Why Central Africa’s banks do not finance Central Africa — and what architecture could make them
I. Anatomy of a rational bank — II. Bankability is not a quality of the project; it is infrastructure — III. Who actually finances young entrepreneurs? — IV. Why the region does not take off: four explanations, three of them insufficient — V. What works elsewhere, and why — VI. A four-tier architecture for Central Africa — VII. What could make this model fail — Conclusion — Statistical annex — Sources
Scope and method. This article covers Central Africa as defined by ECCAS. Readers should know from the outset that this perimeter is politically unstable and statistically heterogeneous: Rwanda announced its withdrawal from the organisation in June 2025, effectively reducing the community to ten members. The six CEMAC countries — Cameroon, Congo, Gabon, Equatorial Guinea, the Central African Republic and Chad — form the only bloc where banking data are produced to a common standard, by BEAC and COBAC, and are therefore genuinely comparable. The DRC, Angola, Burundi, São Tomé and Príncipe and Rwanda operate under distinct monetary regimes; they are treated here as contrast cases, with the caution that requires. Any aggregation of “Central Africa” that claims homogeneity is a writing convenience, not a fact. Unless otherwise stated, amounts are in Central African CFA francs (XAF); EUR 1 = XAF 655.957 under the peg, and dollar equivalents are indicative, converted at roughly XAF 600–610 per US dollar at the relevant reference dates.
Two numbers that should not coexist
At end-December 2024, net treasury balances across CEMAC’s banking system stood at a surplus of XAF 8,245 billion — some USD 13.6 billion, or roughly one third of the sector’s total balance sheet. In the first quarter of 2026, small and medium-sized enterprises across the same zone received XAF 565.9 billion (about USD 930 million) in new lending: 22.5% of all credit disbursed, in economies where SMEs make up the overwhelming majority of the productive base.
These two figures describe the same financial system. They should not coexist. In a functioning credit market, a liquidity surplus of that magnitude flows mechanically into lending to the economy, because idle money costs money. In Central Africa it does not flow. It accumulates, and then it leaves — but towards a single borrower.
That is where serious analysis begins, and it is usually where public debate in the region stops. The dominant narrative reduces to a sentence: banks are timid, they do not understand entrepreneurship, they prefer real estate and import trade, they have abandoned the young. The narrative has one merit — it describes the outcome accurately — and one disqualifying flaw: it confuses a behaviour with its causes, and attributes to bankers’ psychology what is in fact a perfectly legible structure of incentives.
The argument advanced here is different, and less comfortable. Central Africa’s banks behave rationally within the system of relative prices imposed on them. They do not lend to entrepreneurs because an elementary calculation — yield, capital consumption, probability of recovery, exit horizon — tells them to do something else, and because that calculation is not close. As long as the calculation stands, no moral exhortation, no charter of commitment and no regional summit on youth entrepreneurship will change how credit is allocated. Conversely, changing three prices — the price of risk, the price of information and the price of time — would move several trillion CFA francs towards the productive sector without requiring a single banker to become virtuous.
The article proceeds in six steps. It first establishes the anatomy of bank behaviour from BEAC and COBAC data. It then shows that “bankability” in Central Africa is a defective legal construct far more than a question of entrepreneurial quality — with direct consequences for what “support to project promoters” ought to mean. It takes stock, without indulgence, of what actually exists for young entrepreneurs. It examines four competing explanations for the region’s failure to take off, three of which are insufficient. It analyses what works elsewhere and why. Finally, it proposes a four-tier architecture, costed, and sets out the conditions under which it would fail.
I. Anatomy of a rational bank
1.1 The great switch: 2017-2026
Between 2017 and 2024, CEMAC’s banking sector changed business. The single most important figure in this study is this one: according to COBAC’s 2024 annual report, the share of gross bank lending absorbed by governments rose from 24% in 2017 to 61% in 2024. Symmetrically, the share of bank resources channelled to firms and households fell from 74% to 48%.
The dynamics are striking. Financing extended to governments went from XAF 509 billion in 2011 to XAF 5,752 billion in 2024 — an elevenfold increase in thirteen years. Banks’ gross exposure to sovereign treasuries reached XAF 7,642 billion, equal to 405% of their net own funds, against a COBAC prudential ceiling of 25%. The IMF measures the same trajectory differently: CEMAC banks’ cumulative sovereign exposure rose from about 10% of assets in 2015 to nearly 31% by end-2023, with some institutions above 50%.
The regional government securities market, launched by BEAC in November 2011, passed XAF 10,000 billion in outstandings in May 2026 — against a little over XAF 1,000 billion eight years earlier. Commercial banks hold 76.8% of it.
The phenomenon deserves its proper name. This is not a preference for low risk: it is a substitution of clientele. A banking sector that devotes 61% of its lending to a single borrower is no longer a financial intermediary in the classical sense; it is a vehicle for warehousing public debt with a residual retail business attached.
One honest correction is required. The trend has begun to reverse. At end-March 2026, the monetary system’s net claims on governments stood at XAF 11,397.6 billion against XAF 13,655.4 billion of credit to the economy — a ratio of 83.5%, down 4.7 points year on year, with credit to the economy growing 12.5% against 6.5% for claims on governments. In Gabon, the private sector captured 78.4% of bank lending at end-2025, with financing to the state falling 35% in the fourth quarter alone. The switch is therefore not irreversible. But it would take several years at that pace to return to the 2017 structure, and nothing guarantees that the fiscal easing behind it will last.
1.2 The mechanics of crowding out
Why does a credit committee prefer the state? The answer is not a conviction but four parameters, three of which are regulatory or institutional.
Yield. In May 2026, the average cost of borrowing for governments on BEAC’s securities market stood at 9.73%, up from 8.08% in April. Over the same period, the average lending rate applied to SMEs in the zone was 11.00% in the first quarter of 2026. The gross yield pick-up from lending to an SME rather than to a treasury is therefore of the order of 130 basis points — a spread that plainly does not compensate the differential in credit risk, origination cost and recovery cost.
Capital consumption. Domestic sovereign paper carries a 0% risk weight in solvency calculations. An SME loan is weighted at 100%, sometimes more. At constant own funds, a bank can therefore carry a volume of government debt far larger than the volume of private credit it could fund. For a sector in which 22 of 56 institutions were in breach of capital requirements at end-2024, that capital saving is not a detail: it is the binding constraint.
Exit liquidity. A government security is repo-eligible at BEAC and tradable in a secondary market. An SME loan is illiquid until maturity.
Certainty of enforcement. This is the most neglected point. A domestic sovereign default within the CFA franc zone is politically costly and rare; litigation against an SME means a judicial process of uncertain duration and outcome, even under the 2010 OHADA Uniform Act on security interests.
Add the four together and you get not timidity but arithmetic. The Central African or Chadian banker who arbitrages in favour of the Treasury is doing precisely what a London counterpart would do facing the same matrix of returns. The problem is not the banker: it is the matrix.
1.3 What excess liquidity is not
The phrase “excess liquidity” saturates regional debate and misleads it. Four clarifications are in order.
First, this liquidity is not idle: it is already invested. Of the XAF 8,245 billion treasury surplus recorded by BEAC at end-2024 — up XAF 765 billion year on year — 48.8% was held in investment and trading securities, 32% in sight operations and 6.8% in term operations. In other words, what public debate calls “money sleeping in the vaults” is, for nearly half of it, government debt already subscribed. Excess liquidity and sovereign crowding out are not two separate problems: they are two descriptions of the same accounting fact.
Second, liquidity is not capital. A bank can be simultaneously over-liquid and under-capitalised — precisely CEMAC’s situation, where 22 of the 56 institutions recorded by COBAC were in breach on own funds at end-2024 (institution counts range from 53 to 56 depending on source and date, as the perimeters of “banks” and “credit institutions” do not always coincide), and where Chad alone accounted for five of the region’s ten loss-making banks, with a combined net capital shortfall of XAF 247.3 billion (about USD 410 million). A bank without capital cannot lend more, however full its vaults. COBAC Regulation R-2025/02 of 10 December 2025, raising minimum share capital from XAF 10 billion to XAF 25 billion (roughly USD 41 million) for banks and from XAF 1 billion to XAF 4 billion for financial institutions, effective 1 January 2026, acknowledges this reality. An August 2025 study found that 77.4% of the zone’s 53 banks then sat between XAF 10 billion and XAF 20 billion of capital: most of the sector is affected. Recovery plans were due by 30 June 2026, with full compliance targeted for 2029.
Third, the funding base is short. In the third quarter of 2024, loans with maturities of 24 months or less accounted for 83.6% of new lending; medium-term loans (25-60 months) 14.5%, and long-term loans (over 60 months) 1.87%. The cause is structural: deposits are predominantly sight deposits. A bank that transforms demand deposits into seven-year loans takes a liquidity risk that neither its board nor its supervisor will accept. Put differently: even if the entire liquidity surplus were redirected to the economy, it would fund working capital and consumption, not factories. Central Africa’s deficit is not a credit deficit: it is a long-term credit deficit.
Fourth, the cost of risk is real. CEMAC’s non-performing loan ratio stood at 17.4% of gross outstandings at end-March 2025, against 16.6% a year earlier; it was 16.2% at end-2024 against 16.0% in 2023. In 2024, NPLs rose 31.4% in Gabon and 14.5% in Cameroon. A portfolio carrying 17% NPLs imposes defensive pricing and severe selection. And one point that public debate systematically omits: a significant share of those arrears is public in origin. Government payment delays to suppliers asphyxiate otherwise solvent SMEs, which then default to their banks. The state crowds out the private sector from above, by absorbing credit, and from below, by not paying its invoices.
II. Bankability is not a quality of the project: it is infrastructure
2.1 The four missing pillars
The question “how do we make projects bankable?” is badly posed, and the standard answer — “train the project promoters” — addresses the least binding constraint.
A bank does not finance a project. It finances a future cash flow, whose plausibility it verifies, and which it secures against a legal structure enforceable in default. That operation rests on four pillars. In Central Africa, all four are defective.
Pillar 1 — Financial information. The borrower must produce accounts whose accuracy the bank can verify. In a region where informality dominates, that verification is impossible or prohibitively expensive. This is not entrepreneurial laziness: informality is a rational response to a tax burden perceived as arbitrary, in economies where CEMAC’s measured tax take reached only 9.6% of GDP in 2023 against 14.0% in WAEMU. Formalising often means becoming visible to the tax authority before becoming financeable to the bank — an unattractive trade in a company’s early years.
Pillar 2 — Credit history. Until very recently, no credit bureau existed in Central Africa. Creditinfo Central Africa (CICA), licensed by BEAC in December 2025 and formally launched on 20 January 2026 in Douala with IFC support, is the first such infrastructure in the zone. It is arguably the most important financial reform of the region’s decade, and it passed largely unnoticed. Without credit history, every application is treated as a first application: the loyal client and the unknown one face the same grid.
Pillar 3 — Enforceable security. The 2010 OHADA Uniform Act modernised the law of security interests. The problem is no longer the statute: it is the time and cost of enforcing it. A mortgage that cannot be executed in practice is not collateral, it is a document. Banks know this and over-collateralise accordingly, which mechanically excludes any entrepreneur without pre-existing assets — that is, by construction, every young graduate.
Pillar 4 — A solvent downstream market. A project is bankable only if customers can pay. In economies where 31.1% of CEMAC’s 63 million inhabitants lived on less than USD 2.15 a day in 2023 (65.9% in the Central African Republic), and where the principal contracting party — the state — pays late, commercial risk is as decisive as credit risk.
The conclusion is direct: bankability is not a property of the borrower, it is a property of the institutional environment. It cannot be produced through individual training. It is produced through registries, databases, enforcement procedures and risk-sharing mechanisms.
2.2 The pricing paradox: what the data say against intuition
Here is the most counter-intuitive finding of this study, and it deserves to be confronted because it contradicts the dominant narrative.
In the first quarter of 2026, according to BEAC’s report on lending rates, average interest rates applied across CEMAC broke down as follows:
| Borrower category | Average rate (Q1 2026) | Change year on year |
|---|---|---|
| SMEs | 11.00% | +26 bps |
| Other legal entities | 11.70% | +503 bps |
| Large corporates | 11.82% | +258 bps |
| Public administrations | 13.77% | +418 bps |
| Individuals | 17.18% | +341 bps |
Source: BEAC, report on lending rates, Q1 2026.
SMEs enjoy the lowest nominal rate of any borrower category. That is the exact inverse of 2018, when SMEs borrowed at 12.5% against 8.4% for large corporates and 6.9% for public administrations.
Should we conclude that banks now favour SMEs? No — and that is precisely why the figure matters. Three readings apply.
First reading: massive selection bias. The observed rate covers accepted files only. If banks retain only the best-collateralised and most solvent decile of SMEs — typically subsidiaries of larger groups, firms backed by a supply contract with a major counterparty, or companies pledging cash deposits — the measured average rate collapses mechanically. A low rate on a rationed volume is not a signal of generosity: it is a signal of exclusion. The XAF 565.9 billion extended to SMEs represented 22.5% of the total against 58.6% for large corporates: that is the information.
Second reading: the nominal rate hides the real cost. Again per BEAC, fees and charges accounted for 31.9% of the all-in effective rate borne by SMEs, which rank among the categories carrying above-average charges. The headline 11% is not the price paid. The zone’s average effective rate rose from 11.82% in Q4 2025 to 12.36% in Q1 2026, even as BEAC cut its policy rate to 4.50% on 29 June 2026. Monetary transmission to lending is weak — another symptom of the same condition.
Third reading: price is not the adjustment variable. In credit economics, when rationing is quantitative rather than tariff-based, the problem is informational, not a matter of price. This is the Stiglitz-Weiss mechanism: beyond a certain rate, raising the price of credit degrades the average quality of the borrower pool through adverse selection, so the lender prefers to refuse rather than to price. The practical implication for Central Africa is blunt: interest rate subsidy schemes are largely inoperative, because the rate is not what is binding. Tens of billions of CFA francs of public subsidy have been spent in the region lowering rates that were not the problem.
2.3 Support programmes are addressing the wrong constraint
The consequences for the “support to project promoters” industry — ubiquitous in regional public discourse — need to be drawn.
That industry overwhelmingly supplies business-plan training, entrepreneurial coaching and project competitions. These services are not useless, but they attack Pillar 1 (information) on its least binding side, and ignore Pillars 2, 3 and 4 entirely. An impeccable business plan creates neither credit history, nor enforceable security, nor a paying customer.
Support that would measurably move access to credit would do something else:
- Produce enforceable data. Taking a micro-enterprise to kept accounts, a separate operating bank account and twelve months of transaction history — that is, building the file the bank cannot build itself.
- Structure demand, not supply. Anchoring an entrepreneur to a framework contract with a solvent buyer converts a speculative project into a discountable receivable. This is the principle of value-chain finance, and it is the only method that makes an asset-less entrepreneur bankable.
- Carry the guarantee. A support body that brings no credit enhancement brings nothing the bank values.
- Take the first loss. As long as no party accepts the first tranche of loss, the risk stays entirely on the bank’s balance sheet, and the decision stays the same.
The test to apply to every support scheme in the region is this: does it change the credit committee’s decision matrix? If the answer is no, the scheme produces social satisfaction, not financing.
III. Who actually finances young entrepreneurs? A stocktake without indulgence
3.1 What exists, and what it weighs
The region is not devoid of schemes. It is poor in orders of magnitude. Here are the best-documented cases, with their real weight.
Mutuelles Communautaires de Croissance (MC²) — Cameroon, sponsored by Afriland First Bank. Designed by Paul Fokam, the model has the village community provide the capital and run the micro-bank, with Afriland supplying technical supervision and refinancing. The fullest public data, which date from 2015 and would merit updating, indicate roughly 105 units across rural areas and secondary towns, some XAF 145 billion (about USD 240 million) injected into the Cameroonian economy over two decades, claimed repayment rates around 85% and interest rates capped near 15%. This is to date the most original institutional innovation the region has produced: it solves the information problem through social proximity rather than collateral. Its limit is scale — XAF 145 billion cumulative over twenty years against XAF 13,655 billion of credit outstanding to the CEMAC economy at end-March 2026.
La Régionale Bank — Cameroon. A savings-and-credit institution founded in 1993, converted to a universal bank in 2022, positioned on a hybrid agricultural / digital banking model. IFC extended XAF 3 billion (about USD 5 million) in June 2023, with at least 25% earmarked for women and women-led businesses, alongside advisory services on risk management. Structurally this is the most interesting trajectory in the region: a microfinance institution moving upmarket retains the client knowledge universal banks never acquired.
Ecobank Ellevate — Cameroon. A programme dedicated to women’s entrepreneurship, launched in 2021 and upgraded in March 2025 (“Ellevate 2.0”): over USD 2 million (XAF 1.2 billion) disbursed to 1,600 women entrepreneurs in a single year, more than XAF 3 billion cumulatively since 2021, with unsecured loans up to USD 50,000 plus training and networking. The design is sound; its annual volume represents less than 0.01% of the zone’s loan book.
Rawbank — DRC. In March 2026, the DRC’s largest bank closed a USD 265 million package led by IFC: USD 165 million in senior debt (IFC USD 50 million, Proparco USD 50 million, British International Investment USD 25 million, the OPEC Fund USD 20 million, eco.business Fund USD 20 million) and a USD 100 million risk-sharing facility with IFC covering 50%, targeting at least 1,500 additional SMEs over four years. This is the most significant transaction in the region because it combines the two ingredients missing everywhere else: long-dated funding and risk sharing.
FIGA — Republic of Congo. The Fonds d’Impulsion, de Garantie et d’Accompagnement reported for 2025: 9,924 promoters supported under its start-up window, 1,634 project holders mentored and 2,281 beneficiaries of guarantees, for just over XAF 1 billion mobilised in guarantees. Human reach is real; financial coverage is negligible. One billion CFA francs — about USD 1.65 million — of guarantees in an economy whose bank credit is counted in trillions is a scale error of two to three orders of magnitude.
BC-PME — Cameroon. The Banque Camerounaise des PME reported in June 2026 more than XAF 100 billion (about USD 165 million) of loans extended over roughly a decade — some XAF 10 billion a year, in a country whose bank loan book exceeded XAF 5,600 billion by mid-2024.
Public one-stop schemes. In Cameroon, FONIJ/FOGAJEUNE targets 18- to 35-year-olds with tickets of XAF 0.5-5 million (roughly USD 800-8,000), often conditional on three promoters grouping together; the FNE offers financing without bank collateral. These amounts belong to subsistence self-employment, not to the creation of firms capable of hiring.
3.2 The impossible ranking — and the metrics that should be demanded
The question “which are the best banks for young entrepreneurs in Central Africa?” invites an uncomfortable answer: no rigorous ranking is currently possible, because the data that would support one are not published.
No bank in the region discloses the share of its loan book held by firms under three years old, its approval rate by segment, its median ticket size, the share of disbursements made without real security, or its median time to decision. Existing communications concern programmes — that is, discrete envelopes, usually funded by an external development partner — rather than the institution’s ordinary credit policy.
That vacuum is not neutral. It allows a bank to build a reputation for supporting young people on the back of a XAF 2 billion programme while devoting the bulk of its balance sheet to government paper. Regional business journalism should stop relaying envelopes and start demanding the five metrics below, which are the only comparable ones:
- Share of the corporate loan book held by firms less than three years old.
- SME loan approval rate (applications accepted / applications filed).
- Median SME ticket size — the median, not the mean, which a handful of large files distorts.
- Share of disbursements made without real security or with third-party partial guarantee.
- Median time from complete file to decision.
On the basis of publicly available information, three positionings can nonetheless be distinguished:
- Banks that grew out of microfinance (La Régionale, CCA Bank in Cameroon) hold the strongest informational advantage, because they built their client base before becoming banks.
- Banks backed by a multilateral partner (Rawbank and Equity BCDC in the DRC via IFC; La Régionale via IFC) have the risk sharing the others lack. Their SME performance rests largely on that backing — which is itself the demonstration of this article’s thesis.
- Large pan-African and local groups (Ecobank, BGFIBank, Afriland, UBA, Société Générale) run targeted programmes of variable quality whose scale remains marginal relative to their balance sheets.
3.3 A casting error: credit versus equity
One final point, rarely stated plainly: a substantial share of young graduates’ frustration with banks stems from a demand addressed to the wrong window.
A young engineer who wants to launch a technology company is not looking for a loan, even if that is the word used. They are looking for capital: a resource that carries no monthly repayment, that accepts loss, and that is remunerated on future value. A commercial bank structurally cannot supply that: it lends money belonging to its depositors, repayable on demand. Reproaching it for not financing innovation is like reproaching an insurer for not running a venture fund.
The missing link in Central Africa is therefore not solely banking: it is the absence of a private equity and quasi-equity industry. The Bourse des Valeurs Mobilières de l’Afrique Centrale (BVMAC) illustrates the scale of the gap: market capitalisation stood at XAF 479 billion across six listed companies before BGFI Holding’s listing in May 2026 lifted it to XAF 1,658 billion (about USD 2.7 billion) across seven — from roughly 2.4% to more than 5% of regional GDP. The event is positive, but it exposes the fragility: a single listing tripled the market. BVMAC launched an incubator (BVMAC ESPro) in February 2026, operational since 1 June 2026, to prepare companies for listings and bond issues. The direction is right; the scale bears no relation to the need.
IV. Why the region does not take off: four explanations, three of them insufficient
4.1 The resource-rent explanation: correct but incomplete
The most widely held explanation is also the most empirically solid. Natural resources accounted for 46% of CEMAC’s government revenue in 2023 — up to 88% in Equatorial Guinea. A rentier economy produces three well-documented effects: it overvalues the currency and penalises non-oil exports; it makes non-resource taxation politically optional (9.6% of GDP in CEMAC against 14.0% in WAEMU); and it channels the best talent towards rent capture rather than production.
The link to credit is direct. A bank in a rentier economy has no need to build industrial risk-analysis capability: its profitable clientele consists of a handful of large extractive firms, their subcontractors, importers and the state. The skill of appraising a manufacturing SME is acquired only if the market pays for it. Here, it never has.
But the explanation is incomplete. Rwanda, with no oil, recorded private-sector credit at 21.3% of GDP in 2024. Equatorial Guinea, with oil, fell to 5.8% in 2023 from 15.2% in 2019. Rent aggravates; it does not determine.
4.2 The blame-the-banks explanation: partial and lazy
The popular explanation — banks are at fault — is the analytically weakest, for three reasons.
First, banks do not create the conditions of bankability: they observe them. Second, they respond to a system of relative prices they did not set. Third, and above all, intra-regional comparison invalidates the cultural explanation. Gabon lifted the private sector’s share of bank lending to 78.4% at end-2025, with the same banks, the same regulation and the same currency as Chad. What changed in Gabon was the state’s access to alternative funding and the reduction in its draw on bank credit — not a moral conversion among bankers.
One qualification must nonetheless be entered against this defence. Regional banks practise opaque fee pricing — 31.9% of the effective cost borne by SMEs — and chronically under-invest in SME credit analysis, which costs them future clientele. They publish no indicator allowing their contribution to be assessed. And under-investment in origination capacity is a management choice, not a regulatory fate. On that count, the criticism holds.
4.3 The governance explanation: true but hard to act on
Institutional quality, predictability of rules and judicial efficiency are obviously decisive. The analytical difficulty is that this explanation explains everything and prescribes nothing executable in the short run. It also has an empirical weakness: countries of comparable governance quality show markedly different banking penetration and intermediation ratios, which suggests that finer variables — credit infrastructure, market structure, payments regulation — carry substantial weight at given institutional quality.
4.4 The underrated explanation: a thin solvent domestic market and weak integration
This is the explanation I consider most neglected and most decisive.
A firm is financeable only if it has a market. Yet the domestic market of Central Africa, taken country by country, is extremely thin: 63 million people across all of CEMAC, GDP of USD 111.8 billion in 2023, per capita income of USD 3,466 heavily skewed by Equatorial Guinea and Gabon, and 31.1% of the population below USD 2.15 a day. Chad, the CAR and São Tomé do not constitute viable markets for manufacturing at efficient scale.
CEMAC has existed since 1994 and ECCAS since 1983. Intra-regional trade remains among the lowest in the world. Free movement of people and goods is still obstructed by non-tariff barriers, and Rwanda’s announced withdrawal from ECCAS in June 2025 illustrates the bloc’s political fragility.
The link to credit is mechanical: market scale determines project scale, which determines financing scale. A bank will not fund a plant whose break-even assumes 30 million consumers if the firm can reach only five million. As long as regional integration remains theoretical, the stock of genuinely bankable projects will stay structurally limited — and one can reform the banking sector indefinitely without much changing.
This is also what makes the DRC strategically central, and why it is treated separately here. With over 100 million people, a bank loan book of USD 11.355 billion at end-June 2026 (against USD 10.2 billion at end-2025) and roughly 25 million active mobile money users in 2025, the DRC is the only Central African market of critical size. Its handicaps are considerable — dollarisation of around 80% of deposits and loans, extreme geographic concentration (Kinshasa 48% and Haut-Katanga 30% of outstandings, 78% between them), banking penetration of about 27% including mobile. But that is where the question of scale is decided.
4.5 The graduate exodus: what the data actually show
On this point, one rhetorical shortcut must be resisted: the idea that young intellectuals leave because banks refuse to finance their projects. The available data do not support that causal chain as stated.
Afrobarometer’s Round 10 survey (2024) in Cameroon gives the following for 18- to 35-year-olds:
| Indicator | Value |
|---|---|
| Unemployed and actively seeking work | 36% |
| Have considered emigrating | 61% |
| Have at least some secondary education | 80% |
| Believe the country is going in the wrong direction | 77% |
| Rate economic conditions as bad | 65% |
| Approve of government performance on job creation | 19% |
Source: Afrobarometer, dispatch AD1222, Round 10 survey (2024).
Asked about barriers to employment, young people cite inadequate training (23%), a mismatch between skills and available jobs (21%), lack of experience (19%) and weak entrepreneurial skills (18%). Access to finance does not come first. The dominant emigration motive is the search for better job opportunities, not access to credit.
A precise conclusion follows, rather than a slogan. Young people do not leave because a loan application was rejected. They leave because there are no firms capable of employing them at the level of their training. A graduate who emigrates is not fleeing a banker; they are fleeing the absence of a skilled labour market. According to the World Bank, one young person in four across CEMAC is not in employment, education or training, and unemployment reaches 20% in Gabon and Congo; industry employs just 15% of the workforce against 47% in agriculture.
But the loop closes, and this is where credit becomes central again. The firms capable of employing graduates are precisely those that need long-term capital: processing industry, agribusiness, technology-intensive services, health, energy. These are exactly the segments that 1.87% of lending beyond five years cannot finance. Credit is not the direct cause of the graduate exodus; it is the cause of the absence of the firms that would retain them. The distinction is not rhetorical: it changes the policy entirely. Multiplying microcredit windows for young project holders retains no one. Financing the emergence of mid-sized firms does.
One corollary is worth noting, because it opens a funding avenue. The Cameroonian diaspora remits roughly XAF 652 billion a year (about USD 1.1 billion), and the authorities aim to mobilise XAF 2,000 billion over ten years from six million nationals abroad, targeting 500,000 contributors. For reference, Cameroonian national savings are estimated at between XAF 9,900 billion and XAF 20,000 billion over ten years, of which XAF 7,400 billion sits in bank deposits and XAF 2,400 billion in informal circuits. The diaspora is not only a loss: it is a pool of long-term savings — precisely the resource the banking system lacks. It still needs a credible instrument, which presupposes a quality of signature and governance that few states in the region can currently present.
V. What works elsewhere, and why
Four models merit examination. A fifth serves as a warning.
5.1 KfW (Germany): never replace the banker
KfW is the global benchmark for public development banking. Its balance sheet reached EUR 503 billion in 2015, with EUR 79.3 billion in new commitments, of which EUR 20.4 billion for SME promotion. Two features explain its effectiveness.
An implicit federal guarantee gives it AAA status and lets it borrow more cheaply than any commercial bank. It thereby converts sovereign credit quality into cheap long-term funding for the economy — instead of using it to crowd out private credit, as Central African states do.
The house-bank principle (Hausbankprinzip) matters most: KfW does not lend to the end client. It refinances commercial banks, which appraise the risk, distribute the loan and retain it on their books. KfW has no branch network and does not want one. This is not an operational convenience: it is the anti-clientelism mechanism that every African development bank of the 1970s and 1980s lacked.
5.2 KODIT (South Korea): leverage through mutualised guarantee
The Korea Credit Guarantee Fund offers the most instructive guarantee model for Central Africa. Its funding structure is remarkable: roughly 50% of its resources come from a mandatory contribution by banks, assessed on their loan books, 45% from government and 5% from other sources. Banks therefore fund the very mechanism that reduces their risk.
The orders of magnitude: capital of USD 4.7 billion supporting USD 42 billion of guarantees outstanding in 2015 — leverage of about nine times — covering 205,361 firms, 75% of them SMEs. Guarantees outstanding rose from 1.5% of Korean GDP in 2001 to 4.1% in 2013. The effect on credit allocation is striking: the SME share of Korean bank lending rose from 35.7% in 1975 to 76.7% in 2015.
5.3 Development Bank of Nigeria: the wholesale model transposed to Africa
DBN, operational since 2017, is the most compelling African test of the KfW model. It does not lend directly: it refinances participating financial institutions (commercial banks, microfinance banks, merchant banks) and provides them with partial guarantees.
Results as of mid-2026: more than NGN 1 trillion disbursed cumulatively, of which NGN 358 billion in 2025 alone; NGN 512 billion of guarantees issued since 2019, including NGN 233 billion in 2025; over one million end-beneficiary loans, 289,000 of them in 2025; 77% women beneficiaries cumulatively (81% in 2025) and 28% youth; NGN 108 billion channelled to 132,000 MSMEs in the historically underserved northern states; 1.6 million direct and indirect jobs claimed.
The central lesson: a wholesale institution with no retail window can reach a million end borrowers in under a decade, precisely because it does not try to meet them.
5.4 African Guarantee Fund and Equity Bank: private leverage
The agreement signed in May 2025 between the African Guarantee Fund and Equity Bank covers a USD 500 million framework intended to unlock USD 1 billion of new financing in Kenya, Uganda, Rwanda, Tanzania and the DRC, with a first tranche of USD 115 million, explicit targeting of women- and youth-led businesses, and an objective of more than 50,000 jobs. Equity Bank’s chief executive puts the amplification of lending capacity from risk sharing at up to ten times.
Two lessons. First, the mechanism is available in Central Africa: the DRC is included. Second, it rests on a bank that built SME capability before obtaining the guarantee — the guarantee amplifies existing capacity, it does not create it.
5.5 MUDRA (India): the warning
India’s Pradhan Mantri Mudra Yojana distributes unsecured microloans to first-time entrepreneurs. Its outreach results are enormous. So is its cost: the non-performing ratio against outstandings rose from 5.47% in March 2018 to 9.81% in March 2025 for scheduled commercial banks, against 3.60% for India’s MSME sector as a whole.
The lesson is not that unsecured lending should be abandoned. It is that its cost must be budgeted ex ante, explicitly, and must not be parked on bank balance sheets. A youth credit programme without collateral needs a public provision matched to an expected loss rate of 8% to 12%. African schemes reporting 95% repayment rates almost always conceal either extreme rationing or a failure to recognise arrears.
5.6 The internal counter-example: why African development banks failed
The region’s own history must be confronted before anything is proposed. Africa’s public development banks collapsed through the 1980s and 1990s under the weight of bad debt and politically directed lending. In Cameroon, the Banque Camerounaise de Développement, FONADER, FOGAPE and Crédit Agricole du Cameroun were all liquidated. More recently, the Banque Gabonaise de Développement lost its licence in 2024; Togo privatised two failed institutions.
The common cause is identifiable: these institutions lent directly, with governance permeable to political instruction and with no risk sharing alongside a party with an interest in repayment. Any proposal to “create a public bank to finance young people” in Central Africa must answer that objection, or it will reproduce the same disaster. The wholesale model — KfW, DBN — is precisely the historical answer to that failure.
VI. A four-tier architecture for Central Africa
What follows asks no one to be virtuous. It changes the prices that govern the credit decision. It requires costly political trade-offs, and it is presented together with its failure conditions.
Tier 1 — The price of risk: a leveraged regional guarantee fund
Principle. Create a regional credit guarantee mechanism modelled on KODIT, anchored to BDEAC but legally separate, with governance independent of shareholder states.
Funding. Three sources, in the Korean spirit: a mandatory bank contribution assessed on loan books (of the order of 0.1% to 0.2% a year — modest relative to net banking income), a state endowment, and capital from development finance institutions.
Orders of magnitude. Capital of XAF 200 billion (about USD 330 million), at a prudent leverage of eight (against roughly nine for KODIT), would support some XAF 1,600 billion of guarantees outstanding. At a 60% coverage ratio, that would underpin a loan book of around XAF 2,600 billion (about USD 4.3 billion), entirely directed at currently rationed segments — close to 20% of credit presently extended to the CEMAC economy. Relative to regional GDP, guarantees outstanding would represent about 2.4%, against the 4.1% KODIT reached in Korea in 2013: the target is ambitious but remains below a documented precedent. For comparison, Congo’s FIGA mobilised XAF 1 billion of guarantees in 2025.
Non-negotiable rules. Partial coverage of 50% to 70%, never 100%: beyond that, the bank stops appraising the file and moral hazard destroys the mechanism. Guarantee fees priced to actual risk, not free provision. Delegated decision-making to the bank below a threshold, with ex-post control — a fund that re-appraises every file becomes a bottleneck and reconstitutes the very public window one is trying to avoid.
Critical caveat. Economic shocks across CEMAC are strongly correlated, all states being exposed to the same oil cycle. A purely regional fund would therefore concentrate risk rather than mutualise it. It requires an external reinsurance tranche carried by the African Guarantee Fund, the AfDB or Afreximbank to cover tail scenarios.
Tier 2 — The price of information: complete and exploit the credit bureau
CICA, launched in January 2026, must clear three steps to deliver.
Make reporting mandatory and comprehensive for all credit and microfinance institutions, with effective sanction for non-submission. A partial credit bureau is a useless credit bureau.
Build a positive file, not merely a negative one. A registry that records only incidents identifies bad payers without ever rewarding good ones — it tightens rationing instead of easing it.
Integrate alternative data. This is the decisive point and the region’s most under-exploited opportunity. Mobile money represents about 5% of GDP in Cameroon, which alone accounts for the bulk of the Central African market; the DRC counts some 25 million active users. These platforms constitute the largest database of micro and small enterprise cash flows ever assembled in the region — and it is not used to decide credit. A trader collecting XAF 800,000 a month through her mobile account for three years has a more reliable income record than most audited accounts in the region. Converting that history into informational collateral, within a data protection and portability framework, would cost a fraction of what entrepreneurship training programmes cost, for an incomparably greater effect.
Tier 3 — The price of time: a long-term refinancing facility with no retail window
Principle. House at BDEAC a 7-to-10-year refinancing facility that purchases or refinances eligible SME and investment loan portfolios originated by commercial banks. Strict KfW/DBN model: no direct relationship with the end borrower, and the credit risk stays entirely on the originating bank’s balance sheet.
Why this is indispensable. With 83.6% of lending below 24 months and 1.87% beyond five years, no guarantee scheme will suffice: guaranteeing an 18-month loan still does not finance an agro-processing plant. The maturity constraint is, alongside sovereign crowding out, the most structuring feature of the system.
Funding. BDEAC bond issues on regional and international markets, development finance institution lines, and above all mobilisation of long-term institutional savings — social security funds, insurers, sovereign funds — currently invested largely in short-dated government paper. Cameroon’s XAF 2,000 billion diaspora target follows the same logic: these are the only naturally long-dated resources in the region.
Tier 4 — The price of crowding out: loosening the sovereign grip
This is the most political tier, and without it the first three will be partly neutralised. Three levers.
Enforce the existing prudential limit. COBAC caps sovereign exposure at 25% of net own funds; it reached 405% in 2024. Progressively restoring that constraint — or, failing that, introducing a non-zero risk weight on domestic sovereign exposures above a threshold — would immediately change credit committee arbitrage. It is the single most powerful measure in this package, and the politically hardest, because it raises the cost of government funding.
Broaden the investor base for government securities. Banks hold 76.8% of the government securities market. As long as states depend on them to that degree, no authority can seriously constrain that exposure. Developing participation by insurers, pension funds, non-bank institutional investors and the diaspora is therefore a precondition for the preceding reform, not an accompanying measure.
Clear domestic arrears and create a market for public receivables factoring. This is the highest-return measure in the package. An SME holding a certified claim on the Treasury should be able to discount it with its bank at a rate reflecting sovereign risk, not SME risk. That converts public arrears — today one of the leading producers of bank NPLs — into liquid collateral. African factoring volumes rose from about EUR 21.6 billion in 2017 to some EUR 50 billion in 2024, against a continental target of EUR 240 billion; Central Africa is almost entirely absent. The precondition is a credible, fast certification regime for public receivables, which is a matter of administrative will far more than of financial technique.
Sequencing and cost
| Horizon | Measure | Indicative budget cost | Expected effect |
|---|---|---|---|
| 0-12 months | Mandatory reporting to CICA; positive file | Near zero (regulatory) | Lower origination cost; access for first-time borrowers |
| 0-18 months | Certification of public receivables + factoring legal framework | Low | Liquefies arrears; mechanical fall in NPLs |
| 12-36 months | Regional guarantee fund, XAF 200bn capital | Phased endowment; bank levy 0.1-0.2% of loan books | ~XAF 1,600bn guarantees outstanding |
| 12-36 months | Integration of mobile money data into the credit bureau | Low (regulatory + technical) | Extends access to the unbanked |
| 24-60 months | BDEAC long-term refinancing facility | Market-funded, not budgetary | Lengthens maturities |
| Ongoing | Risk-weighting of domestic sovereign exposure; broader investor base | Raises state funding costs | Reallocates credit to the private sector |
VII. What could make this model fail
Analytical honesty requires setting out the failure scenarios, which are numerous and credible.
Capture of the guarantee fund. This is the principal risk. If guarantee allocation becomes an instrument of political patronage, the fund will reproduce the history of Cameroon’s BCD and FOGAPE. The only real protection is procedural: delegation to banks below a threshold, quarterly publication of guarantees outstanding by sector and region, external audit, and above all no individual decision-making power for state representatives on the board.
Free provision. Any guarantee scheme that becomes free, or whose fee is not calibrated to risk, turns into a subsidy, depletes its capital and disappears within five to seven years. It is the most common cause of death for these mechanisms.
A renewed oil shock. BEAC’s 2026 projections — growth of 3.2%, inflation of 2.4%, reserves at 4.72 months of imports, external coverage of 70.7% — depend closely on the oil price. A durable reversal would bring states straight back to the regional bank market and cancel the easing observed since 2025. The reform window is therefore cyclical, and narrow.
Failure of bank recapitalisation. If a significant number of banks fail to reach the XAF 25 billion threshold by 2029, the resulting consolidation will reduce competition in an already concentrated sector — Cameroon, Gabon and Congo accounted for more than 80% of regional balance sheets — and the net effect on SME credit could be negative. The supervisor will have to arbitrate between solidity and competition, and there is no obvious solution.
Absence of solvent demand. This is the most serious objection to the whole argument. If the binding constraint is market thinness rather than financing supply, the package will produce available credit without viable borrowers — and rising NPLs. The answer is not to abandon it but to sequence it: Tiers 2 and 4 (information and crowding out) are robust to this objection, because they improve allocation without forcing volume. Tier 1 (guarantee) must be calibrated prudently and its leverage adjusted to observed loss rates. Credit should not be forced into an economy that has no projects; the obstacles preventing the financing of the projects that do exist should be removed.
Conclusion: change the arbitrage, not the rhetoric
Three propositions deserve to be retained from this analysis, and all three are unwelcome to some part of the regional debate.
Central Africa’s banks are not culpable: they are rational. They respond to a price system that makes financing the state more profitable, less capital-intensive and more legally secure than financing a company. No charter of commitment, no entrepreneurship summit, no communications campaign will change that calculus. Only changing the prices will. This conclusion does not absolve banks of their pricing opacity or their under-investment in credit analysis — but it moves the centre of gravity of responsibility.
Bankability is public infrastructure, not a private skill. Functioning collateral registries, a comprehensive and positive credit bureau, usable alternative data, predictable judicial enforcement, certified and discountable public receivables: that is what manufactures financeable projects. Business-plan training does not. The region’s support industry should draw the consequences for its own usefulness.
The graduate exodus is not a credit problem, it is a firm problem. Young Cameroonians surveyed in 2024 cited training, skills mismatch and lack of experience ahead of financing; 61% considered emigrating to find a job, not a loan. But the firms capable of employing them are exactly those that need seven- to ten-year capital — the resource that represents 1.87% of lending extended. That is how the banking question becomes a question of demography, migration and sovereignty.
The window is open, and it is short. The regional credit bureau has existed since January 2026. Bank recapitalisation is under way. Sovereign crowding out has begun to recede. BVMAC has tripled in size. The oil price, for now, is not squeezing budgets. Taken together, these elements constitute the best configuration Central Africa’s financial sector has seen since the 2014-2016 crisis.
What is missing is neither the money, nor the diagnosis, nor the models: it is the decision to make the state pay the true price of its domestic debt. Everything else follows from that.
Statistical annex
A. Domestic credit to the private sector by banks (% of GDP)
| Country | Latest available year | Value |
|---|---|---|
| Burundi | 2024 | 37.6% * |
| Rwanda | 2024 | 21.3% |
| Cameroon | 2019 | 14.1% |
| Congo, Rep. | 2023 | 13.7% |
| Gabon | 2019 | 13.4% |
| Central African Republic | 2022 | 12.8% |
| Congo, Dem. Rep. | 2023 | 11.2% |
| Chad | 2021 | 8.3% |
| Angola | 2024 | 6.2% |
| São Tomé and Príncipe | 2024 | 6.1% |
| Equatorial Guinea | 2023 | 5.8% |
| For reference: Senegal | 2023 | 29.4% |
| For reference: Côte d’Ivoire | 2023 | 19.0% |
Source: World Bank, indicator FD.AST.PRVT.GD.ZS; Senegal and Côte d’Ivoire per Nkafu Policy Institute (2026).
* Methodological warning. The Burundian figure should not be read as an indicator of financial depth. It reflects a denominator — nominal GDP — compressed by per capita income of USD 355, in a context of 45.5% inflation in April 2025, monetary financing of the fiscal deficit and reserves covering 1.4 months of imports. A high credit-to-GDP ratio can signal macroeconomic distortion as readily as efficient intermediation.
B. CEMAC banking indicators
| Indicator | Value | Date | Source |
|---|---|---|---|
| Government share of gross bank lending | 61% (vs 24% in 2017) | 2024 | COBAC |
| Treasury exposure / net own funds | 405% (limit: 25%) | 2024 | COBAC |
| Claims on governments / credit to the economy | 83.5% (88.2% a year earlier) | March 2026 | BEAC |
| Government securities outstanding | XAF 10,020.5bn (≈ USD 16.5bn) | May 2026 | BEAC |
| Bank share of government securities holdings | 76.8% | March 2026 | BEAC |
| Non-performing loans | 17.4% of gross outstandings | March 2025 | BEAC |
| Loans under 24 months | 83.6% of new lending | Q3 2024 | BEAC |
| Loans over 60 months | 1.87% | Q3 2024 | BEAC |
| SME share of credit disbursed | 22.5% (XAF 565.9bn) | Q1 2026 | BEAC |
| Average SME rate / fees as share of effective rate | 11.00% / 31.9% | Q1 2026 | BEAC |
| BEAC policy rate | 4.50% | June 2026 | BEAC |
| Banks in breach of capital requirements | 22 of 56 | End-2024 | COBAC |
| Minimum share capital for banks | XAF 25bn (from XAF 10bn) | Jan. 2026 | COBAC R-2025/02 |
| Net treasury position of the banking sector | +XAF 8,245bn (≈ USD 13.6bn), 48.8% in securities | End-2024 | BEAC |
C. CEMAC macroeconomic markers
| Indicator | Value | Year |
|---|---|---|
| Population | 63 million | 2023 |
| GDP | USD 111.8bn | 2023 |
| Per capita income | USD 3,466 | 2023 |
| Growth | 1.7% (WAEMU: 3.4%; SSA: 2.9%) | 2023 |
| Projected growth | 3.2% | 2026 |
| Government revenue from natural resources | 46% (Equatorial Guinea: 88%) | 2023 |
| Tax take | 9.6% of GDP (WAEMU: 14.0%) | 2023 |
| Public debt | 52.8% of GDP (Congo: 96%; Gabon: 70.5%) | 2023 |
| Population below USD 2.15 a day | 31.1% (CAR: 65.9%) | 2023 |
| Young people not in employment, education or training | 1 in 4 | 2023 |
| Unemployment | 20% in Gabon and Congo | 2023 |
| Industrial / agricultural employment | 15% / 47% | 2023 |
