Which balance sheet is your PPP on? — The Infra-Fiscal Radar, Issue 07

Lithuania · inside the line

Eurostat put €1.5 billion of military PPPs on the government’s books.

Ex-ante advice of 28 July: four PPP contracts for two new military bases (Rūdninkai, Kairiai), €1,532 million, availability-paid, with the EIB lending €540 million to the private vehicles. The same contract model as the three 2021 base contracts (€98.2 million) that Eurostat had accepted off balance sheet. This time: on.

€1.53 bn

recorded on the government balance sheet — the same contract model sat off it at €98 million in 2021. Scale and context changed the answer, not the clause

The reasoning is the perimeter argument in one page. War is a force-majeure event; the partner may terminate and be compensated in full; the ministry decides what is built, who uses it and when it is exposed. In Eurostat’s words: “a financing mechanism rather than a genuine risk-sharing partnership” — “government investment with private financing, not private investment with government payments.”

Source · Lithuania · Eurostat ex-ante advice, statistical treatment of PPP contracts for military bases, 28 Jul 2026

Mexico · one document, two perimeters

The risk chapter says 96.6 billion. The annex says 3,166.9.

The 2027 economic criteria (8 September) price the State’s contingent exposure to CFE’s Pidiregas projects at about 96.6 billion pesos over some twenty years — “limited risk”. Twenty-three pages later the annex lists the same programme’s conditional Pidiregas — capacity and energy charges owed to independent producers — at 3,166.9 billion pesos contracted, 1,380.8 billion still to pay from 2027.

96.6 vs 3,166.9

billion pesos — the risk chapter’s exposure and the annex’s commitment stock, same programme. Both are true. They measure different perimeters

Neither number is wrong. The first asks what does the Treasury pay if CFE defaults and the assets are worth nothing; the second asks what has CFE, wholly state-owned, already promised to pay. The document answers both — in different chapters, on different pages, with no line between them.

Source · Mexico · SHCP, Criterios Generales de Política Económica 2027, 8 Sep 2026, pp. 51 & 74

DR Congo · the zero in the stress test

Fourteen per cent of GDP in contingent liabilities. PPPs: 0.0.

The 2026 Article IV, published 8 September, stresses the debt path with a tailored contingent-liability shock of 14.0% of GDP: 7.0 for incomplete SOE coverage, 2.0 for public institutions that do not report, 5.0 for the financial sector — and 0.0 for PPPs, where the framework’s default is 35% of the PPP stock.

0.0

of GDP for PPPs inside a 14%-of-GDP stress test — not because the risk is nil, but because there is no measured stock for the 35% to apply to

The same report carries a structural benchmark to rewrite the PPP law and hand PPP fiscal risk to the Ministry of Budget, and names Inga 3 — the flagship hydro project — as “a potential source of fiscal and debt risks” with “complex and partially non-concessional financing”. The largest project in the pipeline is in the text. In the stress test it weighs zero.

Source · DR Congo · IMF Country Report 2026/246, 2026 Article IV, DSA Text Table 1, 8 Sep 2026

Mauritius · the anchor and the rule

The debt limit went from 60% to 80%. Debt is at 88.5%.

IMF fiscal-responsibility advice, published 16 September: the 2008 Public Debt Management Act set a 60% ceiling on public-sector debt; it was revised in 2010, 2017, 2020 and to 80% in 2024 — and “breached repeatedly since its introduction”. Public-sector debt: 88.5% of GDP at the end of FY24/25.

60 → 80 → 88.5

per cent of GDP: the original limit, the revised limit, the outturn. A ceiling that moves to meet the debt is not a ceiling

The mismatch is in the perimeters. The anchor covers the whole public sector; the operational rule covers only budgetary central government — leaving 156 extrabudgetary units, about a quarter of central-government spending, outside the rule that steers the budget. Guarantees stand at 4.3% of GDP; two have already been called.

Source · Mauritius · IMF TA Report 2026/074, Fiscal Responsibility Legislation, 16 Sep 2026, ¶4, ¶20–21

Europe · the second table

Where the perimeter is published, the off-balance number is small.

Every January Eurostat publishes what sits outside general government. Liabilities of public corporations: Germany 84.4% of GDP, Netherlands 73.1%, Luxembourg 65%, France 61.9% — mostly public banks. Off-balance PPPs: below 2% of GDP in every member state, Portugal highest at 1.2%, nine countries at zero.

84.4% vs 1.2%

of GDP — the largest public-corporation stock outside the perimeter, and the largest off-balance PPP stock. The line is published, so hiding behind it buys little

And as of this week the perimeter becomes a rating input for low-income countries too: the IMF–World Bank review of the LIC debt framework proposes a debt-data confidence flag on “whether the full public sector perimeter is captured”. Read the two facts together: where the second table exists, the second number is small.

Source · Eurostat · Contingent liabilities, 2024 data, Jan 2026 · IMF/WB LIC-DSF Review, 21 Sep 2026

The through-line

The anchor is only as honest as its perimeter.

Pulled inside in Vilnius, split across two chapters in Mexico City, carried at zero in Kinshasa, outrun in Port Louis — and published in full, every January, in Brussels. Same liabilities, five lines. Every one of them was drawn by someone; none of them was drawn by the contract:

VfMvalue for money

Compare procurement routes on the same perimeter, or the comparison is the loophole.

FAROPPP fiscal risk

The commitment stock of every signed contract — inside the risk statement, before the stress test.

GFSaccrual accounting

Substance over form, contract by contract: on the balance sheet when the State carries the risk.

Thursday: The Anchor and the Perimeter — what a debt anchor measures, what walks around it, and what the same rule says when it is run on the whole balance sheet.