## The IMF‑World Bank Climate Policy Assessment Tool (CPAT) — Working Paper No. WP/2023/128 (selected sections) ## Source details **Canonical URL:** [The IMF‑World Bank Climate Policy Assessment Tool (CPAT) — Working Paper No. WP/2023/128 (selected sections)](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023128-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023128-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023128-print-pdf.pdf.json) --- ### 1. Background - Urgency and targets - Stabilizing the global climate requires rapid cuts in greenhouse gas (GHG) emissions this decade. - Achieving the Paris Agreement’s target of limiting global warming to 1.5 to 2oC above pre-industrial levels requires cutting global emissions of carbon dioxide (CO2) and other GHGs by 25 to 50 percent below 2019 levels by 2030, followed by a rapid transition to ‘net zero’ emissions sometime shortly after 2050. - Around 130 countries—accounting for about 90 percent of global GHG emissions—have proposed or set targets to be ‘net-zero’ emitters sometime around mid-century. - Current NDCs would only reduce emissions by about 12 percent by 2030 compared with 2019. - Under BAU without new or tightened policies, global emissions are likely to rise by 5 percent in 2030 compared with 2019. - Policy ambition gaps and pricing - Limiting warming to 2oC requires new measures equivalent to a global average carbon price exceeding $75 per ton of CO2-equivalent (CO2e) by 2030. - The current price from explicit carbon pricing schemes is only $5 per ton CO2e. - Fossil fuel subsidies (accounting for undercharging for both supply and environmental costs) were estimated at $5.9 trillion in 2020, equivalent to 6.8 percent of global GDP. - Barriers and need for quantitative tools - Political economy obstacles and pervasive knowledge/data gaps hinder implementation. - Country-level quantitative assessments are required covering energy systems, GHGs, revenues, GDP, and household/industry incidence across instruments and policy mixes. - CPAT overview - CPAT is a spreadsheet-based ‘model of models’ for rapid estimation of mitigation policy effects for over 200 countries. - Coverage: over 200 countries accounting for more than 95 percent of global GHG emissions. - Outputs: energy production/consumption/trade/prices; emissions; revenues; GDP and welfare; industry and household incidence (deciles, urban vs. rural, horizontal equity); development co-benefits (local air pollution and health impacts). - Policies evaluated: carbon taxes, ETSs, fossil fuel subsidy reform, price liberalization, fuel/electricity taxes, VAT changes, regulations, feebates, renewables subsidies, public investment, and policy mixes. - CPAT collates comprehensive datasets including energy consumption/prices, GHGs, local air pollutants, elasticities, environmental costs, NDCs, and decile-level household consumption data for more than 65 countries. - Emphasis on a ‘just transition’ by estimating impacts on poverty, equity, and welfare across income groups and urban/rural households. ### 2. Mitigation Module — structure and key mechanics - Scope and sectoral/fuel disaggregation - Energy consumption split into 15 fuels and electricity sources produced or consumed by 17 sectors. - Energy sources: coal, natural gas, gasoline, diesel, kerosene, liquified petroleum gas (LPG), jet fuel, other oil products, electricity, wind, solar, hydro, other renewables, nuclear, and biomass. - Sectors include power generation; transport; buildings; industries (multiple subsectors); other energy use; non-energy use. - Dynamic energy demand formulation (symbols preserved) - Core equation: (1) 퐸푡 = (푢푡/푢푡−1 ∙ ℎ푡/ℎ푡−1) ∙ 퐸푡−1 - 푢푡/푢푡−1 = (퐺퐷푃푡/퐺퐷푃푡−1)^푣푡 ∙ (ℎ푡∙푝푡 / ℎ푡−1∙푝푡−1)^휂푢 - ℎ푡/ℎ푡−1 = (1+훼)^{−1} ∙ (푝푡/푝푡−1)^휂ℎ - Definitions: 퐸푡 (demand), 푢푡 (usage of energy-consuming capital), ℎ푡 (energy consumption rate), 푣푡 (income elasticity), 푝푡 (price), 휂푢 (usage price elasticity), 휂ℎ (efficiency price elasticity), 0 < 훼 < 1 (autonomous efficiency improvement). - Total own-price elasticity decomposition: (2) 휂퐸 = 휂푢(−) + 휂ℎ(−) + 휂ℎ휂푢(+); rebound effect: (3) Rebound = − (휂푢 ∙ 휂ℎ) / (휂푢 + 휂ℎ). - Power sector modelling - Two alternatives: elasticity-based model and technoeconomic (hybrid) model with stock, dispatch, investment decisions, LCOE, retirement, storage needs. - Elasticity-based generation: cost-weighted shares with generation response via conditional own-price elasticities ε_i (coal elasticity set at -0.7; other fuels -0.5 in CPAT). - Technoeconomic model uses dispatch and investment logit allocation: (6) x_{f,t} / Σ_f x_{f,t} = e^{−K.c_i} / Σ_f e^{−K.c_i}. - Energy supply and price formation - Fossil fuel supply curves assumed perfectly elastic within policy ranges; unit production costs fixed within period. - Retail price: (7) 푝푅 = (푝푆 + 휏푒푥푐푖푠푒) (1 + 휏푉퐴푇) with options for pass-through scalars. - Elasticities and autonomous change - Income elasticities: 32 base elasticities adjusted by income per capita inverse-U relation; example base income elasticities provided by sector and fuel. - Own-price elasticities decomposed into usage and efficiency components; selected values preserved (e.g., implied total demand elasticity for gasoline in Transport: -0.62). - Autonomous rate of technological change set between 0.5 and 1 percent per year; exogenous efficiency improvement table includes values such as Coal: Transport 1.0%, Residential 0.5%, etc. - Emissions estimation and NDC processing - Emissions factors from IIASA (GAINS); GHGs covered: CO2, CH4, N2O, F-gases. - Non-energy emissions and LULUCF handled with specified assumptions (LULUCF linear decline for net-emitting countries). - NDCs converted into percent reductions vs. BAU in 2030 excluding LULUCF for comparability; ranges averaged at midpoint when reported. ### 3. Policy instruments modelled - Carbon taxes - Modeled as (CO2 emissions factor) × (tax rate on CO2), can be comprehensive or partial; passed forward increases fossil fuel prices and electricity prices via pure abatement costs and per-unit emission taxes. - ETSs - Modeled as a virtual tax (shadow price) with scalar 0.9 applied to emissions price to reflect smaller behavioral response vs. equivalent tax. - Other instruments - Taxes on individual fuels, electricity, methane fees. - Renewable subsidies via feed-in subsidy reducing per-unit generation cost. - VAT harmonization options. - Emission rate and energy efficiency regulations via shadow pricing (affect efficiency margin, induce rebound captured). - Feebates modeled through shadow prices added to ht component. - Fossil fuel subsidy reform using IMF dataset (consumer and producer subsidies). - Energy price liberalization with pass-through buckets (0.25, 0.5, 0.75, 1.0). - Policy mixes - Combinations possible (carbon pricing + subsidy reform + liberalization + renewables); many targeted sectoral instruments and public investments currently excluded but may be added. ### 4. Impacts on revenues, GDP, and welfare - Revenues - Direct revenues: carbon price × CO2 emissions covered; existing energy taxes × consumption; net impacts account for subsidies and renewable subsidies. - Revenue recycling options: public investment, targeted transfers, current spending, PIT/CIT cuts, or mixes. - GDP impacts - Estimated using fiscal multipliers from MFMOD and empirical studies; fiscal multipliers aggregated at country income-group level; GDP impacts lagged one year. - Aggregate evidence: GDP impacts of mitigation policies are small or ambiguous and vary with revenue recycling. - Welfare and efficiency costs - Welfare-equivalent losses estimated from integrals under marginal abatement cost schedules and include effects from pre-existing fuel price distortions and domestic environmental co-benefits. - CPAT does not yet fully capture additional welfare effects from revenue recycling efficiency gains or some broader fiscal interactions. ### 5. Planned improvements and caveats (Box 1) - Planned enhancements - More granular sectoral representation, dynamic capital turnover models for transport and buildings, refined industry-sectoral models (steel, chemicals, cement), improved GDP and trade effects, IMF-MARIO input-output database integration, and improved welfare effects to include fiscal distortions and informality. - Linkages with external models - CPAT envisioned to link to Macro-Fiscal Model (MFMOD), ENVISAGE CGE, FTT sectoral models, FARI, and others. - Key caveats (current and in development) - Non-linear responses to large policy changes abstracted; limited endogenized learning-by-doing spillovers; limited international feedbacks on fuel prices; simplified GDP impacts and limited country-specific fiscal/dynamics; sectors currently de-coupled but planned integration; price elasticities assumed realized within one year (users can adjust); initial CPAT 1.0 has specific limitations on international linkages, pass-through assumptions, and GDP effect granularity. ### 6. Transport module — reduced-form model and co-benefits - Dashboard and user interaction - User selects country, policy scenario, strength/coverage, complementary policies, revenue allocation, and parameter customizations; Manual Inputs allowed; outputs delivered within seconds with key charts (emissions vs NDCs, fiscal impacts, GDP impacts, incidence, avoided premature deaths, welfare decomposition). - Reduced-form transport impacts - Transport treated within the broader mitigation framework: fuel price changes affect VKT, congestion, accidents, and road maintenance. - Example numerical outputs: time series for energy demand by fuel (mtoe), prices ($/GJ, $/kwh, $/liter), renewable shares (%), GHG indicators (mtCO2e), and economic indicators (real 2018 US$bn and current account %GDP). - Road transport co-benefits module - Channels: congestion, road accidents, road damage. - Elasticities estimated via regressions: VKT short-run -0.28; long-run -0.56. Congestion short-run -0.34; long-run -0.81. Road fatalities short-run -0.61; long-run -0.44. Road damage long-run -0.44. - Marginal external cost of congestion uses VOT = 60 percent of nationwide average market wage in 2020. - Road maintenance externalities assumed 50 percent of total maintenance costs attributed to diesel consumption (scaled by driving portion of diesel elasticity). - Caveats - Endogeneity and data quality limitations; electrification impacts not explicitly modelled; proxy taxes on VKT may be preferable to fuel duties for some externalities. ### 7. Distribution module — methodology and outputs - Industry incidence - Input-output (GTAP-10) based microsimulation for 59 non-energy sectors across 120 countries; Leontief inverse used to compute sectoral input cost changes from energy price changes (equation (15): 푒_g = f (퐼 − 퐴)^{-1}, f ∉ g). - Default: full pass-through of input cost increases to consumer prices (user-adjustable). - Outputs: input/output price impacts, shares in GVA, output, household demand, and exports; pass-through options allow imperfect pass-through coefficients. - Household incidence - HBS-based decile-level analysis (over 65 countries): households grouped into population-weighted per-capita consumption deciles; budget shares π_{t d g} mapped to 8 energy + 14 non-energy CPAT categories. - Consumption burden computed by (16) ∑_g π_{t d g} ∙ ρ_{t d g} where ρ_{t d g} are relative price increases. - Direct impacts (energy price changes) and indirect impacts (non-energy goods prices via sectoral intensities) summed to total effect. - Behavioral adjustments and revenue recycling - Two behavioral approaches: uniform scaling reconciling mitigation vs distribution revenues; or decile- and product-specific elasticities (USDA-derived) under CES utility functions. - Four recycling modes: targeted transfers, public investment in infrastructure, scaling existing social protection, and reducing PIT liabilities (targeted exemption, personal allowance, proportional compensation formulas preserved as in text). - Transfers modelled with targeting, coverage, and leakage parameters; infrastructure transfers targeted to households without access to specified infrastructure. - Welfare metrics and reporting - Burdens measured as welfare-equivalent losses (consumer surplus losses), including extra expenditures and value of forgone consumption and deadweight losses. - Outputs include absolute LCU impacts by decile, cumulative revenues required to compensate specified bottom shares, urban/rural splits, within-decile (25th/75th percentile) horizontal equity. - Limitations - Structural change not fully captured (sector shares held constant and scaled with GDP); default full pass-through may misstate incidence in internationally competitive sectors; additional channels (capital/labor distributional shifts, local pollution benefits concentration) not fully modelled. ### 8. Air pollution and other co‑benefits (Annex III highlights) - Key health burden facts - 4.5 million deaths from outdoor air pollution in 2019. - 92 and 8 percent of outdoor air pollution deaths are due to PM2.5 and ozone, respectively. - 60 percent of outdoor air pollution deaths are from burning of fossil fuels. - Indoor air pollution caused a further 2.3 million deaths. - Road accidents cause about 1.3 million deaths per year, with 94 percent in low- and middle-income countries. - Total external costs from all road externalities estimated at almost $1 trillion in 2020, with two-thirds coming from congestion alone. - Air pollution co-benefits methodology - Four-step approach: emissions estimation by fuel/sector → translate to concentrations/population exposure via intake fractions and TM5-FASST averaged → map exposure to health burdens using GBD 2019 exposure-response curves by age/disease → monetize avoided mortality using VSL approach (approx. US$4.6 million per death avoided for 2020 in average OECD country, extrapolated by income). - Two concentration approaches: - Intake fraction method using spatial power-plant and urban databases (data for 164 countries, infer for others). - TM5-FASST source-receptor emulator downscaled to country matrices and augmented by local source apportionment. - CPAT provides five methods to estimate emissions-to-health relationships and allows sensitivity analysis. - Monetization and uncertainty - Mortality valuation draws on OECD (2012) meta-analysis; VSL ~ US$4.6 million in 2020 OECD average, adjustable by income elasticity. - Caveats: temporal/geographical averaging, uncertainty in emissions→concentration→health links. - Co-benefits aggregation - Air pollution and transport modules produce monetized co-benefits (health, congestion, accidents, maintenance) that can be compared to abatement costs to estimate net welfare impacts. ### 9. Annex technical details and selected parameter values - Units and sector mapping - Electricity in kWh; coal and natural gas in GJ; road fuels in liters; oil in bbl; other renewables and nuclear in ktoe. - Power sector elasticities and autonomous improvements - Coal generation price elasticity: -0.7; other generation fuels: -0.5. - Annual autonomous productivity improvement: Coal 0.5%; Natural gas/nuclear/hydro 1%; Renewables 5% (costs halve every 15 years). - Elasticities tables (selected preserved values) - Example implied total demand elasticities (by fuel and sector memo): - Gasoline (Transport, Residential, Industries, Services): -0.62, -0.79, -0.81, -1.04 - Electricity implied totals: -0.34, -0.42, -0.42, -0.68 - Simple averages across sectors: Transport -0.54; Residential -0.57; Industries -0.67; Services -0.80 - Rebound effect ranges: 13 to 26 percent across fuels/sectors (implied). - Non-energy and methane sectors - Methane ~20 percent of global GHGs in 2020; ~40 percent from fossil fuel extract/distribution, ~40 percent from agriculture, remainder from waste/other. - Methane-emitting sectors modelled using base-year emission factors, BAU production projections (elasticities: agriculture population 0.90, per capita income 0.15), and MACC-based policy responses. - Revenues and price formation details - Supply price composition, VAT estimation, consumer subsidy formulations, and projection formulas preserved (e.g., rp_t = sp_t + VAT_t + exo_t). - CPAT averages international price projections from IMF WEO and World Bank (2022a) by default. ### 10. Example numerical and chart outputs (as presented) - Air pollution figures (Figure 6 examples): - PM2.5 = 254 Kton, NOx = 1147 Kton, SO2 = 449 Kton by source. - Ambient PM2.5 (μg/m3) time series 2018–2034 with Urban and Rural baselines and carbon tax scenarios; WHO Guidelines reference shown. - Cumulative avoided deaths by age groups across 2020–2035 (Number of deaths). - Annual monetized welfare benefits and percent GDP welfare benefits time series decomposed into Economic costs, Transport co-benefits, Air pollution co-benefits, Climate benefits, and Net. - Energy and emissions example (Figure 4): - Example scenario: US$50 Carbon Price/ton CO2e by 2030 (Unspecified Country) showing energy demand time series (2018–2035), prices ($/GJ, $/kwh, $/liter), renewable shares at 2018, 2023, 2028, 2033 (%), GHG comparisons to Linear pathway to NZE, Baseline, Carbon tax, Unconditional NDC, Conditional NDC, and economic indicators in real 2018 US$bn and current account (%GDP). ### 11. Policy implications and recommendations (conclusions) - Rapid and unprecedented decarbonization across countries is required to stabilize the global climate. - Policy instruments: carbon pricing (taxes and ETSs), fossil fuel subsidy reform, energy market reform, renewable subsidies, feebates, green public investment, regulations, VAT harmonization, and mixes tailored to country circumstances. - Design reforms to support growth, poverty alleviation, equity, environmental quality, and energy access. - Include just transition measures: retraining, relocation, financial support for displaced workers. - Address technology market failures via prizes, R&D support, and advance market commitments. - Revenue recycling (per-capita transfers, labor tax reductions, climate investments) can enhance political acceptability; CPAT quantifies incidence to inform design but qualitative political economy analysis remains important. *Source: IMF‑WB Climate Policy Assessment Tool (CPAT), selected sections from WP/2023/128* ### 1. BACKGROUND ________________________________________________________________________________________ 3 ### 1. BACKGROUND ### Urgency of emission reductions and Paris targets - Stabilizing the global climate requires rapid cuts in greenhouse gas (GHG) emissions this decade. - Achieving the Paris Agreement’s target of limiting global warming to 1.5 to 2oC above pre-industrial levels requires cutting global emissions of carbon dioxide (CO2) and other GHGs by 25 to 50 percent below 2019 levels by 2030, followed by a rapid transition to ‘net zero’ emissions sometime shortly after 2050. - To date, around 130 countries—accounting for about 90 percent of global GHG emissions—have proposed or set targets to be ‘net-zero’ emitters sometime around mid-century. - Current NDCs would only reduce emissions by about 12 percent by 2030 compared with 2019. This is less than half of the emissions cuts needed for 2oC and less than a quarter of the emissions cuts needed for 1.5oC warming. - Under business as usual (BAU) without new or tightened policies, global emissions are likely to rise by 5 percent in 2030 compared with 2019. ### Policy ambition gaps and required instruments - Limiting warming to 2oC requires new measures equivalent to a global average carbon price exceeding $75 per ton of CO2-equivalent (CO2e) by 2030. - The current price from explicit carbon pricing schemes is only $5 per ton CO2e. - Fossil fuel subsidies (accounting for undercharging for both supply and environmental costs) were estimated at $5.9 trillion in 2020, equivalent to 6.8 percent of global GDP. ### Barriers to implementation - Political economy obstacles can be significant, but pervasive gaps in knowledge and data also hold back policymakers from implementing needed policies. - Many government departments lack knowledge about climate policies and their impacts; these knowledge gaps contribute to climate policy implementation gaps. ### Need for new tools and country-level quantitative analysis - Designing and implementing effective and sustained climate mitigation policies requires quantitative, evidence-based, country-level assessments of impacts on: - energy systems (supply, demand, and prices); - CO2 and other GHG emissions; - revenues from existing and new energy taxes; - economic output; and - household and industry incidence. - Assessment must cover tradeoffs among instruments: carbon taxes, emissions trading systems (ETSs), electricity or individual fuel taxes, emission rate and energy efficiency regulations, feebates, renewables subsidies, public investments, and combinations (‘policy mixes’). - No prior tool offered such estimation with near comprehensive country coverage. ### The IMF-World Bank Climate Policy Assessment Tool (CPAT) - CPAT is a spreadsheet-based ‘model of models’ allowing rapid estimation of effects of mitigation policies for over 200 countries. - CPAT is the result of a multi-year collaboration between IMF and World Bank staff and evolved from an earlier IMF tool; the paper describes the first version (‘CPAT 1.0’). - CPAT helps governments design and implement climate mitigation strategies by allowing: - Quantification of many impacts: energy production, consumption, trade, and prices; emissions of local and global pollutants including reductions needed to achieve NDCs; GDP and economic welfare; revenues; industry incidence (across many sectors); household incidence (across deciles, urban vs. rural samples, and horizontal equity); and development co-benefits (local air pollution and health impacts). - Evaluation of many mitigation policies: carbon taxes, ETSs, fossil fuel subsidy reform, energy price liberalization, electricity and fuel taxes, removal of preferential VAT rates for fuels, energy efficiency and emission rate regulations, feebates, clean technology subsidies, and combinations of these policies. - Coverage for many countries: CPAT covers over 200 countries accounting for more than 95 percent of global GHG emissions. CPAT’s input data is complete and there is no need for external data inputs (though users can incorporate their own data or parameter assumptions). - A transparent, user-friendly, and consistent framework: results are presented rapidly via a chart-driven interface, allowing experimentation and sensitivity analyses. - CPAT emphasizes a ‘just transition’ by estimating impacts on poverty, equity, and welfare across income groups and between urban and rural households, including within-decile (horizontal) equity. - CPAT is parametrized to be broadly in the mid-range of ex ante models and parameterized to ex post empirical literature; it is streamlined with transparent underlying parameters adjustable for sensitivity analyses. - CPAT enables cross-country analysis and consistent comparisons of mitigation ambition for over 200 countries, converting varied NDCs into a single comparable metric (required emissions reductions vs. BAU). - CPAT collates new comprehensive datasets including energy consumption and prices; GHGs; local air pollutants; price and income elasticities; environmental costs; NDCs; and comparable decile-level household consumption data for more than 65 countries. ### CPAT applications and integration in IMF work - CPAT has been used extensively by IMF staff in: - Country-level analyses in Article IV Reports, country-specific working papers, technical assistance, Climate Macro Assessment Programs (CMAPs), and in support of the Resilience and Sustainability Trust (RST). - Regional analyses (e.g., policies and impacts of global energy price shocks). - Global analyses (e.g., implications of global economic crisis for mitigation, gaps in global climate policy, and proposals for an international carbon price floor). - Thematic policy analyses for the IMF’s Staff Climate Note (SCN) series (quantified comparisons of mitigation instruments, fossil fuel subsidy reform, methane taxes, and carbon price equivalence of mitigation policies). - Training (e.g., “Macroeconomics of Climate Change” by IMF Institute for Capacity Development). - CPAT is part of a broader climate policy analysis toolbox: other models (DSGE, CGE, agent-based, macroeconometric), MRIO frameworks, innovation and engineering models, and integrated assessment models (IAMs) provide complementary analyses beyond CPAT’s 15-year horizon and scope. ### Organization of the paper - The paper summarizes CPAT 1.0’s methodology including mathematical representation, functional forms, data sources, and key parameter values. - Structure: - Section 2: overview of CPAT’s structure. - Section 3: mitigation module (energy, emissions, economic impacts). - Section 4: distribution module (household and industry incidence). - Section 5: co-benefits modules (air pollution and transport). - Section 6: conclusion. - More detailed descriptions and technical annexes are provided later in the paper. *Italicized source: IMF-WB Climate Policy Assessment Tool (CPAT), 1. BACKGROUND* ### 4. Transport module – a reduced-form model for estimating the impacts of motor fuel price changes on ### 4. Transport module – a reduced-form model for estimating the impacts of motor fuel price changes on ### Dashboard and user interaction - The user interacts primarily with the ‘Dashboard’ (Figure 3), a chart-driven, user-friendly interface. - User-selectable options: - Country of interest. - Mitigation ‘policy scenario’ (e.g., carbon or energy taxes). - Strength/coverage of the policy (across fuels and sectors). - Complementary policies (e.g., fossil fuel subsidy reform, energy price liberalization, feed-in subsidies for renewables). - Allocation of any revenues raised or saved to: tax reductions, current spending, public investment, or transfers. - Customization of key parameters (e.g., price and income elasticities). - CPAT does not require external data to function for the countries covered, but users can input external data (e.g., domestic energy prices) in the ‘Manual Inputs’ tab. - Outputs are delivered within seconds: the Dashboard shows six key charts and over 100 more detailed charts. - Key charts include: GHG emissions projections compared with NDCs in the BAU and policy scenarios; net changes in fiscal revenues by fuel source in the policy scenario; GDP impacts by component; incidence impacts on household consumption deciles; averted premature deaths from improvements in air quality and road safety; and net changes in welfare by component (abatement costs less monetized externality benefits from climate and health/transport co-benefits). ### Mitigation module (core) - Description: - A reduced-form macro-energy model producing country-by-country projections. - Outputs: energy demand and supply, prices, CO2 and other GHG emissions by fuel and sector, revenues, GDP, abatement costs, welfare impacts, and additional metrics. - Estimated under the BAU scenario and multiple mitigation policies. - Policies modeled include: carbon taxes, ETSs, fossil fuel subsidy reform, fuel/electricity taxes, energy price liberalization, renewables subsidies and feed-in tariffs, VAT harmonization, energy efficiency and emission rate standards, feebates, methane fees, and policy mixes. - Example outputs (Figure 4) include: - Energy demand by fuel and impacts on 2030 energy prices. - Renewable shares of power generation, changes in generation by source, and changes in annual investment in power capacity. - GHGs vs. targets, GHG by sector, and industrial CO2 emissions. - Revenues raised by fuel, net impacts on GDP levels by reform year, and current account balance from reduced fuel imports. - Additional outputs (around 50 other charts) include: gaps to socially optimal price levels; national emissions and electricity capacity, investment and generation by energy source; impacts on trade of energy goods; sectoral decarbonization targets (in NDCs); impacts on revenues from changes in taxes and subsidies on fuels, electricity, and renewables; impacts on GDP over time and by policy change (taxes, expenditures, investments, or transfers); GHG emissions by sector, gas, and fuel; energy-related CO2 emissions by sector, industry, and fuel; and graphical displays of key inputs such as growth forecasts, global energy prices, and price and income elasticities. ### Distribution module (incidence) - Purpose: estimates incidence impacts from climate mitigation policy on industries (many sectors) and households (across and within deciles). - Industry impacts: - Changes in energy prices affect industry input costs and production costs. - Impacts estimated for 59 non-energy sectors (e.g., steel, cement, chemicals). - Example outputs (Figure 5, Panel A): cumulative CO2 emissions and gross value added (GVA); emissions intensity of production (tCO2 per $m GVA); price (cost) impacts on 20 of 59 most affected industries. - Industry analysis requires input-output (IO) tables, harmonized for 120 countries to date. - Household impacts: - Detailed budget-share information is used to estimate direct effects (from changes in energy prices) and indirect effects (from changes in prices of non-energy goods and services). - Effects are estimated at the decile level and between urban and rural regions. - Outputs include BAU energy consumption (percent of total by decile), initial impact on household consumption (absolute LCU by decile), and cumulative revenues needed to compensate given household deciles. - Household analysis requires household budget surveys (HBSs), harmonized for over 65 countries to date. - Net incidence: - Net effects account for revenue recycling (e.g., PIT cuts or cash transfers). - Around 25 figures available: impacts on industrial input and output prices (for 59 sectors); composition of household consumption of energy and non-energy goods/services by decile; absolute consumption effects before and after revenue recycling by decile and urban/rural sub-samples; cumulative share of revenues required to compensate given household deciles (e.g., bottom 10 percent); changes in inequality and horizontal equity (median, 25th, and 75th-percentile impacts within each decile for all, urban, and rural households). ### Air pollution and transport modules (development co-benefits) - Purpose: capture welfare spillovers from climate policy on health, congestion, and road safety. - Air pollution: - Fossil fuel combustion creates local pollutants like PM2.5 and low-lying ozone (O3). - These pollutants contributed to 4.5 million premature deaths (in 2019) from outdoor air pollution (IHME 2020). - The air pollution module estimates benefits by disease, age group, location, and source using several methods. - Example outputs (Figure 6, Panel A): baseline emissions of PM2.5, NOx and SO2 by source; impacts on urban and rural PM2.5 concentrations; cumulative avoided deaths by age group. - Around 50 other charts: relative risk of diseases; emissions factors; baseline and changes in deaths by type (indoor, outdoor, ozone), sector, disease, and age group (infants, children, working age, and 65+); changes in morbidity (years lived with disease and disability adjusted-life years); GDP losses due to air pollution; avoided lost wages; savings in health expenditures by payee (government, private, and donors). - Transport: - Increases in road fuel prices tend to reduce road accidents, congestion, and associated external costs. - The transport module estimates changes in congestion and road maintenance costs and other road sector impacts. - Example outputs (Figure 6, Panel B): congestion as a share of travel time; total road maintenance costs; annual monetized welfare benefits from reform. - Charts include: changes in external costs from reduced congestion, road accidents, and maintenance. - Combined welfare assessment: - The air pollution and transport modules provide monetized co-benefits that can be compared against abatement costs to estimate net welfare impacts from policies (including climate benefits, air pollution co-benefits, and transport co-benefits). ### Example numerical outputs and indicators shown in figures - Energy stocks and flows and price indicators in Figure 4 (example for US$50 Carbon Price/ton CO2e by 2030, Unspecified Country) display: - Time series from 2018 through 2035 for energy demand by fuel (mtoe) and prices ($/GJ, $/kwh, $/liter). - Renewable shares of power generation across 2018, 2023, 2028, 2033 (%). - GHG indicators including mtCO2e scales and comparisons to Linear pathway to NZE, Baseline, Carbon tax, Unconditional NDC, Conditional NDC. - Economic indicators including real 2018 US$bn impacts and current account balance (%GDP). - Air pollution and transport figures (Figure 6) report: - PM2.5 = 254 Kton, NOx = 1147 Kton, SO2 = 449 Kton by source. - Ambient PM2.5 (μg/m3) time series from 2018 through 2034 with Urban and Rural baselines and carbon tax scenarios; WHO Guidelines reference shown. - Cumulative avoided deaths by age groups across 2020–2035 (Number of deaths). - Annual monetized welfare benefits and percent GDP welfare benefits (costs) time series decomposed into Economic costs, Transport co-benefits, Air pollution co-benefits, Climate benefits, and Net. *Source: IMF staff using CPAT. International Monetary Fund.* ### 3. Mitigation Module ### 3. Mitigation Module ### Modeling the BAU and Policy Scenario - The mitigation module contains a ‘business-as-usual’ (BAU) and policy scenario to estimate impacts on energy demand, emissions, revenues, GDP, and welfare. - Energy consumption is split into 15 fuels and electricity sources produced or consumed by 17 sectors. - Energy sources: coal, natural gas, gasoline, diesel, kerosene, liquified petroleum gas (LPG), jet fuel, other oil products, electricity, wind, solar, hydro, other renewables, nuclear, and biomass. - Sectors (consistent with UNFCCC): power generation; transport (road, rail, shipping, and aviation, including domestic and international); buildings (residential, food & forestry, public & private services); industries (mining & chemicals, iron & steel, other metals, machinery, cement, other manufacturing, construction, fuel transformation & transport); other energy use; non-energy use. - Projections are based on: - GDP projections (see Annex I: GDP). - Domestic energy prices and projections of future international energy prices (Annex I: Energy prices and International and domestic energy price projections). - Assumptions about income elasticity of demand and own-price elasticity of demand for fuels and electricity (Annex I: Own-price elasticities of demand for energy products consumed by households and firms and Income elasticities of energy demand). - Assumptions on rates of technological change due to exogenous efficiency improvements and changes in cost and productivity of key low-carbon technologies like renewables. - Data sources and parameterization: - Energy demand and production data: IEA (2022a), Enerdata (2022), and other sources. - GDP projections: latest IMF forecasts. - Energy taxes, subsidies, and prices: compiled from publicly available and IMF sources, with inputs from proprietary and third-party sources. - International energy prices projected using an average of WB and IMF projections for coal, oil, and natural gas, then projected to domestic prices using empirically estimated pass-through by country. - Elasticities calibrated via literature review (Annex I) yielding estimates broadly in line with mid-range BAU emissions and policy scenario responsiveness implied by other models. - BAU assumptions: - Current fuel taxes/subsidies and carbon pricing are held constant in real terms (assumes countries do not add to or strengthen existing mitigation policies). - International energy supply is able to meet demand with exogenous international fuel prices. ### Power Sector Modeling - Two power supply models: - Elasticity-based model: estimates changes in the generation mix in response to relative price changes (fuel and other costs). - Hybrid technology-explicit (‘technoeconomic’) model: incorporates explicit stock of power generation assets with investment and dispatch decisions. - Technoeconomic model features: - Uses projections of levelized costs of electricity (LCOE), assumptions on retirement rates, capacity factors, physical/economic limitations, and increasing need for storage. - Makes forward-looking investments in new capacity while dispatching existing assets. - Electricity prices vary for industrial and residential users, determining electricity demand. - Default assumptions: - Hydroelectric capacity is assumed fixed. - Nuclear may be phased up (with a lag) in countries which already have fission reactors. ### Policies Modeled in the Policy Scenario - Price-based policies: - Carbon pricing (carbon taxes and ETSs), fuel and electricity taxes, fossil fuel subsidy reform, energy market reform such as price liberalization, VAT reform. - Renewable subsidies: - Feed-in tariffs (equivalent to a renewable production tax credit) for renewable power generation. - Regulatory policies: - Emission rate standards, energy efficiency standards, and ‘feebates’ (taxes on carbon intensive goods used to fund subsidies on low-carbon goods). - Policy mixes: - Combinations such as a carbon tax with fossil fuel subsidy reform, energy price liberalization, VAT harmonization, and renewable subsidies. - Modeling of regulations: - Non-price policies (regulations) are modeled using a shadow pricing approach to impact efficiency of energy-consuming capital goods without directly impacting consumer energy demand like price-based policies. ### Impacts on Energy, Emissions, and Achievement of NDCs - Policy impacts on fuel use and emissions depend on: - (i) impacts on energy prices, and (ii) price responsiveness of fuels by sector. - Sectoral responses: - Industry, buildings, transport: price changes incentivize shifts to more efficient and cleaner technologies and direct fuel demand reductions. - Power sector: investments shift generation from coal and natural gas towards low-carbon technologies (solar, wind), subject to scale-up limits and increasing storage needs; fuel-based power becoming more expensive raises electricity prices and dampens power demand and overall generation. - Emissions estimation: - Total GHGs and local air pollutants estimated via emissions factors by fuel, country, and sector provided by IIASA (GAINS model). - Greenhouse gases covered: CO2, CH4, N2O, and F-gases (HFCs, PFCs, SF6, NF3) — included in UNFCCC inventories except NF3. - Local air pollutants: PM2.5 and ozone (O3), including PM sources from black carbon (BC), organic carbon (OC), VOCs, CO, NOx, and SO2. - Local air pollutant health impacts (premature deaths) and localized warming/cooling effects are estimated by the air pollution module. - Non-energy emissions: - Included: LULUCF; agriculture; industrial processes; waste; other sources. - Historical GHGs compiled using UNFCCC, EDGAR, FAO, and national sources. - LULUCF emissions assumed to decline steadily for countries with positive emissions and be flat for countries with negative emissions. - Industrial process emissions scale with energy-CO2. - Agricultural CO2 scales with population and per-capita income; waste emissions scale with population. - Methane from agriculture, waste, and extractives estimated using country-specific emissions factors, assuming autonomous technical change, GDP growth and, in policy scenario, marginal abatement cost curves. - Default: non-CO2, non-methane GHGs assumed to change at the same rate as energy emissions (this assumption can be switched off). - NDC assessment: - Mitigation pledges converted into percent reductions vs. BAU in 2030 defined in terms of GHGs excluding LULUCF. - Enables estimation of whether a country’s target is likely to be met under the policy scenario (or baseline for non-binding pledges) and comparisons of mitigation ambition across countries. - Latest forecasts for NDCs and CPAT emissions projections available on IMF’s Climate Indicators Dashboard. ### Impacts on Revenues, GDP, and Welfare - Revenues: - Estimated by comparing total revenue from fuel and electricity taxes, net of outlays on fuel or renewable subsidies, in BAU versus policy scenario. - Revenue-raising policies: carbon taxes, ETSs with auctioned allowances, increases in energy excises, VAT harmonization, reductions in fossil fuel subsidies. - Revenue-reducing policies: expenditures on renewable subsidies, green public investments, most regulations. - Feebates can be revenue-raising, neutral, or reducing depending on design. - Users can recycle revenues to public investment, (targeted) transfers, current spending, cuts in personal income and/or corporate taxes, or mixes thereof; for revenue-reducing reforms users can choose tax bases to raise taxes to ensure revenue-neutrality. - GDP impacts: - Estimated for each country and year using fiscal multipliers extracted from external models and empirical studies. - Policy impacts on GDP vary notably with how revenues are recycled. - Reductions in PIT and increases in public investment tend to be more supportive to GDP than increasing transfers or current government expenditures. - Aggregate empirical evidence: GDP impacts of mitigation policies are small (slightly positive or negative) or ambiguous. - Rebound effects of GDP changes on energy consumption and emissions are small in practice. - Welfare impacts: - Estimated using long-established formulas from public finance literature reflecting integrals under marginal abatement cost schedules and efficiency effects from compounding/offsetting pre-existing distortions from fuel taxes/subsidies. - CPAT conservatively does not currently capture additional welfare effects from revenue recycling and broader fiscal interactions. - Domestic benefits from reduced environmental costs of fuel use (development co-benefits) such as reductions in premature mortality from local air pollution, traffic accidents, and congestion are estimated separately by the air pollution and transport modules and used in welfare calculations. *Source: 3. Mitigation Module — wpiea2023128-print-pdf* ### Box 1. Planned Improvements in CPAT Mitigation Module ### Box 1. Planned Improvements in CPAT Mitigation Module ### Planned enhancements to the mitigation module - More granular representation of energy-consuming sectors, their technologies, and sectoral policies to reflect the growing sectoral approach in mitigation policy (Black and others, 2022a). - Incorporation of separately developed dynamic capital turnover models for transport and buildings that: - Include a dynamic capital stock allowing modeling of policies such as tightening of emission rate standards (for new or existing vehicles and buildings) and green industrial policies (e.g., subsidization of newer technologies). 56 - Allow quantification of spillover impacts of technology policies on costs due to learning curve effects 57 and the impact of capital vintages on optimal mitigation strategies. 58 - Development of more refined industry- and activity-specific sectoral models for: - Industrial sectors like steel, chemicals, and cement. - Agriculture and forestry. - Enhanced modeling of economic impacts, policy coverage, and international linkages: - Better representation of GDP and international trade effects, notably for industrial sectors and fossil fuel exporting countries. - Incorporation of planned policies (for example for nuclear in power and efficiency in buildings) to reflect governments’ existing plans. - Improved production structure tables through use of the IMF’s forthcoming Multi-Analytical Regional Input-Output (IMF-MARIO) database. - Improved welfare-effects estimation by incorporating: - Distortions in the fiscal system (Parry and others, 1999). - Informality and other channels (Bento and others, 2018; Heine and Black, 2019). ### Linkages with external models - CPAT is envisioned to increasingly allow linkages with external models to either provide outputs to or consider inputs from, including: - Macro-econometric models such as the Macro-Fiscal Model (MFMOD; Burns and others 2019). - Computable general equilibrium (CGE) models like IMF’s ENVISAGE (IMF-ENV; Chateau and others 2022). - Sectoral models such as the Future Technology Transformations models (FTT; Mercure 2012, Mercure and others 2018, Knobloch and others 2019, Vercoulen and others 2019). - The IMF’s Fiscal Analysis of Resource Industries model (FARI; Luca and Mesa Puyo 2016). - Other relevant models. ### Caveats to CPAT’s mitigation module (current and in development) - General caveats (some to be addressed in planned improvements): - Non-linear responses to large policy changes are abstracted from. Example: a large increase in emissions prices could facilitate rapid adoption of CCS or direct air capture despite high uncertainty in future costs. 59 - The module does not capture widescale technological change induced by climate policy that may imply higher price elasticities and more positive impacts on GDP. 60 - Learning-by-doing spillovers in low-carbon technologies are not endogenized: the model includes assumptions on learning rates for key technologies, but policy in one country is not assumed to impact global learning rates and assumptions may be conservative (renewables example: costs of solar declined 90 percent between 2010 and 2020). 61 - Feedback from fuels markets is limited: possibility of upward-sloping fuel supply curves 62 and international fuel price changes resulting from simultaneous reforms in large countries are abstracted from, though parameter values are chosen for broad consistency with more detailed energy models (see Annex I). - Specific caveats for the initial iteration (‘CPAT 1.0’): - Limited international linkages. 63 - CPAT accounts for changes in fuel and electricity imports and exports and changes in trade in GDP estimates, but coverage of traded products is limited. - This limitation prevents explicit analysis of implications of border carbon adjustments (BCAs), though additions are planned. - Simplified impacts from policy changes on GDP. - GDP impacts are estimated to account for general equilibrium effects (e.g., changes in employment, balances of payments, monetary factors) to adjust the forecasted growth path. 64 - Fiscal multipliers are currently aggregated at the region and income-group level; country-specific circumstances (e.g., debt distress) are not currently included. 65 - Economic effects do not account for interactions between mitigation policies and distortions in the broader fiscal system (which can reduce policy costs, e.g., via revenue recycling into broader tax reductions). - GDP impacts from changes in informality, induced technical change, or local air pollution (for example on productivity) are not included but could be substantive. 66 - Sectors are de-coupled at present but will become increasingly integrated in future updates (see Box 1). - Global decarbonization requires cutting emissions in power generation while electrifying end-uses, creating inter-sectoral linkages (example: electric vehicles adding modestly to electricity demand; hydrogen for industry). - Future updates will add interactions between electrified sectors and power demand. - Price elasticities: potential mis-specification in timing and magnitude. - CPAT assumes impacts of prices on energy use are fully realized within one year. 67 - This may overstate short-term responsiveness since firms and households take time to adjust, but is considered reasonable as policies are phased over several years. 68 - In practice, this assumption primarily affects the energy intensity component of elasticities, accounting for roughly half of responsiveness; one power sector supply model in CPAT accounts for short-term limits on new investment; dynamic capital turnover models for transport and buildings distinguish policies affecting new vs. existing capital. - Initial evidence suggests price elasticities used may be too low in the long run: - Empirical elasticity studies typically examine market-fluctuation-induced price changes, but policy-induced price changes may elicit larger responses due to higher salience and expected permanence (examples: Li and others 2014; Anderrson 2019; Moore and others 2021). - Users can adjust price elasticities. *IMF Working Paper — Box 1. Planned Improvements in CPAT Mitigation Module* ### 4. Distribution Module ### 4. Distribution Module ### Overview - Focus: distributional impacts of climate mitigation policies on households and firms, including energy intensive, trade exposed (EITE) sectors. - CPAT distribution module scope: impacts estimated for 59 non-energy economic sectors across 120 countries. - Purpose: quantify changes in firms’ input costs and output prices, and household welfare impacts, to inform policy design (compensation, BCAs, carbon price floors). ### Distributional impacts on households - General patterns: - In low- and middle-income countries, carbon pricing policies (before revenue recycling) tend to be moderately progressive due to more concentrated grid access and ownership of energy-intensive goods among higher-income households. - In high-income countries, changes in energy prices tend to be regressive because ownership of energy-intensive goods tends to be broader. - Revenues raised or saved can make reforms pro-poor and equity-enhancing overall; examples of recycling uses: cash transfers, social safety nets, investments in education and health. - Non-pricing mitigation policies do not raise revenues and may erode the tax base for existing energy taxes, limiting options for revenue recycling. - Net impacts are estimated accounting for revenue recycling through PIT reductions, transfers, and public expenditures. - Vertical and horizontal distribution: impacts can be estimated across consumption deciles (vertical) and within deciles (horizontal, 25th and 75th percentiles), and separately for rural and urban households. ### Impacts on firms and EITE sectors - Policymaker concern: impacts on exporting or import-competing firms, especially EITE industries (e.g., steel, cement, chemicals), via increased input costs, potential loss of market share, and carbon leakage. - CPAT outputs for industries: - Changes in firms’ input costs and output prices by industry/sector. - Shares of each industry/sector in gross value added (GVA), total output, household demand, and exports. - Results available for 59 sectors and 8 aggregated CPAT sectors. - Pass-through: users can apply imperfect pass-through coefficients to distinguish input (producer) and output (consumer) price changes. ### Methodology: microsimulation and data inputs - Approach: cost-push microsimulation combining household budget surveys (HBSs) with input-output (IO) tables. - HBS coverage: data on household budget shares obtained from HBSs for, so far, over 65 countries; households grouped into population-weighted, per-capita consumption deciles. - CPAT-compatible classification: - 8 energy goods: coal, electricity, natural gas, oil, gasoline, diesel, kerosene, LPG. - 14 non-energy goods/services: appliances, chemicals, clothing, communications, education, food, health services, housing, other, paper, pharmaceuticals, recreation and tourism, transportation equipment, public transportation. - IO data: sectoral energy intensities derived from global IO tables from the Global Trade Analysis Project (GTAP)-10 database (data for year 2014 across 65 sectors); covers five fossil fuels: coal, electricity, oil, natural gas, and petroleum products. - Price pass-through default: assumed full pass-through of price changes onto consumer prices (flat/perfectly elastic energy supply curves); options exist to relax this default. ### Impacts before and after revenue recycling and behavioral responses - First-order (before recycling) impacts: - Models direct effects (household consumption of fuels and electricity) and indirect effects (increases in non-energy goods and services prices due to higher energy input costs). - Estimates increases in household expenditures and losses in consumer surplus (“burdens”). - Behavioral adjustments: - Two approaches: 1. Uniform scaling of impacts across deciles by the ratio of revenues raised per the mitigation module to revenues raised based on HBS data (accounts for behavioral and structural change by reconciling IO and energy balances). 2. Decile- and product-specific price elasticities of demand derived from country-level data (USDA) applied assuming CES utility functions. - Revenue recycling modes (four simulated): 1. New or existing targeted transfers (user-defined targeted percentiles and targeting inefficiency). 2. Transfers towards public investment in infrastructure (assumed to target parts of income distribution without access to clean water, electricity, sanitation, information technologies, or public transport). 3. Scaling up an existing social protection scheme (benefits proportional to existing social protection schemes). 4. Reducing effective PIT liabilities (proportional or lump-sum reductions, or exempting deciles entirely). - Additional features: - Transfer schemes available for population segments below international poverty lines. - Module estimates share of revenues required to compensate parts of the income distribution (e.g., the bottom two deciles). - Outcomes reported as shares of pre-policy consumption and in absolute per-capita monetary terms at the decile level and for rural/urban subsamples. - For vertical distribution outputs, user can choose between decile mean and median HBS inputs; horizontal impacts estimated for 25th and 75th percentiles within each decile. ### Welfare measures - Burdens: total welfare-equivalent losses measured as losses in consumer surplus, which include: - Extra household expenditures due to higher prices (first-order). - Value of forgone consumption induced by price changes, net of reduced spending (second-order). - Deadweight losses are included in total welfare-equivalent losses. ### Caveats and limitations - Structural change: the share of each sector in total consumption and output is held constant over time (scaled with GDP), so long-run changes in relative production structure may not be fully captured. - Pass-through and incidence: - Default full pass-through assumption may not hold for firms competing internationally; if higher energy prices are passed backward into lower producer prices, incidence could shift to firm owners (lower capital returns) or workers (lower wages). - Impacts of imperfect pass-through are only partly accounted for; options exist to relax the full pass-through assumption. - Other unmodeled channels: - If fossil fuel–intensive industries are capital-intensive, climate policies may increase returns to labor, potentially hurting households with larger capital income shares. - Poorer households may benefit relatively more from reductions in local air pollution if they live in more polluted areas. - More research is required to assess the importance of these additional channels. *IMF Working Papers — IMF-WB Climate Policy Assessment Tool (CPAT), Section 4: Distribution Module* ### 4.5 million deaths from outdoor air pollution in 2019, with 92 and 8 percent due to PM ### 4.5 million deaths from outdoor air pollution in 2019, with 92 and 8 percent due to PM2.5 and ozone, respectively, and 60 percent from burning of fossil fuels ### Key findings - 4.5 million deaths from outdoor air pollution in 2019. - 92 and 8 percent of outdoor air pollution deaths are due to PM2.5 and ozone, respectively. - 60 percent of outdoor air pollution deaths are from burning of fossil fuels. - Indoor air pollution caused a further 2.3 million deaths. - Road accidents cause about 1.3 million deaths per year, with 94 percent in low- and middle-income countries. - Total external costs from all road externalities are estimated at almost $1 trillion in 2020, with two-thirds coming from congestion alone. - Global GHG emissions must be cut by 25 to 50 percent this decade to be on track with limiting warming to well below 2oC, and ideally 1.5oC, above pre-industrial levels. ### CPAT: methodology for air pollution co-benefits - CPAT quantifies mortality, morbidity, and economic costs of local health damages from fossil fuel use for each country in four main steps: - Step 1: Estimate local air pollutant emissions (PM2.5, SO2, N2O, BC, CO, VOCs and CH4) using energy use by fuel, sector, and scenario. - Step 2: Translate emissions into concentrations of PM2.5 and ozone and population exposure using two main approaches (intake fractions and the TM5-FASST approach), which are then averaged. - Step 3: Map population exposure to PM2.5 and low-lying ozone to health burdens using, by age class, baseline mortality rates and exposure-response curves from the 2019 GBD study; assess jointly impacts of outdoor and indoor air pollution; outputs include mortality and disability-adjusted life years (DALYs). - Step 4: Average the two approaches and value changes in mortality risk using a valuation approach that implies a value of around US$4.6 million per death avoided for 2020 in the average OECD country, extrapolated to other countries based on incomes relative to the OECD and an assumed mortality risk elasticity. ### Details on the two air quality approaches used in CPAT - Intake fraction method: - Estimates the portion of PM2.5 that, on average, is inhaled by exposed populations. - For coal, natural gas, and oil power plants, intake fractions derived using spatial data on power plant locations matched to granular population density data and regression coefficients describing fraction of emissions ingested by population at different distances. - For vehicle, building, industry, and other (ground-level) emissions, intake fractions extrapolated nationwide from a database for over 3,000 urban areas. - Intake fractions tend to be higher in densely populated areas and lower where emission sources are coastally located. - Data is available for 164 countries; intake fractions for other countries are inferred from comparable countries in each region. - The intake fraction is converted to a pollution concentration by scaling by the breathing rate. - TM5-FASST approach: - TM5-FASST is an emulator of the full TM5-Chemical Transport Model (CTM) relating emissions to air quality at source and other receptor locations. - Results are based on a ‘source-receptor’ approach downscaled at the country level and augmented by local source apportionment studies. - The original source-receptor matrices in TM5-FASST are separated into 56 regions which are then downscaled to country-specific matrices and supplemented with local source apportionment studies estimating contributions of sources such as fossil fuels to baseline concentrations. - Air quality modelling accounts for local meteorological and topographical factors but is less granular for fossil-fuel related sources like power plants. ### Valuation and uncertainty - Mortality valuation: - Draws on an OECD (2012) meta-analysis of several hundred studies on health risk valuations. - Implies around US$4.6 million per death avoided for 2020 in the average OECD country. - Extrapolated to other countries based on purchasing power parity and an assumed mortality risk elasticity; CPAT allows adjustments to the income elasticity. - Lost wages from morbidity are included but account for a small portion of total costs. - CPAT provides five methods in total to estimate the emissions-to-health relationship; all have been cross-checked against more complex air quality models and allow sensitivity analysis. - Caveats on uncertainty include: - Temporal and geographical scope: CPAT provides annual, population-averaged estimates; significant intra-annual and urban-rural variation can exist. - Uncertainty in relationships between emissions, concentrations, and health impacts: consensus that PM2.5 and ozone impact health but exact relationships are uncertain. ### Road transport co-benefits module: channels and metrics - Channels through which climate policies affect welfare: - Congestion, road accidents, and road damage. - By raising gasoline and diesel costs, climate policies can reduce vehicle kilometers travelled (VKT) and affect congestion, accidents, and wear and tear. - Congestion: - Measured as time lost due to actual travel speed being slower than a ‘free-flowing’ speed. - CPAT forecasts baseline congestion delays using historic congestion growth rates adjusted for GDP and population growth; policy scenario uses elasticity estimates tied to fuel price changes. - Baseline uses last available year of data (from TomTom) projected forward. - Road accidents: - CPAT provides baseline and policy forecasts using empirical estimates of the link between road fuel prices and road fatalities. - Baseline fatality forecasts use latest available data from OECD, IRF, and UNECE and average past growth rates adjusted for GDP and population growth. - Marginal external costs and welfare benefits: - Marginal external cost of congestion estimated by multiplying average travel delays per VKT by: (i) relationship between marginal and average travel delays based on traffic speed-flow curves; (ii) vehicle occupancy (averaging over cars and buses); (iii) people’s value of travel time (VOT, assumed to be 60 percent of the nationwide average market wage in 2020); (iv) fuel economy; and (v) portion of fuel demand elasticity that comes from reduced driving versus improved fuel economy/shifting to EVs. - Accident externalities per liter measured by apportioning traffic fatalities into external versus internal risks, monetizing them using the mortality valuation approach, extrapolating other components from several country case studies, and dividing by fuel use, scaled by the driving portion of fuel price elasticity. - Road damage: - Baseline road maintenance costs projected from IRF data and average historic growth, using an empirically derived relationship between road maintenance costs and GDP and population growth. - Externalities are assumed to be 50 percent of total maintenance costs, with the entire externality attributed to diesel consumption (scaled by driving portion of diesel fuel price elasticity). - VKT forecasts: - Base VKT from IRF; changes driven by average VKT growth, GDP and population growth, and changes in fuel prices. - Relationships estimated econometrically at country level where data available, differentiating short- and long-run responses. ### Caveats specific to road transport module - Fuel price elasticity estimates are assumed to be causal; endogeneity cannot be entirely ruled out despite country and year fixed effects. - Data quality may affect results; where country data are unavailable, IMF region and income group averages are used. - Impacts of electrification of road transport (plug-in and hybrid EVs) are not currently modelled explicitly; EVs create driving-related externalities that are not priced by petroleum product taxes. Future iterations are expected to address this. - Proxy taxes on driving-related externalities (e.g., per-VKT charges) may be preferable to road fuel duties for pricing road externalities; CPAT currently allows for taxes imposed on an energy consumption basis and could include targeted policies in future updates. ### Policy implications and recommendations (from conclusion) - Stabilizing the global climate requires climate mitigation reforms across countries and unprecedented rates of decarbonization. - Policy instruments include carbon pricing (carbon taxes and ETSs), fossil fuel subsidy reform, energy market reform and price liberalization, renewable energy subsidies, feebates, green public investments, regulations, VAT harmonization, and mixes thereof. - Analytical tools like CPAT can help policymakers design reform packages that accelerate decarbonization while supporting growth, poverty alleviation, equity, environmental quality, and energy access. - Reforms should include measures to facilitate a ‘just transition’ and ‘deep decarbonization’: - Policies for retraining, relocation, and financial support for displaced workers (e.g., in coal mining regions). - Additional policies to address technology-related market failures, including prizes, support for basic research, and advance market commitments. - Political economy factors matter for reform durability; CPAT can inform likely political acceptability by quantifying incidence impacts on industries and households, but separate qualitative analyses (e.g., public opinion surveys) are also valuable. - Evidence suggests policy diffusion across countries; recycling revenues through per-capita transfers, labor tax reductions, or expenditures towards climate mitigation or adaptation projects can enhance acceptability. *IMF WORKING PAPERS — IMF-WB Climate Policy Assessment Tool (CPAT), INTERNATIONAL MONETARY FUND* ### Annex I – Technical Details: Mitigation Module ### Annex I – Technical Details: Mitigation Module ### Overview of model structure - CPAT’s mitigation module uses production-based emissions inventories and does not include emissions embodied in imported goods.100 - Five main energy-consuming sectors: - Power: generation of both electricity and district heating and distribution; supplies households and firms. - Industry: eight subsectors — mining & chemicals, iron & steel, non-ferrous metals, machinery, cement, construction, fuel transformation & transportation, and other manufacturing.102 - Transportation: road (mostly gasoline from light-duty vehicles and diesel from heavy-duty vehicles), rail (mostly diesel), domestic aviation (mostly jet fuel), domestic shipping (mostly diesel and fuel oil).103 - Buildings: primary (excluding electricity) energy demand in residential, industrial, and commercial buildings and services; includes energy use in agriculture and forestry as in national GHG inventories. - Other: miscellaneous emissions not captured in other sectors. - Energy sources distinguished: - Fossil fuels: coal, natural gas, gasoline, road diesel, liquified petroleum gas (LPG), kerosene, (domestic) jet fuel, and ‘other oil products’. - Electricity: generated by fossil fuels, renewables (wind, solar, hydro, biomass, other renewables such as geothermal; grid or off-grid), nuclear energy, and imports/exports. - Units: - Electricity measured in kilowatt hours (kWh). - Coal and natural gas in gigajoules (GJ). - Road fuels in liters (l). - Oil and other oil products in barrels of oil (bbl). - Other energy sources (nuclear and other renewables) in kilotons of oil equivalent (ktoe). ### Energy demand: general formulation and interpretation - Core dynamic formulation (all symbols preserved from source): - (1) 퐸푡 = (푢푡/푢푡−1 ∙ ℎ푡/ℎ푡−1) ∙ 퐸푡−1; - 푢푡/푢푡−1 = (퐺퐷푃푡/퐺퐷푃푡−1)^푣푡 ∙ (ℎ푡∙푝푡 / ℎ푡−1∙푝푡−1)^휂푢; - ℎ푡/ℎ푡−1 = (1+훼)^{−1} ∙ (푝푡/푝푡−1)^휂ℎ. - Definitions: - 퐸푡: demand at time t for a specific energy good in a particular sector. - 푢푡: usage of energy-consuming capital goods. - ℎ푡: energy consumption rate of capital goods (inverse of energy efficiency). - 푣푡: income elasticity for the energy good (may change over time). - 푝푡: price for energy in the sector. - 휂푢: price elasticity of demand for the usage of energy-consuming capital goods. - 휂ℎ < 1: price elasticity of the energy consumption rate. - 0 < 훼 < 1: autonomous rate of efficiency improvements for energy-consuming capital goods. - Sectoral interpretations: - Industry: 퐸푡 = use of coal, natural gas, oil, electricity = industrial output × fuel/electricity use per unit output. - Transport: 퐸푡 = vehicle use (VKT) × liters of fuel or electricity use per VKT. - Buildings: 퐸푡 = building stock × fuel/electricity use per unit time (space heating/cooling, lighting, cooking). - Price elasticity decomposition and total own-price elasticity: - Rewritten elasticity result shows total own price elasticity 휂퐸 composed of three terms: - (2) 휂퐸 = 휂푢 (−) + 휂ℎ (−) + 휂ℎ휂푢 (+) - Interpretation in transport: - 휂푢: elasticity of VKT with respect to fuel price (reductions in VKT and vehicle stock). - 휂ℎ: elasticity of fuel consumption rate with respect to fuel price (shifting to more efficient vehicles and EVs). - 휂ℎ휂푢: product term reflecting rebound where reductions in marginal fuel cost can slightly increase usage. - Rebound effect definition: - (3) 푅푒푏표푢푛푑 푒푓푓푒푐푡 = − (휂푢 ∙ 휂ℎ) / (휂푢 + 휂ℎ) - Industrial sector nuance: - 휂푢 reflects changes in consumer demand as higher energy costs are passed through to consumer prices, with pass-through varying by industry subcategory according to energy intensity. ### Energy supply: fossil fuels and power sector - Fossil fuels: - Supply curves for all fossil fuels and countries are assumed perfectly elastic over the range of climate mitigation policies; unit production costs fixed within a period. - Rationale: international markets (esp. oil), elastic domestic coal production over longer term, and reasonable approximation for natural gas given market fragmentation. - Caveat: simultaneous large actions by energy-consuming countries could exert downward pressure on international fuel prices; this is not explicitly modelled in CPAT. - Power sector: two alternative modelling approaches - Elasticity-based supply model: static, approximates changes in electricity generation based on relative price changes across fuels; parameterizable to econometric evidence on coal and fuel price elasticities. - Techno-economic (“engineering”) hybrid model: incorporates power system complexities (system reliability, storage, generation asset turnover, non-linearities such as early retirement of coal assets). Users can choose either model or take an average. - Elasticity-based electricity supply details: - Unit cost of producing electricity at industry level, c: - (4) c = Σ_i θ_i ∙ g_i - θ_i: share of fuel i in total generation. - g_i: full cost of producing and delivering a unit of electricity using fuel i. - Generation shares formula: - (5) θ_i = θ^0_i ∙ { (ĝ_i)^{ε_i} + Σ_{j≠i} θ^0_j [1 − (ĝ_j)^{ε_j}] / (1 − θ^0_j) } - ĝ indicates proportionate change in unit costs relative to BAU. - ε_i < 0: conditional own-price elasticity of generation for fuel i (percent reduction per one-percent increase in generation cost). - Generation costs include variable (fuel, labor) and fixed costs (investment, maintenance, transmission, distribution) annualized, with costs declining at a fixed annual rate for technological improvements. - Techno-economic electricity supply model details: - Two primary optimization decisions: 1. Dispatch decision: given existing assets and system needs (reliability, storage, peaking), determine which assets generate electricity—driven by variable costs and contractual obligations. 2. Investment decision: annual additions depend on retirements and demand; guided by forward-looking LCOE and physical limits (scaleup of wind/solar) and system stability needs (short- and long-term storage). - Dispatch procedure: - Power demand estimated via energy demand equation and separated by sector. - Inflexible capacity (solar, wind, hydro) dispatched first at fixed historical capacity factors. - Semi-flexible assets (nuclear, biomass, oil) next and can be ramped down. - Remaining demand allocated to fully flexible capacity (coal, natural gas) by relative variable costs using a logit formula. - Investment procedure: - Existing capacity retired per country-specific schedules for coal and linear retirement for others; shortfalls trigger investment. - Additions allocated via a logit formula based on forward-looking LCOE, subject to supply constraints on scaleup of solar and wind111 and system stability (storage grows with variable renewable share). - Logit allocation formula: - (6) x_{f,t} / Σ_f x_{f,t} = e^{−K.c_i} / Σ_f e^{−K.c_i} - x = inv (investment) or gen (generation) allocated to generation type f. - c_i: relative cost of generation type f (total LCOE for investment or variable costs for dispatch, including taxes net of subsidies). - Calibration and adaptability: - Historical generation shares calibrated to recent observed data; technoeconomic model is highly adaptable and user-tunable (including forcing specific investment schedules). - Operational assumptions: - Transmission losses, net exports and power industry own use are fixed proportions of power demand based on historical data. - Investments based on gap between current year’s demand and previous year’s capacity less retirements. - All generation sources scalable within a year except nuclear and hydro where new assets come online after 7 years.110 - Supply constraints for solar and wind vary by country or default to a percentage of previous year’s total capacity for solar and wind.111 ### Market equilibrium and prices for all energy sectors - Retail price faced by consumers: - (7) 푝푅 = (푝푆 + 휏푒푥푐푖푠푒) (1 + 휏푉퐴푇) - 푝푆: supply price (fixed within period). - 휏푒푥푐푖푠푒: pre-existing excise (or other) tax on fuel use; negative if consumer-side fuel subsidies exist. Pre-existing carbon taxes and/or ETS permit prices are incorporated into 휏푒푥푐푖푠푒. - 휏푉퐴푇: rate of value-added (or general consumption) tax applied if consumed at household level.113 - Pass-through options: - CPAT allows optional scalars to reflect different assumptions about pass-through of carbon pricing into higher prices for fuels, electricity, and industrial products. - Pass-through may be less than 100 percent due to institutional price setting, market power, or limited ability to pass higher costs into product prices because of international competition. ### Mitigation policy options (intro) - CPAT can model several climate mitigation policies including explicit carbon pricing.114 *Source: Annex I – Technical Details: Mitigation Module (wpiea2023128-print-pdf)* ### 1. Carbon taxes. This policy could represent a carbon tax applied to the supply of fossil fuels in ### 1. Carbon taxes ### Carbon tax design and mechanics - Modeled as an additional charge on pre-existing taxes for a particular fuel equal to: (CO2 emissions factor for that fuel) × (tax rate on CO2). - Can be: - Comprehensive: applied to all fuels and sectors. - Partial: exemptions for individual fuels and sectors (with option to phase out exemptions over time). - To the extent carbon taxes are passed forward, they are reflected in higher fossil fuel prices. - Increase in electricity prices decomposes into: - (i) pure abatement costs: increase in generation costs per unit due to shifting to cleaner, but costlier, generation fuels; - (ii) the tax on remaining emissions per unit of production (or carbon charges on fossil fuel inputs per unit of production). ### Interaction with other factors (from adjacent analysis) - Renewables technologies are more metals- and minerals- intensive than non-renewables, implying rising metals/minerals costs increase relative cost of new renewables investment. Empirical adjustment referenced: - Projected capital expenditure costs for investment in new renewable and non-renewable capacity were upscaled by 10 percent and 5 percent in 2022 respectively, declining to a 5 percent and 2.5 percent permanent increase in 2030 compared with previous projections. ### Emissions Trading Schemes (ETSs) ### Modelling approach in CPAT - ETSs are modelled by their virtual tax, or "shadow price" equivalent: equivalent carbon charges on the fuels used in sectors to which the ETS applies. - CPAT is deterministic and does not capture uncertainty over emissions prices associated with ETSs. - A scalar adjustment, default value 0.9, is applied to the emissions price, implying a moderately smaller behavioral response from the ETS compared with an equivalent carbon tax at the same rate. ### Interpretation of the scalar (0.9) — possible representations - (i) exclusion of small-emitting firms from an ETS applied downstream to large firms in the power and industry sectors; - (ii) higher price uncertainty under an ETS compared with a tax, potentially damping investment incentives for low-carbon technologies; - (iii) grandfathering of allowances to incumbent firms, creating barriers to new entrants and potentially forestalling innovation. ### Empirical note on exclusions - Small-scale emitters exempted from the EU ETS in 2020 accounted for 6.9 percent of EU-wide CO2 emissions (cited as an example of the modest scale of exclusions). ### Other policies modelled in CPAT ### 3. Taxes on individual fuels, electricity, and methane emissions - Modeled as a new, or an increase in an existing, fuel or electricity tax. - User can choose to exempt specific fuels or energy-consuming sectors. - For methane, user can choose which major sources the fee applies to. ### 4. Renewable energy subsidies - Modeled via a feed-in subsidy that provides a proportionate reduction in per-unit generation cost for renewables. - Subsidies for other clean energy technologies such as electric vehicles (EV) are not currently modelled, pending future transport-sector model improvements. ### 5. VAT harmonization - User can harmonize VAT on fuels and electricity where countries have preferential rates. - In corrective taxation scenarios, user can choose to impose VAT on top of the sum of supply costs plus taxes including taxes for externalities. ### 6. Emission rate / energy efficiency regulations - Modeled through a shadow pricing approach: such regulations impact the efficiency margin rather than direct demand response. - CPAT can implicitly model: - CO2 emission rate standards (e.g., per kWh of power generation, per unit of production for individual industries, or per VKT for vehicles); - Energy efficiency standards (e.g., for electricity demand and energy use in industry, transport, and buildings). - These regulations reduce emissions or energy intensity without the same demand response as carbon pricing, and produce a moderately offsetting increase in emissions through the "rebound effect", which CPAT captures. ### 7. Feebates - Feebates are "fee and rebate" regimes that provide a revenue-neutral sliding scale of fees on above-average emission activities/products and rebates for below-average emission activities/products. - Incorporated in CPAT through shadow prices; in model terms, a feebate adds a shadow price to pt in the expression for ht only (whereas a fuel tax or carbon price increases pt in both ut and ht). ### 8. Fossil fuel subsidy reform - CPAT uses IMF dataset on fossil fuel subsidies by fuel product, sector, and country. - Can model partial reforms (removing explicit subsidies) or full reforms (also applying corrective taxes to internalize external costs, i.e., implicit subsidies). ### 9. Energy price liberalization - CPAT contains estimates of pass-through rates for individual fuels in sectors for many countries, based on regressions of domestic fuel price changes versus international prices. - Pass-through rates are bucketed assuming rates of 0.25, 0.5, 0.75 or 1.0 (though for most fuels and countries a 1.0 pass-through rate is assumed). - For countries with price controls, price controls can be phased out gradually in the policy scenario, impacting energy, emissions, and revenues. *IMF WORKING PAPERS — Climate Policy Assessment Tool (CPAT) — Extract: "1. Carbon taxes."* ### 10. Policy mixes. Various policy combinations are possible in CPAT, including carbon pricing, fuel tax ### 10. Policy mixes. Various policy combinations are possible in CPAT, including carbon pricing, fuel tax ### Policy mixes and exclusions - CPAT can combine carbon pricing, fuel tax changes, or regulations with fossil fuel subsidy reform, energy price liberalization, and/or renewable subsidies. - Future iterations will allow a more diverse range of policy mixes to reflect diverse country strategies. - Not currently included in CPAT (but may affect energy consumption and emissions): public investments (e.g., smart grids, public transportation), low-carbon fuel standards, biofuel mandates, building codes, incentives for specific technologies (e.g., geothermal power, nuclear, carbon capture and storage, CCS), emission rate standards for non-road vehicles, measures for extractive industries (e.g., moratoria on extraction, charges on production or fugitive emissions), mitigation instruments beyond the energy sector, and broader R&D-promotion policies. In many cases these omitted policies have only modest impacts on emissions. ### Energy Sector: Key assumptions - Energy consumption by sector and country for the latest available year compiled from IEA, Enerdata and the UN. - Electricity demand modelled separately for buildings, industry, and transport, focusing on domestic generation (including exported generation where fuels are combusted domestically, but not imported generation). - For household and industry fuels: demand projected forward using equation (1). For fuels used in power generation: consumption projected with elasticity- and engineering-based models and then averaged in the default case. - Real GDP projections use IMF’s latest WEO estimates (five years of projected GDP) extended into the long term using IMF-ENV CGE model; deviations in GDP in the policy scenario estimated using described approach. ### Income elasticities of energy demand (base and adjustments) - CPAT uses 32 ‘base’ income elasticities covering eight energy sources and four sectors. - Base income elasticities derived from Burke and Csereklyei (2016) and a dataset of over 250 empirical studies and over 4,500 observations. - Global average total income elasticity observed during 1960-2010: 0.74. - Income elasticities are adjusted for income per capita using an inverse-U relation with per-capita income (derived from Gertler and others (2016)), peaking around lower-middle income status (at around $3,000 per capita) and declining toward developed-country maximum (around $22,000 per capita). - Early projection-year adjustments: scalars interact with income elasticities so energy-related emissions in early years match GHG inventories; these vary by country and year. - Example base income elasticities (sector rows: Transport, Residential, Industries, Services): - Coal: 0.00, 0.40, 0.50, 0.70 - Natural gas: 0.50, 0.60, 1.00, 1.00 - Gasoline: 0.70, 0.50, 0.50, 0.50 - Diesel: 0.60, 0.50, 0.50, 0.50 - Oil (other oil products): 0.80, 0.50, 0.90, 1.20 - Biomass: 0.00, 0.10, 0.10, 0.10 - Renewables: 1.00, 0.75, 0.75, 1.00 - Electricity: 1.20, 0.75, 0.75, 1.10 ### Own-price elasticities of demand (final demand consumed by households and firms) - Parameterization follows a three-step approach: meta-study targets → sectoral inference → sense check with in-house database. - Initial meta-study: Labandeira and others (2017) (~2,000 elasticities from 430 studies). - Developing-country elasticities adjusted slightly higher (about 10 percent), except for diesel. - Cross-price elasticities to capture leakage to biomass: - Biomass wrt LPG: 0.25 - Biomass wrt kerosene: 0.25 - Gasoline wrt diesel: 0.25 - Own-price elasticities used in CPAT (usage and efficiency components, and implied total demand elasticities). Selected values (by fuel; sectors: Transport, Residential, Industries, Services): - Own-price elasticities of demand - usage: - Coal: -0.27, -0.27, -0.27, -0.37 - Natural gas: -0.47, -0.37, -0.47, -0.57 - Gasoline: -0.37, -0.47, -0.37, -0.57 - Diesel: -0.20, -0.20, -0.60, -0.47 - Renewables: -0.47, -0.47, -0.47, -0.67 - Electricity: -0.17, -0.27, -0.27, -0.47 - Own-price elasticities of demand - efficiency: - Coal: -0.40, -0.40, -0.40, -0.60 - Natural gas: -0.40, -0.50, -0.50, -1.00 - Gasoline: -0.40, -0.60, -0.70, -1.10 - Diesel: -0.30, -0.30, -1.20, -0.40 - Electricity: -0.20, -0.20, -0.20, -0.40 - Memo: implied total demand elasticities (selected): - Coal: -0.56, -0.56, -0.56, -0.75 - Natural gas: -0.68, -0.69, -0.74, -1.00 - Gasoline: -0.62, -0.79, -0.81, -1.04 - Diesel: -0.44, -0.44, -1.08, -0.68 - Electricity: -0.34, -0.42, -0.42, -0.68 - Simple averages across sectors: Transport -0.54; Residential -0.57; Industries -0.67; Services -0.80 - Price elasticities used in CPAT calibrated so weighted averages for developed and developing countries are within 10 percent of target elasticities. - Sense checks: - Comparison with large empirical database (weighted by inverse t-statistics and by fuel-share of global emissions). - Model intercomparisons of BAU energy consumption and emissions versus other models — baseline projections and price responsiveness broadly in line with others. - Rebound effect calculations: formula (8) and implied simple average rebound effect across fuels and sectors ranges from 13 to 26 percent. - Note: CPAT uses ‘long-term’ elasticities; elasticities may be conservative because policy-induced price changes can elicit stronger responses than market-induced changes and non-linearities may exist at very high prices. ### Autonomous rate of technological change and exogenous efficiency improvements - Autonomous rate set between 0.5 and 1 percent per year for each fuel-sector pair. - Table of exogenous efficiency improvements (annual rates) by sector and fuel (Transport, Residential, Industries, Services): - Coal: 1.0%, 0.5%, 0.5%, 0.5% - Natural gas: 1.0%, 1.0%, 1.0%, 1.0% - Oil Products: 1.0%, 0.5%, 0.5%, 0.5% - Biomass: 1.0%, 0.5%, 0.5%, 0.5% - Renewables: 1.0%, 0.5%, 0.5%, 0.5% - Electricity: 1.0%, 0.5%, 0.5%, 0.5% - Empirical examples cited: - Fuel economy of vehicles: annual decrease in energy consumed per kilometer travelled of 1.7 percent for cars and light trucks from 2000 to 2020. - Efficiency improvements of residential buildings: annual average of 1.3 percent between 2000 and 2020. - Note: compositional changes (e.g., larger vehicles, larger residences) have partially offset efficiency improvements. ### Power sector: elasticity-based power supply model and generation assumptions - Generation shares by fuel taken for over 200 countries from IEA, Enerdata, UN and other sources. - Own-price elasticities for generation fuels (conditional on total electricity output): - CPAT assumes coal generation price elasticity of -0.7 for all countries. - For other generation fuels, elasticity assumed to be -0.5. - Rationale and empirical context: - Survey suggested short-run coal price elasticity between -0.15 and -0.6. - US simulations suggest coal price elasticity ~ -0.15 in some models; other studies show larger responsiveness. - EIA (2014) example: $34 per ton carbon tax (raising coal prices by ~150 percent) reduces US coal use by 32 percent in 2040; $85 per ton carbon tax reduces coal use by 90 percent. - A China study reports coal price elasticities of -0.3 to -0.7. - CPAT assumes declining costs of renewable energy increase price responsiveness of coal use and can induce technological innovation, hence adoption of a larger (absolute) elasticity for coal. - Annual rate of autonomous productivity improvement at power plants: - Coal: 0.5 percent - Natural gas, nuclear, hydro: 1 percent - Renewables: 5 percent (i.e., costs halve every 15 years) - Non-fuel generation costs (O&M, capital, system integration) based on a large dataset across countries; system integration costs higher for variable renewables. ### Energy prices, taxes, and subsidies - Historical energy prices sourced from IMF and World Bank country desks, IEA, Eurostat, Enerdata, Global Petrol Prices, and others. - Retail prices collected for many countries and fuels; missing values assumed equal to supply costs with VAT included for residential use if VAT is charged. - Supply prices for fossil fuels calculated as consumption-weighted average of imported and domestic prices: - ps = θ pint + (1−θ) pdom - pint = pimp + mint; pdom = cext + mext - Definitions: θ is import share (0≤θ≤1); pint price received by domestic firms that import; pdom production cost plus taxes and margins; pimp international reference price; mint and mext margins; cext domestic extraction cost plus taxes. - Supply-price projection assumptions: margins and production costs fixed in real terms; taxes on production and import prices change based on international reference prices. CPAT averages international price projections from IMF WEO and World Bank (2022a) by default. - Fuel taxes: - Estimated by price-gap approach: difference between retail and supply prices; implicitly reflect excises, VAT provisions, regulated price distortions, and carbon pricing. - Pre-existing carbon taxes and ETS charges estimated from sources (converted from $/tCO2 to $/energy unit using CO2 emissions factors) and kept flat in baseline (user can tweak). - New carbon tax or ETS converted similarly; excise taxes can be added on top or separately per user input. - Fossil fuel subsidies: - CPAT includes explicit producer subsidies (embedded in supply costs, s_i^Prod) and consumer subsidies (in excise/other taxes component, s_i^Cons). - Producer-side subsidies estimated externally and apportioned between global versus domestic supply cost impacts. - Consumer-side subsidies modelled as s_i^Cons = s_i^Fix + s_i^Var · Δp_i^Int; variable part estimated using panel regressions and pass-through buckets of 0%, 25%, 50%, 75%, 100% (rounded up). - Total consumer-side subsidy capped at average level of previous years’ per-unit consumer-side subsidies. - VAT: - VAT estimated as VAT rate times VAT base. VAT rates from IMF Fiscal Affairs Department Tax Policy Rates Database and IEA Energy Prices and Taxes Dataset (2022a). - VAT rates for non-residential energy consumption assumed zero (firms can reclaim VAT). - VAT payment calculated using one of four formulas depending on user choices (VAT reform option and whether externalities are included in VAT base). - Domestic price projection formulas (selected): - rp_t = sp_t + VAT_t + exo_t - sp_t = sp_0^Fix + (sp_{t−1} − sp_0^Fix) · Δp_t^Int + θ · p_s_t - exo_t = ecp_t + ncp_t + fap_t + s_i^Cons - ecp_t = existing carbon price component; ncp_t = new policy component (new carbon pricing or new excise); fap_t = fixed/ad-valorem balancing item; s_i^Cons = consumer subsidy. - For electricity: supply cost is domestic production cost or cost-recovery price evaluated at domestic fuel prices; electricity prices projected using changes in fuel prices and generation shares averaged over the two power supply models. ### Impacts of policies and targets: emissions and revenues - CO2 emissions from fossil fuel combustion = consumption of each fossil fuel product × CO2 emissions factor, aggregated across sectors and fuels. - Industrial process CO2 (e.g., cement) scale with energy-CO2 in industrial sector. - Agricultural CO2 scale with population and per capita incomes; waste and other emissions scale with population. - Methane emissions (agriculture, waste, extractives) estimated using country-specific emissions factors, assuming autonomous technical change, GDP growth, and marginal abatement cost curves. - Default policy scenario: non-CO2 GHGs, except for methane, assumed to change at the same rate as energy emissions. - Revenues from climate mitigation policies: - Direct revenues from carbon pricing = carbon price × CO2 emissions to which applied (for ETSs, fraction of allowances auctioned included). - Revenues from pre-existing energy taxes = fuel tax rate × fuel consumption applied, aggregated across fuels and sectors, plus electricity taxes where applicable. - Indirect revenue losses from base erosion included implicitly (difference between BAU revenues and BAU tax rates applied to policy-scenario consumption). - Regulations and revenue-neutral feebates: no direct revenue impact; indirect revenue loss possible due to erosion of pre-existing energy tax bases. - Renewable and clean technology subsidies: direct revenue loss = subsidy rate × base plus indirect revenue losses from pre-existing energy taxes. - CPAT does not currently include carbon capture and storage (CSS) or emissions from non-energy use of fuels in lifecycle emissions. *Source: wpiea2023128-print-pdf (IMF Working Paper — Climate Policy Assessment Tool (CPAT), section 10).* ### Annex III discusses the estimation of the monetized domestic environmental costs from fuel use. The ### Annex III: Estimation of Monetized Domestic Environmental Costs from Fuel Use ### Domestic environmental co-benefits - Domestic environmental co-benefits of mitigation policies are calculated by the induced reductions in use of a fuel product in a particular sector, multiplied by the corresponding domestic environmental cost per unit, and aggregated across sectors and fuels. - Total co-benefits are the emission reduction times the co-benefit per ton of CO2 reduced. - Climate benefits from cutting emissions are not included in co-benefits as they vary substantially across countries, though studies suggest these benefits would swamp pure abatement costs at the global level. - Rennert and others (2022) estimate the discounted flow of global climate benefits at $185 per ton of CO2 reduced; under a global carbon price of $75 per ton, this would imply climate benefits five times pure abatement costs (per IMF staff calculations). - Co-benefits also include reductions in traffic congestion and accident externalities from higher road fuel prices. ### Welfare or efficiency costs and net economic benefits (CPAT) - In CPAT, welfare costs (sometimes called deadweight losses or excess burdens) reflect losses in producer and consumer surplus in fossil fuel markets and are interpreted as the annualized costs of using cleaner, but costlier technologies, and of reducing energy consumption below preferred levels. - Efficiency costs are calculated using second-order approximations based on long-established public finance formulas and include three components in CPAT and one additional component not yet in CPAT: - Pure abatement costs: - Reflect (1) the annualized costs of adopting cleaner but more expensive technologies, net of any savings in lifetime energy costs and avoided investment in emissions-intensive technologies; and (2) costs to households and firms from reduced energy use. - Pure abatement costs reflect integrals under marginal abatement cost schedules and, for moderate emissions reductions, are measured with reasonable confidence. - Additional welfare effects from pre-existing fuel price distortions: - If carbon mitigation policies reduce use of a fuel subject to a pre-excise tax there is an additional welfare cost equal to the tax wedge times the reduction in fuel use (or a welfare gain if there is a pre-existing fuel subsidy). These effects are computed using outputs from CPAT. - Domestic environmental co-benefits (see above). - Potential fiscal benefits (not in current CPAT): - Reflect economic efficiency gains from productive use of carbon pricing revenues—revenue raised times the efficiency benefit per dollar recycled. - Component is smaller if some revenue is used for transfers to compensate low-income households. - Offsetting effect: higher production costs and consumer prices lower the real returns to work effort and investment, which can deter labor supply and capital accumulation; this effect can dominate at higher levels of emissions abatement when the tax base for carbon pricing is narrower. ### GDP impacts and fiscal multipliers - CPAT models impacts of mitigation policies on GDP using fiscal multipliers. - Fiscal multipliers are defined here as the percent change in GDP in subsequent years of the policy from a percent change in tax increases/cuts and/or spending increases/cuts from a one percent of GDP change in energy taxes. - Multipliers are extracted for over 100 countries from the WB’s main macrostructural model (MFMOD) and from empirically estimated multipliers using a long dataset on the relationship between changes in taxes and GDP over time. - Multipliers are estimated based on a one percentage point of GDP change in taxes (on energy, goods and services, personal income tax or corporate income tax) or expenditures (public investment, transfers, expenditures on goods and services or public wages), and are aggregated at the country income-group level. - Impacts on GDP are lagged by one year (the impact of a policy reform in year 0 affects GDP from year 1 onwards). - Future enhancements envisaged: incorporation of fiscal multipliers from more models and empirical studies, with emphasis on disaggregation of channels through which climate mitigation policies affect GDP. ### Nationally Determined Contributions (NDCs) processing in CPAT - NDC targets vary in reporting format (percentage reduction from historical year, absolute level, percentage reduction relative to assumed BAU); sectoral and GHG coverage varies (notably inclusion/exclusion of LULUCF). - CPAT converts and presents targets consistently by estimating implied target levels of emissions in 2030 where absolute levels are not reported. - For targets stated as percentage reduction versus BAU in a future year, CPAT assumes equivalence to CPAT’s BAU. - Adjustments are made for coverage of GHGs and LULUCF and differences in time periods. - Targets are presented as target level for total GHG emissions (usually excluding LULUCF) in 2030 and can be shown as percentage reduction compared with CPAT’s BAU for cross-country comparison. - Multiple targets shown: conditional and unconditional NDCs treated separately; targets with and without LULUCF treated separately. - For NDC targets given as a range, CPAT assumes an average corresponding to the midpoint of the range. - If an estimated NDC target level is above CPAT’s estimated BAU, CPAT assumes the target is non-binding. - For EU countries, national targets include sectors covered by EU ETS and Effort Sharing Regulation; reductions in both ETS and ESR sectors are assumed to increase in stringency to achieve the EU’s revised target of 55 percent reduction in GHGs compared with 1990 levels at a similar level as required to achieve the EU’s previous target of 40 percent reduction compared with 1990 levels. - Implied NDC estimates are periodically updated as countries revise or clarify their targets to the UNFCCC. ### Non-energy sectors included in CPAT - Sectors beyond energy included (per UNFCCC inventories): - Industrial processes and product use – emissions predominately from the production of cement. - Non-energy agriculture – emissions of methane, mostly from ruminants (cattle, sheep) and rice farming. - Energy agriculture – the energy consumed to power agricultural facilities and other activities. - Land use, land-use change and forestry (LULUCF) – emissions from land use change, notably the CO2 lost from deforestation. - Waste – emissions from waste in dumps, notably methane. - Fossil fuel extraction and distribution – emissions from the production and transport of oil, natural gas, and coal. - Other – catch-all category (e.g., emissions from the military). - Methane context and shares: - Globally, methane accounted for about 20 percent of global GHGs in 2020, of which about 40 percent was from fossil fuel extraction and distribution, about 40 percent from agriculture, and the remainder from waste and other sources. - Modelling approach differs for predominately methane-emitting sectors (extractives, non-energy agriculture, waste) and other sectors (energy agriculture, industrial processes and product use, LULUCF). ### Modelling of methane-emitting sectors (non-energy agriculture, extractives, waste) - Three-step approach in CPAT: 1. Base-year emissions factors are calculated as emissions divided by production. - Base year emissions come from UNFCCC data for non-energy agriculture and waste; an average of UNFCCC and IEA (2022b) for extractives. - Production sources: IEA and Enerdata energy balances for extractives, FAO (FAOSTAT 2022) for agriculture, WB for waste (WB 2022b). 2. Project emissions factors and production forward in BAU: - Emissions factors assume an annual natural decline of 0.25 to 0.50 percent for all but coal, reflecting autonomous technological improvements. - Production of agriculture changes with GDP per capita and population using elasticities of 0.15 and 0.9, respectively. - Production of waste changes with real GDP, using elasticities that vary by country income-grouping (lower-income countries have higher elasticities around 1.0 compared to 0.9 for advanced economies). - Production of oil, natural gas, and coal changes with global demand from the IEA BAU (assumed to be the ‘Stated Policies Scenario’, SPS). - Total BAU emissions are calculated by multiplying production and emissions-intensity. 3. Estimate policy-induced changes to emissions-intensity using marginal abatement cost curves (MACCs): - Constant elasticity specifications imply changes in emissions factors occur at a decreasing rate at higher emissions prices (MACCs are convex). - Elasticity values are initially estimated at country and sector-specific level (using EPA MACCs) then adjusted across countries so global percentage reductions in emissions intensity in response to pricing align with midrange estimates from recent studies. - MACCs are assumed to have a zero intercept, ruling out broader market failures (e.g., firms not exploiting investments that are profitable in the absence of mitigation policy). - Production under the policy scenario is assumed unchanged to reflect revenue-neutral policy design and limited pass-through in imperfect markets. - In general, demand responses are small and equal around 2 and 20 percent of the emissions reduction from a globally coordinated methane fee of USD 70 per ton of CO2e (see Parry and others, 2022). ### Land use, land-use change and forestry (LULUCF) - LULUCF is conservatively assumed constant in the baseline for all countries (both net-sink and net-emitting countries). - Historical data for developed countries is taken from UNFCCC submissions (latest data at time of writing available for year 2019); data for developing countries based on FAO estimates (Tubiello and others 2021). - LULUCF emissions are projected forward assuming a linear reduction during 2020-50 based on SSP5-8.5 scenario changes, which is a 2.5 percent reduction in net emissions per year in the period between 2020-50. - Note: SSP5-8.5 is the ‘worst case’ IPCC scenario; the equivalent compound annual decline under SSP3-7.0 is slightly lower at 2.3 percent between 2020-50. - LULUCF data is highly uncertain with large discrepancies across sources; many models exclude LULUCF for comparability and due to measurement issues. ### Elasticities in non-energy sectors (summary) - Agriculture: - Population elasticity: 0.90 - Per capita income elasticity: 0.15 - Waste: - Population elasticity: 1.00 - Per capita income elasticity: na - Other: - Population elasticity: 1.00 - Per capita income elasticity: na *Source: IMF staff (Annex III, IMF-WB Climate Policy Assessment Tool (CPAT) — working paper content).* ### Annex II – Technical Details: Distribution ### Annex II – Technical Details: Distribution ### Module overview - Analysis is based on changes in energy prices under the climate mitigation policy scenario (relative to BAU) for a given year of interest (e.g., 2030), obtained from the mitigation module. - Price changes are used to estimate impacts on industries and households, accounting for net changes in consumption incidence from revenue recycling. - The distribution module describes: key formulas; estimation of indirect (household incidence) effects; revenue ‘recycling’ ‘modes’; optional features in CPAT’s distribution module. ### Impacts on industries (Direct and Indirect) - Input-output data: - CPAT uses input-output tables sourced from the GTAP-10 database, which contains 2014 data for 65 sectors, including 59 non-energy sectors. - The 65 GTAP sectors include the following five fossil fuels: coal (“coa”), electricity (“ely”), oil (“oil”), natural gas (“gas”, “gdt”) and petroleum products (“p_c”). - Non-energy production energy intensities are estimated as: - (15) 푒_g = f (퐼 − 퐴)^{-1}, f ∉ g - f is a vector assigning an energy intensity coefficient to each non-energy sector. - (퐼 − 퐴)^{-1} is the Leontief inverse matrix. - 퐼 is the identity matrix. - 퐴 is a normalized matrix of technical coefficients based on inter-sectoral product flows from the GTAP database. - Method: - The Leontief inverse of each non-energy sector for each energy input 푒_g is multiplied by the change in the energy input’s price (from the mitigation module). - Yields a first-order estimate of the average increase in input costs faced by firms in non-energy industries in percent of total input costs (includes direct and indirect impacts). - Default assumption: these increases in input costs are fully passed through to output prices faced by households (user-adjustable). - Aggregation: - Input cost increases for 59 non-energy GTAP sectors are aggregated into 14 aggregated sectors with average intensities, 푒_g, and changes in output prices to facilitate household incidence analysis. - For each GTAP energy/fossil fuel sector, f, fossil fuel intensities, 푒_g in Eq. (15) of the 14 CPAT non-energy consumption categories are a weighted average of the energy intensities of the respective non-energy GTAP sectors (using the GTAP IO tables’ household final demand vectors as weights). ### Impacts on households (Direct and Indirect) - Notation and household grouping: - Households grouped into population-weighted, per-capita consumption deciles d = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}. - Budget shares 휋_{t d g} and relative price increases 휌_{t d g} are used to compute impacts. - Consumption burden formula: - (16) ∑_g 휋_{t d g} ∙ 휌_{t d g} - g are (energy, non-energy) goods/services consumed by households. - 휋_{t d g} is the share of decile d’s total consumption spent on good/service g at time t. - 휌_{t d g} is the relative price increase for good/service g caused by the mitigation policy. - Data sources and aggregation: - Household budget shares obtained from national household budget surveys (HBSs), aggregated into 8 energy and 14 non-energy CPAT-compatible categories. - Budget shares computed by dividing total consumption expenditure on each CPAT good/service category by each household’s total consumption expenditure. - Direct impacts (energy prices): - Sector-specific percent price changes in (fossil fuel) energy prices between BAU and policy scenario obtained from the mitigation module. - Example: for a good with a budget share of 2 percent of total household consumption, a 5 percent increase in said good’s price reduces decile d’s consumption by 0.1 percentage points (via equation (16)). - Indirect impacts (non-energy goods/services): - Calculated assuming full pass-through of cost increases to consumer prices domestically (flat/perfectly elastic supply curves). - Non-fuel sector price increases = sum-product of: i) each sector’s intensity in each energy product (fossil fuel); ii) the price increase of each energy product induced by the mitigation policy. - Map sectoral fossil fuel intensities to CPAT consumption categories to re-estimate (16); sum across goods/services to obtain indirect incidence effect. - Total effect: - Sum of direct and indirect effects yields estimate of total effect on consumption, which can be expressed in welfare terms (losses in consumer surplus) and adjusted for imperfect pass-through, behavioral responses, etc. ### Revenue recycling: Targeted transfers - Revenues can be recycled via targeted transfers: new cash transfers or scale-ups of existing social safety nets. - Targeting parameters (user-defined): - the share of total revenues to be used for (new, existing) targeted and/or public spending transfers; - the share of the population to receive the chosen transfer (starting from the bottom of the consumption distribution; C_j^T ≤ p(T) C_j); - the “coverage rate” i.e., the share of the population targeted and actually receiving the transfer; - the “leakage rate” i.e., the share of the untargeted population receiving the transfer. - Modeling: - Transfers modelled as direct, per capita payments averaged by deciles. - Targeted per-capita transfer T for targeted population pop_T is: - (17) T_y = P^M_y / pop^T_y - Survey population scaled to match projected national accounts population, pop, based on IMF WEO vintage: - (18) pop^T_y = (∑ pop_w^T) * θ_pop_y - (19) θ_pop_y = pop_y / ∑ pop_w - Decile-specific average per-capita transfer T^d_y depends on portion of decile population pop^T_{d y} below targeted consumption threshold: - (20) T^d_y = T_y * (∑ pop^T_{d y}) / (∑ pop_{d y}) ### Revenue recycling: Current spending - Option to allocate portion of revenues to general government expenditures (current spending) allocated among: social assistance, insurance, protection, labor support, other social assistance, cash transfers, contributory pensions, public works, in-kind benefits, or cash transfers (based on ASPIRE dataset). - Assumption: schemes scaled proportionately by ∆A for each protection scheme a. - (21) ∆A_{a y} = P^M_y * A_{a y} / A_{a y} (as presented in text: ∆A_{a y} = P^M_y A_{a y} / A_{a y}) - Calculation of total government spending: - CPAT multiplies per-capita quintile transfer amounts by a fifth of the population pop_y to arrive at total spending by quintile, then sums. - Average decile transfers: every two deciles receive same amount as corresponding quintile (e.g., deciles 1 and 2 receive quintile 1 per-capita transfer). - Same approach applies to recycling via existing targeted transfers. ### Revenue recycling: Infrastructure investment - Users can allocate revenues to increase access to: water, sanitation, electricity, information & communication technology (ICT), public transport or average infrastructure access across types. - Portion of decile population receiving transfer for infrastructure i: - (22) pop^T_{i d y} = ∑_{d=1}^{10} pop_{d y} * (1 − Ϝ_{d i}) - Ϝ_{d i} is weighted share of households in decile d who have access to infrastructure category i. - Transfers received only by households without initial access to infrastructure type i (infrastructure access shares Ϝ_{j i} calculated from HBS data). ### Revenue recycling: Personal income tax (PIT) reductions - Distributional effects depend on baseline PIT liabilities by decile. CPAT uses decile shares of PIT liabilities as proxy, obtained from: - Luxembourg Income Study (LIS) “hxitax” variable (available for 26 countries; mostly 2010-2019 except Dominican Republic 2007, Romania 1997, Sweden 2005). - Commitment to Equity (CEQ) “Standard Indicators” database (42 countries; period 2009-2017). - Decile-specific shares in aggregate PIT liabilities are calculated as ratios of decile-specific tax liabilities to sum across deciles. - Missing country observations replaced with mean (or median if user chooses) of country’s World Bank income or regional group. - Calibration of economy-wide PIT paid: - equals product of: i) average PIT-to-GDP ratio during 2010-2019; ii) GDP in year of analysis y. - Analytical expressions: - (23) L_{d c y} = s_{d c} * AVG_PIT_GDP_c * RGDP_{c y} and l_{d c y} = L_{d c y} / P_{d c y} - L_{d c y} and l_{d c y} are total and per-capita PIT liabilities of decile d in country c and year y. - AVG_PIT_GDP_c is 2010-2019 average PIT-to-GDP ratio for country c. - RGDP_{c y} is real GDP in constant 2021 LCU in country c and year y. - P_{d c y} is total population of decile d in country c and year y. - s_{d c y} is share of PIT liability of decile d in country c and year y, proxied by direct/individual income tax data. - Merge with decile-level HBS data assuming 1:1 correspondence between LIS/disposable income and CEQ/market income deciles and consumption deciles from HBSs. When LIS and CEQ overlap, CPAT prioritizes latest-year source. - If user recycles revenues towards “labor tax reductions”, CPAT provides per-capita decile-specific gains under three mutually exclusive PIT liability reduction scenarios: - Targeted Exemption: - Selected HH consumption deciles gain the baseline amount of PIT they pay, conditional on available CP revenues (deciles can be fully exempt). - (24) g_{d c y_TE} = l_{d c y} * I[d = exempt] * I[0 ≤ remaining CT revenues ≤ l_{d c y}] - Indicator functions denote selection and available revenue constraints. - Note: if CP revenues insufficient to fully offset liability, Distribution Module allocates remaining revenue starting from that decile upwards. - Personal Allowance: - PIT liabilities uniformly reduced across PIT-paying population (per-capita lump-sum to working population). - (25) g_{d c y_PE} = min { l_{d c y}, r_{c y} } - r_{c y} is ratio of all available CT revenues to sum of all individuals (maximum mean per-capita gain). - Any remaining revenues equally divided across all individuals and paid as additional gains. - Proportional Compensation: - Each decile receives average per-capita gain increasing with baseline PIT liability. - (26) g_{d c y_PC} = l_{d c y} * f_{c y} - f_{c y} = total LCU available CP revenues / total LCU PIT paid across all deciles. - Caveats and assumptions: - PIT liability calculations abstract from detailed fiscal regime data (tax credits, surtaxes, deductions). - Assumes LIS and CEQ household survey data capture PIT paid across income distribution. - Gains from PIT reductions assumed equally distributed across population sub-groups (working/non-working, adults/children, men/women). - Assumes perfect correspondence between consumption and income deciles. - Assumes no changes in PIT payments/compliance in response to climate mitigation policy. - Urban vs. rural gains scaled by share of urban and rural population in total population. ### User options and optional adjustments - Behavioral and structural change-adjusted incidence: - IO-based analysis assumes fixed technical coefficients and full price pass-through; estimated incidence is an upper bound/short-term estimate. - CPAT computes an adjustment comparing mitigation module revenues P^M_y and distribution module revenues P^D_y (P^M_y < P^D_y). - (27) P^D = ∑ p_j * C_j - (28) P^D_y = P^D * GDP_y * θ_c / ∑ C_j (GDP_y represents expected GDP in year y; θ_c is final consumption expenditure to GDP ratio from WDI) - (29) p_Adj = (P^D_y − P^M_y) / P^D_y, when P^M_y < P^D_y - p_Adj applied multiplicatively across all household consumption deciles to scale down incidence. - Decile-specific price elasticities of demand: - Short-term demand-side adjustments modeled using decile-, country- and consumption category-specific price elasticities from USDA data. - Elasticities provided by country and consumption category (COICOP) for high-, middle- and low-income countries. - CPAT maps elasticities to distribution module consumption categories; uses country reported elasticities for middle deciles and scales upper/lower deciles proportionally based on deviations between low/high and middle-income country elasticities. - Under CES utility functional forms, country-decile-specific price elasticity 휖_use_g modifies budget share C_{j g} as: - (30) C_{j g} * (p_t / p_{t-1})^{휖_use_g} - Imperfect pass-through: - Producers may absorb part of mitigation-induced price increase. - Coefficients γ_g (γ_g ≤ 1) applied to effective tax rate t for sector g: - (31) t_{f g}^* = t_{f g} * γ_g - Coefficients based on Ganapati and others (2020), Neuhoff & Ritz (2019) and Abdallah and others (2020). - Emissions-based adjustment of sectoral price changes: - Recognizes potential mismatch between GTAP monetary energy flows and observed energy flows (from IEA) used in mitigation module. - CPAT can adjust consumer price incidence using theoretical “time-zero” revenue flows per fuel and CPAT sector (revenues that would have been raised if no price-induced adjustments had occurred). - The final demand portion of price incidence (usually around 60%) is taken as static consumption incidence; CPAT allows scaling of initial incidence effects to this level and recomputes estimates. - Cooking-fuel adjusted incidence effects: - Option to exclude the primary fossil fuel used for cooking from simulated mitigation policy for selected deciles (starting from the bottom decile). - Interpreted as a rebate to limit incentives to switch to biomass for cooking. - Primary cooking fuels identified for each country using WHO Household Energy Database. - Households pay the carbon tax/price when buying fuel but receive compensatory transfer for cooking needs. *IMF Working Paper — Annex II – Technical Details: Distribution (CPAT distribution module).* ### Annex III – Technical Details: Co-Benefits ### Annex III – Technical Details: Co-Benefits ### Air pollution co-benefits — emissions factors - Emissions factors for local pollutants are sourced from IIASA. - Baseline dataset: ECLIPSE_V5a_CLE_base. - Coverage: data for 89 countries (regional averages used for others). - Dataset creation date: June 2015. - Temporal coverage: 1990 to 2050 in five-year intervals. - Assumption: emissions factors generally decrease as higher-quality fuel and better control technologies are introduced. - Emissions factors grouped to reflect CPAT sectors and fuels as described in Wagner and others (2020). ### Air pollution co-benefits — concentrations to health outcomes and economic costs - Main health outcomes calculated: - Premature mortality attributed to pollution from fuel use. - Lost disability-adjusted life years (DALYs). - Baseline data sources: WHO; GBD, 2019. - PM2.5 methodology: - CPAT uses an approach based on GBD (2019) to model relationship between pollutant concentrations and health outcomes. - A relative risk (RR) curve determines the air pollution-attributable fraction in the burden of diseases. - RR curves differentiate impacts among age groups: neonatal, post-neonatal, under-15 years old, 15 to 64 years, and 65 years or older. - RR curves differentiate by diseases (e.g., chronic obstructive pulmonary disease, stroke, ischemic heart disease). - Ozone impacts: - Use a similar methodology with RR functions based on Turner and others (2016). - Multiple risk-factor adjustments: - Adjustments are made when multiple risk factors impact the same outcome to avoid double counting and omitting any non-linearity in relationships between exposure and health outcomes. - Household air pollution (HAP): - HAP defined as the use of solid fuels for cooking. - Included because: i) the GBD (2019) framework requires joint estimation of ambient and household pollution; ii) higher fossil fuel prices could lead to households using more solid fuels (e.g., biomass). - When HAP is high relative to ambient pollution, health gains from reducing ambient pollution will be much smaller than when HAP is low or zero; thus, for countries with high HAP, health impacts from carbon pricing can be small. ### Economic costs of outdoor air pollution (externalities) - Work absenteeism: - Leads to wage losses, worker replacement costs, and productivity declines for other workers dependent on the absent labor. - Modeled using approaches from Ostro (1987) and Holland (2014). - Baseline levels of absenteeism from OECD (2020) and WHO (2019). - Market output losses: - Due to working years lost from mortality and years with disability. - CPAT uses a model based on Pandey and others (2021), without adjustments for rural vs. urban samples. - Mortality is assumed to occur up to 20 years from the time of exposure to elevated levels of pollutants. - Health expenditures attributed to pollution: - Total costs due to pollution are a percentage of total costs, based on the percentage of DALYs caused by fuel-related pollution (similar to Preker and others, 2016). - Expected future health care costs at the country level sourced from IHME (2020). ### Road transport co-benefits — VKT, baselines, and projections - Baseline data sources: - Vehicle kilometers traveled (VKT) and road maintenance: World Road Statistics (IRF 2021). - Congestion-related data on free-flowing speeds and actual speeds: TomTom. - Driving-related fatalities: WRS, UNECE, and OECD. - Projection method: - VKT and externalities projected forward using country-specific data. - Apply short- and long-run elasticities with respect to prices and macro-economic indicators. - Separate regressions run for each outcome variable (VKT and each externality). - Estimation approach follows Burke and Nishitateno (2015): within estimator from a static fixed-effects equation represents shorter-run effects; between estimator provides long-run effects. ### Road transport co-benefits — regression specification and elasticities - Regression (Equation (32)): - ln(Y_{c,t}) = α + β ln(p_{c,t}) + γ X_{c,t} + μ_t + μ_c + ε_{c,t} - Where: - Y = outcome variable (VKT, congestion, fatalities, road maintenance) - p = gasoline price - X = a vector of covariates including GDP per capita and population - μ = country and year fixed effects - c = country - t = year - Estimated average elasticities with respect to fuel prices: - VKT: short-run -0.28; long-run -0.56. - Congestion: short-run -0.34; long-run -0.81. - Road accident fatalities: short-run -0.61; long-run -0.44. - Road damage (with respect to diesel prices): short-run elasticity not statistically significant; long-run elasticity on average -0.44. - Note: road damage results are in line with the literature. *Annex III – Technical Details: Co-Benefits, IMF-WB Climate Policy Assessment Tool (CPAT).* ### 396. Elsevier. https://doi.org/10.1016/S0140-6736(20)30752-2. ### The IMF-World Bank Climate Policy Assessment Tool (CPAT): A Model to Help Countries Mitigate Climate Change — Working Paper No. WP/2023/128 (excerpt) ### Nature of this content unit - This content unit is a references/bibliography excerpt from the working paper (pages 63–69). - It compiles citations to academic articles, institutional reports, datasets, and working papers used in the IMF-WB Climate Policy Assessment Tool (CPAT) analysis. - The excerpt does not present new empirical findings, projections, scenarios, or policy recommendations itself; it documents sources underpinning the working paper. ### Thematic coverage indicated by the cited literature - Climate mitigation policy design and evaluation (carbon taxes, emissions trading, policy sequencing, border carbon adjustments). - Energy sector analysis and transition issues (World Energy Outlook, Energy Efficiency, vehicle fuel economy, fleet turnover, methane emissions). - Macroeconomic and fiscal analysis related to climate action (Fiscal Monitor, Fiscal Policies for Paris Climate Strategies, country Selected Issues reports). - Health and pollution impacts and valuation (Global Burden of Disease, IHME tools, WHO source apportionment, studies on absenteeism and health care expenditures). - Non-CO2 greenhouse gases and methane (Global Methane Tracker, coal mining emissions studies, mitigation potential reports). - Technology, innovation, and low-carbon technology transfer (studies on technology diffusion, innovation economics, and policy support for technology transfer). - Modeling approaches and methodological references (input-output analysis, integrated assessment models, marginal abatement cost curves, emission factor documentation). - Empirical studies on energy demand elasticities, fuel price pass-through, and household/sectoral responses to pricing and regulation. ### Types of sources compiled - Peer-reviewed journal articles and meta-analyses. - International organization reports and databases (IMF, IEA, World Bank, IPCC, WHO, IHME, IIASA). - Working papers, technical notes, and policy research papers. - Data repositories and tools cited as inputs (GBD Results Tool; SSP Database; World Bank Carbon Pricing Dashboard; SEDAC GPWv4). *Source: Excerpt of references from The IMF-World Bank Climate Policy Assessment Tool (CPAT): A Model to Help Countries Mitigate Climate Change — Working Paper No. WP/2023/128 (pages 63–69).* --- _Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023128-print-pdf.pdf_