Blended finance is a grammar, not a product

The example below is stylized. It combines structures used in coursework and public case materials; it is not a description of a private term sheet.

Suppose a development finance institution offers a guarantee covering the first 18% of eligible credit losses on a portfolio of loans to mid-sized solar developers. Why 18%? The number should not be decorative. It should follow from the portfolio's expected and stressed loss distribution, recovery assumptions, concentration, the guarantee's coverage rules, and the amount of residual risk a commercial lender can underwrite.

The number matters, and so does its position in the loss waterfall.

The structure is the argument

Blended-finance diagrams often look like solutions to a risk-return puzzle: concessional capital, commercial capital, and a guarantee arranged into one stack. The diagram is only useful if it shows who absorbs which loss, when, and under what conditions.

A funded first-loss tranche is generally junior. It absorbs covered losses before mezzanine or senior capital, subject to the transaction's waterfall. An unfunded first-loss guarantee can create a similar economic protection without becoming a funded tranche on day one. Commercial lenders price and size their exposure against the risk that remains after that protection.

If the first-loss support is thicker than necessary to mobilize the commercial layer, scarce concessional capacity may be wasted. If it is too thin, the commercial layer may still be unable to participate. The relevant threshold is therefore an underwriting result, not a round number chosen for a slide.

Guarantees as conditional promises

A guarantee is a contingent obligation. It usually does not require the guarantor to disburse the covered amount at closing, although fees, reserves, or collateral arrangements may still apply. Payment occurs only after defined triggers and claims procedures are satisfied.

The risk covered must be named precisely. A partial credit guarantee covers specified debt-service or credit losses. It does not, by itself, protect a lender against currency depreciation. Currency risk requires a separate hedge, liquidity facility, local-currency instrument, or guarantee whose terms explicitly cover that exposure.

The risk has not disappeared. It has been allocated to a party that is willing and able to bear it under stated terms.

What "blending" is blending

Capital stacks do not merely blend patience. They combine instruments with different risk positions, return requirements, currencies, tenors, and claims on cash flow. A good structure begins with the asset and asks which mismatch is preventing finance.

If revenue ramps slowly, the debt may need a grace period, longer tenor, or repayment profile matched to cash flow. If revenue is back-loaded, sculpted amortization may protect coverage in early years. If collections are volatile, a reserve or liquidity facility may be appropriate. Each feature should answer a specific risk in the asset or contract.

That sequence matters. Starting with an available pool of concessional capital and then searching for an asset to fit around it can produce a transaction that is blended in form but weak in additionality. The test is not whether public or philanthropic support appears in the structure. The test is whether that support addresses a genuine financing barrier without subsidizing risk the market was already prepared to take.

What to read for

Read a blended-finance term sheet as a set of conditional claims:

• Which losses are eligible, and which party absorbs them first?
• Is support funded at closing or contingent on a later claim?
• What are the coverage cap, triggers, exclusions, fees, and recovery-sharing rules?
• Which currency and tenor does each instrument cover?
• What residual risk remains with the borrower and commercial lenders?
• What evidence supports the claim that the structure mobilizes capital that otherwise would not participate?

The cover memo may call the capital catalytic. The loss waterfall, claims language, and counterfactual will tell you whether that description holds.