Chile built the best affordability envelope in the region and classified it — what the instrument measures, why the half that matters most is assumed rather than projected, and why disclosure is the reform
Author: David Duarte Arancibia, Austral Fiscal & Infrastructure Intelligence Date: September 2026 Audience: ministries of finance and budget offices; PPP units and infrastructure ministries; fiscal councils, supreme audit institutions and transparency councils; the IMF, the World Bank and regional development banks; academics in public finance Series: paper 20 of the Austral series on the fiscal risks of public-private partnerships. It is the direct-commitment counterpart to The Useful Silence, which covers the liabilities that are not recognised at all, and the companion to The PPP That Never Ends, which shows why the headroom this paper measures is less free than it looks. Disclosure and method. The author was head of the public-private partnership unit at Chile's Ministry of Finance and later head of the Public Finance Division at DIPRES, Chile's budget office; he is a co-developer of PFRAM, the IMF/World Bank PPP Fiscal Risk Assessment Model, and was the World Bank's lead on PPP fiscal-risk assessment. He is the founder of Austral, which sells fiscal-risk tools; readers should weigh that interest. This paper describes a Chilean instrument it does not reproduce. The author requested the current version under Chile's access-to-information law and it was refused (§4). He also holds a 2022 working copy from his time in government. No figure from that file appears here. Every quantity in this paper is either public and cited, or explicitly synthetic and generated by a committed script. §5 explains why a paper recommending disclosure cannot be built on the version the State has just withheld.
Every government running a PPP programme faces one question above all others: how much can we still commit? Answering it requires two projections side by side — what the sector's budget will be, and what current contracts have already spoken for. The difference is the affordability envelope, and it is the number that decides whether the next tranche of projects is prudent or reckless.
Almost nobody publishes it. That much this paper can establish, at least for the three largest concession programmes in South America. Whether governments build the envelope internally and simply never report it is a different question, and a harder one — but Chile turns out to be the case that answers it, because Chile does build it, and the instrument is not public.
It is called Gasto Sectorial — Compromisos Futuros, it is maintained by the concessions directorate and the budget office, it runs from 1995 to 2050 in inflation-indexed units, and it carries both halves in one place: the Ministry of Public Works budget line as availability, against current contracts, new contracts and new investment in existing contracts as commitment, disaggregated into fixed subsidies, expropriations, engineering and inspection, construction VAT and operating VAT. The author requested the current version under the access-to-information law. It was refused by decision of the Budget Director on 26 August 2026, under the trade-secrecy exception of Law 20.285, on the ground that the live file "incorporates estimates associated with contractual modifications and compensation for concession contracts that are, as of today, in the process of negotiation with the respective concessionaire," and that disclosure would harm "the negotiating position of the Treasury."
That reason is a good one, and the paper says so. It is also not the whole story, and separating the two is the contribution. One row of the instrument is legitimately reserved. The rest — the payment path of signed contracts, which consists of published supreme decrees, and the ministry's budget, which is published law — carries no exception at all. It is not withheld. It was simply never reported, because the instrument was built as a management model and no one ever designed the disclosure product that should sit on top of it.
The paper then argues that the instrument, admirable as it is, is asymmetric in a way that matters: the committed half is measured contract by contract with citations, while the availability half is projected at a constant assumed growth rate rather than anchored to the medium-term fiscal framework. Using a synthetic country calibrated to published Chilean magnitudes — because the real file cannot be used — it shows what that asymmetry does. Holding the portfolio of signed contracts exactly fixed and varying only the assumption about the budget ceiling, headroom in 2055 ranges from 111 to 297 million units — a factor of 2.7. Expressed as a percentage of the budget the same table varies by only 7 to 9 points, which is why envelopes reported in percentages conceal precisely the fragility that matters.
A second finding concerns VAT, and it is a distinction ministries routinely collapse. Operating VAT dominates the commitment — 66% of the sectoral budget claim in the synthetic case — and it is fiscally neutral, since the ministry pays it and the treasury collects it. But the ministry's budget line must appropriate it regardless, where it competes with everything else in the portfolio. Reporting only the gross figure overstates fiscal pressure; reporting only the net conceals the real budget constraint. Both columns belong in the table, because they answer the questions of two different ministers.
Finally, the regional comparison inverts the paper's own starting assumption. Colombia publishes more than Chile on this dimension: an annual APP quota expressed as a share of GDP on a declining path from 0.700% in 2026 to 0.300% from 2045, with a year-by-year table by sector to 2056 and project-level programmed spending, all in the Medium-Term Fiscal Framework. Peru caps the accumulated stock of firm and quantifiable contingent commitments, net of revenue, at 12% of GDP. The three ask different questions — Peru "how much have we used?", Colombia "how much may I authorise this year?", Chile "how much space remains and when does it open?" — and Chile's is the right question and the only one that cannot be answered from outside the State.
The recommendation follows, and it is deliberately unambitious. Do not publish the model: it is re-estimated several times a year and a snapshot of a live model is not transparency. Publish a standardised annual extract at a fixed date, with stable definitions, alongside the budget — the payment path of signed contracts, the negotiation provisions collapsed into a single aggregate or released as each process closes, the availability side anchored to the medium-term framework and shown as scenarios rather than a line, and both gross and net of VAT. Chile is closer to this than almost any country. What it lacks is the last step, the one that turns a management tool into a public account, and that step is cheap.
Keywords: public-private partnerships · affordability · fiscal space · budget commitments · fiscal transparency · access to information · contingent liabilities · medium-term fiscal framework · vigencias futuras · Chile · Colombia · Peru.
Feedback & discussion. Published openly for challenge. Comment at
https://austral-intelligence.com/research; email [email protected].
A minister of public works has a pipeline. A minister of finance has a constraint. The meeting between them turns on a single number: how much can still be committed without pre-spending a budget that future governments will need for something else?
The question is harder than it sounds, because PPP commitments are long and budgets are short. A concession contract runs twenty-five to fifty years. A medium-term fiscal framework runs three to five. Committing to a stream of availability payments in 2027 places a claim on the budgets of 2040 and 2050, which no current framework projects and no current government will administer.
This is not a new observation. The author made it in 2015, in a master's thesis on the fiscal management of PPP contingent liabilities, in one sentence that has aged well:
"A PPP contract lasts, most of the time, more than 20 years; however, the medium-term fiscal framework does not go beyond 4 years."
Eleven years later that gap is still the central unsolved problem of PPP affordability, and this paper is an attempt to say precisely what it would take to close it.
The series has approached the surrounding terrain from several directions. The Useful Silence argues that governments have a standing incentive not to recognise PPP liabilities at all, because a fiscal rule written on a measured aggregate rewards any contract form that escapes the measure. The Pipeline Under the Ceiling shows what happens to a budget when committed payments accumulate under a spending rule. This paper takes the case where the government is not hiding anything — where the commitments are recognised, measured and projected competently — and asks whether the resulting number can be seen by anyone outside the building.
In the best-instrumented country in the region, the answer is no.
The term is used loosely, so it is worth fixing.
An affordability envelope is the difference, year by year and over the full horizon of the longest contract, between two projections:
Headroom is the difference. It is not a stock and it is not a ratio; it is a path. Its shape matters more than its level, because the practical question is never "is the programme affordable" in the abstract but "how much can be signed this year, and what does that do to 2040."
Three things distinguish an envelope from the instruments more commonly found.
It is not a debt ratio. Measuring PPP commitments against GDP tells a finance ministry whether it has a solvency problem. It does not tell a works ministry whether it can tender three hospitals.
It is not a cap. A ceiling — on the stock of commitments, or on annual authorisations — is a discipline device. It tells you when you have hit the wall. It does not tell you how far away the wall is, or when it moves.
And it is not a contingent-liability report. Those value what might be paid under guarantees. An envelope is mostly about what will be paid under signed contracts, which is a larger number and a more certain one.
The apparatus for managing PPP fiscal risk is well developed, and this paper is not claiming a void where there is a body of work. Irwin's Government Guarantees (World Bank, 2007) set the template for valuing risk in privately financed infrastructure. PFRAM, the PPP Fiscal Risk Assessment Model built by the IMF and the World Bank, projects a project's effect on the deficit, on debt and on the stock of contingent liabilities, with sensitivity analysis on macro and project parameters. The World Bank's fiscal commitments and contingent liabilities (FCCL) framework runs from identification and analysis through fiscal-impact assessment, budgeting and monitoring. The IMF Fiscal Transparency Code devotes a full pillar to fiscal risk and treats as best practice the identification, quantification and annual disclosure of guarantees and of total obligations under PPP contracts. For Latin America specifically, Reyes-Tagle's work at the Inter-American Development Bank surveys the fiscal impact of the region's programmes.
Two of those the author helped build, which is a reason to be careful rather than a reason to be quiet.
Set against that apparatus, the gap this paper identifies is specific and, once stated, hard to unsee.
These frameworks are project-facing and stock-facing. The envelope is portfolio-facing and flow-facing. PFRAM assesses a project. FCCL asks whether a commitment is affordable at the point of approval — affordability appears in it as a gate, applied case by case. The Transparency Code asks whether the resulting obligations are disclosed, and the natural artefact is a stock: total obligations, exposure as a share of GDP.
None of them asks the government to state, and publish, how much room is left, and in which years. A ministry can pass every one of these tests — sound project appraisal, a well-run affordability gate, complete annual disclosure of obligations, a competent contingent-liability model — and still be unable to answer the question a minister actually asks before authorising the next tranche.
That is the gap. It is not a failure of any of these instruments on their own terms; it is a product that nobody's terms currently require.
Chile publishes a contingent-liability report of high quality; its Informe de Pasivos Contingentes models minimum-revenue guarantees with a stochastic traffic model developed with World Bank technical assistance. That report is not an envelope and does not claim to be. The envelope is a different instrument, and Chile has one.
The model is called Gasto Sectorial — Compromisos Futuros. Its name and its existence are matters of public record because the Budget Director named it in a formal decision (§4). It is maintained for the concessions directorate and the budget office, it is an electronic spreadsheet with summary and detail sheets, and it is updated several times a year.
Its architecture — as distinct from its contents — is this. The horizon runs from 1995 to 2050, in unidades de fomento, Chile's inflation-indexed unit of account. The summary sheet carries, as rows:
| row | what it measures | |
|---|---|---|
| — | Ministry of Public Works budget | availability: the ceiling of the sectoral appropriation |
| I | operating subtitles (personnel, goods and services, investment complement) | running costs |
| II | current contracts | commitment under signed contracts |
| · current and pending commitments | fixed subsidies to construction and to operation | |
| · expropriations | ||
| · investment complement (studies, engineering and inspection) | ||
| · construction VAT / operating VAT | fiscally neutral, binding on the budget line | |
| III | new contracts | the pipeline: what is planned for tender |
| IV | new investment in concessioned contracts | expansions to contracts already in force |
| — | (reserved) | provisions for contract modifications under negotiation and for dispute resolution |
Headroom is availability minus the sum of I, II, III, IV and the provisions — which is why the reserved row is not a footnote but a term in the subtraction, and why the file cannot simply be released with one column blanked and the totals left standing.
Two features deserve emphasis, because they are rarer than they should be.
It projects fifty-six years. Not the three of a medium-term framework, and not the ten of a long-term sustainability annex. It runs to the horizon of the longest contract, which is the only horizon at which the question makes sense.
It disaggregates by concept, not just by project. Separating operating VAT from construction VAT from expropriations from fixed subsidies is what allows a user to distinguish claims that are fiscally real from claims that are budgetary but neutral — the distinction §8 turns on, and one that most reporting collapses.
Chile, in short, has built the instrument this series has been asking for — and it is not among the things it publishes.
On 29 July 2026 the author filed a request under Law 20.285, Chile's access-to-information statute, for the current version of the file, its detail by contract, its methodological assumptions, and its update frequency. The request was routed from the concessions directorate to the budget office.
It was refused on 26 August 2026, by Official Letter 2049 of the Budget Directorate, signed by the Budget Director. The operative reasoning, in the decision's own words:
"The instrument indicated, in its current version, incorporates estimates associated with contractual modifications and compensation corresponding to concession contracts that are, as of today, in the process of negotiation with the respective concessionaire. In relation to that specific information, this Directorate configures the ground of reserve or secrecy provided in Article 21 No. 4 of Law No. 20.285, inasmuch as its disclosure would entail an eventual prejudice to the negotiating position of the Treasury and the consequent impairment of the country's economic or commercial interest in those processes. Accordingly, it is not appropriate to deliver the file 'Gasto Sectorial — Compromisos Futuros' in its internal working version."
This is a serious and defensible reason, and it should be recognised as such rather than treated as evasion. If a government has estimated what it expects to pay to settle a live renegotiation, publishing that estimate hands the counterparty the State's reservation price. Any negotiator would say the same. The exception exists for exactly this.
The decision then does something the paper takes seriously in the opposite direction. Applying the divisibility principle of Article 11(e), which requires that a document not be withheld in full when only part is reserved, it points the requester to where the non-reserved information can be found: "the commitments and amounts recorded in the tender specifications and in the respective supreme decrees of award and/or modification of each concession contract in force," on the concessions directorate's website.
That is an accurate statement of where the underlying data live. It is also, read carefully, an admission with consequences. The State is saying that the commitment side of the envelope is public information — distributed across thousands of decrees, but public. What is not available is the aggregation. The number does not exist outside the building not because it is secret, but because nobody publishes the sum of public parts.
Three further checks establish that the instrument appears in no routine publication.
So the position is precise: one row of the instrument is legitimately reserved; the rest is not withheld, it is simply unreported. Those are different failures with different remedies, and conflating them has kept the discussion stuck.
The author holds a 2022 working copy of the file from his time in government. No figure from it appears in this paper.
The reasons are worth stating, because the temptation runs the other way and a reader is entitled to know the constraint under which the argument was built.
First, and decisively: a paper recommending that a State publish an instrument cannot be built on leaking the version that State has just declined to publish. Whatever the legal position of a four-year-old copy, the argument of §11 would not survive its own method.
Second, the reserved ground is specific and the prudent line is not to litigate its edges using the file itself.
Third — and this is the part that makes the constraint bearable — the architecture is not what is reserved. Row names reveal no negotiating position. And the paper's actual claims are structural: that the committed half is measured while the availability half is assumed; that VAT dominates the budget claim while being fiscally neutral; that headroom is highly sensitive to a ceiling assumption. None of those requires the file's values. They require a model that reproduces the mechanism, which §7 supplies and commits as a script.
Where public Chilean magnitudes exist, they are used and cited: the concessions directorate reports a committed investment stock of USD 32,598 million at mid-2025 across 80 active contracts, and the contingent liabilities report gives maximum minimum-revenue-guarantee exposure of about 0.87% of GDP with a present value of expected payments an order of magnitude smaller. Those anchor the synthetic country's scale. They do not substitute for the envelope, because they are not it.
Here is the paper's central analytical claim about the instrument, and it is a criticism made from inside a broadly admiring view.
The two halves of the envelope are not built to the same standard.
The committed half is excellent. Each line traces to a signed contract, a published award decree or a tender specification. It is disaggregated by concept. It runs the full horizon. It is, in the strict sense, measured.
The availability half is a projection of the sectoral budget at an assumed constant growth rate. It is not anchored to the medium-term fiscal framework, not derived from the structural balance rule, not presented as a set of scenarios, and not reconciled with any published fiscal path. It is, in the strict sense, assumed.
That asymmetry would be defensible if the sectoral budget were a stable series. It is not. Over the executed period the Chilean public works budget has moved between substantial annual contractions and substantial annual expansions — the historical record contains year-on-year changes of well over ten per cent in both directions. A constant growth rate is not a description of that series; it is a placeholder standing where a projection should be.
And because headroom is a difference, an error in the ceiling passes through undamped. If the commitment path is right to the peso and the availability path is wrong by a percentage point of annual growth compounded over thirty years, the headroom is wrong by a multiple, not by a margin.
This is a fixable defect, and it is the paper's main technical recommendation. The medium-term framework exists. Its scenarios are published. Anchoring the availability side to them — and showing headroom as a fan rather than a line — is an afternoon's work for the team that already maintains the model.
To show what the asymmetry does without using the withheld file, this section builds a synthetic country. Every parameter below is illustrative. They are calibrated to the order of magnitude of published Chilean sources so that the conclusions are credible, not so that they reproduce Chile. The generating script is committed (Appendix A).
The model runs 2026–2055. Each contract carries two distinct flows, and keeping them separate is essential to §8: the direct payment the State makes, and the operating revenue base on which the State pays VAT. Tolled road concessions pay little subsidy and carry a large revenue base; social infrastructure — hospitals, prisons — is the reverse, with the State paying nearly everything and no user revenue at all. The portfolio mixes both, with first-generation contracts expiring on a staggered schedule and two re-tendered contracts running long.
Headroom in millions of units available to commit. The portfolio of signed contracts is identical across all four rows. The only thing that changes is the rule used for availability.
| scenario | real growth | 2030 | 2040 | 2050 | 2055 |
|---|---|---|---|---|---|
| flat 3.0% — a constant assumed rate | 3.0% | 88.5 | 133.3 | 188.3 | 220.7 |
| anchored to trend growth | 1.8% | 83.3 | 110.4 | 138.5 | 152.8 |
| fiscal consolidation | 0.8% | 79.2 | 93.8 | 106.1 | 111.0 |
| sustained expansion | 4.0% | 92.9 | 155.2 | 241.3 | 296.9 |
In 2040 headroom runs from 94 to 155 — a factor of 1.7. In 2055, from 111 to 297 — a factor of 2.7.
And here is the presentational point that matters as much as the result. Expressed as a percentage of the sectoral budget, that same table varies by only seven to nine points, because numerator and denominator move together. In absolute units — which is what a ministry can actually commit — the assumption doubles or fails to double the available space.
An envelope reported in percentages therefore conceals exactly the fragility a reader needs to see. That is not a hypothetical criticism: percentage-of-budget and percentage-of-GDP are the standard presentations, and both have this property.
The second finding is a distinction that ministries collapse constantly, in both directions.
In the synthetic portfolio, VAT — overwhelmingly operating VAT — is 66.3% of what the sectoral budget must appropriate over 2026–2055, rising as direct subsidy payments run off.
That share is not a free parameter, and the reason matters for whether the model can be trusted. It is not set anywhere in the script; it falls out of the portfolio's composition — the mix of tolled and social assets, and the ratio of each contract's revenue base to its direct subsidy. A reader who thinks the share is too high can change those inputs and re-run. The claim the paper makes is structural, not numerical: wherever a programme is dominated by user-financed assets, operating VAT will be the largest single claim on the sectoral budget line, because it accrues on concession revenue rather than on State payments.
The reason is worth spelling out. Operating VAT accrues on the revenue of the concession — the whole toll take, over decades — not on the State's payments to the concessionaire. A programme dominated by user-financed roads will therefore generate a VAT claim on the sectoral budget far larger than its subsidy claim.
Now the distinction:
Reporting only the gross figure overstates fiscal pressure. Reporting only the net conceals the budget constraint the sectoral minister actually faces. Neither is a rounding difference: the two columns differ by a factor of roughly three.
So the envelope needs both, and they should be labelled by whose question they answer. The gross column is the public works minister's. The net column is the finance minister's. An instrument that reports only one has silently chosen a side in an argument the two of them are having.
This paper began from the assumption that Chile has the instrument and the neighbours do not. On the disclosure dimension that assumption is false, and correcting it changes the recommendation from an abstraction into a copy-and-paste.
Under Article 26 of Law 1508 of 2012, the national planning council, on the prior opinion of the fiscal policy council, sets each year the annual limit of authorisations to commit future budget years for PPP projects, alongside the primary surplus target. Those authorisations — vigencias futuras — are then reported in the Medium-Term Fiscal Framework.
The 2025 framework authorised COP 175.9 trillion of future-year commitments for 2026–2054, of which 40.8% are PPP — some COP 71.7 trillion. The fiscal policy council extended PPP authorisations to 2056 on 13 June 2025.
And the disclosure is granular in exactly the way an envelope requires. The PPP quota is set as a share of GDP on a declining path — 0.700% in 2026 falling to 0.300% from 2045 — with allocation by sector, published as a year-by-year table to 2056, accompanied by project-level programmed spending for 26 transport projects across 2026–2053.
That is not a cap alone. It is a cap with a path, published annually, at project level.
Peru's Law 32441, published on 16 September 2025 and in force from 25 December 2025, provides that the accumulated stock of firm and quantifiable contingent commitments, net of revenue, of the non-financial public sector under PPP contracts may not exceed 12% of GDP, reviewable every three years by supreme decree. The same law permits terms of up to 60 years.
A stock cap is a solvency guard. It is not a path, and it does not answer the sequencing question.
| Chile | Colombia | Peru | |
|---|---|---|---|
| Annual path of commitments | 1995–2050, by concept | 2026–2056, by sector and project | — |
| Ceiling | implicit (sectoral budget) | PPP quota, 0.7% → 0.3% of GDP | stock, 12% of GDP |
| Projected budget availability | yes, to 2050 | framework horizon | — |
| Published? | No | Yes, annually | Yes (the cap) |
| Maximum contract term | 50 years | — | 60 years |
The three are answering different questions, and the differences are not degrees of quality:
Chile has the best model and the worst disclosure. Colombia has the disclosure. That is a more uncomfortable finding than the one this paper set out to report, and a more useful one.
One qualification, from the companion paper, is important enough to state here because it bears directly on how the envelope should be read.
The PPP That Never Ends shows that under Article 25 of Chile's concessions law, when a concession expires the works must be granted in concession again; reversion to the State is an exception requiring a reasoned presidential decree. Across eighteen Chilean assets concessioned more than once, the successor contract carries a median of 3.25 times the original official budget, and 3.07 times per kilometre of the same corridor.
The consequence for this paper is direct. The headroom that an envelope shows opening as contracts run off is, in Chile, legally pre-committed to be re-tendered on the same asset. It is not free space for new projects. A ministry reading the declining commitment path as room to commit elsewhere is reading a number the statute has already spent.
There is a second, technical consequence. Most recent Chilean second-round concessions are awarded by least-present-value-of-revenue auction, in which the contract ends when the bid revenue is collected rather than on a date. For those contracts the run-off date in the envelope is not a date at all; it is a function of realised traffic. An envelope that projects them to their statutory maximum is using a worst case as a point estimate.
Neither point is a criticism of the Chilean instrument, which handles signed contracts properly. Both are arguments that an envelope needs a re-tendering assumption stated on its face — and that leaving it implicit produces a headroom number that is systematically too generous.
The obvious recommendation — "publish the spreadsheet" — is wrong, and the reason is worth being precise about.
The file is re-estimated several times a year. Provisions are updated as each negotiation advances, contract modifications enter, tenders are awarded. A model that is continuously re-estimated cannot function as a transparency artefact: a copy obtained today is stale before it can be analysed, and no reader could replicate or verify it against anything.
So the gap is not a missing version. It is a missing product: a standardised extract, at a fixed date, with stable definitions. That is what turns a management model into a public account, and it is what Colombia already does.
Four design decisions:
1. An annual extract at a fixed date, published with the budget. The payment path of contracts in force, year by year, to the horizon of the longest contract, disaggregated by concept. This carries no ground of reserve: signed contracts and published decrees. It is the largest part of the instrument and the least contentious.
2. Negotiation provisions collapsed into a single aggregate, or released as each process closes. The Article 21(4) exception protects one row; today it buries the other ten. A single total — or publication with a lag tied to the close of each negotiation — delivers the envelope without exposing any position against any counterparty. This is the one recommendation that requires the State to accept a small cost, and it is small.
3. The availability side anchored to the medium-term framework and published as scenarios. Not a constant assumed rate, and not a single line. §7 shows why: it is the half that cannot be measured contract by contract, so it should be presented with its uncertainty visible rather than hidden in a growth assumption.
4. Gross and net of VAT, in adjacent columns, labelled by whose question each answers.
Chile is far closer to this than almost any country. It has the model, the horizon, the disaggregation and the institutional home. What it lacks is the last step — and that step costs a template and a decision, not a system.
The central instrument is described, not shown. §5 explains why. A reader wanting to verify the architecture has three independent routes: the Budget Directorate's own decision names the file and its reserved content; the divisibility ruling states where the underlying commitment data are published; and the synthetic model in §7 is committed and re-runnable. None of that is the same as inspecting the file, and the paper does not pretend otherwise.
The quantitative results are synthetic. The 2.7× sensitivity and the 66% VAT share are properties of a model calibrated to published magnitudes, not measurements of Chile. They demonstrate a mechanism. Anyone asserting them as Chilean facts would be misreading the paper, and §7 says so at the top.
The claim that the availability side is a constant assumed rate is a structural description, made from the architecture rather than demonstrated with the file's values. It is the paper's most consequential factual claim about the instrument and the one most worth independent confirmation — most directly by asking the Budget Directorate to state the projection convention, which involves no reserved information whatever.
The Colombian and Peruvian comparisons rest on published frameworks and secondary reporting of them, not on replication of either country's underlying models. The instruments are characterised at the level of design — horizon, granularity, whether a path is published — which is the level the argument needs, but a specialist in either country would add nuance this paper does not.
And the argument is Chile-anchored. It is a proof of existence — that a complete envelope has been built and can be built — plus a comparison. It is not a survey. Whether other programmes maintain equivalent instruments internally is an open question, and one that could be answered with a short questionnaire to PPP units.
For finance ministries and budget offices.
For PPP units and infrastructure ministries.
For transparency councils and supreme audit institutions.
For standard-setters and international financial institutions.
The synthetic model. costa_andina.py generates every quantity in §7 and §8 and writes data/costa_andina_envolvente.csv. Four availability scenarios (constant 3.0%, trend-anchored 1.8%, consolidation 0.8%, expansion 4.0%) over 2026–2055, applied to an identical portfolio of eight stylised contracts, each carrying a direct payment stream and, where applicable, an operating revenue base on which VAT accrues. Negotiation provisions are modelled as a decaying fraction of the current-contract commitment, to show their weight without asserting their level. All parameters are illustrative.
Chilean public sources used. Concessions directorate quarterly report, second quarter 2025 (committed investment stock of USD 32,598 million; 80 contracts in force at 30 June 2025) and its spreadsheet annexes through the first quarter of 2026, verified to contain operational statistics only. Budget Directorate, Informe de Pasivos Contingentes 2025 (minimum-revenue-guarantee exposure; expected-payment present values; the statement that as concessions extinguish "space is freed to incorporate the mechanism in new contracts"). Duarte Arancibia, David, Pasivos contingentes en contratos de asociación público-privada, master's thesis, 2015, §7.3, for the horizon-mismatch quotation.
The access-to-information record. Request AE002T0003996, filed 29 July 2026 with the concessions directorate and routed to the Budget Directorate. Refused by Official Letter 2049 of the Budget Directorate, 26 August 2026, signed by the Budget Director, under Article 21 No. 4 of Law 20.285, with a divisibility referral under Article 11(e). Quotations in §4 are translated from the decision; the Spanish original is on file.
Colombia. Law 1508 of 2012, Article 26. Medium-Term Fiscal Framework 2025 (COP 175.9 trillion of future-year authorisations for 2026–2054, of which 40.8% PPP); extension of PPP authorisations to 2056 by the fiscal policy council, 13 June 2025; PPP quota expressed as a share of GDP on a declining path from 0.700% (2026) to 0.300% (from 2045), published by sector to 2056 with project-level programming for 26 transport projects, 2026–2053. These figures are drawn from the published framework and from specialist secondary reporting of it, not from replication of the underlying model.
Peru. Law 32441, published 16 September 2025, in force from 25 December 2025 with Supreme Decree 316-2025-EF; cap of 12% of GDP on the accumulated stock of firm and quantifiable contingent commitments net of revenue for the non-financial public sector; maximum term 60 years.
Literature and frameworks. Irwin, Timothy, Government Guarantees: Allocating and Valuing Risk in Privately Financed Infrastructure Projects, World Bank, 2007. International Monetary Fund and World Bank, PPP Fiscal Risk Assessment Model (PFRAM) 2.0 User Manual, September 2019. World Bank, fiscal commitments and contingent liabilities (FCCL) framework — guidelines and technical manual, whose components run from identification and analysis through fiscal-impact assessment, budgeting for affordability and monitoring. International Monetary Fund, Fiscal Transparency Code (Pillar III, fiscal risk analysis and management) and Fiscal Transparency Handbook, 2018. Reyes-Tagle, Gerardo (ed.), Fiscal Impact of PPPs in Latin America and the Caribbean, Inter-American Development Bank. These are characterised at the level of what each instrument asks of a government, from published descriptions of scope; the paper does not replicate any of their methodologies.
Companion paper. The Article 25 and least-present-value-of-revenue findings in §10 are established in The PPP That Never Ends (Austral series, paper 22), with their own committed artifacts.
What is deliberately absent. No value from the author's 2022 working copy of Gasto Sectorial — Compromisos Futuros appears anywhere in this paper, in the scripts, or in the committed data. See §5.
Affordability envelope — the year-by-year difference between projected budget availability and commitments already made, over the horizon of the longest contract.
Article 21 No. 4 (Law 20.285) — the ground under Chile's access-to-information law permitting reserve where disclosure would harm the economic or commercial interest of the country.
Article 11(e) (Law 20.285) — the divisibility principle: a document may not be withheld in full where only part of it is reserved.
Availability payment — a payment to a concessionaire for making an asset available, independent of usage.
Headroom — availability minus commitment; the space remaining to commit new projects.
LPVR (least present value of revenue) — an auction design in which the concession ends when a bid present value of revenue has been collected, producing a variable term.
Medium-term fiscal framework (MTFF) — a government's multi-year projection of revenue, expenditure and balance, typically three to five years.
Minimum revenue guarantee (MRG / IMG) — a guarantee that the State will top up concession revenue to a floor.
UF (unidad de fomento) — Chilean unit of account indexed to consumer prices.
Vigencias futuras — Colombian authorisations to commit future budget years; the instrument through which PPP commitments are approved and reported.