What kind of emissions does our analysis cover?

This analysis combines two kinds of emissions:

Territorial emissions:  calculated based on the emissions produced within a geographic boundary. These include emissions associated with household consumption, capital investments and government expenditure in a country. They consider emissions from products produced for exports but do not consider emissions associated with imported products. 

Net emissions embodied in trade: emissions associated with imported products minus emissions associated with exported products

The national level consumption emissions (that we consider in this analysis) are calculated as the territorial emissions plus the net emissions embodied in trade.

How does the analysis treat investments?

While investment emissions are included in our analysis, we do not separate them out. Allocating investment emissions to a specific income group is challenging. Many assets are held by institutional investors who invest on behalf of a wide array of individuals and firms. Only the individuals and firms with the largest holdings have a significant influence on the institutional investors. Additionally, physical investments (i.e. gross fixed capital formation) are different from financial investments. Understanding investments is a part of our next set of research.

How do we allocate country-level consumption emissions to individuals?

We divide national consumption emissions among individuals based solely on their income. We assume that above a certain base level of emissions (floor) and below a certain maximum emissions level (ceiling), emissions increase in proportion to income. We assume that individuals in the same income percentile in each country generate the same per capita emissions. 

Where does our historical data come from?

We take historical gross domestic product (GDP) and population data at 2021 USD in purchasing power parity (PPP) terms from the World Bank's World Development Indicators¹, gap-filled with data from the Penn World Tables².

We obtain the share of national income (i.e. GDP) for each income percentile from the World Inequality Database (WID).^3,4 WID systematically combines national accounts, survey, wealth and fiscal data to represent income as accurately as possible given the lack of precise information at the top income levels due to underreporting. Although the income data used for deriving the income shares is based on adults over 20 years of age, we assume that a given income share was applicable to all individuals within the respective percentile group.

Data on national emissions between 1990 and 2022 are taken from the Global Carbon Atlas.^5,6 For global statistics, we use the total global emissions for 1990 to 2023 from the Global Carbon Budget.⁷ We used fossil carbon dioxide (CO2) consumption emissions determined by adding together territorial emissions and trade emissions. We do not consider non-CO2 emissions and emissions from land-use change (e.g., emissions from clearing forests for agriculture) due to limited publicly available data in this regard.

Why did we compare household incomes on PPP as opposed to MER?

In this analysis, we compared household incomes in different countries' currencies using purchasing power parity (PPP), which considers differences in the cost of living between countries, as opposed to market exchange rates (MER), which only accounts for currency exchange rates. The 2018 World Inequality Report⁸ explains that using market exchange rates, the richest global 1% have four times as much income as the bottom 50%, whereas using PPP exchange rates, they have twice as much. However, PPP is the standard adopted in the World Inequality Database. Therefore, our decision to compare incomes on a PPP basis makes richer and poorer countries seem more equal than they may be.

Why is it important to set a floor and ceiling?

We assume that all humans use resources for their survival, and therefore emit a minimum amount of carbon to live (an emissions floor). We also know that resource use is limited by its availability and production capacity, and so there is a determinate amount of carbon emissions generated each year. Given that carbon emissions are finite, we assumed a household has a maximum amount of carbon emissions that it produces (an emissions ceiling). This upper limit is informed by literature on very high-income carbon footprints.^9-16

How do we set the minimum emissions level (floor)?

When allocating national consumption emissions across income groups, we apply a country-specific per capita emissions floor below which we assume an individual's emissions will not fall. This floor is the emissions associated with an income equal to 30% of the median income of a country. This income level is one-half of the level defined for the European Union's risk-of-poverty threshold (60% of the median income).¹⁷ We reason that the actual minimum emissions level should be set well below that of a household at risk of poverty in a high-income region and therefore picked one-half of the threshold. We tested our analysis with the higher floor of 60% of the median income. The changes in the global distribution of carbon inequality were within a couple of percentage points – a statistically insignificant difference.

How do we set the maximum emissions level (ceiling)?

When allocating national consumption emissions across income groups, we apply a global emissions ceiling above which we assume per capita emissions do not rise regardless of income. In our previous analysis, we assumed a conservative ceiling of 300 tons of carbon dioxide per capita. In the 2025 analysis, the maximum emissions ceiling was raised from 300 to 3000 tons of carbon dioxide per capita to reflect new evidence of the carbon footprint of the richest individuals in literature.^14-16 Even with the increase, this will limit the perceived growth of emissions at the highest income levels.

How do emissions change with income? What elasticity number did we use in our analysis?

We assume that between the floor and ceiling discussed above, emissions rise in relation to income and that the relationship can be expressed as an elasticity of emissions with respect to income. Depending on income-dependent consumption behaviour in a given country, emissions may grow faster than income (elasticity >1), in proportion to income (e=1), or more slowly than income (elasticity <1).

In our analysis we applied an elasticity of 1 to the income ranges between the floor and ceiling. In other words, we use a piecewise constant elasticity:

elasticity = 0.0 (low income: i.e. income < 30% national median income)

elasticity = 1.0 (medium income: i.e. above lower and below higher income)

elasticity = 0.0 (high income: i.e. income such that emissions > 3000 tons of CO₂/capita)

Thus, our analysis is not equivalent to assuming that the dependence of emissions on income is characterized by an elasticity of 1.0. If one can define an effective elasticity as the weighted average across the population of local elasticity, then our methodology yields an effective elasticity that varies by country and is generally about 0.82. The effective elasticity is a less progressive elasticity than has been used in previous studies.^18-20 This suggests that we err on the side of underestimating carbon inequality.

How are the income levels defined?

The following table presents the average and minimum income per capita for each income group on a global basis. To calculate this, we multiplied the share of national income for each income percentile (using data from the WID) by the GDP (in 2021 USD PPP) to get the income for each population percentile in each country. We sorted the population percentiles across all countries by income, grouped populations together by the income groups shown in the table below, and determined the minimum and average per capita income for each group.

Note that the household income is given in 2021 USD PPP, meaning that the income is adjusted for the differences in purchasing power for a given currency. This allows us to compare the spending potential of households across countries.

Table 1: Population, income and CO₂ emissions per income group, 2019

How is the income distributed to children?

The income shares that we use from WID are for adults over 20 years of age. However, we assume that this income is being spent on the consumption of everyone including children within a particular income group. Since our analysis looks at how income translates to consumption and then to emissions, it is reasonable to assume that the income is applicable to children as well.

References

1. World Bank. (2023). World Development Indicators. https://databank.worldbank.org/source/world-development-indicators
2. Groningen Growth and Development Centre. (2023). Penn World Table version 10.01. https://doi.org/10.34894/QT5BCC
3. World Inequality Lab. (2025). Data – WID – World Inequality Database. https://wid.world/data/
4. Alvaredo, F., Atkinson, A. B., Chancel, L., Piketty, T., Saez, E., & Zucman, G. (2018). Distributional National Accounts Guidelines: Methods and Concepts Used in WID.World. https://wid.world/document/dinaguidelines-v1/
5. Global Carbon Project. (2025). Carbon Emissions. Global Carbon Atlas. Accessed March 2025. https://globalcarbonatlas.org/emissions/carbon-emissions/
6. Andrew, R. M., & Peters, G. P. (2025). The Global Carbon Project’s fossil CO₂ emissions dataset (Version 251022) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.17417124
7. Global Carbon Project. (2024). National fossil carbon emissions v2024 [xlsx]. (Friedlingstein et al., 2024b, ESSD). https://globalcarbonbudgetdata.org/latest-data.html
8. Alvaredo, F., Chancel, L., Piketty, T., Saez, E., & Zucman, G. (2018). World Inequality Report 2018. World Inequality Lab. https://wir2018.wid.world/
9. Climate Analytics & NewClimate Institute. (2021). Climate Action Tracker. https://climateactiontracker.org/
10. Ummel, K. (2014). Who Pollutes? A Household-Level Database of America’s Greenhouse Gas Footprint. Center for Global Development. http://www.ssrn.com/abstract=2622751
11. Chancel, L., & Piketty, T. (2015). Carbon and Inequality: From Kyoto to Paris. Paris School of Economics. https://doi.org/10.13140/RG.2.1.3536.0082
12. Otto, I. M., Kim, K. M., Dubrovsky, N., & Lucht, W. (2019). Shift the focus from the super-poor to the super-rich. Nature Climate Change, 9, 82–84. https://doi.org/10.1038/s41558-019-0402-3
13. Gössling, S. (2019). Celebrities, air travel, and social norms. Annals of Tourism Research, 79, 102775. https://doi.org/10.1016/j.annals.2019.102775
14. Barros, B., & Wilk, R. (2021). The outsized carbon footprints of the super-rich. Sustainability: Science, Practice and Policy, 17(1), 316–322. https://doi.org/10.1080/15487733.2021.1949847
15. Chancel, L. (2022). Global carbon inequality over 1990–2019. Nature Sustainability, 5, 931–938. https://doi.org/10.1038/s41893-022-00955-z
16. Starr, J., Nicolson, C., Ash, M., Markowitz, E. M., & Moran, D. (2023). Income-based US household carbon footprints (1990–2019) offer new insights on emissions inequality and climate finance. PLoS Climate, 2(8), e0000190. https://doi.org/10.1371/journal.pclm.0000190
17. Eurostat. (n.d.). Glossary: At-risk-of-poverty rate. https://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:At-risk-of-poverty_rate
18. Dorband, I. I., Jakob, M., Kalkuhl, M., & Steckel, J. C. (2019). Poverty and distributional effects of carbon pricing in low- and middle-income countries – A global comparative analysis. World Development, 115, 246–257. https://doi.org/10.1016/j.worlddev.2018.11.015
19. Hubacek, K., Baiocchi, G., Feng, K., Muñoz Castillo, R., Sun, L., & Xue, J. (2017). Global carbon inequality. Energy, Ecology & Environment, 2, 361–369. https://doi.org/10.1007/s40974-017-0072-9
20. Oswald, Y., Owen, A., & Steinberger, J. K. (2020). Large inequality in international and intranational energy footprints between income groups and across consumption categories. Nature Energy, 5, 231–239. https://doi.org/10.1038/s41560-020-0579-8