Blended finance must solve a defined market failure

Technical assistance is the narrower intervention when the binding obstacle is missing capability or dependable information rather than the price of capital. A sponsor may need supervisors who can control quality, an environmental and social management system that survives staff turnover, or production records that reconcile good output to collections. Paying those work packages can improve the evidence on which every financier depends. A permanent interest subsidy would not, by itself, create any of them.

The workstream should be separate from the debt documentation and separately costed. Its budget should identify the deliverable, the executing party, the procurement route, the venture's own contribution, the acceptance evidence and the recurrent cost after handover. A report delivered to a sponsor is not the same as a capability operating inside the factory. Acceptance should test use: a supervisor completes an assessment, a meter register reconciles to source readings, or the safeguard owner closes an action with supporting records.

Preparation finance still needs instrument discipline. GCF's Project Preparation Facility describes grants, repayable grants and equity for eligible Accredited Entities and Project Specific Assessment Approach applicants. It assigns responsibilities for procurement, oversight, use and reporting, and states that support leading to a feasible income-generating project should generally be repayable on terms determined during proposal development and review.[S08] This demonstrates why preparation support is not automatically a non-repayable operating subsidy. It does not establish that GFT, the venture or an unnamed partner qualifies for that route.

The content of the assistance should follow the actual constraint. GFT's Cameroon textile programme provides design context through its feasibility, performance-management system and digital-strategy components. That programme is not evidence that a factory has implemented the systems or achieved results. For a proposed blended structure, GFT would translate the relevant components into procurable work packages: supervisor development, manual production records, metering definitions, data reconciliation, safeguard responsibilities and assurance readiness.

Environmental and social work illustrates the ownership test. IFC Performance Standard 1 applies where environmental and social risks and impacts exist, with management through an appropriate management system.[S22] The accompanying Guidance Notes explain the Performance Standards and good practice, but do not replace them.[S34] Technical assistance should therefore leave named operating responsibilities, procedures, records and corrective-action routines. A lender pack prepared externally and never adopted by management would leave the capability gap intact.

The minimum-concession test also applies to technical assistance. Fund only the increment needed to overcome the defined barrier. Do not pay for a broad corporate consultancy programme because a small metering or training gap is difficult to isolate. Where the work produces private value that the sponsor can reasonably finance, consider sponsorship, cost-sharing or repayment. Where it produces verifiable public learning, define what may be disclosed without exposing legitimately confidential operating information.

A concession should last only as long as the obstacle that justified it. The review mechanism should therefore follow evidence, not the calendar alone. Relevant signals include accepted machinery, repeatable collections, reconciled production and cash records, completed safeguard actions, demonstrated management capability and credible expressions of commercial financing interest. Each signal needs an owner, a verification method and a decision consequence. An improving narrative without those records is not an exit trigger.

The no-repeat-subsidy case belongs in the original model. Remove future concessional disbursements that are not contractually committed, then test operating cash, debt service, safeguard spending, maintenance and working capital. Include any recurring verification and reporting costs that remain after support ends. If the venture survives only by assuming another donor, the proposal has not established commercial sustainability. IFC's principle that impact should contribute towards commercial viability points directly to this test.[S04]

Exit does not always mean immediate repayment or refinancing. A preparation grant can end with accepted handover evidence. A technical-assistance agreement can close when the venture operates the new capability and funds its recurrent cost. A subordinated instrument can amortise or remain until its agreed maturity without receiving a further public contribution. A guarantee can expire as the covered exposure falls. Refinancing is only one possibility, and it should be counted only when actual terms support it.

Exhibit 36 shows an illustrative review path. The starting concessional share of 10 percent comes from the paper's hypothetical XAF 100 million junior contribution within XAF 1,000 million of capital. Later shares are constructed choices, not forecasts. They decline only if new financing rounds satisfy the stated evidence gates. A falling share must never be produced merely by increasing the denominator while the absolute subsidy continues to grow.

Every gate needs three possible decisions, not just continuation and exit. Reduce support when the evidence shows a smaller residual barrier. Redesign the instrument when the barrier persists but its form has changed. Stop when the operating case remains weak or the public cost exceeds the justified benefit. Automatic reduction on a fixed date could destroy a useful demonstration before its evidence exists. Automatic renewal would conceal whether the demonstration has worked.

The final exit review should revisit the original counterfactual. Ask whether commercial providers now accept the exposure, which verified evidence changed their decision, and whether later projects can obtain comparable terms without the same contribution. If the answers remain unclear, retain the uncertainty in the record. A completed disbursement is not proof that a market has been reinforced.

Transparency makes the concession inspectable without requiring publication of every commercially sensitive contract. The disclosure record should reveal why support is needed, what it is worth under the stated method, who benefits, what private finance changes, and when support ends. Detailed customer prices, personal information, security credentials and protected supplier terms can remain controlled. The rationale, valuation basis and public exposure should not disappear behind that confidentiality.

IFC states that it discloses subsidy levels and their justification in its Summary of Investment Information, and that an internal blended-finance committee checks adherence to its principles.[S04] GFT treats that as an institutional governance example, not as a rule binding this venture or evidence that any provider will accept its proposal. The transferable lesson is to separate recommendation from approval: the team designing the concession should not be the only team testing its necessity and incidence.

Exhibit 37 proposes a disclosure record that keeps the main judgements visible. Amounts should distinguish committed, available and disbursed states. Beneficiaries should include indirect incidence where a lender's loss or sponsor's return changes. Results should carry their period, source and verification status. Confidentiality exceptions should identify the protected field and the reason rather than suppress the entire analysis.

Mobilisation reporting needs especially strict definitions. Sponsor equity is not automatically capital mobilised by the concession, because it may have been committed before support was considered. DFI own-account finance should be shown separately from private commercial finance and from donor resources. The concessional contribution itself is never private mobilisation. If several interventions occur together, the method should state whether capital is attributed wholly, proportionally or only descriptively. Totals must reconcile to the sources-and-uses statement without counting one commitment twice.

Commitment states matter as much as categories. An indicative discussion, an approved amount, a signed facility and a cash disbursement are not interchangeable. Report each on its actual date and retain undrawn conditions. The OECD's observation about relatively limited private mobilisation cautions against presenting a large headline ratio as self-evident success.[S03] A smaller amount of genuinely additional finance may be more useful than a large total that includes capital already present.

Development reporting must remain on a separate axis. The venture can report jobs, production, resource use and local value only with their own baselines and methods. AIMM's explicit separation of expected development impact from additionality reinforces that boundary.[S26] Achieving an operating result after receiving support does not prove that the concession caused the whole result, just as obtaining commercial finance does not prove development value.

GFT's recommended approval sequence is therefore restrictive. Reject blending when the market failure is undefined, the unsupported case already works, or the operating model remains structurally weak. Where a genuine barrier remains, choose the smallest intervention that removes it, disclose its incidence, fund the associated measurement, and make continued support conditional on verified progress. Concessional capital then becomes a bounded instrument for changing a financing decision, not a permanent feature of the business model.