant. The result is a paradox: the most economically important firms are the least bankable, not because they are uncreditworthy, but because the system cannot see them.
A risk machine built to say no
South Africa's banks are not failing at their job; they are succeeding at the wrong one. The sector is among the most sophisticated and well-capitalised in any emerging market, and also among the most concentrated. The four largest banks hold roughly 83% of banking assets, and the five largest account for about 90% of deposits and lending3. Concentration breeds a particular temperament: when a handful of incumbents compete for the same prime, low-risk customers, the rational strategy is to protect the book, not to extend the frontier.
The data shows the retreat. The proportion of bank lending directed to SMEs fell from 28% in 2010 to 12% in 20204. And here is the tell: South Africa's ratio of non-performing loans to SMEs is unusually low by international standards. That is not a badge of prudence; it is a symptom of over-selection. As the Competition Commission has observed, a low SME default rate signals that institutions are "relatively more risk-averse," screening out viable, innovative, higher-growth firms before they ever reach a loan committee5. A bank that never lends to a business that fails is a bank that is not lending to enough businesses that would have succeeded.
The mechanism is the risk model itself. Traditional underwriting demands audited financial statements, tax compliance history, collateral and an existing banking relationship. Yet historically more than half of South African SMEs keep no formal financial records1. The model does not assess these firms badly; it cannot assess them at all, and so defaults to rejection. Backward-looking risk, built on the documented past, is structurally blind to the forward-looking enterprise. This is the binding constraint, and it is philosophical before it is technical.
Rewiring risk: from risk-averse to risk-leaning
The first thing commercial banks must change is not a system but a stance. A risk-averse institution asks, "What could go wrong, and how do I avoid it?" A risk-leaning institution asks, "What could go right, and how do I price for it?" The difference is not recklessness; it is a willingness to underwrite the future on evidence the past did not record. For a country that needs to build scaling enterprises, risk-leaning is not a luxury; it is the job.
The enabling technology already exists. Where audited statements are absent, banks can underwrite on the data the enterprise generates every day: bank-transaction flows, payments and point-of-sale records, supplier and tax data, mobile and platform signals, even behavioural and alternative data. McKinsey's analysis of South African SME lending found that a digital, data-driven credit assessment can pre-screen applicants to achieve more than a 90% approval rate, lift the model's discriminatory power (Gini) toward 70%, and cut both risk costs and operating costs simultaneously1. Better data does not mean looser lending; it means sharper lending: more approvals and better selection at once.
The philosophical shift is to treat data as collateral. A predictable cash-flow stream is, in economic substance, a more reliable security than a depreciating fixed asset that is slow and costly to seize. Once a bank can see the business, it can price the risk; once it can price the risk, it can say yes. The instruments follow the stance: cash-flow-based lending, revenue-based finance, dynamic limits that grow with the firm, and embedded credit at the point of transaction. The constraint was never that these firms are uncreditworthy. It was that the bank was looking in the wrong direction.
But rewiring risk has a limit. Even the sharpest data engine is still a commercial balance sheet, and a commercial balance sheet has a horizon. It must match assets to liabilities, hold capital against risk, and return value to depositors and shareholders on a defined timeline. There is a class of enterprise (early-stage, long-gestation, capital-hungry, nation-building) whose payoff lies beyond that horizon. For those firms, no amount of better data makes a return-maximising bank the right holder of the risk. That is the work of the second engine.
Patient capital and the developmental mandate
South Africa's deepest financing problem is not the price of money or even access to it; it is the tenor and temperament of money. Building a scaling enterprise takes years of patient, risk-absorbing capital that expects to wait. The commercial market, by design, is impatient: it rewards liquidity, quick exits and predictable returns. The "missing middle" (firms too large for microfinance, too small or too unproven for private equity, too long-horizon for a bank) exists precisely in the gap that patient capital is meant to fill11.
This is the proper role of development finance institutions. A DFI is not a slow bank; it is a different instrument with a different objective function. Where a commercial bank optimises for risk-adjusted return on a defined horizon, a DFI optimises for development and impact (jobs, industrialisation, transformation, exports, green transition) over a much longer one. It can therefore hold risks, and wait for returns, that no commercial lender rationally would. That patience is not inefficiency; it is the entire point.
South Africa is unusually well-endowed with these institutions. The Industrial Development Corporation (IDC) funds industrialisation; the Development Bank of Southern Africa (DBSA) funds infrastructure; the National Empowerment Fund (NEF) funds transformation; the Small Enterprise Finance Agency (sefa), an IDC subsidiary, reaches small firms with facilities from R500 to R5-million; and the Land Bank funds agriculture10. Collectively, South Africa's major DFIs have disbursed more capital in Southern Africa than all international DFIs combined10. The raw material of a patient-capital system is already on the ground.
The problem is that the two engines run on separate tracks. DFIs, lending directly, lack the origination reach, distribution and day-to-day customer data of the commercial banks. Banks, lending commercially, lack the patient, risk-absorbing capital and developmental mandate of the DFIs. Each holds half of what the builder needs. The release of patient capital at scale depends on connecting them, which is exactly what the most successful developing economies have done.
How China, Brazil, Germany and India release patient capital
The model that unlocks patient capital is not theoretical. Across very different political economies, the highest-performing systems share one design principle: the development bank supplies and absorbs risk on the capital, while the commercial bank originates, distributes and prices it. The DFI does not compete with the banking system; it leverages it.
Brazil's BNDES is the clearest template. Founded in 1952 and operating with no retail branches of its own, BNDES channels long-term development credit largely through on-lending: accredited commercial, cooperative and investment banks distribute its financing lines to firms across the country. The bank's own research finds that when it works through intermediaries' risk-evaluation practices and networks rather than lending large volumes directly, its credit is associated with higher growth6. Germany's KfW formalises this as the "house-bank principle": KfW never lends directly to the SME but refinances the loan at favourable rates through the borrower's own local bank, which keeps the customer relationship and a share of the risk7. The development capital is patient; the origination stays commercial.
China's CDB demonstrates the sheer scale the model can reach: the policy bank held on the order of US$2.7-trillion in assets and is a defining source of long-horizon development finance8. It is worth being precise here: recent scholarship has challenged the romantic view of Chinese "patient capital," noting that much of it is in fact carefully collateralised and risk-mitigated rather than indefinitely forgiving8. The lesson South Africa should take is not unconditional patience but structured patience: long tenors deliberately paired with disciplined risk management. India's MUDRA, a refinancing agency under the SME-focused SIDBI, shows the reach of pure intermediation: MUDRA lends to no entrepreneur directly, instead refinancing banks, NBFCs and microfinance institutions, channelling cumulative lending of tens of trillions of rupees to hundreds of millions of borrower accounts9.
Four institutions, four political systems, one architecture: patient capital from the development bank, commercial origination from the banking system, connected by on-lending and guarantees. South Africa has both halves. It has simply never built the connector at scale.
| Institution | Country | How it works | Lesson for South Africa |
|---|---|---|---|
| BNDES | Brazil | No branches; on-lends long-term credit through accredited commercial & cooperative banks that handle origination and risk. | Working through banks' networks beats lending directly, and is linked to higher firm growth. |
| KfW | Germany | "House-bank principle": refinances the SME's own local bank at favourable rates; the bank keeps the relationship and skin in the game. | Patient capital can be wholesale; let the commercial bank own the customer and share the risk. |
| CDB | China | Vast policy bank (~US$2.7tn assets) supplying long-horizon development finance at national scale. | Scale is achievable, but pair long tenors with disciplined, structured risk management, not blank-cheque patience. |
| MUDRA / SIDBI | India | Refinancing agency: lends to no entrepreneur directly, refinances banks, NBFCs and MFIs for last-mile micro-credit. | A pure refinance/guarantee layer can extend reach to millions without building a retail footprint. |
An operating model to release patient capital at scale
Put the two engines together and the architecture becomes clear. Development finance institutions supply patient, risk-absorbing capital against a developmental mandate. Commercial banks supply origination, distribution, customer relationships and a data-driven risk engine against a commercial discipline. A blended-finance layer connects them, translating the DFI's appetite for impact and patience into terms a commercial balance sheet can hold. The builder, at the end of the chain, finally meets a financial system designed to back rather than to screen them.
Three connectors do the work. First, wholesale on-lending: DFIs refinance banks' SME books at long tenors and concessional rates, lengthening the patience of money the bank then originates and services: the BNDES and KfW mechanism. Second, risk-sharing guarantees and first-loss capital: a DFI (or concessional donor) absorbs the first tranche of losses, so the bank can lend further up the risk curve than its own capital rules allow: the blended-finance mechanism that crowds private money in12. Third, a shared data and impact layer: the data engine that lets the bank price risk also lets the DFI measure development impact, aligning the two objective functions on the same enterprise.
The capital then arranges itself as a stack. Concessional and first-loss capital from DFIs and donors sits at the bottom, taking the earliest hits. Subordinated, patient DFI capital sits in the middle. Senior commercial bank capital sits on top, protected by the layers beneath it and therefore willing to lend at scale. Each rand of patient capital at the base mobilises several rands of commercial capital above it: the leverage that turns a limited public balance sheet into a large flow of private credit.
What changes for each party
Commercial banks
Gain: de-risked access to a large, underserved market and longer-tenor funding. Give up: the comfort of collateral-only lending. Must build: a data-driven, risk-leaning credit engine and DFI partnership capability.
DFIs
Gain: origination reach and leverage, each rand mobilises several of private capital. Give up: direct-lending turf and control. Must build: wholesale, guarantee and first-loss products designed to crowd banks in, not out.
Entrepreneurs
Gain: patient, appropriately priced capital that backs the future, not just the balance sheet. Give up: little. Must build: the data trail (transactions, payments, records) that makes them visible to a risk-leaning lender.
Banking the next economy
South Africa's commercial banking is indeed due a refresh, but the refresh is not cosmetic. It is a change of stance and a change of architecture. The stance is to move from risk-averse to risk-leaning: to treat data as collateral, to underwrite the future on the evidence the present generates, and to price risk rather than avoid it. The architecture is to stop running development finance and commercial finance on separate tracks, and to wire them together so that patient capital flows through the banking system to the enterprises that will build the next economy.
Neither engine is sufficient alone. A risk-leaning bank without patient capital will still hit the horizon of its own balance sheet. A patient DFI without commercial origination will still fail to reach the firms that need it. Run together (connected by on-lending, guarantees and a shared data-and-impact layer) they become a system that does what neither can do apart: release patient capital into the missing middle, at scale, on terms a builder can use. The countries that have built scaling enterprises did not choose between their banks and their development institutions. They made them partners.