## 1fraea2022002

## Source details

**Canonical URL:** [1fraea2022002](https://www.imf.org/-/media/files/publications/cr/2022/english/1fraea2022002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2022/english/1fraea2022002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2022/english/1fraea2022002.pdf.json)

---

### Executive summary: fiscal context and objectives
- Historical debt trajectory and drivers:
  - Public debt increased almost uninterruptedly over the last four decades, reaching close to 100 percent of GDP just before the COVID-19 crisis.
  - Recurrent spending expansions during recessions were only partially reversed, contributing to the long-term rise in the debt ratio.
  - Exception: sustained reduction in the spending ratio in the mid to late 1990s, leading to euro adoption, when the primary balance registered a surplus for a few years.
- Interest rate and debt service dynamics:
  - Nominal yield of 10-year sovereign bonds declined from above 10 percent in the early 1980s to -0.1 percent in 2020.
  - The effective interest rate reached an historical minimum of 1.1 percent in 2020.
  - Interest expenses as a share of GDP rose up to the late-1990s and then declined as the interest rate-growth differential fell.

### COVID-19 impact and recent fiscal response
- Fiscal response magnitude and effects:
  - Authorities’ total envelope of emergency and recovery programs around 28 percent of GDP (including below-the-line measures and public guarantees).
  - Above-the-line measures with impact on the fiscal balance: more than 3 percent of GDP in 2020 and 4 percent of GDP in 2021.
  - Debt ratio is expected to have increased by about 17 percentage points of GDP between 2019 and 2021.
- Medium-term unchanged-policies projections (October 2021 WEO):
  - Medium-term primary deficit projected to remain about 1¼ percent of GDP larger compared to pre-crisis projections.
  - Overall balance projected to remain ¾ percent of GDP worse than before the crisis.
  - Under unchanged policies, the deficit would decline moderately while remaining above 3 percent over the medium term and public debt would remain on a mildly increasing path.
- Cross-country position:
  - France’s overall fiscal deficit projected to remain one of the largest among euro area countries by 2026, behind only Belgium and Spain.
  - Projected difference with Germany’s debt ratio in 2026 expected to exceed 50 percentage points of GDP.

### Anchor, timing, and pace of fiscal consolidation (model-based)
- Policy consensus and institutional context:
  - Broad consensus for a gradual yet significant spending-based consolidation over the medium term.
  - Authorities’ pre-crisis medium-term objective was a (structural) deficit of 0.4 percent of GDP.
- Model diagnostics and recommended anchor:
  - Structural stochastic model features: forward-looking benevolent government; fiscal multiplier effects; automatic stabilizers; hysteresis from recessions; implementation lags; interest rate increasing with debt; stochastic risk of losing market access.
  - Model solution implies targeting a primary surplus of about ½ percent of GDP in the steady state.
  - Recommended transition: gradual, extending to around 2029-30 to reach the medium-term anchor.
- Recommended consolidation pace and profile:
  - For 2022: model recommends an improvement of 2⅔ percent of GDP in the structural primary balance (aligned with October 2021 WEO unchanged-policies projection of an improvement of 2.6 percent of GDP).
  - Starting in 2023: recommended annual average improvement of about 0.8 percent of GDP over 2023–26 (cumulative improvement of 3.3 percent of GDP).
  - Model recommends annual pace slightly more than ½ percent of GDP larger than staff projections under unchanged policies (about 0.4 percent of GDP per year larger over 2023–24 and around ¾ percent of GDP per year larger over 2025–26).
  - Recommended efforts continue beyond the medium term at a softer pace: average annual pace about ¼ percent of GDP over 2027–29 and falling below 0.1 percent thereafter.
  - Total effort required over the medium term: around 4¼ percent of GDP.
  - Profile is relatively frontloaded: about ¾ of the recommended adjustment within 2023–26.
- Debt outcomes under recommended path:
  - Public debt would peak in 2023 and start declining thereafter.
  - By 2029, public debt would reach about 107 percent of GDP.
  - Public debt would reach its pre-crisis level by 2036.

### Sensitivity to calibrations and alternative scenarios
- Long-run real effective interest rate:
  - Baseline assumes long-run real effective interest rate converges to 1.9 percent.
  - Five alternative calibrations considered, ranging from slightly less than 0.9 percent (2015–19 average) to 2.8 percent (2000–09 average).
  - Medium-term recommendation for the structural primary balance ranges between a deficit of 0.4 percent of GDP to a surplus of 1½ percent of GDP across these calibrations—each case implies tighter stance than unchanged policies.
  - Near-term recommendation looser than unchanged policies in baseline, but beyond 2023 all calibrations yield a tighter stance than unchanged policies.
- Other parameter variations (selected):
  - Debt sensitivity to interest rates: baseline 휶휶 = 1.5 percent; alternative = 2.5 percent. Medium-term primary balance broadly unchanged; 2022 recommendation slightly tighter (by close to 0.1 percent of GDP).
  - Debt threshold for 50 percent market-access risk: baseline 풅풅 = 150 percent of GDP; alternative = 200 percent of GDP. Increasing threshold implies lower medium-term primary balance and looser 2022 stance by small quantitative differences (0.03 and 0.08 percent of GDP respectively).
  - Fiscal multiplier: baseline multiplier m1 = 0.98; alternative m1 = 0.5. If multiplier = 0.5, recommended medium-term primary balance slightly larger, with a more gradual near-term consolidation (recommended stance 2 percentage points of GDP looser than baseline in 2022).
  - Long-term growth: baseline = 1.3 percent; alternative = 1.6 percent. If long-term growth = 1.6 percent, recommended fiscal stance would be looser by 0.5 percent of GDP in the near term.

### Fiscal planning implications and summary recommendations
- Medium-term anchor:
  - Urgent need to rebuild fiscal space with a medium-term anchor at a primary surplus of ½ percent of GDP.
- Design and sequencing:
  - Recommended consolidation is spending-based, gradual but significant, with a frontloaded effort concentrated in 2023–26.
  - Policy design should account for uncertainty in interest rate and growth trajectories and consider institutional measures to improve compliance with medium-term objectives.
- Institutional recommendations (high level):
  - Consider measures to improve compliance and credibility of multiyear targets (see detailed framework section below).

*Prepared by Bertrand Gruss (AFR, formerly EUR) and Sebastian Weber (MCM, formerly EUR); content as presented in the source PDF.*

---

### A strong fiscal framework should help — role, evidence, and reform agenda
- Role of the fiscal framework:
  - The fiscal framework comprises fiscal rules and targets and the institutions responsible for the design, implementation, and oversight of fiscal policy.
  - Analysis focuses primarily on fiscal rules and the institutions charged with oversight.
- Evidence on compliance and forecast bias:
  - France has complied on average in about 1/3 of the cases with a strict numerical definition across four EU rules (expenditure, fiscal deficit, structural balance, and debt rules).
  - Low compliance concentrated in cyclical downturns during the early-2000s and the GFC; in the last decade missing targets has become unrelated to the business cycle.
  - France tends to under-predict expenditures in the near-term and over-predict revenues over the medium-term, resulting in a consistent upward bias for the projected fiscal balance beyond the current year.
  - Medium-term fiscal projections for France have been consistently more optimistic compared to outcome and to peers; the 2-year ahead forecast error is biased toward over-optimism in almost all years and is consistently more than for the median EU country.
  - Empirically, overoptimism in medium-term fiscal projections is robustly associated with higher debt, even after conditioning on actual budget balances and output growth surprises.
- Determinants and empirical approach:
  - Literature identifies three main determinants of fiscal forecast errors: (1) fiscal institutions, (2) political factors, and (3) economic conditions.
  - Two panel regression approaches used: one OLS across covariates; a second conditional regression analyzing rule design elements when an expenditure rule is in place.
- Key empirical findings on rule design:
  - No clear evidence that rule complexity (number of fiscal rules) reduces forecast bias.
  - Ownership of rules (fraction of national rules) potentially reduces overoptimism, though estimates lose significance with full controls.
  - Having a national expenditure rule significantly reduces forecast error bias by limiting under-projection of expenditures.
  - Multi-year expenditure ceilings help reduce overoptimism bias.
  - Rules backed by constitutional legal nature or political/coalition agreements perform better than rules based on statutory law.
  - Monitoring of rules outside the government and wider coverage of expenditure rules (including the whole general government, GG) are associated with lower over-optimism.
  - Presence of a formal enforcement procedure appears counter-productive in joint regressions (potentially due to collinearity), though not when examined individually.
  - All effects become stronger at longer forecast horizons.
- Characteristics of France’s current framework and gaps:
  - France has multiple fiscal rules (expenditure, fiscal balance, debt, revenue); the revenue rule applies only to a subset of items and is unrelated to specific EU rules.
  - Multiyear targets in France’s Stability Programs have been treated as non-binding in the past with important deviations.
  - Coverage of some fiscal rules is incomplete: limited levels of government covered and excluded instruments (e.g., tax expenditures).
  - The High Council of Public Finances (Haut Conseil des finances publiques, HCFP) monitors compliance but with limited full-time staffing, powers, and primarily ex-post involvement.
- Recommended reforms to improve credibility and compliance:
  - Revamped expenditure rule:
    - Cover the entire general government.
    - Apply over a longer horizon with explicit multi-year nominal targets.
    - Ensure strong political and legal endorsement for the multi-year targets.
    - Vote multi-year objectives during the annual budget cycle and explicitly account for how deviations from previous objectives are offset (including in the current budget).
  - Strengthen monitoring and oversight:
    - Require timely execution data for all subsectors of the general government.
    - Establish comprehensive monitoring by a non-government entity (e.g., the fiscal council) with enhanced powers.
    - Integrate the fiscal council more strongly into the budget process to provide timely assessment of macroeconomic and fiscal forecasts.
    - Bolster the fiscal council’s ability to endorse government projections by extending its role into providing independent forecasts, estimates of yields of planned discretionary policies, and debt sustainability analysis—requiring adequate staffing and resources.
  - Provide debt sustainability analysis beyond the five-year horizon to underpin multi-year expenditure ceilings and budget planning.
- Simulations and medium-term consolidation target:
  - Counter-factual analysis: adopting a system comparable to Denmark would have implied an average forecast bias over 2005–19 of zero.
  - Structural stochastic model simulations suggest aiming for a primary surplus of about ½ percent of GDP over a period of seven to eight years.
  - Model indicates a mildly expansionary stance in 2022 (net of projected withdrawal of crisis-related support) followed by significant consolidation efforts until the medium-term target is reached around 2029-30.

### Appendix I — model calibration (selected parameters)
- Discount factor 휷휷: 0.99
- Risk aversion 흈흈: 2
- Labor elasticity 휼휼: 1/0.3
- Weight of labor ξ: 1
- Fiscal multiplier when the gap is null m1: 0.98 (Baseline), 0.5 (Alternative)
- Fiscal multiplier sensitivity to shocks m2: 3
- Automatic stabilizers (semi-elasticity of primary balance to gap) a: 0.61
- Adjustment cost of fiscal policy χ: 3
- Real effective interest rate (steady state): 1.9% (Baseline); 0.9% (average 2015-19); 1% (average 2010-19); 2.8% (average 2000-09) (Alternatives)
- Effect of debt level on the risk premium 휶휶: 1.5% (Baseline); 2.5% (Alternative)
- Effect of debt change on the risk premium 휶휶2: 0.5%
- Debt level at which the risk to lose market access is 50% 풅풅: 150% (Baseline); 200% (Alternative)
- Effect of debt change on the risk to lose market access d2: 1
- Adverse scenario coefficient in case of loss of market access d3: -1%
- Potential GDP growth: 1.3% (Baseline); 1.6% (Alternative)
- Shock persistence 흆흆: 0.66
- Shock size: 2.6%
- Hysteresis: 10% (Baseline); 0% (Alternative)
- Hysteresis threshold: -1%

### Occupational pathways for the climate transition in France — key findings
- Emissions and targets:
  - France’s total greenhouse gas emissions in 2019 represented a 17.7 percent decline from 1990 levels.
  - EU collective target: net greenhouse gas emissions reduction of at least 55 percent by 2030 compared to 1990 levels.
  - National plans include: carbon-free energy system by 2050; increase energy efficiency to halve energy consumption; decrease non-energy emissions by 38 percent in farming and 60 percent in industry (both compared to 2015 levels); increase carbon sinks.
  - NECP envisages up to 500,000 new jobs by 2030, mostly in tertiary sectors.
- Sectoral and occupational patterns:
  - Employment and labor hours in carbon-intensive sectors have been secularly declining relative to low-carbon sectors.
  - High carbon-intensity sectors (per worker): electricity/gas, water supply/sewerage, mining, transportation, manufacturing, agriculture.
  - These high carbon sectors employ less than a tenth of all workers.
  - Craft and machinery workers more prevalent in high carbon sectors; specialized occupations concentrated in carbon-intensive sectors may face greater transition difficulty.
- Green jobs (Onemev taxonomy) — scope and characteristics:
  - Sixty-two occupations classified: ten green occupations ≈ 0.5 percent of total employment; fifty-two greening occupations ≈ 14 percent of total employment.
  - Together these account for about 15 percent of employment in France.
  - Women represent about 17 percent of green and greening jobs.
  - Share of youth in green jobs: about 24 percent (versus 20 percent across all jobs).
  - Education: about 54 percent of green jobs require a qualification of the baccalaureate or above (compared to 62 percent across all jobs).
  - In 2019, vacancies for green occupations represented about 17 percent of all job offers; 14 percent of all jobseekers belonged to green occupations.
  - Firms reported hiring difficulties: 60 percent of green-related recruitment projects were difficult recruitments versus 50 percent for all occupations combined.
  - Sectoral concentration of greening jobs: construction 46 percent; transport 23 percent.
- Labor market implications from prior evidence:
  - Clean air policy in the United States: affected workers incurred an earnings loss of 20 percent (Walker, 2013).
  - Carbon-tax in British Columbia: short-lived increase in unemployment and gradual wage effects via lower wages for new hires (Yip, 2020).
  - Implication: aggregate costs of climate reallocations may be limited, but costs to certain sectors/workers can be large due to displacement and weak job matches.

### Occupational mobility networks, similarity measures, and transition prospects
- Methodology and data:
  - Occupation classification uses Onemev and probability of occurrence in polluting industries (Vona et. al., 2018).
  - Skills and competence rely on ESCO; work characteristics supplemented with O*NET.
  - Polluting occupations defined as those with a probability of working in polluting sectors seven times higher than in any other job; polluting sectors are those in the 95th percentile of sector-wise pollution intensity for at least three pollutants including CO2.
  - Small overlap (~3 percent of all ESCO occupations) between green and polluting occupations; overlapping occupations are classified as brown.
- Box 1 — measuring occupation similarity (method summary):
  - Uses natural language processing on ESCO essential and optional skills to measure overlap.
  - Uses vector similarity metrics on ESCO skills hierarchy to capture alignment of work activities.
  - Uses vector similarity on physical, interpersonal and structural work characteristics from O*NET.
  - Occupation-pair similarity is the simple average of the three similarity measures.
- Descriptive network findings:
  - Polluting occupations tend to be clustered in the periphery; green occupations are more dispersed.
  - Polluting occupations with larger labor share (physical/chemical technicians, engineers, transporters) are more integrated with lower transition needs.
  - Green occupations are diverse and some situated alongside polluting occupations, suggesting feasible occupational pathways in some cases.
- Job-transition match statistics:
  - Median job match for polluting-to-green transitions is about 4 matches (heterogeneous across occupations).
  - Example heterogeneity: mining engineers have below-median job-transition match to green occupations; chemical engineers have above-median matches.
  - It is easier to transition between green occupations and from other occupations to green than from polluting to green.
- Role of training, experience, and job-zone:
  - O*NET job-zones:
    - Job-zones 1 and 2: low education (up-to a high school diploma), some previous experience, low training.
    - Job-zones 3 and 4: vocational or bachelor’s degree, requisite previous experience, high training.
    - Job-zone 5: graduate degree or more, high work experience.
  - Median polluting-to-green job matches are relatively higher for polluting occupations in job zones 3 and 4.
  - Regression: a one standard deviation increase in a polluting occupation’s training or work-experience level is significantly associated with an increased green job match by about 0.2 standard deviations.
  - Level of education is less strongly associated with possible green transitions.
- Automation and remote-work readiness:
  - Polluting occupations are less likely to be automated and have less remote-work capability.
  - Green occupations can be more easily automated with higher ability for remote-working.
  - Implication: polluting occupations are less at risk from automation but face more obstacles shifting to new ways of working, which could lower future job mobility if workers are not appropriately trained.

### Job multipliers, carbon adjustments, and sectoral targeting
- Framework:
  - Semi-closed input–output model with partially endogenized consumption to quantify employment impacts of carbon emissions reduction.
  - Three employment multipliers: direct, indirect (Leontief backward linkages), induced (forward linkages via household demand).
  - Carbon-adjusted employment multipliers: direct/indirect/induced multipliers multiplied by output per unit of carbon emissions.
- Empirical trends (2008–2019):
  - Between 2008 and 2019 direct job multipliers fell in all sectors.
  - Indirect and induced multipliers decreased relatively more in carbon-intensive sectors compared to low-carbon sectors.
  - For the majority of sectors, the combination of indirect and induced effects was greater than direct effects.
  - Falling employment multipliers for high-carbon sectors driven by a fall in induced household demand for products from other high-carbon sectors.
- Reported numerical figures (jobs per million of output, figure values):
  - Top figure sequence: 8.3 7.1 7.0 6.3 6.6 5.6 6.8 5.6 3.9 3.2 3.9 2.1
  - Lower panel breakdown sequence: 8.3 7.1 7.0 6.3 2.6 1.9 3.1 2.3 4.0 3.7 3.7 3.3 1.3 1.0 1.3 0.6 2.6 2.3 2.6 1.4
  - Environmental job multipliers (affected jobs per unit of carbon emissions) per sector (sequence) preceding sector names: 00.5 11.5 22.5 33.5
  - Sectors listed in same order: MANUFACTURE OF PETROLEUM; WATER SUPPLY/ SEWERAGE; MANUFACTURE OF CHEMICALS; MINING; TRANSPORTATION AND STORAGE; MANUFACTURE OF FOOD; ARTS, ENTERTAINMENT AND RECREATION; CONSTRUCTION; MANUFACTURE OF ELECTRICALS; TELECOMMUNICATIONS; REAL ESTATE; OTHER MANUFACTURING; RESEARCH AND DEVELOPMENT; HUMAN HEALTH ACTIVITIES; COMPUTER PROGRAMMING/INFORMATION
- Interpretation:
  - Tertiary and service-oriented sectors such as IT, R&D, health services, telecommunications, have high carbon-adjusted employment multipliers.
  - Sectors such as mining, transportation, chemicals manufacturing have low environmental employment multipliers.
  - Manufacturing of electronics and machinery have relatively high employment effects per unit of carbon emissions due to strong economy-wide integration.

### Policy considerations and institutional measures for labor transitions
- Skills and training:
  - Training and work experience (but not necessarily education) are associated with improved job matches; policies should emphasize training and experience dissemination.
  - Improve dissemination of work experience (e.g., employer ratings or evaluations) to improve employment outcomes in entry-level labor markets; example: subsidizing firms for hiring young workers (as used under the French recovery plan).
  - Governments and private sector could regulate skills-based measures of worker performance (employment tests or simulated work experience) to avoid stigmatization of previous high-carbon job contexts.
- Data-driven career-path strategies:
  - Adopt a data-driven strategy using granular skills matches to map career pathways for workers affected by the climate transition; public-sector agencies can use these maps to aid job seekers.
- Sectoral investment and pedagogy:
  - Invest in sectors with potential to create sustainable green jobs or support evolution of skills in affected sectors.
  - Link new training programs and pedagogy to digital tools given the shift to new ways of working and low remote-work readiness in carbon-intensive industries.
  - Emphasize development of inter-sectoral qualifications to avoid occupations being locked-in by non-transferable sector-specific skills (ILO, 2018).
- Private sector and institutional support:
  - Work with private sector to raise attractiveness of green jobs to increase labor supply.
  - Strengthen institutional setup: provide legal frameworks for green skills and training or establish multi-year green growth plans.
  - Consider bolstering long-term support programs for displaced workers unable to transition easily.

*Source: IMF staff analysis in 1fraea2022002.*

### References ____________________________________________________________________________ 18

### 1fraea2022002 - References ____________________________________________________________________________ 18

### Appendix / Contents overview
- Appendix sections and chapter headings listed:  
  - I. Model Calibration and Additional Results ___________________________________________ 19  
  - AN ANATOMY OF OCCUPATIONAL PATHWAYS FOR THE CLIMATE TRANSITION IN FRANCE ___________________________________________________________________________ 21  
  - A. Introduction _______________________________________________________________________ 21  
  - B. The Sectoral and Occupational Dimension of the Climate Transition _______________ 22  
  - C. Green Jobs in France _______________________________________________________________ 23  
  - D. Occupational Mobility Networks to Analyze Job Transition Pathways ______________ 25  
  - E. Additional Insights on Desirable Transitions from Job Multipliers __________________ 30  
  - F. Conclusion and Policy Considerations ______________________________________________ 33  
  - BOX 1. Measuring Occupation Similarity ___________________________________________________ 27

### Executive summary: France — fiscal context and objectives
- Historical debt trajectory and drivers:
  - Public debt increased almost uninterruptedly over the last four decades, reaching close to 100 percent of GDP just before the COVID-19 crisis.
  - Recurrent spending expansions during recessions were only partially reversed, contributing to the long-term rise in the debt ratio.
  - Exception: sustained reduction in the spending ratio in the mid to late 1990s, leading to euro adoption, when the primary balance registered a surplus for a few years.
- Interest rate and debt service dynamics:
  - Nominal yield of 10-year sovereign bonds declined from above 10 percent in the early 1980s to -0.1 percent in 2020.
  - The effective interest rate reached an historical minimum of 1.1 percent in 2020.
  - Interest expenses as a share of GDP rose up to the late-1990s and then declined as the interest rate-growth differential fell.

### COVID-19 impact and recent fiscal response
- Fiscal response magnitude and effects:
  - Authorities’ total envelope of emergency and recovery programs around 28 percent of GDP (including below-the-line measures and public guarantees).
  - Above-the-line measures with impact on the fiscal balance: more than 3 percent of GDP in 2020 and 4 percent of GDP in 2021.
  - Debt ratio is expected to have increased by about 17 percentage points of GDP between 2019 and 2021.
- Medium-term unchanged-policies projections (October 2021 WEO):
  - Medium-term primary deficit projected to remain about 1¼ percent of GDP larger compared to pre-crisis projections.
  - Overall balance projected to remain ¾ percent of GDP worse than before the crisis.
  - Under unchanged policies, the deficit would decline moderately while remaining above 3 percent over the medium term and public debt would remain on a mildly increasing path.
- Cross-country position:
  - France’s overall fiscal deficit projected to remain one of the largest among euro area countries by 2026, behind only Belgium and Spain.
  - Projected difference with Germany’s debt ratio in 2026 expected to exceed 50 percentage points of GDP.

### Anchor, timing, and pace of fiscal consolidation (model-based)
- Policy consensus and institutional context:
  - Broad consensus for a gradual yet significant spending-based consolidation over the medium term.
  - Authorities’ pre-crisis medium-term objective was a (structural) deficit of 0.4 percent of GDP.
- Structural stochastic model diagnostic:
  - Model features: forward-looking benevolent government; fiscal multiplier effects; automatic stabilizers; hysteresis from recessions; implementation lags; interest rate increasing with debt; stochastic risk of losing market access.
  - Model solution implies targeting a primary surplus of about ½ percent of GDP in the steady state.
  - Recommended transition: gradual, extending to around 2029-30 to reach the medium-term anchor.
- Recommended near-term and medium-term consolidation pace:
  - For 2022: model recommends an improvement of 2⅔ percent of GDP in the structural primary balance (aligned with October 2021 WEO unchanged-policies projection of an improvement of 2.6 percent of GDP).
  - Starting in 2023: recommended annual average improvement of about 0.8 percent of GDP over 2023–26 (cumulative improvement of 3.3 percent of GDP).
  - Model recommends annual pace slightly more than ½ percent of GDP larger than staff projections under unchanged policies (about 0.4 percent of GDP per year larger over 2023–24 and around ¾ percent of GDP per year larger over 2025–26).
  - Recommended efforts continue beyond the medium term at a softer pace: average annual pace about ¼ percent of GDP over 2027–29 and falling below 0.1 percent thereafter.
  - Total effort required over the medium term: around 4¼ percent of GDP.
  - Profile is relatively frontloaded: about ¾ of the recommended adjustment within 2023–26.
- Debt outcomes under recommended path:
  - Public debt would peak in 2023 and start declining thereafter.
  - By 2029, public debt would reach about 107 percent of GDP.
  - Public debt would reach its pre-crisis level by 2036.

### Sensitivity to calibrations and alternative scenarios
- Long-run real effective interest rate:
  - Baseline assumes long-run real effective interest rate converges to 1.9 percent.
  - Five alternative calibrations considered, ranging from slightly less than 0.9 percent (2015–19 average) to 2.8 percent (2000–09 average).
  - Medium-term recommendation for the structural primary balance ranges between a deficit of 0.4 percent of GDP to a surplus of 1½ percent of GDP across these calibrations—each case implies tighter stance than unchanged policies.
  - Near-term recommendation looser than unchanged policies in baseline, but beyond 2023 all calibrations yield a tighter stance than unchanged policies.
- Other parameter variations:
  - Debt sensitivity to interest rates: if sensitivity is 2.5 percent instead of 1.5 percent, medium-term primary balance broadly unchanged; 2022 recommendation slightly tighter (by close to 0.1 percent of GDP).
  - Debt threshold for 50 percent market-access risk: baseline 150 percent of GDP; increasing threshold to 200 percent of GDP implies lower medium-term primary balance and looser 2022 stance, with small quantitative differences (by 0.03 and 0.08 percent of GDP respectively).
  - Fiscal multiplier: baseline multiplier 0.98; if multiplier = 0.5, recommended medium-term primary balance slightly larger, with a more gradual near-term consolidation (recommended stance 2 percentage points of GDP looser than baseline in 2022).
  - Long-term growth: baseline 1.3 percent; if long-term growth = 1.6 percent, recommended fiscal stance would be looser by 0.5 percent of GDP in the near term.

### Fiscal planning implications and recommendations (summary)
- Model implies urgent need to rebuild fiscal space with a medium-term anchor at a primary surplus of ½ percent of GDP.
- Recommended consolidation is spending-based, gradual but significant, with a frontloaded effort concentrated in 2023–26.
- Policy design should account for uncertainty in interest rate and growth trajectories and consider institutional measures to improve compliance with medium-term objectives.

*Prepared by Bertrand Gruss (AFR, formerly EUR) and Sebastian Weber (MCM, formerly EUR); content as presented in the source PDF.*

### 16.      A strong fiscal framework should help

### 16.      A strong fiscal framework should help

### Role of the fiscal framework
- The fiscal framework comprises fiscal rules and targets and the institutions responsible for the design, implementation, and oversight of fiscal policy.
- The analysis focuses primarily on fiscal rules and the institutions charged with oversight.

### Evidence on compliance with EU fiscal rules and forecast bias
- France has historically fallen short of its commitments to fiscal targets under the EU rules.
- Replicating Larch and Santacroce (2020) for the four EU rules (expenditure, fiscal deficit, structural balance, and debt rules), France has complied on average in about 1/3 of the cases with a strict numerical definition.
- Low compliance was concentrated in cyclical downturns during the early-2000s and the GFC, but in the last decade missing the targets has become unrelated to the business cycle.
- France tends to under-predict expenditures in the near-term and over-predict revenues over the medium-term, placing it among countries with a consistent upward bias for the projected fiscal balance beyond the current year.
- Medium-term fiscal projections for France have been consistently more optimistic compared to outcome and to peers; the 2-year ahead forecast error is biased toward over-optimism in almost all years and is consistently more than for the median EU country.
- Empirically, overoptimism in medium-term fiscal projections is robustly associated with higher debt, even after conditioning on actual budget balances and output growth surprises.

### Determinants of overoptimism (literature and empirical approach)
- Literature identifies three main determinants of fiscal forecast errors: (1) fiscal institutions, (2) political factors, and (3) economic conditions.
- Two panel regression approaches are used:
  - First OLS: estimates effect of fiscal rules, political factors, and economic conditions on k-years ahead forecast errors of fiscal balance, expenditure, and revenue to GDP ratios. Covariates include: expenditure rule dummy, number of fiscal rules in place, fraction of national fiscal rules, extent of government majority, election year dummy, current account openness, GDP growth forecast error, education level, year dummies.
  - Second regression (conditional on an expenditure rule being in place): analyzes fiscal balance forecast error depending on rule design elements and supporting institutions. Covariates include dummies for (1) existence of multi-year expenditure ceilings, (2) legal basis of fiscal rules, (3) existence of monitoring outside the general government, (4) sectoral coverage of the rule, and (5) existence of a formal enforcement procedure.

### Key empirical findings on fiscal rules and design elements
- No clear evidence that rule complexity (number of fiscal rules) reduces forecast bias.
- Ownership of rules (fraction of national rules) potentially reduces overoptimism in fiscal balance projections, though estimates are not significant at conventional confidence levels when all controls are included.
- Having a national expenditure rule significantly reduces forecast error bias by limiting under-projection of expenditures.
- Multi-year expenditure ceilings help reduce overoptimism bias.
- Rules backed by constitutional legal nature or political/coalition agreements perform better than rules based on statutory law.
- Monitoring of rules outside the government (by non-government bodies) and wider coverage of expenditure rules (including the whole general government, GG) are associated with lower over-optimism.
- The presence of a formal enforcement procedure appears counter-productive in the joint regressions, though this does not hold when enforcement is examined individually; the counter-productive result may reflect collinearity among regressors and correlation with over-projecting revenues.
- All the effects become stronger at longer forecast horizons.

### Characteristics of France’s current framework and institutional gaps
- France has multiple fiscal rules in place, including expenditure, fiscal balance, debt and revenue rules; the revenue rule applies only to a subset of items and is unrelated to specific EU rules.
- Multiyear targets in France’s Stability Programs have been treated as non-binding in the past with important deviations from targets observed.
- Coverage of some fiscal rules in France is incomplete: limited levels of government covered (e.g., central government only) and excluded instruments (e.g., tax expenditures).
- The High Council of Public Finances (Haut Conseil des finances publiques, HCFP) monitors compliance but its full-time staffing, powers, and involvement in the budget process are limited and primarily ex-post in nature.

### Policy implications and recommended reforms
- A revamped expenditure rule could improve compliance:
  - Cover the entire general government.
  - Apply over a longer horizon with explicit multi-year nominal targets.
  - Ensure strong political and legal endorsement for the multi-year targets.
  - Vote multi-year objectives during the annual budget cycle and explicitly account for how deviations from previous objectives are offset (including in the current budget) to enforce medium-term commitments.
- Strengthen monitoring and oversight:
  - Require timely execution data for all subsectors of the general government.
  - Establish comprehensive monitoring by a non-government entity (e.g., the fiscal council) with enhanced powers.
  - Integrate the fiscal council more strongly into the budget process so it can provide timely assessment of macroeconomic and fiscal forecasts.
  - Bolster the fiscal council’s ability to endorse government projections by extending its role into providing independent forecasts, estimates of yields of planned discretionary policies, and debt sustainability analysis—requiring adequate staffing and resources.
- Provide debt sustainability analysis beyond the five-year horizon to underpin multi-year expenditure ceilings and budget planning.

### Simulations and medium-term consolidation target
- Counter-factual analysis indicates that adopting a system comparable to Denmark (the country in the sample with the lowest medium-term forecast errors) would have implied an average forecast bias over 2005–19 of zero.
- Simulations using a structural stochastic model suggest France should aim for a primary surplus of about ½ percent of GDP over a period of seven to eight years.
  - The model indicates aiming for a medium-term primary surplus of ½ percent of GDP would provide sufficient space for fiscal policy to stabilize economic cycles given France’s historical distribution of shocks.
  - Taking into account initial conditions in 2021, the model recommends a mildly expansionary stance in 2022 (net of projected withdrawal of crisis-related support) followed by significant consolidation efforts until the medium-term target is reached around 2029-30.

### Conclusion
- France’s public debt has been increasing for over four decades and is projected to remain high and diverge from euro area peers over the medium term.
- Sustained debt buildup reflects that increases in spending during recessions were only partially redressed thereafter.
- Successive governments have been excessively optimistic relative to peers in projecting medium-term fiscal balances, leading often to breaches of EU fiscal commitments.
- Reforming the fiscal framework—especially a national multiyear expenditure rule backed by political consensus, covering the whole general government, combined with strong external monitoring and an empowered fiscal council—could improve credibility of medium-term consolidation and help contain debt going forward.

*Source: IMF staff analysis in “16.      A strong fiscal framework should help,” from the provided IMF chapter PDF.*

### Appendix I. Model Calibration and Additional Results

### Appendix I. Model Calibration and Additional Results

### Model calibration (Appendix Table 1)
- Discount factor 휷휷: 0.99
- Risk aversion 흈흈: 2
- Labor elasticity 휼휼: 1/0.3
- Weight of labor ξ: 1
- Fiscal multiplier when the gap is null m1: 0.98 (Baseline), 0.5 (Alternative)
- Fiscal multiplier sensitivity to shocks m2: 3
- Automatic stabilizers (semi-elasticity of primary balance to gap) a: 0.61
- Adjustment cost of fiscal policy χ: 3
- Real effective interest rate (steady state): 1.9% (Baseline); 0.9% (average 2015-19); 1% (average 2010-19); 2.8% (average 2000-09) (Alternatives)
- Effect of debt level on the risk premium 휶휶: 1.5% (Baseline); 2.5% (Alternative)
- Effect of debt change on the risk premium 휶휶2: 0.5%
- Debt level at which the risk to lose market access is 50% 풅풅: 150% (Baseline); 200% (Alternative)
- Debt limit accuracy d1: 3
- Effect of debt change on the risk to lose market access d2: 1
- Adverse scenario coefficient in case of loss of market access d3: -1%
- Potential GDP growth: 1.3% (Baseline); 1.6% (Alternative)
- Shock persistence 흆흆: 0.66
- Shock size: 2.6%
- Hysteresis: 10% (Baseline); 0% (Alternative)
- Hysteresis threshold: -1%

- Note: "1/ See Country Report No. 19/246 for model equations and definitions of the different parameters."

### Key findings — Occupational pathways for the climate transition in France (Introduction and overview)
- France’s total greenhouse gas emissions in 2019 represented a 17.7 percent decline from 1990 levels.
- EU collective target: net greenhouse gas emissions reduction of at least 55 percent by 2030 compared to 1990 levels.
- National plans include: carbon-free energy system by 2050; increase energy efficiency to halve energy consumption; decrease non-energy emissions by 38 percent in farming and 60 percent in industry (both compared to 2015 levels); increase carbon sinks.
- National Energy and Climate Plan (NECP) envisages up to 500,000 new jobs by 2030, mostly in tertiary sectors.
- Evidence and trends:
  - Employment and labor hours in carbon-intensive sectors have been secularly declining relative to low-carbon sectors.
  - Compensation per employee trends have remained similar between high-carbon and low-carbon sectors.
  - Trends suggest extensive-margin adjustments in high-carbon sectors: reduced labor requirements while preserving wages.
  - Net-neutral employment impact in aggregate suggested by jobs growth in low-carbon sectors.

### Labor market effects and displacement risks
- Prior evidence:
  - Clean air policy in the United States: affected workers incurred an earnings loss of 20 percent (Walker, 2013), driven by nonemployment and lower future earnings; affected workers experienced more rapid earnings recovery relative to typical displacements.
  - Carbon-tax in British Columbia: short-lived increase in unemployment driven by increased unemployment spells; aggregate wages fell gradually via lower wages for new hires while incumbent worker wages remained stable (Yip, 2020).
- Implication: Aggregate costs of climate reallocations may be limited, but costs to certain sectors/workers can be large due to displacement and weak job matches.

### Sectoral and occupational dimensions
- High carbon-intensity sectors (per worker): electricity/gas, water supply/sewerage, mining, transportation, manufacturing, agriculture.
- These high carbon sectors employ less than a tenth of all workers, suggesting reallocation needs may concern a small share of total labor force.
- Not all jobs within carbon-intensive sectors are threatened; some occupations (e.g., R&D of green technologies within transportation) may persist or grow.
- Occupational concentration:
  - Craft and machinery workers are more prevalent in high carbon sectors (agriculture, mining, manufacturing, electricity, water).
  - Specialized occupations concentrated in carbon-intensive sectors may face greater difficulty in transitioning.

### Green jobs in France (scope, characteristics, labor market dynamics)
- Definition and scope (Onemev taxonomy):
  - "Green" occupations: directly aim to protect the environment (e.g., forest ranger, water technician).
  - "Greening" occupations: do not have a direct environmental purpose but evolve to integrate environmental skills (e.g., architect, thermal insulation installer).
- Coverage:
  - Total occupations classified: sixty-two occupations (ten green occupations ≈ 0.5 percent of total employment; fifty-two greening occupations ≈ 14 percent of total employment).
  - Together these account for about 15 percent of employment in France.
- Demographic and skill profile:
  - Women represent about 17 percent of green and greening jobs.
  - Share of youth in green jobs: about 24 percent (versus 20 percent across all jobs).
  - Education: about 54 percent of green jobs require a qualification of the baccalaureate or above (compared to 62 percent across all jobs).
- Labor demand and supply:
  - In 2019, vacancies for green occupations represented about 17 percent of all job offers; 14 percent of all jobseekers belonged to green occupations.
  - Positive job recruitment gap indicates employers face supply constraints for green hires.
  - Firms report hiring difficulties: in 2019, firms reported that 60 percent of green-related recruitment projects were difficult recruitments versus 50 percent for all occupations combined.
  - Recruitment profile: green job offers targeted a higher proportion of manual skilled and specialized workers (reflecting concentration in construction: 46 percent of greening jobs; transport: 23 percent).

### Occupational mobility networks and skills matching
- Methodology components:
  - Occupation classification: occupations classified into "green", "polluting", and "other" using Onemev and probability of occurrence in polluting industries (Vona et. al., 2018).
  - Skills and competence: primary reliance on the European Commission’s ESCO (European multilingual classification of Skills, Competences, Qualifications and Occupations).
  - Work characteristics: supplemented with O*NET data for physical and social work-context factors (e.g., interpersonal relationships, physical proximity, duration of work, degree of automation).
- Classification details and caveats:
  - Polluting occupations defined as those with a probability of working in polluting sectors seven times higher than in any other job; polluting sectors are those in the 95th percentile of sector-wise pollution intensity (measured in emissions per worker) for at least three pollutants including CO2.
  - Small overlap (~3 percent of all ESCO occupations) between green and polluting occupations; overlapping occupations are classified as brown.
  - ESCO occupations link to essential and optional knowledge, skills and competences; skills can be cognitive or practical; competence refers to broader adaptive ability.
  - Matching limitations: attributes from Onemev and O*NET are crosswalked to ESCO occupations; matching is approximate and should be regarded as approximate.

*Source: Appendix I. Model Calibration and Additional Results (1fraea2022002).*

### Box 1. Measuring Occupation Similarity

### Box 1. Measuring Occupation Similarity

### Methodology for measuring occupation similarity
- Kanders et. al. 2020 methodology uses skills information in ESCO and work characteristics in O*NET to produce pairwise occupation similarity measures by:
  - Using natural language processing techniques - a subfield of machine learning – to measure the overlap of essential and optional skills for each occupation pair. Essential skills are those skills that that are usually required when working in an occupation, independent of the work context or the employer. Optional skills refer to skills that may be required or occur when working in an occupation depending on the employer, working context or country.
  - Using vector similarity metrics, to measure the alignment of typical work activities performed in each occupation, provided by the ESCO skills hierarchy. The ESCO hierarchy specifies a categorization of skills at several levels so that skills in the same hierarchical category can be inferred to belong to the same type of work activity.
  - Using vector similarity measures, to measure the overlap between physical, interpersonal and structural work characteristics for each occupation pair.
  - Taking a simple average of the above-three similarity measures for each occupation pair.

### Occupational mobility network and descriptive findings
- The skills-based occupational mobility network (based on the occupational similarity matrix from Kanders et. al., 2021) shows:
  - Polluting occupations tend to be clustered in the periphery, whereas green occupations are more dispersed.
  - Network nodes plotted at ISCO 4-digit occupation level; node size proportional to employment; links represent similarity of skills, knowledge, competencies, work-activities, work features, education and experience.
  - Polluting occupations (e.g., plant & machinery and craft related trades) are clustered and less integrated; polluting occupations with larger labor share (e.g., physical/chemical technicians, engineers, transporters) are more integrated with lower transition needs.
  - Green occupations are diverse (civil engineers, skilled agriculture) and some are situated alongside polluting occupations (e.g., plant & machinery and craft related trades), suggesting overlap and feasible occupational pathways.

### Job-transition possibilities and measured job-matches
- Computation: number of job-matches per occupation = sum of pair-wise occupational similarity across each occupation type (green, polluting, other).
- Key descriptive findings:
  - The median job match for polluting-to-green occupation transitions is low – about 4 matches (but there is some heterogeneity).
  - Example heterogeneity: mining engineers have a low (below-median) job-transition match into green occupations compared to chemical engineers who have high (above-median) matches.
  - It is easier to transition between green occupations and from other occupations to green.
  - For transiting into an “other” occupation, polluting occupations tend to have relatively lower job matches, but they are higher than their transition possibility into green occupations.
  - Illustration: mining engineers (polluting occupation) have the highest job-match with geologists (green occupation) and mechanical engineers (other occupation) but no match with that of horticulturalists (green occupation) or dentists (other occupation).

### Role of on-the-job training, work experience, and job-zones
- O*NET job-zone classification:
  - Job-zones 1 and 2: typically require low education (up-to a high school diploma), some previous work experience and a low-level of training (a few months or through an apprenticeship program).
  - Job-zones 3 and 4: typically require a vocational or bachelor’s degree, requisite previous work experience and a high level of training (several years of on-the-job and/or vocational training).
  - Job-zone 5: typically require a graduate degree or more, high level of work experience and some training (if skills were not acquired through education or experience).
- Findings:
  - Median number of polluting-to-green job matches are relatively higher for polluting occupations in job zones 3 and 4 (medium education but high training level).
  - Regression exercise (as reported) regresses training, experience and education for each polluting occupation o, on its job-match similarity with every other green occupation s and repeats for other non-green occupations. The specification is reported as:
    �  퐽퐽퐽퐽퐽퐽퐽퐽퐽퐽퐽퐽   푐푐ℎ
    표표,푠푠
    푠푠∈퐺퐺퐺퐺퐺퐺퐺퐺퐺퐺, 푂푂푡푡ℎ퐺퐺퐺퐺
    =훽훽푇푇푇푇 퐽퐽푇푇푇푇푇푇푇푇푇푇
    표표
    +훾훾퐸퐸훾훾훾훾
    표표
    +휃휃퐸퐸퐸퐸퐸퐸
    표표
    +  휀휀
    표표
  - Standardized-effect results:
    - A one standard deviation increase in a polluting occupation’s training or work-experience level is significantly associated with an increased green job match by about 0.2 standard deviations.
    - The level of education is less strongly associated with possible green transitions.

### Automation, remote-work readiness, and implications for transitions
- Using the occupation-specific automation index (Kanders et al., 2020) and remote-working index (del Rio-Chanona et al. 2020):
  - Polluting occupations are less likely to be automated and have less capabilities of remote-working.
  - Green occupations can be more easily automated with higher ability of remote-working.
  - Implication: polluting occupations are less at risk from automation but face more obstacles in shifting to new ways of working, which could lower future job mobility if workers are not appropriately trained.

### Job multipliers, carbon adjustments, and sectoral targeting
- Framework: semi-closed input–output model with partially endogenized consumption to quantify employment impacts of carbon emissions reduction. Three employment multipliers derived:
  - Direct job multiplier: increase in the number of direct jobs in a sector from a unit increase in its output,
  - Indirect job multiplier (Leontief backward linkages): increase in number of jobs in other sectors from a unit increase in a sector’s output,
  - Induced job multipliers (forward linkages): increase in number of jobs from increased household demand (through labor compensation) feeding into all sectors from a unit increase in a sector’s output.
- Augmentation: multiply direct, indirect and induced multipliers by output per unit of carbon emissions to obtain carbon-adjusted employment multipliers.
- Empirical findings (2008–2019):
  - Between 2008 and 2019 direct job multipliers fell in all sectors.
  - Indirect and induced multipliers decreased relatively more in carbon-intensive sectors compared to low-carbon sectors.
  - For the majority of sectors, the combination of indirect and induced effects was also greater than direct effects.
  - Falling employment multipliers for high-carbon sectors are mostly driven by a fall in induced household demand for products from other high-carbon sectors.
- Reported job-multiplier numbers in figures (jobs per million of output):
  - Top figure values (2008 vs 2019, Low-carbon vs High-carbon, Direct/Indirect/Induced):
    - 8.3 7.1 7.0 6.3 6.6 5.6 6.8 5.6 3.9 3.2 3.9 2.1
  - Lower panel breakdown (2008 vs 2019, components):
    - 8.3 7.1 7.0 6.3 2.6 1.9 3.1 2.3 4.0 3.7 3.7 3.3 1.3 1.0 1.3 0.6 2.6 2.3 2.6 1.4
  - Environmental job multipliers (affected jobs per unit of carbon emissions) per sector listed in figure (values presented as sequence heading into sector names):
    - 00.5 11.5 22.5 33.5
    - Sectors shown (in same order): MANUFACTURE OF PETROLEUM; WATER SUPPLY/ SEWERAGE; MANUFACTURE OF CHEMICALS; MINING; TRANSPORTATION AND STORAGE; MANUFACTURE OF FOOD; ARTS, ENTERTAINMENT AND RECREATION; CONSTRUCTION; MANUFACTURE OF ELECTRICALS; TELECOMMUNICATIONS; REAL ESTATE; OTHER MANUFACTURING; RESEARCH AND DEVELOPMENT; HUMAN HEALTH ACTIVITIES; COMPUTER PROGRAMMING/INFORMATION
  - Interpretation from text:
    - Tertiary and service-oriented sectors such as IT, R&D, health services, telecommunications, have high carbon adjusted employment multipliers.
    - Sectors such as mining, transportation, chemicals manufacturing have low environmental employment multipliers.
    - Manufacturing of electronics and machinery have relatively high employment effects per unit of carbon emissions change due to strong economy-wide integration.

### Policy considerations and institutional measures
- Summary of recommendations:
  - Training and work experience (but not necessarily education) are associated with improved job matches; policies should emphasize training and experience dissemination.
  - Improve dissemination of work experience (e.g., employer ratings or evaluations) to improve employment outcomes in entry-level labor markets; example policy: subsidizing firms for hiring young workers (as used under the French recovery plan).
  - Governments and private sector could regulate skills-based measures of worker performance (employment tests or simulated work experience) to avoid stigmatization of previous (high-carbon) job contexts.
  - Adopt a data-driven strategy using granular skills matches to map career pathways for workers affected by the climate transition; public-sector agencies can use these maps to aid job seekers.
  - Invest in sectors with potential to create sustainable green jobs (as identified in Section E) or support evolution of skills in affected sectors.
  - Link new training programs and pedagogy to digital tools given shift to new ways of working and low remote-work readiness in carbon-intensive industries.
  - Emphasize development of inter-sectoral qualifications to avoid occupations being locked-in by non-transferable sector-specific skills (ILO, 2018).
  - Work with private sector to raise attractiveness of green jobs to increase labor supply.
  - Strengthen institutional setup: examples noted include providing legal frameworks for green skills and training or establishing multi-year green growth plans.
  - Consider bolstering long-term support programs for displaced workers unable to transition easily.

*Italic: Source: 1fraea2022002 - Box 1. Measuring Occupation Similarity (extracted content from provided PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1fraea2022002.pdf_