A sizing-and-recognition discipline for hybrid PPPs and blended finance — why the size of the subsidy, not the decision to grant one, is the fiscal question, with a reference implementation on the Austral platform
Author: David Duarte Arancibia, Austral Fiscal & Infrastructure Intelligence Date: July 2026 Audience: Ministries of Finance and PPP units; multilateral development bank (MDB) infrastructure and blended-finance desks; supreme audit institutions; sector regulators; concessional-finance and DFI teams; fiscal-transparency researchers Companion software: Austral platform —
https://platform.austral-intelligence.comSeries position: A policy companion to Sizing the Viability Gap Fund (Paper 1), which supplies the optimisation engine this paper argues from. It draws the risk-allocation side from The Optimal Risk-Retention Frontier (Paper 11) and the balance-sheet side from The Recognition Rule (Paper 10). Occasioned by the World Bank Group's Hybrid PPP Initiative and the blog "Hybrid PPPs: Turning risk into viability" (18 December 2024). Disclosure: the author was, in prior roles, head of the Public-Private Partnerships unit at Chile's Ministry of Finance and part of the World Bank's fiscal-commitments team, and is a co-developer of the IMF/World Bank PFRAM. He is the founder of Austral, which builds the VGF-sizing, risk-allocation and recognition tools this paper says a blended-finance initiative should carry — a commercial interest the reader should weigh. The paper is offered as constructive analysis of a welcome initiative, not as a criticism of the institution; it argues that the initiative is right in direction and needs a sizing-and-recognition layer to spend concessional money well. Every claim rests on public method and a reproducible engine.
The World Bank Group's Hybrid PPP Initiative — blending concessional finance with private capital and a "One World Bank" package of advisory and preparation support — is a welcome and correct response to a real problem: economically sound infrastructure projects that stall because their cash flows cannot cover debt service. Its workhorse instrument, the Viability Gap Fund (VGF), is the right one. But the initiative, as presented, decides that a subsidy is warranted without addressing the only question a finance ministry must actually answer: how big should it be?
This paper makes three arguments, using nothing beyond established method and a reproducible engine. First, the popular framing treats "the viability gap" as an input — a given shortfall to be bridged — when it is an output: the residual that remains after the risk allocation and financial structure have been optimised. A large share of apparent viability gaps are bankability gaps — a risk-allocation problem, cured by placing construction and other manageable risks with the party that manages them — misdiagnosed as funding gaps and treated with a subsidy. Blended finance systematically conflates the two, and the cost of the confusion is concessional money spent to paper over a gap a better allocation would have closed for free. On a worked $620M user-pays concession, the optimiser returns a required VGF of 0% — the project is bankable with no subsidy at all, on its baseline risk allocation: the gap was not a given.
Second, once the gap is genuinely positive, its size is computable, not a matter of judgment. Sizing the VGF is a constrained optimisation — the risk allocation and financial structure as design variables, the VGF percentage as the principal knob, and bankability, Value for Money and fiscal affordability as binding constraints — solved for the smallest grant that clears all three at once. Hand-sizing, the current practice, systematically overshoots, and overshooting is not conservative: it is a windfall to the private partner paid out of the scarcest money a government has.
Third, a blended commitment is a fiscal commitment that must be recognised. A VGF stacked with concessional terms and guarantees creates a contingent-liability position that belongs on the augmented public balance sheet, not off it because the money came cheap. An initiative that blends public, private and concessional capital is precisely the setting in which the size of the subsidy and the residual risk the state retains must be computed and disclosed, not assumed.
The policy lead is a single discipline: minimise before you fund, size the residual, recognise what remains. The Hybrid PPP Initiative is a good product missing an analytical layer; this paper specifies the layer and points to the open, reproducible tools that implement it.
Keywords: Viability Gap Funding · hybrid PPP · blended finance · concessional finance · bankability · risk allocation · public-private partnership · fiscal commitments and contingent liabilities (FCCL) · recognition · Value for Money · Ministry of Finance · multilateral development banks.
Feedback & discussion. Published openly for challenge by MDB blended-finance teams, ministries of finance, and the academic community — corrections, counter-examples and country cases especially welcome. Comment at
https://austral-intelligence.com/research; email the author at [email protected].
Closing the infrastructure gap — the figure most often quoted is upward of US$15 trillion by 2040 — will not happen on public balance sheets alone, and the last decade's answer has been blended finance: concessional public capital deployed to mobilise private capital into projects that would not otherwise reach financial close. The World Bank Group's Hybrid PPP Initiative is a clean articulation of this idea for the public–private partnership (PPP) context. It combines concessional financing with private investment and, importantly, a "One World Bank" package — IDA/IBRD concessional money alongside IFC transaction advisory and preparation support — so that a stretched ministry receives structuring capacity and capital in one offer rather than across fragmented windows.
Three things in this are right, and worth saying plainly before the argument turns critical. The diagnosis is right: many projects whose benefits exceed their costs still fail to close because their revenues do not cover debt service, and the construction phase reads to investors as unacceptably risky. The instrument is right: the Viability Gap Fund — an upfront grant, conventionally expressed as a percentage of capital expenditure, first institutionalised at scale in India's 2006 VGF scheme (Government of India, 2006) and now standard across the MDB toolkit — is the correct tool for a genuine revenue shortfall. And the packaging is right: reducing the World Bank's own fragmentation, and making the concessional-plus-advisory offer a named, promotable product, lowers a real barrier for governments that lack preparation capacity.
The critique that follows is therefore not that the initiative is wrong. It is that a good initiative is missing the number at its centre — and that the omission is not incidental but the exact place where scarce concessional money is wasted.
The initiative's own narrative contains the confusion it needs to resolve. It opens with the construction phase, which investors find "particularly high risk," and moves, within a sentence, to a subsidy as the remedy. But these are two different gaps, and only one of them is a funding gap.
Treating a bankability gap with a funding instrument is the characteristic error of the blended-finance moment. It is expensive twice over. First, it overpays: a grant is committed to buy down a risk that a contractual reallocation would have transferred for nothing. Second, it mistargets: the concessional money lands as a general sweetener to the project's returns rather than as the price of a specific, retained risk — so neither the ministry nor the auditor can later say what the public paid for. The Retention Frontier (Paper 11) makes the allocation side of this quantitative — the efficient share of each risk the state should keep, and the premium the private side charges to bear what it cannot manage; the point here is prior and simpler: you cannot size a funding subsidy correctly until you have separated it from the risk problem it is often standing in for.
The deepest issue is a matter of sequence, and it is where the language of "bridging the gap" quietly misleads. It casts the viability gap as an input — a fixed quantity, sitting in the project, to be spanned by a subsidy. It is nothing of the kind. The gap is an output: the residual that remains after the risk allocation and the financial structure have been chosen. Change the allocation, and the gap changes. Change the tenor, the gearing, the guarantee package, and it changes again. There is no single "the gap" to bridge; there is a gap conditional on a structure, and most of the structures a first pass produces are not the ones that minimise it.
This inverts the reflex. The disciplined sequence is:
"Minimise before you fund" is not an efficiency nicety. It is the difference between a subsidy that pays for an irreducible shortfall and a subsidy that pays for the analyst's failure to look for the structure that would have removed it.
Once the gap is genuinely positive, its size is not a matter of seasoned judgment; it is a constrained optimisation, and treating it as anything softer is how ministries overpay. The design variables are the project's risk allocation and financial structure; the VGF percentage is the principal sizing knob; and the constraints are the three gates every PPP must clear — bankable (the lenders' cover ratios hold), Value-for-Money-positive (cheaper as a PPP than as public procurement), and fiscally affordable (the commitment fits the government's envelope over the contract term). The object is the smallest VGF at which all three bind simultaneously.
Sizing the Viability Gap Fund (Paper 1) states this problem formally and ships a reference implementation. Its practical contribution is that the practitioner can pose the question that matches the actual decision rather than a single canned one: the smallest grant that makes the project bankable; the smallest lifetime fiscal cost; the most attractive bid within a fixed fiscal envelope; the smallest grant that survives a stress catalogue; or the whole trade-off curve between them. Each returns a defensible, auditable number, with signed reports built for supreme audit institutions — not a hand-sized figure defended by experience.
The contrast with practice is stark. Hand-sizing does not err symmetrically: faced with uncertainty, the practitioner rounds the subsidy up, because an under-funded project visibly fails to close while an over-funded one closes and looks like a success. So the systematic bias is toward overshooting — and an over-sized VGF is not a prudent buffer. It is a transfer of the scarcest money a government commands, concessional finance, into the private partner's returns, invisible precisely because the project closed.
Paper 1's worked example is calibrated to a real project: the Accra–Tema Motorway Extension, a US$620 million user-pays toll concession — the kind of economically sound but financing-challenged project the Hybrid PPP Initiative exists to serve. Run through the optimiser, the answer the engine returns is VGF = 0%, reached on the project's baseline risk allocation with no reallocation needed: the project is bankable with no grant, placing effectively zero direct fiscal commitment on the government. (The point is not that a clever reallocation closed the gap here — none was required — but that the reflexive assumption of a gap was simply wrong.)
The lesson is not that subsidies are never needed — many genuine funding gaps are positive and a VGF is exactly right for them. The lesson is that "the viability gap" is not a given. A project that the reflex would have flagged for concessional support turns out, when the number is actually computed, to need none. Had the reflex prevailed — observe a financing difficulty, reach for the fund — the public would have committed concessional money to a project that did not require it, and called the result a success. The number that prevents that mistake is the minimised gap, and it exists only if someone computes it before the cheque is drawn.
Where the minimised gap is positive, the same engine returns the smallest grant that clears the three gates, and the argument of §4 applies: that number, not a rounded-up estimate, is what the initiative should carry.
There is an incentive edge to all of this that a fiscal authority cannot ignore. A viability gap frequently traces to a weakness in project preparation — an optimistic demand forecast, an incomplete design, an unresolved land or permitting risk, a risk the public side should have retired before tender. When the response to that gap is simply to fund it, the subsidy quietly pays for the preparation failure that produced it — and, worse, teaches the sponsor that optimistic preparation is costless, because the fund will close whatever gap it leaves.
This is the soft-budget-constraint logic that Engel, Fischer and Galetovic identified at the heart of PPP renegotiation, arriving here one stage earlier — at financial close rather than during operations. The Austral series has documented its downstream form: renegotiation is not exceptional but the base case (63% of Chilean concessions), and it concentrates where the contracting authority that prepared the project badly is also the actor advocating to accommodate the claim. A blended-finance facility that funds gaps without first asking whether the gap is earned — irreducible after honest structuring — or manufactured — an artefact of thin preparation — is a soft-budget machine at the front end of the project as well as the back. "Minimise before you fund" is, among other things, the discipline that stops the fund from subsidising the very behaviour it should be pricing out.
The final omission is on the balance sheet. Concessional finance is cheap, and cheapness invites the treatment of a blended commitment as something less than debt. It is not. A VGF stacked with concessional terms and, frequently, a guarantee package is contingent-liability creation — a commitment the budget must carry over the contract term and the accounts should recognise, on the same principle by which any PPP obligation is recognised.
Which principle applies is itself contested — The Recognition Rule (Paper 10) sets out how ESA 2010, IPSAS 32 and GFSM 2014 can each score the same deal differently, and how the measure a government most wants to look good on is precisely the one that recognises the least. Blended finance sharpens the temptation: because the money arrived concessional and "developmental," the commitment is especially easy to leave off the fiscal-rule balance sheet even as it binds the state in substance. The consequence is the recognition gap this series has priced elsewhere — the augmented liability that sits off the anchor the law watches while remaining, economically, the sovereign's. A hybrid PPP is a textbook generator of exactly that gap, and an initiative that promotes hybrids without a recognition discipline promotes the gap along with them.
Recognition is not an accounting afterthought to the sizing question; it is its completion. The number the optimiser returns — the minimised, gate-clearing VGF, plus the residual risk the state has explicitly retained — is the fiscal commitment to book. Sizing it correctly and recognising it correctly are the same act, done properly.
The pieces of this argument are established; the composition is not. Five literatures each own part of the problem, and the discipline proposed here is what connects them.
The blended-finance critique has documented, at the aggregate, exactly the failure this paper locates at the project level. Attridge and Engen (2019) find that concessional blending mobilises less private capital than is claimed, concentrates in already-bankable sectors and lower-risk geographies, and — the point that matters here — that the size of the subsidy and its additionality are rarely disclosed or measured. The OECD's DAC Blended Finance Principles (2018) make mobilisation and additionality the test; Mazzucato and colleagues (2025), pointedly titling their critique Mind the Mission, Not the Gap, argue the "gap" framing itself misdirects. What this literature has not done is operationalise the missing number — the smallest project-level subsidy additionality would actually require. It diagnoses the opacity; it does not close it.
The optimal-subsidy and auction literature supplies the mechanism the framing here rests on. Engel, Fischer and Galetovic (2001), in their least-present-value-of-revenue auction, showed that the support a concession needs is not a fixed quantity to be assumed but one a competitive mechanism can reveal — with the contract term itself made endogenous. India's VGF scheme institutionalises the same insight in blunt form: the bid variable is the VGF amount, so the market sizes the grant. Both establish the central claim — the gap is an output, not an input — but from the ex post side: the auction reveals the number once the project is structured and taken to market. The planner still needs to compute the minimised gap ex ante, to choose the structure and set the fiscal envelope before bids are called. That ex ante computation is what Paper 1 supplies and this paper argues should sit inside the blended-finance offer.
The risk-allocation literature explains why so many apparent funding gaps are bankability gaps in disguise. Bing Li and colleagues (2005), surveying UK PFI practice, place macro-level risks with the public sector and construction, design and operation risks with the private partner who manages them — the allocation whose absence manufactures the financing difficulty a subsidy is then asked to cure. Iossa and Martimort (2015) supply the incentive microeconomics: how bundling and residual-control rights shape effort and the cost of risk transfer. This literature establishes that mis-allocation is the first-order lever on financeability; it does not connect the allocation to the size of the residual subsidy — the link the discipline here draws, and that the Retention Frontier (Paper 11) makes quantitative.
The public-sector accounting literature governs the recognition step. IPSAS 32 turns on control — who directs the service and holds the residual interest; Eurostat's ESA 2010 turns on risk — a PPP leaves the government balance sheet only if the private partner bears construction risk plus either availability or demand risk, with guarantees and majority construction financing among the reward factors that pull it back on (Eurostat and EPEC, 2016). Irwin (2007) set out how to value the government guarantees that so often decide the question. The recurring finding — that the treatment is gameable, and the measure a government most wants to look good on recognises the least — is the Recognition Rule (Paper 10) theme; the addition here is that concessional, "developmental" money is the easiest of all to leave off, so a blended commitment is a recognition-gap generator by construction.
The soft-budget-constraint literature supplies the incentive hazard. Kornai (1986), and Kornai, Maskin and Roland (2003), formalised how subsidies that rescue loss-making units erode the very effort and selection that would have avoided the loss — a mechanism since applied to aid, to white elephants, and (by Engel, Fischer and Galetovic, 2014) to PPP renegotiation itself. This paper places the same logic one stage earlier, at financial close: a facility that funds gaps without asking whether they are earned or manufactured softens the budget constraint on project preparation.
Set side by side, the five leave a clean space. Each owns a piece — the critique (subsidies are opaque and over-claimed), the mechanism (competition can reveal the number), the allocation (mis-allocation manufactures the gap), the accounting (the commitment must be recognised), the incentive (funding a gap rewards the failure that made it) — and none composes them into a single project-level discipline the sponsor of a hybrid PPP can run before committing concessional money. Minimise, size, recognise is that composition.
The Hybrid PPP Initiative is a good product missing an analytical layer, and the layer is specifiable. For the initiative to spend concessional money well rather than merely spend it, three things should sit at its core.
For the ministry, before it accepts the money:
For the World Bank Group and other blended-finance providers:
None of this is a brake on the initiative. It is what makes the initiative defensible: a hybrid PPP whose subsidy is minimised, gate-tested and recognised is a far stronger case for concessional support than one whose gap was assumed and whose commitment was left off the books.
This is a policy paper, not a new method: the optimisation it argues from is developed and implemented in Paper 1, the risk-allocation frontier in Paper 11, and the recognition framework in Paper 10; the contribution here is the discipline that composes them — minimise, size, recognise — and its application to the blended-finance moment. Three limitations are worth stating. The worked figure (VGF = 0% on the $620M concession) is a calibrated demonstration of the sizing engine on a real project's parameters, not a claim about any specific transaction in the World Bank's pipeline — and it is deliberately a zero-gap case, which makes the "output, not input" point most sharply but leaves the sizing of a positive minimal grant, though procedurally identical, not separately worked here. The claim that a large share of viability gaps are bankability gaps in disguise is an argued proposition, grounded in the risk-allocation evidence of the series, not a measured population statistic — it names the failure mode the discipline is built to catch, and a register of minimised-versus-assumed gaps would be the way to quantify it. And the recognition treatment is jurisdiction-specific: which standard binds, and how much of a blended commitment it puts on the balance sheet, varies by regime, as Paper 10 details.
The larger point stands across those caveats. The decision to grant a viability subsidy is the easy one, and the one the blended-finance moment has made comfortable. The fiscal question is the size — and the size is an output of analysis, not an input to it. Minimise before you fund, size the residual, recognise what remains. The World Bank is right that hybrid PPPs need viability gap funding. No one should write the cheque before someone computes the number.
Occasioning source
Blended finance and concessional subsidy
Optimal subsidy, auctions, and VGF design
Risk allocation and PPP incentives
Recognition and fiscal accounting
Soft budget constraint
Austral series and tooling