France’s 2027 budget: the price of public debt

France can increase its tax burden, reduce its deficit and still accumulate more debt relative to GDP. Bond-market risk emerges when that trajectory raises the price of future borrowing, reduces the value of outstanding securities and puts pressure on investors who finance those securities with leverage. The budget and the market operate on different clocks.

As of 22 September 2026, France’s 2027 budget remains in preparation. Its presentation to the Council of Ministers is scheduled for 1 October, according to AFP’s report. This analysis examines the preliminary fiscal framework and its financial transmission. S01

A higher tax burden, a smaller deficit and a larger debt ratio

The government framework reported on 19 September combines three changes between 2026 and 2027: taxes and compulsory social contributions would rise from 43.9% to 44.2% of GDP, the public deficit would fall from 5.4% to 5.0%, and public debt would increase from 119.3% to 121.7%. Public expenditure is projected at 56.9% of GDP, down 0.2 percentage points. These are projections, not observed outcomes. S01

There is no accounting contradiction. The deficit measures an annual financing shortfall. Debt is an accumulated stock. Reducing the former does not eliminate the latter. The debt-to-GDP ratio also depends on its denominator: the monetary value of national output.

A higher ratio of taxes and social contributions to GDP can reflect changes to tax bases, exemptions or revenue performance. It does not by itself describe statutory tax rates or each taxpayer’s bill.

The finance ministry attributes the projected increase to reduced tax reliefs and anti-fraud measures. S01 A creditor’s concern is the net revenue, its persistence and when it can actually be collected. An anticipated recovery from enforcement should be distinguished from a well-documented recurring tax base.

Taxes and compulsory social contributions are not total public revenue either. Insee defines them net of uncollectable amounts and tax credits. S19 Subtracting 44.2% from 56.9% therefore does not reconstruct the deficit: it omits other revenue and differences in accounting scope.

The announced €54 billion fiscal effort needs a baseline too. Localtis reports that it includes €45 billion of new measures and €9 billion carried over from previous decisions. An improvement relative to an unchanged-policy path is not necessarily an equivalent year-on-year fall in expenditure. S02

Consider spending expected to increase from 100 to 106. If it reaches 103 instead, spending has increased by 3 and has also been cut by 3 relative to the no-measures scenario. Both statements are true. Assessing the budget therefore requires the baseline, the recurring measures and the collection schedule. Expected revenue and durable revenue are not interchangeable.

Debt can drift upwards without another rise in yields

Assume opening debt of 119.3% of GDP, a total deficit of 5% of GDP and nominal GDP growth of 2.5%. That growth rate is an educational assumption, not an official forecast attributed to the government. With no stock-flow adjustment:

Growth reduces the weight of inherited debt by about 2.91 GDP percentage points. The deficit adds 5 points. The debt ratio therefore increases by 2.09 points. Neither a bond-market panic nor a missed payment is required to produce that result.

The total deficit that would stabilise debt under these assumptions is 2.91% of GDP. At nominal growth of only 1.5%, the stabilising deficit falls to 1.76%. There is no universal accounting threshold at 3%. The threshold depends on the initial debt ratio, nominal growth and transactions that change debt without entering the deficit.

This calculation does not reconstruct the government’s 121.7% projection. That would require its nominal GDP assumptions, stock-flow adjustments and complete reconciliation. Insee explicitly cautions that financial assets and cash movements prevent changes in gross debt from being read directly as the public deficit. S03

For reference, Insee records debt at 117.5% of GDP at the end of March 2026. That quarterly observation is not interchangeable with a forecast for the end of the year. S03

Nominal growth combines changes in real output and GDP prices. It is not real growth alone, nor necessarily real growth plus consumer-price inflation. The Banque de France’s September projections put real growth at 0.4% in 2026 and 0.9% in 2027. Those figures cannot simply be substituted for nominal growth in the debt equation. S04

Inflation is not a costless debt write-off either. Even when it increases nominal GDP and some tax receipts, it can raise spending, the cost of new borrowing and payments on inflation-linked debt. Those securities respond to their contractual index, not simply to the GDP deflator. AFT, France’s debt-management agency, distinguishes bonds indexed to French prices from those indexed to euro-area prices. S05

The financing requirement is larger than the deficit alone

When a government bond matures, the state can issue another bond to repay it. That refinancing does not itself create another deficit: it replaces an old liability. It does, however, require buyers and a new financing price.

The funding requirement consequently includes the cash deficit, principal repayments and other financial operations. A general-government deficit cannot be added indiscriminately to a central-government issuance programme. Their accounting perimeter and timing differ. France’s public debt includes administrations beyond the central state, whereas AFT manages the state’s marketable debt.

AFT’s indicative programme for 2026 provides for €310 billion of medium- and long-term issuance, net of buybacks. It covers the central state’s financing, in a different year from the 2027 framework. S06

Funding composition matters as much as size. Shorter borrowing may reduce some immediate costs but requires more frequent returns to the market. Longer borrowing locks in a cost for more time, potentially at a higher yield demanded by investors. Neither choice removes the trade-off.

At the end of August, AFT reports approximately €2,903.8 billion of marketable state debt, including €219.2 billion of short-term debt, with an overall average remaining maturity of 8 years and 142 days. That perimeter differs from Maastricht general-government debt. S07

Average maturity buys time because legacy fixed coupons are not rewritten every trading day. It does not mean that all debt will be refinanced exactly eight years from now, or that exactly one-eighth matures annually. Only the maturity schedule can establish the rollover profile. Average maturity is also not duration, the measure of a bond price’s sensitivity to yields.

The first shock is a change in the price of existing bonds

An investor compares holding an existing security with buying a new issue. If the yield required on comparable cash flows rises, the price of the old security must fall to remain competitive. Its contractual coupon has not changed; the price paid to receive that coupon adjusts. S08

Take a fictional bond with a face value of 100 and an annual coupon of 3. With ten years remaining, it is worth 88.13 at a required yield of 4.5%. If that yield immediately rises to 5.5%, its value falls to 81.16, a loss of 7.91%. With thirty years remaining, the same conventions produce a 15.75% loss. These are discounted-cash-flow calculations with no default, fees or passage of time between valuations. They are not quotes for a particular French government bond, known as an OAT.

One hundred basis points means one percentage point of yield, not a 1% fall in price. The effect depends on maturity, coupon and the initial yield. Modified duration, the price sensitivity to yield, provides a local approximation: the fractional price change is approximately minus modified duration multiplied by the yield change, expressed as a decimal. For larger movements, exact cash-flow pricing avoids ignoring convexity.

An investor able to hold a bond until repayment is in a different position from one who must sell tomorrow. But holding to maturity assumes no default and the ability to finance the wait. It does not recover a purchase premium above par automatically or guarantee purchasing power. A fund that continually renews its bond portfolio need not have the single maturity date of an individual security. S08 S09

A lower price can, meanwhile, offer a better yield to a new buyer. Falling bond prices do not imply that all buyers have vanished. Repricing may be the condition under which additional buyers are willing to finance the issuer.

French yields do not reflect French fiscal policy alone

AFT’s TEC 10 index rises from 4.19% on 1 September to 4.45% on 21 September, having reached 4.52% on 15 and 16 September. That is a 26-basis-point rise between the first two dates, but a 7-basis-point decline from the observed peak. This constant-maturity index is neither an executable quote for a specific bond nor the average interest rate on the debt stock. S10

Those observations establish a repricing. They do not attribute every basis point to the budget framework. The ECB itself raised all three policy rates by 25 basis points on 10 September, effective from 16 September, taking the deposit rate to 2.50%. S11

Interpreting the French yield requires separating expected euro interest rates, compensation for holding duration, issuer-specific risk and liquidity conditions. This is an analytical framework, not four perfectly observable prices that can be added together without estimation.

The OAT-Bund spread compares French and German yields at a similar horizon. It helps isolate part of the relative French move, but is not a pure probability of default. German pricing also reflects liquidity and demand for securities usable as collateral. Observations must be synchronised and maturities comparable; the result should then be checked against other euro-area issuers and the swap curve. ECB research on repo markets illustrates why collateral matters for financing conditions. S12

The second shock accumulates in future budgets

The fiscal transmission is slower. Fixed-rate securities retain their coupons; maturing debt, new deficits and short-term borrowing gradually expose the state to new market conditions. The yield required today, the average cost of the outstanding stock and the interest expense recognised during the year must therefore be kept separate. S06 S07

For scale, suppose €300 billion of financing permanently costs one percentage point more. The full-year annual interest run rate on that volume would increase by roughly €3 billion. This is not an estimate of France’s 2027 bill: issuance dates, buybacks, Treasury bills, indexation and accounting conventions affect the annual result.

A persistent shock accumulates as maturities pass. Maturity delays the transmission to public budgets. The repayment schedule determines the speed of that transmission.

Separating interest from the rest of the budget gives the complete debt-ratio identity:

The primary surplus is the balance before interest, positive when revenue exceeds non-interest spending. At a high debt ratio, a persistent gap between the effective borrowing cost and nominal growth requires a larger primary effort to stabilise debt. This does not require the yield on every OAT to exceed growth. The accounting relationship concerns the average effective cost and nominal GDP.

The risk becomes self-reinforcing if higher interest expense requires more borrowing, investors demand additional compensation, and more expensive finance weakens economic activity. That is a possible feedback loop, not a rule under which every yield increase must cause the next one.

l0g / local simulator

Two clocks of bond-market risk

Compare a reference market yield with a permanent shock. Public debt responds gradually; the bond price is recalculated immediately.

Educational scenarios, not forecasts for France. Only the opening debt ratio uses a government projection. All other values are adjustable assumptions. The reference scenario does not reproduce the 2027 budget plan.

Scenario assumptions

Permanent yield shock (basis points)100 basis points = 1 percentage point. Range: -200 / 500.

Annual nominal GDP growth (% per year)Output and GDP prices, not real growth alone. Range: -3 / 8.

Annual primary deficit (% of GDP)Excludes interest. Negative value = surplus. Range: -5 / 10.

Detailed assumptions: debt, average rates and bond

Opening debt (% of end-2026 GDP)Government forecast, not an observed outturn. Range: 50 / 250.

Initial effective stock rate (% per year)Assumed average cost, not the ten-year yield. Range: 0 / 10.

Reference market yield (% per year)Constant assumption, not a live quote. Range: 0 / 10.

Annual stock-rate adjustment (% per year)Stylised fraction, not a maturity schedule. Range: 0 / 100.

Fictional bond maturity (whole years)Only affects bond pricing, not public debt. Range: 1 / 50.

Fixed bond coupon (% per year)Face value 100, annual payment. Range: 0 / 10.

Reset assumptions

Additional debt in 2032

+2.77GDP percentage points

Immediate price change

-7.91 %Fictional bond, same valuation date.

Additional interest in 2032

+0.81GDP percentage points

Before: 88.13; after: 81.16. Post-shock yield: 5.50 %.

Simulated public debt, % of GDP. Gap in GDP percentage points.

Year	ReferenceRef.	With shockShock	Gap
2027	122.09	122.24	+0.15
2028	125.15	125.57	+0.43
2029	128.46	129.29	+0.83
2030	131.99	133.35	+1.36
2031	135.75	137.76	+2.01
2032	139.71	142.48	+2.77

Total deficit stabilising the opening debt ratio: 2.91 %
At constant nominal growth, with zero stock-flow adjustment. This threshold is not a policy target.

Calculation assumptions and limitations

Each year, the average stock rate moves towards the market yield by the selected fraction. Interest is applied to opening debt and the primary deficit is then added. The shock is permanent from 2027 to 2032. Growth and the primary deficit do not respond to yields. The adjustment fraction is not an actual Treasury bill or OAT maturity schedule; new borrowing during the year is not priced separately.

New average rate = (1 − fraction) × previous average rate + fraction × market yield

Interest / GDP = average rate × previous debt / (1 + nominal growth)

Debt / GDP = previous debt / (1 + nominal growth) + primary deficit + interest / GDP

The model excludes stock-flow adjustments, inflation-linked debt, cash balances, issue premia, policy responses, defaults and forced sales. Pricing discounts annual coupons and repayment of 100 at maturity, immediately after a coupon date, using a flat yield curve. No real security identifier is used.

Calculated in your browser. This tool sends no inputs and uses no cookies or input storage.

Under the displayed assumptions, a permanent 100-basis-point shock adds 2.77 GDP percentage points to debt in 2032 relative to the reference path. Additional annual interest expense reaches 0.81 GDP percentage points that year. These are not official projections. The primary deficit is constant, growth does not respond and the annual adjustment fraction of 12.5% is a modelling convention, not a figure extracted from France’s maturity schedule.

The reference path itself drifts because it contains no scheduled fiscal correction. The tool separates effects; it does not turn six years of constant parameters into a central forecast. Setting the adjustment fraction to zero preserves the immediate bond-price loss but removes the fiscal gap between the two paths. That is the distinction between the two clocks.

When a valuation loss becomes a demand for cash

Some investors can absorb a decline in market value. It becomes more binding when the security backs borrowing. In a repo, funding is secured against securities under valuation and margin arrangements. Depending on the contract, a fall in collateral value can require additional securities or cash. S12

Suppose collateral worth 100 supports borrowing of 98 after a 2% haircut. If its value falls to 92 and the required haircut rises to 5%, funding capacity falls to 87.40. Maintaining the original borrowing against that collateral alone leaves a 10.60 gap. This example deliberately combines two shocks. It does not claim that those haircuts were applied to French OATs in September 2026.

Without a liquidity buffer or additional collateral, an investor may have to reduce the position. Forced selling can depress prices and generate further margin calls elsewhere, allowing leverage to amplify the need for liquidity.

The Bank of England documented this amplification in UK liability-driven investment funds in 2022. That episode illustrates forced-sale dynamics; French exposures and contracts require a separate assessment. S13

Banks use sovereign debt as an investment and as collateral. In an ECB working paper, Giovanni Dell’Ariccia and his coauthors analyse the interacting channels of sovereign securities holdings, government guarantees and economic activity. S14

Accounting treatment, hedges, funding stability and liability sensitivity determine how a valuation loss affects a bank or insurer. Leverage and liquidity needs can amplify the impact. The EBA’s 2023 analysis accordingly examines carrying amounts, market values and hedges separately. S18

The market is open, but on what terms?

On 17 September, AFT allotted €12.991 billion of medium-term OATs, before any additional non-competitive offers. Bid-to-cover ratios across the four securities ranged from 2.22 to 2.93, with average yields between 3.59% and 4.01%. All four issues found buyers at the published terms. S15

Demand depends on price, issue size and the securities offered. Detecting persistent deterioration requires repeated observation of comparable auctions, price concessions, secondary-market liquidity and repo terms.

The ECB also states that APP and PEPP portfolios are declining and principal repayments are no longer reinvested. S11 Other things equal, lower central-bank absorption leaves more securities for other investors. The clearing price depends on their capacity and willingness to bear the risk.

A rating downgrade can amplify repricing if it crosses a mandate threshold, changes a haircut or reduces the eligible buyer base. It does not automatically force every fund to sell. In the Eurosystem collateral framework, eligibility and haircuts are governed by explicit credit-quality rules. A rating change is not the same thing as universal exclusion. S16

The conditions for ECB intervention

The Transmission Protection Instrument, or TPI, addresses unwarranted, disorderly market dynamics threatening monetary-policy transmission. Activation is a Governing Council judgement, incorporating considerations including fiscal conditions and debt sustainability. It is not a published promise to intervene automatically at a particular spread. S17

A crucial nuance is that the criteria do not make every excessive-deficit procedure an automatic disqualification. They also consider whether effective action has been taken in response to European recommendations. Conversely, the existence of the TPI does not remove the sustainability requirements. S17

The analytical implication is straightforward. Investors cannot treat the instrument as an unconditional guarantee of the price of every OAT they hold. Nor can the budget debate end with “the ECB will pay”. Protecting monetary transmission and neutralising every consequence of a fiscal deficit are different commitments.

The decisive test is the path beyond 2027

Risk cannot be reduced to crossing a symbolic debt ratio or to one day of elevated yields. A credible trajectory could combine recurring revenue, better-controlled spending and preserved growth capacity. It would reduce future financing needs and the compensation required for uncertainty. Replacing durable revenue with one-off transactions might improve a single year without addressing what follows.

The composition and sequence of the adjustment matter. Measures that compress activity also affect receipts and nominal GDP; measures that protect productive capacity may follow a different path. Without explicit behavioural assumptions and fiscal multipliers, attaching a precise growth effect would be unjustified. The simulator does not attempt to do so.

The complete budget should therefore be read by reconciling the primary balance, nominal assumptions, interest expense and forthcoming financing programme. On the market side, useful monitoring combines synchronised spreads, comparable auctions and collateral conditions. Neither an isolated yield nor an agency rating replaces that joint assessment.

More expensive debt burdens future budgets while imposing valuation losses or liquidity needs on some creditors today. The path beyond 2027 will determine how far that transmission can be contained.

Scope, methodology and limitations

The evidence cutoff is 22 September 2026; the latest TEC 10 observation used is 21 September. The budget framework is preliminary. No final opinion on the 2027 budget from France’s fiscal watchdog, the Haut Conseil des finances publiques (HCFP), was used.

Statistical observations, projections and calculations are identified separately. AFT’s debt figures cover the state’s marketable borrowing; general-government ratios include other public administrations. The debt, bond-pricing and repo exercises are explicit illustrations, not forecasts, quotations, observed exposures or investment advice. The sources and conventions below allow the calculations to be reproduced.

Further reading: the repo market and gilts, leverage and the Bank of England.

Sources

1. S01 · Budget: record debt and proposed measures affecting pensionersAFP · 2026-09-19. Reported government framework: debt, deficit, taxes and public spending in 2026 and 2027.
2. S02 · 2027 budget: the announced €54 billion fiscal effortBanque des Territoires / Localtis · 2026-09-18. Breakdown of the effort announced on 17 September: €45 billion new and €9 billion carried over.
3. S03 · Maastricht debt in the first quarter of 2026Insee · 2026-06-25. 117.5% of GDP at end-March; explains why debt changes are not identical to the deficit.
4. S04 · Interim macroeconomic projections, September 2026Banque de France · 2026-09. Real GDP growth: 0.4% in 2026 and 0.9% in 2027.
5. S05 · Introduction to inflation-linked OATsAgence France Trésor · undated. Distinguishes French and euro-area inflation indexation.
6. S06 · Indicative state financing programme for 2026Agence France Trésor · 2025-12-30. €310 billion of medium- and long-term issuance, net of buybacks, for 2026.
7. S07 · Key figures for marketable state debtAgence France Trésor · 2026-09-10. At 31 August: €2,903.8 billion, including €219.2 billion short term; average maturity 8 years and 142 days.
8. S08 · Investor Bulletin: Interest Rate RiskSEC / Investor.gov · 2013-06-26. Relationship between yields, prices, coupons and maturity.
9. S09 · Bond Funds and Income FundsSEC / Investor.gov · undated. Bond funds carry interest-rate and net asset value risk.
10. S10 · Daily TEC 10 indexAgence France Trésor · 2026-09-21. 4.19% on 1 September; 4.52% on 15–16 September; 4.45% on 21 September.
11. S11 · Monetary policy decisionsEuropean Central Bank · 2026-09-10. 25-basis-point increase; deposit rate 2.50% from 16 September; no APP/PEPP principal reinvestment.
12. S12 · Home bias and repo ratesEuropean Central Bank · 2024-06. Repo funding, sovereign collateral and correlation between borrower and collateral.
13. S13 · Financial Stability Report, December 2022Bank of England · 2022-12. Documents margin calls, selling and the 2022 LDI liquidity crisis.
14. S14 · Managing the sovereign-bank nexusGiovanni Dell’Ariccia et al. · ECB WP 2177 · 2018-09-24. pp. 2–3.
15. S15 · €12.991 billion medium-term OAT auctionAgence France Trésor · 2026-09-17. Amounts, yields and bid-to-cover ratios for four securities; settlement 21 September.
16. S16 · Eurosystem credit assessment framework (ECAF)European Central Bank · 2026-04-01. Collateral eligibility rules and credit-quality steps.
17. S17 · Transmission Protection InstrumentEuropean Central Bank · 2022-07-21. TPI purpose, fiscal criteria, sustainability and Governing Council judgement.
18. S18 · Analysis of unrealised losses on banks’ bond holdingsEuropean Banking Authority · 2023-07-28. Separately examines carrying amounts, fair values and hedges.
19. S19 · Definition of taxes and compulsory social contributionsInsee · 2024-05-30. Taxes and actual social contributions, net of tax credits and uncollectable amounts.