Public finances · reproducible analysis · cross-checked sources
Luxembourg borrows little, spends little, and keeps its AAA: debt at —% of GDP, among the lowest in Europe. And yet the European Commission rates its long-term sustainability risk “high” · the worst grade on the scale. How do both truths hold at once? That is the question this dashboard takes apart, piece by piece: the State budget line by line, the debt reread with Domar, Blanchard and Ostry, the European comparison · and the point where it all plays out, ageing. Every figure is recomputed from official sources. No black box, no agenda: enough to make up your own mind. 6 September 2026 edition, ten days before the Chamber of Commerce forum: a fifth part quantifies each reform lever, compares it with what ten European countries have done, reports the positions of the participants, and places an order-of-magnitude simulator in the reader’s hands.
Public debt stands at —% of GDP in — · among the lowest in the Union · but the coalition agreement sets a self-imposed ceiling of 30%. The projections diverge: the government sees it staying below the ceiling, the European Commission sees it crossing as early as 2027. The counter shows both.
—
The central-government budget (the State alone) is in deficit · — bn€ in —. But general government in the Maastricht sense (which adds municipalities and social security) was in surplus (+—% of GDP in —). The bridge between the two is the social-security surplus · a temporary demographic dividend, not a structural strength. That balance is what counts for the AAA and the 30% ceiling · and in 2025 the bridge no longer sufficed: general government itself moved into deficit (−2.0% of GDP).
The bridge, in points of GDP (—): central — + municipalities — + social security — = general government —. This social surplus shrinks as pensions climb · in —, the whole tips to —% of GDP. —
The internationally comparable view: social protection (—% of GDP, of which old age —) dominates, ahead of economic affairs, health and education. In total, Luxembourg spends less than the EU average; function by function, it spends more in half of them (social protection, education, economic affairs) and markedly less on health, defence and general services · while its GDP, inflated by cross-border workers and finance, mechanically understates all of its ratios (see method). Blue bars: Luxembourg; grey bars: EU-27 average.
The national detail, as enacted: of — bn€ in current and capital expenditure, health & social security, education and mobility take the lion's share. The investment (capital) share tells you how much the State is preparing the future rather than maintaining the present.
Excluding third-party accounts and financial operations (financing flows). Perimeter: central government.
of the national budget goes to public investment (capital expenditure) · the rest funds day-to-day operations.
Counter-intuitive: Luxembourg is not a low-tax country. Its levy (taxes + contributions) reaches —% of GDP, above the EU average (—%) · but it spends less, hence the near-balance. Its tax system leans heavily on direct taxes (income and corporate: —% of GDP vs — in the EU), which makes revenue cycle-sensitive · a risk the IMF flags.
Debt service weighs almost nothing · the comparison speaks for itself: —. That is the room for manoeuvre the others no longer have.
A debt's sustainability is read not from its level but from the r − g dynamics: the gap between the interest rate paid on the debt (r ≈ —%) and the economy's nominal growth (g ≈ —%). For Luxembourg, r − g ≈ — points: the primary balance that stabilises the debt comes out at —% of GDP · the country can run a primary deficit of up to ≈ —% of GDP without the ratio rising. Beyond that, debt climbs: the 2025 deficit (−2.0% of GDP) breaches the threshold · and that is exactly what the counter shows. Read with Domar and Blanchard, the 30% ceiling is thus more a choice of credibility than a solvency requirement.
r = interest burden over debt, same vintage (—); g = recent average nominal growth. r is an inherited rate (old coupons) · hence the sensitivity test below: the margin rests on assumptions, not on a theorem.
| Assumptions | r | g | stabilising balance |
|---|
—
Ostry & al. (IMF): fiscal space is the distance to the “debt limit”. With debt at 26%, a track record of balance and r < g, Luxembourg would hold it in abundance (it is not in the original sample) · which the AAA affirmed by all four agencies reflects.
Blanchard (2019) stresses it: r < g is probabilistic (rates can climb back) and does not erase long-term costs. Luxembourg's real constraint is not the level of debt · it is ageing (below).
Attributed paraphrases; full references in the method section. Reinhart-Rogoff's “90% threshold” (2010) is cited as a caution: it did not survive replication (Herndon-Ash-Pollin, 2013).
Here is why a 26% debt is rated “HIGH risk”. On unchanged policies, the cost of ageing rises from — to —% of GDP by 2070 · one of the steepest increases in the EU. Pensions alone gain +8.3 points of GDP (from — to —%) · the largest rise in the EU. The number of contributors per pensioner collapses, and the scheme · in surplus today · swings into deficit. The sustainability indicator S2 = — points (risk threshold: 6) measures the permanent effort required to stabilise debt over an infinite horizon.
Contributions − benefits. Positive today, it turns negative towards the late 2020s and reaches −8% of GDP in 2070. Distinct perimeters: COFOG “old age” (10.2%, 2023) covers all benefits; the Ageing Report's “pensions” (9.2%, 2022), public schemes; the FDC reserve, the general scheme only. —
The pension fund (FDC) weighs in at — bn€ = —× annual benefits (≈ —% of GDP) · more than the entire public debt. It buys roughly two decades of delay, with exhaustion projected —: a cushion of time, not a solution.
Three institutions, three readings of the same trajectory. The government sees debt staying below 30%, the Commission sees it crossing in 2027, and the CNFP puts the 2029 deficit at potentially double the government's (once the 2028 tax reform and the defence ramp-up are factored in). Their point of convergence: the current position is sound, but all of them recommend a spending rule and a pension reform.
—
Same country, same accounts, three trajectories. The gap owes nothing to the business cycle and everything to the underlying assumptions: the tax reform announced for 2028 (≈ €850m a year) and the defence ramp-up (up to +€880m by 2029) appear in the CNFP's path, not in the government's. The lesson in method is worth as much as the conclusion: a budget figure without its assumptions is not information.
| Reading | Deficit at the projection horizon | Debt vs the 30% ceiling | Key assumptions |
|---|
—
The question that gives its title to the Chamber of Commerce's back-to-school forum on 16 September 2026 · “is reform still possible?” · has a documentable answer. Here is the state of play, strictly separating what has been passed, what has been tabled and what has been announced. No position is taken here: only facts and institutional assessments are reported.
Sources : Chambre des Députés (dossiers 8634 et 8676), gouvernement.lu (communiqués du 18 décembre 2025 et du 8 juin 2026), European Commission (rapport pays, 3 juin 2026), FMI (Article IV, 30 juin 2026), CNFP (avril et mai 2026), Chambre de Commerce (avis budgétaire 2026 ; programme du forum du 16 septembre 2026). Programme du forum ↗ · PL 8676 ↗ · vote pensions ↗
Thirteen levers, three families: the pension parameters (six), the spending framework (three), revenue (four). Each line gives the mechanism, the effect quantified by an official or recognised source (IGSS, Commission, OECD, Court of Auditors, IDEA), who carries the effort and the status: passed, tabled, announced or merely assessed. Two levers point towards additional spending (single tax class, cut in corporate tax): they appear on the same footing, because a menu that showed only the savings would not be one. In the background, defence: from 2% of GNI in 2025 towards 5% in 2035 (of which 3.5% for defence in the strict sense), i.e. €2.4bn from 2029 under the Court of Auditors' linear assumption. The positions are reported, never taken as our own.
| Lever | Quantified effect | Who bears the effort | Status | Who recommends / opposes it |
|---|
—
The sliders combine the elasticities published by the IGSS (March 2025 contribution, 2022 technical review) and by the Commission (2024 Ageing Report, 2025 Debt Sustainability Monitor). One convention, and only one: one point of GDP of pension spending in 2070 is worth ≈ 0.7 point of S2, the ratio of two Commission figures (5.8 points of S2 attributed to pensions in the 2025 DSM, for +8.3 points of GDP of spending over 2022-2070 in the 2024 Ageing Report). It is an order-of-magnitude tool, linear by construction: it shows the size of the effort that separates the country from the “high risk” threshold, not how to distribute it.
—
The forum's question has an empirical answer elsewhere in Europe. Ten cases are selected for two reasons: a reform adopted and maintained, and an effect measured since by the OECD, the Commission or the national institution. A common thread: most enshrined the trajectory in law long in advance, five of them by linking it automatically to life expectancy (Sweden, Denmark, the Netherlands, Finland, Estonia); Austria kept to parametric adjustments. Belgium, for its part, passed its reform in 2026: the effect remains to be measured. Luxembourg is shown alongside, with the same indicators.
—
The 16 September programme is known: opening by Carlo Thelen at 9:45 a.m., two morning keynotes (“Why is it so hard to reform?”, Gilles Finchelstein; “Why is it urgent to reform?”, Anne-Sophie Alsif), the conversation with Lex Delles and Vincent Hein's panel at 2 p.m., Gilles Roth's address at 4:30 p.m., then “Reforming Europe” with Christophe Hansen and Thierry Breton. The first edition drew 350 participants. The speakers and the actors in the debate have already taken positions: here is what they said, in reported speech, dated and sourced, enough to gauge that same evening what will have shifted.
Read lever by lever, the apparent diversity of views narrows. On the spending rule and on pensions, the independent assessors (OECD, IMF, Commission, Court of Auditors) say the same thing, in different terms. The disagreements concentrate on indexation, where employer and employee organisations clash head-on, and on the single tax class, which the government supports and the Chamber of Commerce, IDEA and Moody's criticise. Where all converge, the debate is about timing and dosage, no longer about the principle.
—
Rather than a recommendation, a discipline of observation. These four indicators · all public, all revised on known dates · are enough to tell, year after year, whether the Luxembourg paradox is resolving or closing in.
Net primary expenditure 2025: +8.1% outturn against a 5.8% ceiling set by the EU Council · non-compliance found by the CNFP (13 May 2026) and the Commission (country report of 3 June 2026). The pension cash balance and its tipping year: IGSS technical review and Ageing Report. r − g and S2: recomputed/carried through here at every release.
The State alone, versus the whole of State + municipalities + social security (the Maastricht perimeter). Debt, the AAA and the 30% ceiling are judged on the latter; the enacted budget covers only the former.
The budget balance excluding interest. It is what enters debt dynamics: the “stabilising” primary balance is the one that keeps the debt-to-GDP ratio constant.
The gap between the interest rate paid on the debt (r) and nominal GDP growth (g). Negative, it melts the debt ratio away with no fiscal effort; positive, it mechanically adds to it (Domar 1944, Blanchard 2019).
The permanent fiscal effort (in points of GDP) required to bring debt down to 60% by 2070 (S1) or stabilise it over an infinite horizon (S2), ageing included. Beyond 6 points, the Commission rates the risk “high”.
The balance corrected for the economic cycle: it isolates discretionary policy from the ups and downs of the economy. Fragile in Luxembourg, where the output gap of a small, open financial economy is hard to estimate.
The distance between current debt and the limit beyond which a state's fiscal reaction no longer keeps pace with rising interest (Ostry & al., IMF). Luxembourg holds among the largest in the EU · and it is preserved, precisely, by not drawing on it too much.
Debt at —% of GDP, r < g, a marginal interest burden, a fourfold AAA: by the metrics of burden and solvency · debt, interest, rating · Luxembourg has no public-finance problem. The 30% ceiling is chosen discipline more than necessity: that is the reading Domar and Blanchard's arithmetic calls for.
By any metric of the stock of commitments, it is immense: +— points of GDP in ageing costs by 2070, an S2 of — points · a long-term risk among Europe's highest, for one of its lowest debts. The pension reserve buys time; it does not buy the solution.
Debt is only the delayed symptom of an unsettled question: who will pay for ageing, and from when? As long as that question stays open, the remaining social surplus masks an approaching deadline. The debate the institutions document is less about the 30% than about the spending rule and the parameters of the pension system · the point where OECD, IMF, Commission and CNFP converge.
State deficit (~— bn€) ≠ general-government balance: any serious debate starts by stating which balance is meant. Every figure on this dashboard comes from an official source, recomputed with no proprietary model; the economic ideas are attributed and referenced. This dashboard takes no position · it makes the debate falsifiable.
—