## 1. Economy-wide, can a strategy that relies solely on subsidies and public investments be effective even without explicit carbon pricing? Do R&D subsidies change that calculus?

## Source details

**Canonical URL:** [1. Economy-wide, can a strategy that relies solely on subsidies and public investments be effective even without explicit carbon pricing? Do R&D subsidies change that calculus?](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021290-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021290-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021290-print-pdf.pdf.json)

---

### Major research questions
- Economy-wide effectiveness of subsidies and public investments without explicit carbon pricing; role of R&D subsidies.
- Economic cost of rapid versus delayed decarbonization in the power sector.
- Effectiveness of regulation in the power sector.
- Labor displacement associated with greening the power sector.

### Models and relation to literature
- Integrated assessment model with endogenous fuel-specific technological change (application of Barrett (2021), building on Hassler and others (2020) and Golosov and others (2014); featured in IMF (2020)).
- Detailed sectoral model (ENV-IMF) for power sector, timing, regulation, and sectoral re-allocation (related to Van der Mensbrugghe, 2019; Château and others, 2014).
- Related empirical/work modeling work includes Stock and Stuart, 2021.

### Key findings (economy-wide)
- Subsidies cannot substitute for explicit carbon pricing; even a moderate carbon tax greatly economizes on fiscal cost of reducing emissions.
- R&D subsidies are useful and may create positive international spillovers but do not meaningfully change the basic conclusion that carbon pricing is essential.
- Early decarbonization of the power sector is more economically costly initially, but only marginally; environmental benefits are large.
- Renewable energy standards (regulation) can be effective if they provide an appropriate shadow cost of carbon; nonetheless, a carbon tax remains preferable.

### Intuition from simple policy examples (numerical details preserved)
- Carbon tax (coal-only illustrative):
  - Coal CO2 intensity: four tons CO2 per ton oil equivalent.
  - Coal cost: around $80 per ton oil equivalent.
  - A $10 per ton CO2 tax raises coal cost to 80+10*4=$120 per ton oil equivalent (a ~50 percent increase).
  - Elasticity of substitution = 2 → coal use relative to other fuels falls by approximately 55 percent (composition channel).
  - Coal emissions share in the U.S.: around 10 percent → total emissions fall by around 5.5 percent via composition channel.
  - Expenditure share of coal in energy: around 3 percent → 50 percent increase in coal cost raises overall energy cost by around 1.5 percent.
  - Elasticity of demand = 0.25 → quantity channel reduces emissions by around 0.4 percent.
  - Result: total decline in emissions in the order of 5.0 percent.
- Green production subsidy (illustrative):
  - A 30 percent subsidy on green fuels fiscally approximately equivalent to a $10 carbon tax.
  - U.S. annual CO2 production assumed at 5 gT; a $10 carbon tax produces revenues of $45bn (as emissions fall ~10 percent).
  - Under assumed elasticities, a 30 percent subsidy causes near-doubling of renewables market from around $80bn to $150bn.
  - Public cost ≈ 30 percent of $150bn = around $45bn.
  - Composition effect: renewables input share grows 3 percentage points, from 5 percent to about 8 percent.
  - Expenditure share of renewables ≈ 10 percent → energy price falls by around 3 percent.
  - Elasticity of demand = 0.25 → demand for all fuels grows by 0.75 percent.
  - Net impact: emissions decline by 2.25 percent (3 percent reduction via substitution, 0.75 percent increase via quantity channel).
- R&D subsidies:
  - Encourage research and permanent greening but effects are slow.
  - Production subsidies do not alter private production costs permanently and may deter research.
  - Carbon taxes are essential in the short run to bridge the gap until induced technical change takes effect.

### Policy experiments: subsidies-only vs. subsidies-plus-carbon-tax (model scenarios)
- Baseline (no mitigation): emissions start at around 5 gigatons CO2 per year and double by mid-century.
- Administration’s climate goals mapped: emissions below the 2005 level by 50 percent in 2030 and by 90 percent in 2050 (targeting carbon-neutral by 2050).
- Package #1 (P1): production subsidies only
  - Production subsidies: 40 percent in 2020, rising to 70 in 2030, 80 percent in 2040 and 90 percent by 2050.
  - Fiscal outcome: if fully funded by government borrowing, debt-to-GDP would rise by around 30 percent points (text truncated in source).
  - Private consumption will fall as (Ricardian) households save in anticipation of higher future taxes.
- Package #2 (P2): production and R&D subsidies
  - Production subsidies: same trajectory as P1.
  - R&D subsidies: 40 percent in 2020, rising to 70 in 2030, 80 percent in 2040 and 90 percent by 2050.
  - Fiscal/macro: effect of R&D minimal in this framework; government debt in 2050 is less than 1 percentage point lower than P1.
  - R&D front-loads technological change slowly; allows carbon tax to be reduced sooner but not before 2050.
- Package #3 (P3): production and R&D subsidies plus carbon tax
  - Production subsidies: 5 percent in 2020, rising to 20 in 2030, 30 percent in 2040, and 40 percent by 2050.
  - R&D subsidies: 90 percent in 2020, and remaining roughly unchanged to 2050.
  - Carbon tax: $17 per ton in 2020, rising to $40 in 2030, $100 in 2040, and $230 in 2050.
  - Fiscal/macro: carbon tax revenues partially offset subsidies; fiscal cost adds only around 2 percentage points to debt-to-GDP in 2050; household consumption falls only by around 1 percent in 2050.

### Key takeaways (economy-wide)
- For a given fiscal cost, carbon taxes are more effective than production subsidies because they:
  - Discriminate against cheap, dirty fuels; and
  - Activate both composition and quantity channels to reduce emissions.
- Production subsidies and short-run R&D subsidies have limited short-run effectiveness and can increase energy usage via the quantity channel; R&D subsidies have larger long-run impacts but are slow.
- Ambitious near-term targets likely require carbon pricing combined with subsidies and R&D support to manage short-run emissions and induce long-run technical change.
- Supportive fiscal measures are needed to manage labor reallocation and geographically concentrated impacts during the transition.

---

### Power sector: timing, regulation, technology, and labor reallocation

### Model, baseline assumptions, and electricity mix facts
- Model: IMF-ENV, a global, dynamic, sectoral CGE model.
- BAU (no new policies after 2021) and EV penetration scenario: EVs reach 100 percent of the stock of cars by 2050.
- 2020 electricity mix: fossil fuels share 64 percent
  - Natural gas: 44 percent in 2020.
  - Coal: 19 percent in 2020.
- BAU in 2050 projections:
  - Gas: 43 percent.
  - Coal: 11 percent.
  - Wind: increases from 7 to 15 percent.
  - Solar: increases from 3 to 11 percent.

### Decarbonization scenarios (both target 90 percent non-fossil electricity)
- Early decarbonization:
  - Reach 90 percent non-fossil generation by 2035.
  - Subsidies (feed-in-tariffs for wind and solar) rise to 30 percent by 2035, then progressively phased out during 2035-45; regulation constraint remains thereafter.
- Delayed decarbonization:
  - Reach 90 percent non-fossil generation by 2045.
  - Subsidies rise to 30 percent by 2045, then progressively phased out; regulation constraint remains.

### Box 1: power sector representation and regulation
- Electricity provider chooses optimal mix across sources (solar, hydro, nuclear, wind, oil power, gas power, coal power, other renewables).
- Production function F(.) is nested CES of generation by sources.
- Regulation is an additional constraint imposing minimum share of non-fossil power (Φ) increasing to 90 percent in 2035 (early scenario) or 2045 (delayed).
- Regulation does not discriminate across fossil fuels; reductions occur evenly across gas and coal.
- Subsidies directly boost solar and wind more than other non-fossil technologies.

### Emissions outcomes and timing
- Under either early or delayed policy, power sector and total emissions substantially and permanently lower compared to EV scenario.
- If action is delayed, power-sector emissions are the same only after 2045; cumulative emissions over 2021-2050 are higher when delayed.
- Power sector emissions reduced by more than 80 percent (from more than 1200 tons of CO2 today to less than 280 tons in 2050).  
  - Note: not reduced by 90 percent because regulation targets non-fossil share, not carbon content.

### Economic impacts and dynamics (Box 1 results)
- GDP losses peak when regulation becomes fully operational:
  - Losses up to about 1 percent when regulation becomes fully operational (2035 under early decarbonization or 2045 under delayed decarbonization).
  - Thereafter GDP losses decrease, reaching around 0.4 percent in 2050.
- Costs rise initially because regulation becomes stringent and rising subsidies create price distortions despite capital reallocation.
- After subsidies phased out, GDP costs decrease; net present value of economic costs from faster transition is marginal.
- Carbon tax is marginally more efficient and easier to administer than regulation constraint.
- Current model omitted co-benefits of emission reductions; including co-benefits would lower transition costs.
- Employment moderately decreased overall.

### Prices and international linkages (selected figures)
- BAU international price projections (2021–2050):
  - International liquid fossil fuel prices projected to grow by 0.8% per year.
  - International gaseous fossil fuel prices projected to grow by 0.35% per year.
  - International coal prices decline by 0.1% per year.
- Under EV penetration (relative to BAU) by 2050:
  - US production prices of refined and crude oil are lower by 4% and 8%, respectively.
  - US electricity production price rises by 5%, raising coal price by 7%; gas prices roughly unchanged.
- Under early decarbonization (relative to EV scenario) in 2050:
  - Demand for coal and gas by power sector falls drastically; their production ends up 3% and 2.5% lower, respectively.
  - Electricity consumer prices fall up to 7% in 2035 as subsidies roll out; after phasing out subsidies in 2035, consumer and electricity prices increase by about 6% in 2050.

### Sectoral impacts and employment (selected quantitative outcomes)
- Early decarbonization relative to EV penetration:
  - Coal extraction: production cut of more than 50 percent.
  - Gas distribution and manufacture: decline by more than 40 percent.
  - Natural gas production: declines by 7 percent (reduced domestic demand largely offset by export rise).
  - Transportation sector activities: reduced by 5 percent.
  - Business services: production reduced by 1.5 percent (temporary; largely disappears once electricity subsidies phased out by 2050).
- Employment changes (relative to 2020 levels, in 2050):
  - Coal extraction: around than 70,000 job losses.
  - Gas manufacturing and distribution: around 100,000 job losses.
  - Employment in electricity sectors almost fully offsets job losses in fossil fuel sectors in model outcomes.
- Labor policy implications:
  - Need for reallocation policies: retraining, geographic mobility support, targeted redevelopment.
  - Transition not frictionless; actual unemployment could rise more than model predicts due to labor market rigidities and skills mismatches.
  - Quality of jobs created versus lost may differ (bargaining power, benefits).

### Sensitivity analysis (percent deviations from EV penetration scenario)
- Fiscal financing: wage tax vs lump sum tax (real GDP, household consumption, employment, wages, CO2 fossil combustion)
  - Real GDP (wage tax financing): 2025 -0.2; 2035 -1.1; 2050 -0.4
  - Real GDP (lump sum financing): 2025 0.1; 2035 -0.9; 2050 -0.1
  - Household Consumption (wage tax financing): 2025 -0.3; 2035 -1.5; 2050 -0.6
  - Household Consumption (lump sum financing): 2025 0.1; 2035 -1.3; 2050 -0.3
  - Employment (wage tax financing): 2025 -0.1; 2035 -0.5; 2050 -0.1
  - Employment (lump sum financing): 2025 0.2; 2035 -0.2; 2050 0.2
  - CO2 fossil combustion (wage tax financing): 2025 -7; 2035 -32; 2050 -33
  - CO2 fossil combustion (lump sum financing): 2025 -7; 2035 -32; 2050 -33
- Decomposition of early decarbonization into instruments (percent deviations):
  - Real GDP:
    - Subsidies only: 2025 -0.15; 2035 -0.68; 2050 -0.02
    - Regulation only: 2025 -0.05; 2035 -0.43; 2050 -0.39
    - Early decarbonization: 2025 -0.20; 2035 -1.11; 2050 -0.41
  - Household Consumption:
    - Subsidies only: 2025 -0.21; 2035 -0.86; 2050 0.00
    - Regulation only: 2025 -0.07; 2035 -0.67; 2050 -0.64
    - Early decarbonization: 2025 -0.28; 2035 -1.53; 2050 -0.64
  - Employment:
    - Subsidies only: 2025 -0.08; 2035 -0.29; 2050 0.01
    - Regulation only: 2025 -0.02; 2035 -0.17; 2050 -0.11
    - Early decarbonization: 2025 -0.10; 2035 -0.45; 2050 -0.10
  - CO2 from fossil combustion:
    - Subsidies only: 2025 -5; 2035 -17; 2050 -1
    - Regulation only: 2025 -2; 2035 -14; 2050 -32
    - Early decarbonization: 2025 -7; 2035 -32; 2050 -33
- Interpretation:
  - Lump sum taxation yields more benign aggregate impacts (especially short run) while achieving same electricity mix and CO2 reductions.
  - Short-run adverse impacts on GDP, consumption, and employment come mostly from distortive electricity subsidies.
  - Regulation becomes more important in the medium term; subsidies are mostly phased out by 2050.

### Shadow cost of regulation (Box on measuring via carbon tax)
- Carbon tax adjusted endogenously to replicate power-sector emissions under the regulatory early-decarbonization scenario.
- Implied carbon tax trajectory: rises to US$44/tCO2 by 2035 and US$76/tCO2 by 2050.
- Fiscal and distributional differences:
  - Income tax adjustment to balance budget is 2 percentage points lower in 2035 under carbon tax scenario than under regulation.
  - Carbon tax finances half of solar and wind subsidies in carbon tax scenario.
  - Carbon tax is more efficient because it discriminates across fossil fuels by carbon content; regulation reduces fossil fuels uniformly.
  - Carbon taxes may impose uneven household costs; flanking measures recommended to protect low-income households.

---

### Fiscal, policy, and implementation context for U.S. climate agenda

### Recent U.S. targets and policy emphasis
- Re-joined Paris Agreement.
- April 2021 NDC: reduce U.S. net greenhouse gas emissions by 50-52 percent below 2005 levels in 2030.
- Additional announced goals: a net-zero economy by 2050 and 100 percent clean electricity generation by 2035.
- U.S. approach emphasizes investment and subsidies coupled with stricter regulations rather than explicit carbon pricing.

### Planned/legislated investments and spending (selected figures)
- Infrastructure Investment and Jobs Act: around $190bn investments in power infrastructure, public transport and rail services, as well as electric vehicles.
- Build Back Better Act (yet to be signed into law): $460bn planned for clean energy.
- Administration planning to spend around $70bn on environmental adaptation measures focused on resilience against droughts, heat, floods, and wildfires.

### Table of selected legislative spending items (as presented)
- Infrastructure Package (US bn):
  - Public Transport 105
    - of which Rail 66
  - Electric Vehicles 7.5
  - Electric Grid 65
  - Clean up Superfund and brownfield sites, reclaim abandoned mine land and cap orphaned oil and gas wells 21
  - Replace Lead Pipes 15
  - Resilience against droughts, heat, floods, and wildfires 50A
  - TOTAL 263.5
- Build Back Better (US bn):
  - Clean Energy 460.5
    - of which tax credits 320
  - Electric Vehicles 21.5
  - Green House Gas Reduction Fund 29
  - Home Energy Efficiency 19
  - Workforce development for climate resilience 20A
  - TOTAL 550

### Constraints and risks to success
- Build Back Better Act not yet law; final details may change.
- Public budgets cannot bear all needed climate investments alone; private sector participation is critical.
- Most targets require more than the current political term to realize; policy continuity is an important risk.

### Policy recommendations (from chapter context)
- Administration should follow through with a combination of subsidies, investments, regulations, and explicit carbon pricing.
- A subsidies-and-investments–only package will be very fiscally costly and inefficient.
- R&D subsidies are useful but insufficient on their own or combined only with other subsidies.
- Early power-sector decarbonization is critical and not very costly; it allows phasing out expensive subsidies sooner and reduces cumulative emissions.
- Regulation is useful in the power sector (easier to administer and enforce) but a carbon tax remains more efficient; any carbon pricing should be paired with flanking measures to address distributional concerns.
- Supportive fiscal policies needed to cushion geographically concentrated impacts and facilitate labor reallocation (retraining, mobility, hiring incentives).

*Italic: Source — wpiea2021290-print-pdf (excerpt provided).*

### 1.      Economy-wide, can a strategy that relies solely on subsidies and public investments

### 1.      Economy-wide, can a strategy that relies solely on subsidies and public investments be effective even without explicit carbon pricing? Do R&D subsidies change that calculus?

### Major research questions
- 1. Economy-wide, can a strategy that relies solely on subsidies and public investments be effective even without explicit carbon pricing? Do R&D subsidies change that calculus?
- 2. Regarding the power sector, how economically costly is it to decarbonize the sector quickly versus transitioning over a longer horizon?
- 3. Can regulation in the power sector provide effective incentives?
- 4. And, finally, what would be the labor displacement associated with greening the power sector?

### Models and relation to literature
- An economy-wide integrated assessment model with endogenous fuel-specific technological change is used to assess the impact of subsidies, including R&D, and to compare outcomes to those with carbon taxes. This model is an application of Barrett (2021), building on Hassler and others (2020) and Golosov and others (2014), and was featured in IMF (2020). Integrated assessment models were pioneered by Nordhaus (see the review in Nordhaus, 2011).
- A detailed sectoral model (ENV-IMF) is used to study the power sector, timing of decarbonization, the use of regulation, and implied sectoral re-allocation. The ENV-IMF model is related to “ENVISAGE” (Van der Mensbrugghe, 2019) and the “OECD ENV-Linkages Model” (Château and others, 2014).
- Related work includes Stock and Stuart, 2021, which assesses a wider set of policies for the U.S. power sector and their optimality in a more detailed model.

### Key findings
- 1. On an economy-wide basis, subsidies cannot substitute for explicit carbon pricing and even a moderate carbon tax can greatly economize on the overall fiscal cost of reducing emissions (reducing the need to raise other taxes to pay for the subsidies). R&D subsidies can play a useful role, and may create positive international spillovers, but do not meaningfully change this basic conclusion.
- 2. Early decarbonization of the power sector is more economically costly initially, but only marginally. The benefits for the environment are, however, large.
- 3. Regulations in the form of a renewable energy standard can be effective in the power sector insofar as they provide an appropriate shadow cost to carbon. Nonetheless, a carbon tax would still be preferable.

*Source: wpiea2021290-print-pdf - 1.      Economy-wide, can a strategy that relies solely on subsidies and public investments be effective even without explicit carbon pricing? Do R&D subsidies change that calculus?*

### 4. Finally, during the transition to green power labor would have to shift across sectors.

### 4. Finally, during the transition to green power labor would have to shift across sectors.

### Labor reallocation and supportive fiscal policy
- During the transition to green power labor would have to shift across sectors.
- Supportive fiscal policies would likely be needed to:
  - cushion the impact of some of the most geographically concentrated effects; and
  - facilitate the reallocation of labor (potentially through interventions to support to acquire new skills, move geographically, incentivize hiring).

### Federal focus and limits; role of states
- The paper focuses on federal policies; detailed analysis of state-level policies is outside its scope.
- Two reasons state-level analysis matters:
  - The extent to which any federal climate policy is binding will vary considerably across states, as some jurisdictions have much more ambitious local climate policies already in place.
  - The federal government may have a limited set of levers in several crucial areas (power and transportation sectors) where states influence investment choices (e.g., allocations to public transport relative to new roads, or more renewable energy relative to gas-fired power plants).

### Structure of the paper
- Section II: reviews the U.S. administration’s recent climate announcements and puts them in context.
- Section III: assesses economy-wide climate policy options including comparing subsidies-only and subsidies-plus-taxes packages.
- Section IV: studies the power sector in detail.
- Section V: concludes.

### Ambitious climate plans and recent actions (key facts and figures)
- The U.S. is the world’s largest economy and the second-largest greenhouse gas emitter after China.
- In 2019 U.S. emissions – total and per capita – were close to those in 1990.
- Electricity generation and transportation represent on average around 31 and 26 percent of total emissions, respectively.
- Historical policy context:
  - U.S. pledged to lower 2020 emissions by 17 percent compared to 2005 levels (Copenhagen accord), but commitments generally not backed by sufficiently ambitious policies.
  - The Clean Power Plan (CPP) aimed to reduce 2030 carbon emissions from the power sector by 32 percent below their 2005 levels.
  - The pre-2017 administration’s NDC for 2025 aimed to reduce emissions by 26 to 28 percent below 2005.
  - From January 2017 to January 2021 Brookings’ Deregulation tracker counts 74 actions that weakened environmental protection.
  - Policy rollbacks from 2017–2021 might have added up to 1.8 gigatons of CO2-equivalent by 2035 (Pitt et al., 2020).
- New administration pledges and targets:
  - Re-joined the Paris Agreement.
  - April 2021 NDC aims to reduce U.S. net greenhouse gas emissions by 50-52 percent below 2005 levels in 2030.
  - Additional announced goals: a net-zero economy by 2050 and 100 percent clean electricity generation by 2035.
- U.S. approach emphasizes investment and subsidies coupled with stricter regulations rather than explicit carbon pricing.
- Planned/legislated investments and spending:
  - Infrastructure Investment and Jobs Act includes around $190bn investments in power infrastructure, public transport and rail services, as well as electric vehicles.
  - Build Back Better Act (yet to be signed into law) has $460bn planned for clean energy.
  - Administration planning to spend around $70bn on environmental adaptation measures focused on resilience against droughts, heat, floods, and wildfires.
- Recent regulatory and executive actions:
  - Administration announced that 50 percent of all new passenger cars and light trucks sold in 2030 should be “zero-emission” through new Federal fuel economy standards.
  - Executive actions include halting new oil and natural gas leases on public lands and waters; establishment of a Climate Change Support Office within the Department of State; revocation of a key crude pipeline permit; and development of a government-wide strategy to assess and mitigate climate-related financial risks.
- State-level activity:
  - By August 2021, 31 states and two territories have active renewable or clean energy requirements; three states and one territory have set voluntary renewable energy goals.
- Constraints and risks to success:
  - Build Back Better Act is yet to be put into law and its final details may change.
  - Public budgets cannot bear all needed climate investments alone; private sector participation is critical.
  - Most targets require more than the current political term to realize; policy continuity is an important risk to outcomes.

### Table of selected legislative spending items (as presented)
- Infrastructure Package (US bn):
  - Public Transport 105
    - of which Rail 66
  - Electric Vehicles 7.5
  - Electric Grid 65
  - Clean up Superfund and brownfield sites, reclaim abandoned mine land and cap orphaned oil and gas wells 21
  - Replace Lead Pipes 15
  - Resilience against droughts, heat, floods, and wildfires 50A
  - TOTAL 263.5
- Build Back Better (US bn):
  - Clean Energy 460.5
    - of which tax credits 320
  - Electric Vehicles 21.5
  - Green House Gas Reduction Fund 29
  - Home Energy Efficiency 19
  - Workforce development for climate resilience 20A
  - TOTAL 550

### Subsidies vs. carbon pricing: model framework and economic channels
- Modeling framework:
  - Application of Barrett (2021), building on Hassler and others (2020): an integrated assessment model where carbon emissions raise temperatures which reduce productivity.
  - Energy, capital, and labor are inputs to aggregate output (GDP); energy production generates emissions; eight fuel types modeled: coal, conventional oil, unconventional (fracked) oil, natural gas, renewables, hydro power, and nuclear (first four are fossil fuels with positive carbon intensity).
  - Direction of technical change is endogenous: research incentives depend on market size, affecting long-run technology sets.
- Two economic channels determine emissions:
  - Quantity channel: changes in total energy use change emissions; trade-off between output and emissions arises because energy is an input to GDP.
  - Composition channel: changes in fuel mix for a fixed energy quantity change emissions; depends on relative fuel prices and elasticity of substitution between fuels (typically estimated to be relatively elastic).
- Implications:
  - Pure composition effects do not impact quantity of energy used and thus induce no trade-off between emissions and output.
  - Policies that change relative prices (taxes, subsidies) operate through both channels, with differing net effects.

### Intuition through simple policy examples (numerical details preserved)
- Carbon tax example (coal-only, illustrative):
  - Coal produces four tons of CO2 per ton oil equivalent.
  - Coal cost around $80 per ton oil equivalent.
  - A $10 per ton CO2 tax raises cost from $80 to 80+10*4=$120 per ton oil equivalent (a ~50 percent increase).
  - With elasticity of substitution = 2, coal use relative to other fuels falls by approximately 55 percent via the composition channel.
  - Coal emissions share in the U.S. is around 10 percent → total emissions fall by around 5.5 percent due to composition channel alone.
  - Expenditure share of coal in energy is around 3 percent → a 50 percent increase in coal cost raises overall energy cost by around 1.5 percent.
  - With elasticity of demand = 0.25, total emissions fall via the quantity channel by around 0.4 percent.
  - Result: total decline in emissions in the order of 5.0 percent (composition + quantity effects).
  - Interpretation: carbon taxes are effective because they sharply penalize cheap, dirty fuels; most of the reduction comes from composition effects; welfare effects are typically small since households’ energy usage falls only modestly via the quantity channel.
- Green production subsidy example:
  - A 30 percent subsidy on green fuels is fiscally approximately equivalent to a $10 carbon tax (revenues/expenditure similar in magnitude).
  - U.S. annual CO2 production assumed at 5 gT; a $10 carbon tax produces revenues of $45bn (as emissions fall ~10 percent in response).
  - Under assumed elasticities, a 30 percent subsidy causes near-doubling of renewables market from around $80bn to $150bn.
  - Public cost ≈ 30 percent of $150bn = around $45bn.
  - Subsidies make energy cheaper overall; composition and quantity effects work in opposite directions.
  - In the example:
    - Composition effect: input share of renewables grows 3 percentage points, from 5 percent to about 8 percent.
    - Expenditure share of renewables ≈ 10 percent → price of energy overall falls by around 3 percent.
    - With elasticity of demand = 0.25, demand for all fuels grows by 0.75 percent.
    - Net impact: emissions decline by 2.25 percent, split into a 3 percent reduction via the substitution channel and a 0.75 percent increase via the quantity channel.
  - R&D subsidies differ from production subsidies:
    - Production subsidies do not alter private production costs permanently and may deter research.
    - R&D subsidies encourage research and permanent greening of technologies, but effects are slow.
    - Conclusion: carbon taxes are essential in the short run to bridge the gap until induced technical change takes effect.

### Policy experiments: subsidies-only vs. subsidies-plus-carbon-tax (model scenarios)
- Baseline (no mitigation policies):
  - Emissions continue to rise, starting at around 5 gigatons CO2 per year and doubling by mid-century.
- Administration’s climate goals mapped into emissions path:
  - Emissions below the 2005 level by 50 percent in 2030 and by 90 percent in 2050 (targeting carbon-neutral by 2050).
- Policy packages considered (summary reference to Table 2 in the source):
  - Package #1 (P1): subsidies for production of renewables alone, starting at around 40 percent and rising to over 90 percent by 2050.
    - Such high subsidies are necessary because subsidies are relatively ineffective in reducing emissions in the short run (they lower the overall cost of energy, increasing usage and undermining composition shifts).
    - Fiscal outcome: If fully funded by government borrowing, the debt-to-GDP ratio would rise by around 30 percent points (text truncated thereafter in source).

### Key takeaways from the analysis
- For a given fiscal cost, carbon taxes are more effective than production subsidies at reducing emissions because:
  - They discriminate more sharply against cheap, dirty fuels; and
  - Composition and quantity effects both act to reduce emissions under taxes.
- Production subsidies (and short-run R&D subsidies) have limited short-run effectiveness and can increase energy usage via the quantity channel; R&D subsidies can have larger long-run impacts but operate slowly.
- Meeting ambitious near-term targets likely requires carbon pricing in combination with subsidies and R&D support to manage short-run emissions and induce long-run technical change.
- Supportive fiscal measures are needed to manage labor reallocation and geographically concentrated impacts during the transition.

*Italic: Source — wpiea2021290-print-pdf, section content provided.*

### 2050.  As a result, private consumption will f all as (Ricardian) households save in

### wpiea2021290-print-pdf - 2050.  As a result, private consumption will f all as (Ricardian) households save in

### Policy packages and their macroeconomic effects
- Package #1 (P1): production subsidies only
  - Production subsidies: 40 percent in 2020, rising to 70 in 2030, 80 percent in 2040 and 90 percent by 2050.
  - Fiscal/macro effects:
    - Subsidy-led approach is fiscally costly, driving government debt up.
    - Private consumption will fall as (Ricardian) households save in anticipation of higher future taxes to pay off the public debt.
- Package #2 (P2): production and R&D subsidies
  - Production subsidies: 40 percent in 2020, rising to 70 in 2030, 80 percent in 2040 and 90 percent by 2050.
  - R&D subsidies: 40 percent in 2020, rising to 70 in 2030, 80 percent in 2040 and 90 percent by 2050.
  - Fiscal/macro effects:
    - Effect of R&D is minimal in this framework; emissions targets met at an only fractionally reduced fiscal cost.
    - Government debt in 2050 is less than 1 percentage point lower than the pure subsidy case.
    - R&D subsidies have a larger effect on the long-run direction of technical change, allowing the carbon tax to be reduced sooner (although still after 2050); effect is slow because the starting share of renewables in energy is small.
- Package #3 (P3): production and R&D subsidies plus carbon tax
  - Production subsidies: 5 percent in 2020, rising to 20 in 2030, 30 percent in 2040, and 40 percent by 2050.
  - R&D subsidies: 90 percent in 2020, and remaining roughly unchanged to 2050.
  - Carbon tax: $17 per ton in 2020, rising to $40 in 2030, $100 in 2040, and $230 in 2050.
  - Fiscal/macro effects:
    - Revenues from the carbon tax are used to partially offset subsidy expenditures.
    - Carbon tax is very effective at reducing emissions, with particularly large marginal effect when the tax is low.
    - Research subsidies optimally front-loaded to maximize long-horizon effectiveness.
    - Resulting fiscal cost adds only around 2 percentage points to the debt-to-GDP ratio in 2050.
    - Household consumption falls only by around 1 percent in 2050.

### Environmental and economic impacts (comparative findings)
- By construction, all three packages imply the same emissions reduction.
- Roles of instruments in P3:
  - Carbon tax: drives down emissions fast, exploiting composition and quantity channels.
  - R&D subsidies: lower long-run cost of reducing emissions by shifting technological base toward clean fuels; improve long-run trade-offs and allow carbon taxes to be scaled back faster than under production subsidies alone.
  - Production subsidy: cushions impact of higher carbon tax on energy prices faced by households.
- Policy conclusion:
  - Packages relying on subsidies are fiscally costly, driving debt up and reducing household consumption as revenue sources need to be found to cover subsidy costs.
  - When supplemented with a moderate carbon tax, same emissions outcomes can be delivered at a much lower economic and fiscal cost.

### Energy price dynamics and technology effects (model results)
- Baseline (no policy):
  - Energy prices fall at around 1 percent per year given efficiency gains.
  - Renewable energy prices fall at around 1.5 percent per year due to research lowering costs.
  - These relative declines are insufficient to spur a significant green shift in energy composition.
- Policy impacts on prices:
  - P2 (production and R&D subsidies): price of energy falls much faster – over 2.5 percent per year on average – as higher energy usage increases energy-research and drives down input costs.
  - In subsidy examples, the cost of renewables falls more slowly than the average:
    - With production subsidies alone, renewables fall at around 1.3 percent per year.
    - With both R&D and production subsidies, price of renewable falls at 1.5 percent per year (but still more slowly than average energy price).
  - Introduction of carbon taxes:
    - Spurs shift away from fossil fuels toward renewables, increasing renewables’ market share and stimulating research.
    - Price of clean energy falls nearer 2.5 percent per year.
    - Despite this, there is an overall increase in the energy price of 0.5 percent per year in the short run due to the carbon tax.
  - Induced green innovation and shift into cheaper green fuels means trend quickly turns around; within 10 years, the average price of energy is falling once more.

### Greening the power sector: scenarios, assumptions, and key quantitative facts
- Model and baseline assumptions:
  - Analysis uses the IMF-ENV model, a global, dynamic, and sectoral Computable General Equilibrium (CGE) model.
  - Business as usual (BAU) scenario assumes no new climate and energy policies after 2021.
  - EV penetration scenario (baseline for power-sector analysis) assumes EVs reach 100 percent of the stock of cars by 2050.
- Electricity mix statistics (historical and BAU projections):
  - In 2020, the share of fossil fuels in power sector was 64 percent:
    - Natural gas: 44 percent in 2020.
    - Coal: 19 percent in 2020.
  - Under BAU in 2050:
    - Gas accounts for 43 percent.
    - Coal decreases to 11 percent.
    - Wind increases from 7 to 15 percent.
    - Solar increases from 3 to 11 percent.
- Two decarbonization policy scenarios (both target 90 percent non-fossil electricity):
  - Early decarbonization:
    - Reach 90 percent non-fossil generation by 2035.
    - Subsidies (feed-in-tariffs for wind and solar) rise to 30 percent by 2035, then progressively phased out during 2035-45; regulation constraint remains in place thereafter.
  - Delayed decarbonization:
    - Reach 90 percent non-fossil generation by 2045.
    - Subsidies rise to 30 percent by 2045, then progressively phased out thereafter; regulation constraint remains in place.
- Table 3 scenario summary (selected entries preserved verbatim):
  - Business as usual (BAU): EV penetration; Electricity Expansion from Electric vehicle, serves as a baseline for policy scenarios.
  - Early decarbonization:
    - “EV penetration” plus:
    - Subsidies in the form of feed-in tariffs for solar and wind rise to 30 percent by 2035, then are progressively phased out.
    - Regulation constraint that non-fossil fuel power equal to at least 90 percent of total power in 2035.
  - Delayed decarbonization:
    - “EV penetration” plus:
    - Subsidies in the form of feed-in tariffs for solar and wind rise to 30 percent by 2045, then are progressively phased out.
    - Regulation constraint that non-fossil fuel power equal to at least 90 percent of total power in 2045.
  - Sensitivity Analysis (Box 2 and Box 3) entries include Subsidies only; Regulation only; Carbon Price (carbon tax for power sector endogenously determined to reproduce the same CO2 path as “Early decarbonization”; revenues used to partially offset subsidies).
- Fiscal balancing rule:
  - In all policy scenarios, government balances the budget in each period by adjusting wage income tax rates to maintain the same expenditures as in the EV penetration scenario.
- EV penetration implications:
  - As EV penetration progresses, transport sector decarbonizes but electricity demand rises.
  - Higher electricity demand can increase emissions from the power sector relative to baseline in the short term because it extends the useful life of older fossil-fuel generators.
  - Greater EV penetration on its own is not sufficient, in the next few years, to substantially mitigate GHG emissions in the US; success requires concurrent decarbonization of the power sector.

*Source: wpiea2021290-print-pdf (excerpt provided).*

### Box 1. Power Sector and regulation on electricity mix in the ENV-IMF  modelling framework

### Box 1. Power Sector and regulation on electricity mix in the ENV-IMF  modelling framework

### Model representation of electricity supply
- Representative electricity provider chooses optimal mix of electricity generation combining:
  - Xely: electricity production = XTD (demand for electricity transmission and distribution services) + X(powa) (demands for power generation from primary energy sources: solar, hydro, nuclear, wind, solar, other renewables, oil power, gas power, coal power).
- Production function F(.) is a nested CES function of electricity generated by the various primary energy sources.
- Regulation is modelled as an additional constraint imposing a minimum share of non-fossil power generation (Φ) which is gradually increasing to 90 percent in 2035.

### Policy scenarios and implementation timing
- Early decarbonization scenario: regulation drives non-fossil-fuel powered electricity to reach the regulated 90 percent of total generation by 2035.
- Delayed decarbonization scenario: the same 90 percent target is reached by 2045 only, by construction.
- Regulation constraint does not discriminate across fossil fuels; reduction is even across gas and coal.
- Subsidies (price incentives) directly boost solar and wind power, increasing power generated by both more than other non-fossil technologies.

### Emissions outcomes
- Under either policy scenario, both power sector and total emissions are substantially and permanently reduced compared to the EV scenario.
- If action is delayed, the level of emissions of the power sector is the same after 2045 only; cumulative emissions over 2021-2050 are higher when delayed.
- Power sector emissions are reduced by more than 80 percent (from more than 1200 tons of CO2 today to less than 280 tons in 2050).  
  - Note: emissions are not reduced by 90 percent because the regulation does not target carbon content and therefore does not discriminate between coal and gas power.

### Economic (GDP) impacts and dynamics
- GDP losses peak when the regulation constraint becomes fully operational:
  - Losses up to about 1 percent when the regulation becomes fully operational (2035 under early decarbonization or 2045 under delayed decarbonization).
  - Thereafter GDP losses decrease, reaching around 0.4 percent in 2050.
- Costs rise initially because the regulation becomes more stringent and because rising subsidies create price distortions despite capital reallocation.
- After subsidies are phased out, GDP costs decrease over time.
- Net present value of economic costs from a faster transition is marginal; delaying decarbonization is not justified by large economic savings.
- Carbon tax is noted as easier to administer and marginally more efficient than the regulation constraint.
- The current model version does not include co-benefits of reducing emissions, which would reduce transition costs if included.
- Employment is moderately decreased overall.

### Prices and international linkages
- Commodity and factor prices are endogenous; international prices drive domestic liquid and gaseous fossil fuel prices.
- Under BAU international price projections from 2021 to 2050:
  - International liquid fossil fuel prices projected to grow by 0.8% per year.
  - International gaseous fossil fuel prices projected to grow by 0.35% per year.
  - International coal prices decline by 0.1% per year.
- Under EV penetration and policy scenarios, international prices are marginally affected (other countries assumed not to act).
- US price effects (relative to BAU or EV scenarios):
  - Under EV penetration: US production prices of refined and crude oil are lower by 4% and 8%, respectively, in 2050 relative to BAU.
  - Under EV penetration: US electricity production price rises by 5%, which raises demand for energy used by the power sector, resulting in a 7% increase in the price for coal; gas prices roughly unchanged.
  - Under early decarbonization: demand for coal and gas by the power sector falls drastically; their production ends up 3% and 2.5% lower in 2050 relative to the EV scenario.
  - Consumer prices for fossil fuels rise with production prices but to a lesser degree due to taxation offsets.
  - Electricity consumption prices fall up to 7% in 2035 as subsidies are rolled out (relative to EV scenario).
  - After phasing out subsidies in 2035, consumer and electricity prices increase by about 6% in 2050 (relative to EV scenario).

### Sectoral impacts and employment
- Early decarbonization produces strong reallocation: fossil fuel power almost shut down in 2035 while renewable power is strongly boosted.
- Key sectoral outcomes (relative to EV penetration scenario):
  - Coal extraction: production cut of more than 50 percent.
  - Gas distribution and manufacture: decline by more than 40 percent.
  - Natural gas production: declines by 7 percent (large fall in domestic demand largely offset by export rise).
  - Transportation sector activities: reduced by 5 percent in response to increasing fuel price for electric vehicles.
  - Business services: production reduced by 1.5 percent (temporary; largely disappears once electricity subsidies phased out by 2050).
- Employment changes:
  - Employment generally follows activity changes but to a lesser degree (wage bill adjustments absorb part of changes).
  - Large employment reductions in some sectors:
    - Coal extraction: around than 70,000 job losses in 2050 relative to 2020 level.
    - Gas manufacturing and distribution: around 100,000 job losses in 2050 relative to 2020 level.
  - Employment in electricity sectors almost fully offsets job losses in fossil fuel sectors.
- Policy implications for labor market:
  - Need for efforts to reallocate labor from shrinking sectors to other areas (retraining, geographic mobility, targeted redevelopment).
  - Transition will not be frictionless; actual unemployment could rise more than model predicts due to labor market rigidities and skills mismatches.
  - Quality of jobs created versus lost may differ based on bargaining power, benefits, and other factors.

### Sensitivity analysis and decomposition of early decarbonization (Box 2)
- Sensitivity to financing assumptions (numbers are percent deviations from the EV penetration scenario):

  Budget balance through adjustments to wage taxes vs lump sum taxes
  - Real GDP:
    - Wage tax financing: 2025 -0.2; 2035 -1.1; 2050 -0.4
    - Lump sum financing: 2025 0.1; 2035 -0.9; 2050 -0.1
  - Household Consumption:
    - Wage tax financing: 2025 -0.3; 2035 -1.5; 2050 -0.6
    - Lump sum financing: 2025 0.1; 2035 -1.3; 2050 -0.3
  - Gross Investment:
    - Wage tax financing: 2025 0.0; 2035 -0.1; 2050 -0.1
    - Lump sum financing: 2025 0.1; 2035 -0.1; 2050 0.0
  - Employment:
    - Wage tax financing: 2025 -0.1; 2035 -0.5; 2050 -0.1
    - Lump sum financing: 2025 0.2; 2035 -0.2; 2050 0.2
  - Real gross wage:
    - Wage tax financing: 2025 -0.3; 2035 -1.1; 2050 -0.3
    - Lump sum financing: 2025 -0.8; 2035 -1.2; 2050 -0.4
  - CO2 fossil combustion:
    - Wage tax financing: 2025 -7; 2035 -32; 2050 -33
    - Lump sum financing: 2025 -7; 2035 -32; 2050 -33

- Interpretation:
  - Lump sum taxation generates more benign aggregate impacts (especially short run) while achieving same electricity mix and CO2 reductions.
  - Distributional impacts differ across financing assumptions but cannot be assessed by the model (representative consumer).

- Decomposition of early decarbonization into instruments (numbers are percent deviations from the EV penetration scenario):

  Subsidies only, Regulation only, Early decarbonization
  - Real GDP:
    - Subsidies only: 2025 -0.15; 2035 -0.68; 2050 -0.02
    - Regulation only: 2025 -0.05; 2035 -0.43; 2050 -0.39
    - Early decarbonization: 2025 -0.20; 2035 -1.11; 2050 -0.41
  - Household Consumption:
    - Subsidies only: 2025 -0.21; 2035 -0.86; 2050 0.00
    - Regulation only: 2025 -0.07; 2035 -0.67; 2050 -0.64
    - Early decarbonization: 2025 -0.28; 2035 -1.53; 2050 -0.64
  - Employment:
    - Subsidies only: 2025 -0.08; 2035 -0.29; 2050 0.01
    - Regulation only: 2025 -0.02; 2035 -0.17; 2050 -0.11
    - Early decarbonization: 2025 -0.10; 2035 -0.45; 2050 -0.10
  - CO2 from fossil combustion (percent deviations):
    - Subsidies only: 2025 -5; 2035 -17; 2050 -1
    - Regulation only: 2025 -2; 2035 -14; 2050 -32
    - Early decarbonization: 2025 -7; 2035 -32; 2050 -33

- Interpretation:
  - Short-run adverse impacts on GDP, consumption, and employment come mostly from distortive electricity subsidies.
  - Regulation becomes more important in the medium term.
  - In 2035, subsidies and regulation contribute close to equally to CO2 decline; subsidies are more costly in GDP and employment terms due to wage taxation used to finance them.
  - By 2050, subsidies are largely phased out and lingering effects stem from regulation.

*Source: IMF-ENV model simulations (from the provided PDF content).*

### Box 1. Measuring the shadow cost of regulation with a carbon tax

### Box 1. Measuring the shadow cost of regulation with a carbon tax

### Method and scenario design
- The shadow cost of regulation is approximated with a carbon tax by endogenously adjusting the carbon tax so that power sector emissions follow the same path as in the early decarbonization scenario.
- Under the carbon tax scenario the carbon tax is set to replicate the emissions trajectory of the regulatory constraint in the early decarbonization scenario.

### Carbon price trajectory implied by the regulatory constraint
- The carbon tax implied by the regulatory constraint in the early decarbonization scenario gradually rises to US$44/tCO2 by 2035 and US$76/tCO2 by 2050.

### Fiscal and distributional differences between carbon tax and regulation
- Fiscal incidence:
  - Under the carbon tax scenario the adjustment in income tax rates to balance the budget is lower than in the decarbonization (regulation) scenario: 2 percentage points lower in 2035.
  - The carbon tax finances half of the solar and wind subsidies under the carbon tax scenario.
- Efficiency and allocation:
  - The carbon tax is more efficient because it discriminates across fossil fuel sources according to carbon content.
  - The regulation constraint does not discriminate across fossil fuel sources according to carbon content; under the regulatory constraint fossil fuel sources are uniformly reduced.
- Distributional concerns:
  - Carbon taxes may impose uneven costs across different households.
  - Any package that contains carbon taxes could include flanking measures for low-income households to encourage social acceptability and mitigate unequal effects on household consumption.

### Policy implications and recommendations (from the broader chapter context)
- The administration needs to follow-through with policies: subsidies, investments, regulations and explicit carbon pricing.
- A policy package relying only on subsidies and investments will be very fiscally costly and will not achieve climate goals efficiently.
- R&D subsidies are useful but cannot deliver the targets on their own or combined only with other subsidies.
- Decarbonizing the power sector early is critical and not very costly; early action allows phasing out expensive subsidies sooner and reduces cumulative emissions.
- Regulation can be very useful in the power sector because it is better understood and easier to administer and enforce.
- The green transition will likely create winners and losers in the short-term; supporting policies are required to cushion economic effects, sustain social support, and prevent stop-go climate policy dynamics.

*Source: IMF-ENV model simulations and accompanying discussion in Box 1 of the cited paper.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021290-print-pdf.pdf_