GHG data by country tier

What each tier starts from¶

Tier 0: The U.S. GHG model¶

The U.S. has the most detailed emissions data, and its model is built on all 405 common sectors.

National totals. The U.S. greenhouse gas inventory sets the national totals for 2024, by emitting activity and gas. These come from the University of Maryland’s GHG inventory analysis Desai et al., 2026, which follows the tables of EPA’s Inventory of U.S. Greenhouse Gas Emissions and Sinks, with detail from EPA’s inventory for 2023 where it is needed (for example, agricultural soils and fluorinated gases) U.S. Environmental Protection Agency, 2025. The inventory is on a territorial basis.

Placing the totals on industries. The inventory names an industry for some emissions and not others, so each emitting activity is placed on industries by one of these routes:

• Directly. Where the inventory names the industry, as for most process and fugitive emissions (petroleum and natural gas systems, cement, and other industrial processes), the emissions go to that industry.

• Facility reports. Most fuel burned at industrial plants is placed using what the plants themselves report: emissions reported to EPA’s GHGRP U.S. Environmental Protection Agency, 2026, and to the National Emissions Inventory U.S. Environmental Protection Agency, 2025, matched to one record per site. Power plants come from eGRID for 2024 Cornerstone Sustainability Data Initiative, 2026. For manufacturing natural gas, coal, and petroleum fuels, each industry’s share comes from facility reports where they cover at least 80% of the industry’s combustion, and from the Manufacturing Energy Consumption Survey (2022) where they don’t U.S. Energy Information Administration, 2025.

• Physical fuel use. Fuel burned in vehicles and equipment is placed on the industries that use it, based on fuel volumes. Highway fuel is split by vehicle owner using Federal Highway Administration data for 2024 Federal Highway Administration, 2025, with federal fleets from GSA’s Federal Fleet Report U.S. General Services Administration, 2025 and private trucks by business type from the Census Bureau’s Vehicle Inventory and Use Survey (2021) U.S. Census Bureau & Bureau of Transportation Statistics, 2023. Industrial diesel is split across agriculture, construction, mining, and manufacturing from fuel sales and survey data U.S. Energy Information Administration, 2022U.S. Energy Information Administration, 2025.

• Purchases. The remaining emissions are allocated to the industries that buy the relevant fuel or product, in proportion to their purchases in the U.S. Use table.

Households. Emissions the inventory places in homes and personal vehicles go to households and are not in B (see The GHG extension: overview).

Tier 1: National sources¶

Each Tier 1 country’s emissions come from its own national data.

China’s data are on a territorial basis, and the UK’s on a residence basis. Japan’s 3EID data are built on the same 2020 IO table and reconciled to Japan’s national GHG inventory.

China has no published emissions table matching its IO table, so we build one:

• CO₂ from fossil fuels and cement sets the totals, from Our World in Data Ritchie et al., 2023, and is placed on sectors by each sector’s physical fuel use from China’s national energy statistics.

• Power-sector CO₂ is set to match China’s official 2023 grid emission factor Ministry of Ecology and Environment of the People's Republic of China & National Bureau of Statistics of China, 2025.

• CH₄ and N₂O come from the Community Emissions Data System (CEDS) Hoesly et al., 2025Hoesly et al., 2018, and farm CH₄ and N₂O from China’s first Biennial Transparency Report People's Republic of China, 2024.

• Fluorinated gases come from China’s first Biennial Transparency Report, brought forward to 2023 with EDGAR trends Crippa et al., 2025.

Household fuel use is not separated in China’s national totals, so its CO₂ is spread over the industries rather than left out.

South Korea’s source covers six gases. The missing gas (NF₃) is estimated from the sector’s CO₂ using the U.S. ratio.

Each country’s emissions are then mapped to the common sectors with structural reflection, the same as its IO table. Each national total is kept.

Tier 2: FIGARO and ICIO emissions data¶

Tier 2 countries take their emissions from the same source as their IO table, so emissions and output line up sector by sector. The two sources differ a lot in detail.

FIGARO countries¶

FIGARO emissions come from two kinds of underlying data:

• Air emissions accounts (30 countries). The 27 EU member states report emissions by industry in their air emissions accounts Eurostat, 2026. For Norway, Switzerland, and Türkiye, Eurostat fills in the industry split.

• EDGAR estimates (10 countries). For Argentina, Australia, Brazil, Canada, India, Indonesia, Mexico, Russia, Saudi Arabia, and South Africa, Eurostat estimates emissions by industry from EDGAR Crippa et al., 2025Schaaf et al., 2025. EDGAR groups some activities together, so we correct the split. Within mining, refining, and the power and gas utilities, CO₂, CH₄, and N₂O are re-split using CEDS shares Hoesly et al., 2025, holding the combined total fixed. Shares for steel, chemicals, and non-metallic minerals are also reset to CEDS, and process CO₂ is kept off trade and service industries.

For both groups, FIGARO’s industries are first converted to products using FIGARO’s supply table, to match the product-by-product IO table. The same broad sectors split in the IO table are split here too:

• Electricity, gas, and heat are separated using Eurostat energy balances Eurostat, 2026 and CEDS gas distribution methane.

• Agriculture is split into farm sectors using FAOSTAT CH₄ and N₂O FAO, 2025 and Eurostat agricultural accounts Eurostat, 2026.

• Mining, non-metallic minerals, and basic metals are split as in the IO table.

The emissions are built on the 2025 FIGARO tables, while the FIGARO IO tables are the 2026 edition. Only each country’s direct emissions by industry are used, so the difference matters mainly for non-EU countries, where Eurostat estimates the emissions.

The products are then mapped to the common sectors with ABSR. The last gas group, SF₆ with NF₃, is split using the U.S. share of SF₆ in each sector.

ICIO countries¶

The OECD data give one CO₂e total for each of the 50 industries, with no split by gas. Farming is split out first, using FAOSTAT CH₄ and N₂O FAO, 2025, before the totals are mapped to the common sectors with ABSR. The split by gas comes entirely from the U.S. mix for each sector. Electricity is mapped as one sector, with no separate split for gas and heat.

Tier 3: Template emissions¶

Tier 3 countries have no emissions data matched to an IO table. Their direct emission intensities start from the same template as their input recipe, a blend of Mexico (30%), Malaysia (30%), and Tunisia (40%). The template uses those countries’ CO₂ emissions by industry for 2018, from the OECD’s 2021 edition of CO₂ emissions embodied in international trade OECD, 2021. Other gases are added using the U.S. ratio of total CO₂e to CO₂ in each sector.

Each country then adjusts the template two ways:

• Overall level. All intensities are scaled by the country’s fossil CO₂ per unit of GDP relative to the template’s. Fossil CO₂ is from EDGAR (version 6.0) for 2018 Crippa et al., 2021Monforti-Ferrario et al., 2021, and GDP from the IMF World Economic Outlook (October 2021) International Monetary Fund, 2021.

• Electricity. The electricity sector uses the country’s own power-sector CO₂ from EDGAR for 2018, divided by the value of its electricity output: 2018 generation from the U.S. Energy Information Administration U.S. Energy Information Administration, 2026 times an average electricity price from the World Bank.

Rest of world¶

The rest-of-world region’s direct emission intensities are the output-weighted average of four Tier 2 countries: Colombia and Chile (from ICIO) and Mexico and Türkiye (from FIGARO). See The IO model: overview for how the region is used.

Data years¶

Emissions data are published on different schedules, and the model uses the most recent for each country.

Emissions stay at their data year. For the data year of each country’s IO table, see Methods by country.