Loading...
Loading...
A structural diagnosis of the catalytic claim.
Blended finance promises to mobilise private capital by using concessional capital to de-risk commercial returns.
The premise is elegant: a tranche of concessional or philanthropic capital absorbs first-loss exposure, commercial investors supply the bulk, and the concessional dollar is leveraged into a multiple of private capital. As markets matured, the ratio of private to concessional capital — the mobilisation ratio — was meant to climb until the subsidy could withdraw.
The problem: after a decade, the ratio has stalled. In emerging-market infrastructure — the domain where blended finance is most heavily promoted — each dollar of blended finance has mobilised on the order of forty cents of private capital. A sub-unitary ratio does not just fall short of the promise; it inverts it.
The underperformance is structural, not operational: single-cycle, liability-bearing instruments are being applied to multi-cycle, relational objectives. From that one mismatch, three failure modes follow — and each would persist under flawless execution.
Blended finance occupies an unusual position: it is at once the most institutionally endorsed instrument for channelling private money toward public goods, and one of the least able to demonstrate that it does so. Multilateral development banks, bilateral donors, foundations and standard-setters have spent more than a decade building the scaffolding around it — facilities, principles, taxonomies, deal databases — on the strength of a genuinely elegant premise.
That premise has not held, and the gap is no longer a matter of interpretation. Annual commitment volumes have stayed broadly flat. The market has been described — by its own standard-setters, not its critics — as a “cottage industry” of bespoke, fragmented, non-standardised interventions that has “not scaled as rapidly as hoped” and “mobilised relatively limited private finance.” That is a remarkable admission from the institutions whose endorsement underwrites the field.
Confronted with the evidence, the field has responded almost entirely in operational terms: the pipelines are too thin, the managers too inexperienced, the documentation too bespoke, the facilities too small. Each remedy implies the underperformance is a transitional condition that accumulated practice will dissolve. But when a field assembles its best practitioners, refines its documentation and tightens its standards for ten years and the central ratio does not move, the binding constraint is unlikely to lie in execution.
Every financial instrument carries an implicit theory of time. A senior loan assumes a single cycle: capital is deployed, principal and interest are returned within a defined term, and at maturity the instrument is extinguished. Equity assumes a single, longer cycle terminating in an exit. These are single-cycle, liability-bearing instruments — built to be deployed once, to carry a defined claim, and to resolve at a terminal date. The arc from deployment to resolution is not incidental to the instrument; it is the instrument.
Blended finance assembles its structures almost entirely out of these inherited components — tranches, guarantees, first-loss layers, mezzanine and anchor positions. The concessional layer, for all the developmental language attached to it, is in cycle terms simply a differently-priced participant in the same single arc. Its concessionality changes the price it accepts; it does not change the shape of the arc it travels.
The objective is shaped differently. Building a market, establishing a track record, shifting investor perceptions of a frontier risk — none of these is realised at the resolution of any single transaction. They are realised, if at all, across a sequence of transactions, each meant to make the next one cheaper. The developmental return does not crystallise at a maturity date; it compounds across repetition. These are multi-cycle, relational objectives.
This is the inherited architecture problem, statable in one sentence: a multi-cycle, relational objective is being pursued with single-cycle, liability-bearing instruments. It is a temporal mismatch built into the instrument class, not a defect in any particular deal — the instruments resolve where the objective does not, and extinguish themselves at precisely the point where the developmental work is supposed to begin compounding.
The first consequence of fitting a multi-cycle objective into a single-cycle instrument is that the objective's long horizon collapses onto the instrument's short one. Two clocks are running. The objective's clock runs across cycles with no natural stopping point until the market can stand without concession. The instrument's clock runs from deployment to maturity and stops there. Because only the instrument's clock is structurally defined, it becomes the operative clock for the whole undertaking.
Concessional capital placed in a first-loss position is, by design, the capital most exposed to consumption within the cycle — absorbing the losses that make the senior tranches investable. When the cycle resolves, it has done its work precisely by being spent. There is no structural provision for it to persist into a second cycle, because the instrument it inhabits has no second cycle. This is horizon collapse: the concessional dollar is single-use not as a matter of policy, but as a property of the architecture.
If the binding constraint were execution quality, the field's defining metric ought to improve as the same intermediaries run successive funds: the second fund should need less concession than the first. Yet each fund tends to re-open its concessional case from scratch, raising a fresh first-loss layer to crowd in a fresh cohort of commercial capital — because the prior fund's concessional layer was extinguished at resolution rather than carried forward as a recyclable base. The field re-solves the first-cycle problem again and again, not because its practitioners fail to learn, but because its instruments cannot hold a position across the boundary where the learning would accrue.
The central verb of blended finance is de-risk. The second failure mode is that, under a single-cycle instrument, de-risking and catalysing are not the same act — and the architecture is biased toward the first while the field's vocabulary treats them as synonyms. This is the de-risking paradox.
To subsidise a transaction is to improve the returns available within it — absorb a loss, cap a downside, guarantee a payment so the risk-adjusted return clears a commercial threshold this time. To catalyse is something else: to change the conditions of the next cycle, so the same risk is priced lower, or borne by the market directly, or requires less concession, or none. Subsidy operates within a cycle; catalysis operates across the boundary between cycles. A transaction can be flawlessly de-risked, close on schedule, return capital to its senior investors, and leave the cost of capital for the next comparable transaction exactly where it was.
A single-cycle instrument can only observe and reward the present transaction, because the present transaction is the only thing inside its horizon. It has no representation of the next cycle and therefore no way to register whether the next cycle's conditions improved. So it rewards the thing it can see: co-investment at deployment, a crowding-in ratio booked at close. The catalytic claim is asserted; the subsidy is what is actually purchased.
The paradox sharpens under the field's own leverage metric. High mobilisation ratios are easiest to obtain where private appetite already exists, so maximising the headline ratio pushes concessional capital toward lower-risk transactions in more developed markets and away from the frontier contexts where market-creation is genuinely needed. The OECD's own guidance now concedes this gradient. Optimising the metric and optimising for catalysis are not merely different — under a single-cycle instrument they pull in opposite directions.
The first two failure modes concern what the instrument does. The third concerns what it lets us see — and it is the most consequential, because it determines whether the other two can ever be detected. The mobilisation ratio, and with it nearly the entire evidentiary apparatus of the field, is computed at the point of deployment: how much private capital was committed alongside the concessional capital in this transaction, at this close. This is measurement at the deployment boundary, and it is the natural measurement boundary of a single-cycle instrument, because deployment and resolution are the only two events the instrument defines.
But the catalytic claim is a claim about cycles, not deployments. The developmentally decisive questions — did the cost of capital for the next comparable transaction fall? was the concessional layer recovered and redeployed rather than consumed? did the market learn to price the risk with less subsidy? — are each statements about the relationship between one cycle and the next, visible only across the boundary the single-cycle instrument does not represent. The field is measuring at the one location where its central claim is guaranteed to be invisible.
This renders the catalytic thesis structurally unfalsifiable in the field's own terms. A mobilisation ratio can be computed for every transaction ever done, refined and reported to any decimal place, and it will still say nothing about catalysis — because catalysis is defined over a domain the metric does not span. A number measured in the wrong place cannot be made into the right number by measuring it more carefully. The verification gap is located in where the measurement is taken, not in how well it is taken — which is why no improvement in reporting discipline can close it.
If the diagnosis is correct, the corrective cannot be a better-run blended fund, because the failures are not failures of running. It must be an instrument whose cycle structure matches the cycle structure of the objective. Because each failure mode is a single-cycle instrument betraying a multi-cycle objective in a particular way, the conditions are simply the three modes inverted:
Multi-cycle by construction. Inverts horizon collapse. Concessional capital must persist across the boundary between cycles — recovered and redeployed rather than consumed at resolution — so the second cycle begins from the position the first established, not from scratch. Not a longer maturity date, but an architecture in which the capital outlives any individual transaction.
Non-extractive recovery. Inverts the de-risking paradox. Recovery of capital across cycles must be structurally distinguished from a priced return on a single cycle. The capital must return — to be redeployed — but its return must not be characterised as the yield that drives the subsidy-versus-catalysis confusion, so the incentive to optimise the within-cycle ratio dissolves rather than merely being discouraged.
Cross-cycle measurement. Inverts measurement at the deployment boundary. The instrument must carry its own cross-cycle measurement boundary, making the relationship between cycles observable as a property of the instrument itself. If it represents the second cycle, the change in cost and risk of capital between cycles becomes a quantity it can register — and the catalytic claim becomes falsifiable in its own terms.
These are not independent design preferences. They are one requirement viewed from three angles: that the instrument's theory of time match the objective's. Instruments in the recoverable-grant and charitable working-capital family are one route that meets these conditions — the Charitable Asset Deployment Agreement (CADA) is specified to satisfy them directly.
Read the CADA explainerThe stagnation is too consistent across geographies and vintages, too indifferent to a decade of improving operational sophistication, and too tightly coupled to the basic shape of the instrument to be a maturity problem. Standardising the documentation of single-cycle instruments does not give them a second cycle; it makes the first cycle tidier. Each failure mode would persist under flawless execution, which is the signature of a structural cause rather than an operational one.
No. The intuition behind blended finance — that public or philanthropic capital can change the terms on which private capital engages a hard problem — is sound, and this analysis leaves it entirely intact. The argument is narrower: the dominant form in which that intuition was institutionalised inherited a single-cycle, liability-bearing architecture mismatched to the multi-cycle objectives it was asked to serve. The problem is the form, not the intuition.
The mission-oriented literature locates the underperformance in a gap between the instrument and public purpose, and calls for capital to be steered more deliberately toward mission. That is correct as far as it goes, but it identifies the wrong layer: redirecting and re-governing a single-cycle instrument leaves it single-cycle. The misalignment lives further down, in the cycle structure of the instrument itself — the same move that clarified the recurrent disappointment of public–private partnerships as a property of the contractual form, not of the parties.
The full diagnosis, and the instrument family the conditions point to — recoverable-grant and charitable working capital.
View PapersA multi-cycle, non-extractive instrument in the field today — concessional capital that recycles rather than being consumed.
Live Deployments