Blended Finance Structuring: Capital Stacks, First-Loss Capital and Risk-Sharing

The problem

Blended finance is frequently described as a way to make an unbankable project bankable. That framing is imprecise and leads to poor structuring decisions. Blended finance should solve a specific, identified risk problem, not subsidise weak project economics. Concessional capital deployed against the wrong barrier does not mobilise commercial investors; it simply recycles donor money without building a repeatable market.

Who this is for

This is written for DFI transaction teams, foundation and philanthropic capital allocators, and project sponsors negotiating a capital stack across investor classes. It assumes the reader already understands standard project finance terms and is deciding how, specifically, to size and price a concessional layer.

Why this matters now

SDG-aligned sectors, climate adaptation, nature-based solutions, food systems, still face financing gaps in the hundreds of billions of dollars annually that public and philanthropic capital cannot close alone. Blended structures are the primary mechanism for mobilising commercial capital at that scale, but the track record is mixed: many facilities have deployed concessional capital without measurably shifting the volume of private capital that would otherwise flow to the sector, because the concessional layer was sized to make a deal politically presentable rather than to solve a specific market failure.

The instrument: what a capital stack actually does

A blended structure layers capital by risk tolerance and return expectation. Each layer exists to solve a distinct problem, and conflating them is the most common structuring error:

• First-loss or junior capital. Absorbs early losses so senior investors take a materially cleaner risk profile. Typically provided by a foundation, DFI, or government facility, sized to the specific loss scenario it is meant to cover, not to an arbitrary percentage of the deal.
• Guarantees. Political risk, partial credit, or currency guarantees that de-risk a specific exposure without tying up as much capital as direct first-loss investment. A guarantee is the right tool when the barrier is a discrete, identifiable event risk rather than ongoing credit risk.
• Concessional (below-market) debt or equity. Lowers the blended cost of capital for the whole structure. This is the layer most often oversized relative to what the market failure actually requires.
• Technical assistance funding. Grant capital that builds project pipeline and institutional capacity, frequently the difference between a project being investable at all versus merely conceptual.
• Senior commercial debt or equity. Has to clear a normal commercial return hurdle. If this layer cannot be priced competitively even after the concessional and guarantee layers are in place, the structure has not actually mobilised private capital, it has substituted for it.

Transaction logic: diagnosing the barrier before choosing the instrument

Structuring starts with identifying which specific barrier is blocking commercial capital, then selecting the instrument built for that barrier. If the underlying risk is currency volatility, a first-loss tranche does not fix it, a currency hedge or guarantee does. If the constraint is a thin pipeline of bankable projects, a guarantee facility sitting on top of an empty pipeline accomplishes nothing; the correct tool is technical assistance capital to build pipeline. Defaulting to whichever blended-finance instrument is currently fashionable, rather than diagnosing the actual barrier, is the single most common reason blends fail to mobilise capital at scale.

Impact measurement and what makes a blend genuinely SDG-aligned

Genuine SDG alignment requires an impact thesis specific enough to measure against a defined target, not a portfolio tagged with SDG icons after the fact. That requires governance built into the transaction itself: reporting covenants with defined KPIs, impact-linked pricing where appropriate (the sustainability-linked loan model, where the interest rate moves with KPI performance), and a mandate for the fund manager or SPV that survives changes in personnel. Structuring for impact is a legal and financial exercise executed at term-sheet stage, not a marketing exercise executed after close.

Risks, limitations, and greenwashing concerns

• Concessionality creep. Sizing the concessional layer to the deal's political optics rather than the actual risk gap invites justified criticism that public and philanthropic money is subsidising returns that private capital would have accepted anyway.
• Additionality claims that don't hold up. If the commercial capital in the structure would have been deployed regardless of the concessional layer, the blend has not mobilised anything; it has replaced capital that was already available.
• KPIs disconnected from the core transaction. An impact target bolted onto a deal without pricing or covenant consequences is not a credible SDG claim.

Illustrative example: sizing a first-loss tranche to the actual risk

Consider, illustratively, a renewable energy project in a frontier market where the commercial barrier is not credit risk but currency convertibility risk. A poorly diagnosed structure might apply a 20% first-loss equity tranche across the whole capital stack, tying up scarce concessional capital against a risk it doesn't address. A correctly diagnosed structure instead applies a targeted currency guarantee sized to the specific convertibility exposure, leaving the first-loss layer smaller and available for the next transaction. The difference in capital efficiency, and in how many projects a fixed pool of concessional capital can support, is substantial. This example is illustrative only.

Conclusion

The transactions that mobilise real, recurring private capital are the ones where the concessional layer was sized against a diagnosed, specific market failure and designed to exit once that failure was resolved. Every other structuring choice, capital stack seniority, guarantee scope, KPI design, follows from that diagnosis rather than preceding it.