Doughnut of social and planetary boundaries monitors a world out of balance – Nature

This section summarizes how we select and analyse social indicators with respect to minimum social standards in the Doughnut’s social foundation and ecological indicators with respect to planetary boundaries and downscaled per capita boundaries in the Doughnut’s ecological ceiling. A detailed description of methods and data for each indicator is provided in the Supplementary Information.

Conceptual framework

There have been two previous iterations of the Doughnut framework: the first version was published in 2012 (ref. ¹) and the second in 2017 (refs. ^2,3). Since the 2017 version of the Doughnut was published, there have been substantial advances in the global availability of data on social outcomes, especially those indicators monitored in the global indicator framework of the UN SDGs⁵¹. There have likewise been substantial advances in planetary boundaries science, which have been synthesized at the global scale in the 2023 iteration of the planetary boundaries framework¹³ and have also seen widespread uptake across scientific disciplines and society²⁶. In this Analysis, we draw on these social and ecological advances in understanding to renew and strengthen the Doughnut framework, through a revision of its dimensions and indicators and the inclusion of available time-series data at the global scale. Building on previous efforts to ‘downscale’ the Doughnut^31,32, we also develop an approach that disaggregates the global Doughnut in a cross-scale comparable framework to reveal inequalities in social shortfall and ecological overshoot across country clusters. See ref. ⁵² for a comparison with previous iterations of the Doughnut.

Global time-series data

We collected time-series data over the 2000–2022 period from publicly available international databases, such as the UN’s SDG Indicators Database¹⁴ and scientific sources referenced in the most recent planetary boundaries update¹³, among others (see Supplementary Information for all data sources). The first year considered in our analysis is 2000, which is the earliest year available in the SDG Indicators Database¹⁴ and coincides with our ambition to monitor humanity’s trajectory in the twenty-first century. For comparability and continuity, we aimed to collect time-series data using the same indicators as Raworth^2,3 in the 2017 version of the Doughnut, unless such data were unavailable or more relevant alternative indicators have since become available.

We collected global estimates directly from international databases for the planetary boundary indicators and calculated population-based estimates from large samples of national values for the social indicators, which were also collected from published and publicly available international databases. For the social indicators, we have included large samples of countries throughout, but we note that there is a varying number of countries with data available per indicator, with a median (and interquartile range) of 151 (124–187) countries included and a 90% (83–99%) share of the world’s population (Supplementary Table 1). We acknowledge that indicators measured on the basis of different methods, data sources, system boundaries and country coverage may not be fully comparable, but the holistic nature of the Doughnut makes this unavoidable for the time being, given current data availability.

Furthermore, we note that our use of data published by international data sources does not explore uncertainties within the historical estimates of each social and ecological indicator nor do we fully account for possible shifts within indicator-specific reporting systems. We do, however, document all data sources used in our Analysis in Supplementary Tables 1 and 2, providing explicit references to relevant metadata documentation for each social and ecological indicator issued by its respective data provider. Overall, we include a maximum of two indicators for each dimension and select 22 social indicators and 13 ecological indicators in total (compared with 20 social indicators and ten ecological indicators in the 2017 version).

Establishing the social foundation

To establish the social foundation, we began with the list of social dimensions and indicators in the 2017 Doughnut and investigated their current relevance in light of advances in the global indicator framework monitoring the SDGs, following six criteria adapted from Raworth², which postulate that social indicators should:

• Be globally relevant and serve as effective proxies for broader concerns in their respective dimensions.

• Focus on the worst-off, measuring deprivations as a percentage of the population (as opposed to measuring averages).

• Be measured with sufficiently recent data, with extensive international coverage and with time-series monitoring.

• Have officially recognized thresholds of minimum acceptable standards.

• Be monitored in the global indicator framework of the UN SDGs.

• Include at most two indicators for each dimension of the social foundation.

We considered hundreds of indicators in publicly available international datasets, reached out to relevant experts and used our own expertise to narrow down a draft set of social dimensions and indicators that most closely aligned with these ideal criteria. We conducted an internal review of this draft set of dimensions and indicators with colleagues holding a range of perspectives: some with insight on the broad constituents of human flourishing, others with a focus on specific indicators and data sources (see ‘Acknowledgements’). We are grateful for the comments and suggestions shared by colleagues during the internal review process but we make no claims of a representative or exhaustive set of relevant insights nor do we claim that all internal reviewers agree with all of our choices—we take full responsibility for the final indicator selection, as shown in Table 1. We acknowledge that the social indicators that we have selected as best-available proxies for broader concerns in their respective dimensions are open to debate, given that other indicators exist, although in practice we have found the six criteria listed above to be limiting. See Supplementary Discussion 1 for further details on each social indicator and Supplementary Table 1 for data sources.

Establishing the ecological ceiling

At the global scale, the dimensions of the Doughnut’s ecological ceiling are defined by the nine critical Earth-system processes specified in the planetary boundaries framework^13,53,54. The planetary boundaries framework has seen a very large uptake across academic disciplines, policy circles and the general public²⁶ but it has also been the subject of considerable scientific debate, which has surfaced many critiques⁵⁵ and responses since the original formulation in 2009. Although it is beyond the scope of our study to provide an in-depth critical assessment of this enormous body of literature, we emphasize that our use of the planetary boundaries to define the Doughnut’s ecological ceiling is not taken lightly or uncritically—it is an intentional choice based on our understanding of the state of the art in Earth-systems science and, crucially, our aligned commitment to iteratively incorporate new knowledge as that science continually advances. We acknowledge that the limitations of the planetary boundaries framework could affect our results and may require further investigation.

Furthermore, we note that several of the terms used in the planetary boundaries framework have been altered in the Doughnut with the aims to: (1) align terminology across the social and ecological domains (that is, we generally refer to ‘dimensions’ and ‘indicators’ throughout, rather than ‘Earth-system processes’ and ‘control variables’) and (2) make them more accessible to a non-technical audience.

To establish the ecological ceiling, we began with the list of ecological indicators in the 2023 planetary boundaries update and assessed their applicability considering our ambition to monitor time series according to two criteria, namely, that ecological indicators should:

• Be part of the 2023 update to the planetary boundaries framework, tracking Earth-system processes in the same units, ideally with the same data sources.

• Be measured with sufficiently recent published data, with global coverage and with time-series monitoring.

We considered dozens of indicators in published data sources from international databases and the scientific literature, reached out to relevant experts and used our own expertise to narrow down a draft set of ecological indicators that most closely matched these criteria for the nine ecological dimensions of the planetary boundaries. We conducted an internal review of this draft set of dimensions and indicators with colleagues holding a range of perspectives: some with insight on planetary boundaries science, others with a focus on specific indicators and data sources (see ‘Acknowledgements’). We are grateful for the comments and suggestions shared by colleagues but we do not claim that the internal reviewers cover a representative or exhaustive set of relevant expertise nor do we claim that all reviewers would agree with all of the indicators we have chosen. We take full responsibility for the final indicator selection, as shown in Table 2.

Notably, the chemical pollution indicator is an exception to the criteria above, for which we propose to focus on the production of hazardous chemicals. This metric is distinct from but related to the (unquantified) indicator in the planetary boundaries framework on the ‘percentage of synthetic chemicals released to the environment without adequate safety testing’¹³. Our approach combines a focus on the total production of chemicals, which is in line with the landmark study of Persson et al.⁵⁶ on novel entities, with calls for nature-based chemicals that are conducive to life, rather than hazardous to health and environment^57,58. See Supplementary Discussion 2 for further details on each ecological indicator and Supplementary Table 2 for data sources.

Calculating social shortfall and ecological overshoot

In our Analysis, social indicators are presented relative to their extent below the social foundation and ecological indicators are presented relative to their extent beyond the ecological ceiling. For the social indicators, each one is already expressed in percentage terms as the proportion of the global population falling below its respective minimum standard. These percentages can therefore be interpreted directly as the normalized extent of social shortfall between 0% (no shortfall) and 100% (complete shortfall) each year.

For the ecological indicators, which are expressed in absolute units, the normalization procedure scales each indicator–boundary pair by assigning the pre-industrial Holocene baseline a value of zero, divides each indicator value by its respective planetary boundary and then subtracts one from each normalized ratio (see Supplementary Table 2 for pre-industrial Holocene baselines provided by the planetary boundaries framework¹³). In percentage terms, the normalized extent of ecological overshoot has a lower bound of −100% (the pre-industrial baseline), with 0% indicating no overshoot, and no upper bound for values greater than zero.

In mathematical terms, the general formula for the normalized ecological overshoot in a given year is given by overshoot_t = (x_t − x_base)/(x* − x_base) − 1, in which x_t is the ecological indicator in year t, x* is the planetary boundary and x_base is the pre-industrial baseline. However, three indicator–boundary pairs are exceptions to this general formula (aragonite saturation state, forest area and stratospheric ozone), as they are each framed inversely, for which a decrease in the indicator value implies worsening ecological conditions (rather than an increase). The normalization formula to express overshoot in comparable terms for these inverted indicator–boundary pairs is given by \({{\rm{overshoot}}}_{t}=\left(1-\frac{{x}_{t}}{{x}_{{\rm{base}}}}\right)/\left(1-\frac{x* }{{x}_{{\rm{base}}}}\right)-1\).

Determining historical and scenario trends

We estimated historical trends over time for each indicator of social shortfall and ecological overshoot using ordinary least squares regression and two-sided hypothesis tests with a linear model, or y = β₀ + β₁t, in which t is a year index with base year 2000, y is the indicator and β₀ and β₁ are the regression coefficients (intercept and slope, respectively). See Supplementary Data for the full regression model results for each indicator, including estimated coefficients, robust (heteroskedasticity-consistent and autocorrelation-consistent) standard errors, P-values and coefficients of determination (adjusted-R²).

We derived simple indicator-specific scenario pathways to eliminate social shortfall by 2030 and ecological overshoot by 2050 by calculating the annual linear rate of change between 2022 levels and zero for each indicator over its respective period (that is, 8 years for social indicators, 28 years for ecological indicators). We derived these simple scenario pathways to offer a rough sense of the level of ambition required to live within the Doughnut by mid-century, in comparison with the historical trends estimated empirically. Although social shortfall trends and ecological overshoot trends are both expressed in percentage points per year, we note that the values are not directly comparable across the social and ecological domains because they are scaled differently: the former are expressed in a 0–100 scale, whereas the latter are scaled with respect to pre-industrial Holocene baselines that are defined differently for different boundaries.

The illustrative aspiration to eliminate social shortfall in the Doughnut by 2030 is broadly aligned with the social ambition of the SDGs, which all UN member states have committed to achieving by 2030. Unlike the social indicators, we opted against a 2030 aspiration to eliminate ecological overshoot in the Doughnut because there is minimal political support to achieve the specific ecological targets defined by the planetary boundaries framework by this date. Instead, we derived an illustrative 2050 aspiration that is consistent with the political ambition to achieve ‘net-zero’ greenhouse gas emissions by this date⁵⁹ and is broadly aligned with a 50% probability of keeping global heating below 1.5 °C (ref. ⁶⁰)—a level that reduces the risk of triggering tipping points in the Earth-system, such as major ice-sheet collapse and near-complete coral mortality (although the possibility of crossing such thresholds cannot be ruled out above 1 °C heating, which has already occurred)⁶¹.

Disaggregating the global Doughnut

The global Doughnut was disaggregated into three country clusters based on average levels of annual gross national income (GNI) per capita over the 2000–2022 period, using data collected from the Human Development Report⁶² from the United Nations Development Programme (UNDP), available for 193 countries. GNI per capita is expressed in international dollars (Int-$) at 2017 prices, which means that national values are adjusted for inflation and for differences in living costs between countries. We defined country clusters by income percentile thresholds as follows: poorest 40% of countries (less than Int-$ 8,100 per capita; N = 78), middle 40% of countries (between Int-$ 8,100 and 33,200 per capita; N = 77) and richest 20% of countries (more than Int-$ 33,200 per capita; N = 38). See Extended Data Fig. 7 for country-specific details.

For the social foundation, we calculated the proportion of the population falling below minimum social standards within each country cluster using the same indicators, methods and data sources as the global Doughnut. As such, levels of social shortfall are directly comparable across the global and country-cluster scales. We use a consistent set of 193 countries to define each country cluster but not all countries have data available for all indicators (as noted in more detail in the ‘Global time-series data’ subsection). We analysed social shortfall by country cluster in the year 2017 to enable comparison with the national environmental footprint data collected from ref. ¹⁵, which is only available for this single year (except for the public transport indicator, for which we include 2020 values owing to a lack of earlier data).

For the ecological ceiling, the planetary boundaries are related to critical Earth-system processes, which cannot be disaggregated to smaller scales in a directly comparable manner. However, because the original planetary boundaries framework was proposed in 2009, scholars have been developing methods to translate these Earth-system indicators into finite global resource budgets informed by planetary boundaries science, which can be allocated to individual countries according to a sharing principle. After more than a decade of applied research, a general translation procedure has been established with well-known limitations, acknowledgement of uncertainties and discussion of ethical implications of distinct sharing principles, among other themes³⁸.

To translate the global Doughnut’s ecological ceiling to the country-cluster scale, we collected equality-based per capita boundaries for the year 2017 from a recently published study¹⁵, which are related to four of the Doughnut’s ecological dimensions (climate change, nutrient pollution, freshwater disruption and biodiversity breakdown). To compare country-cluster performance with respect to these per capita boundaries, we collected national data for six consumption-based environmental footprint indicators available for 168 countries in 2017 from the same source¹⁵ (carbon dioxide, phosphorus, nitrogen, blue water, species loss and HANPP) and calculated a population-weighted average of each national per capita environmental footprint within each country cluster. The environmental footprints are calculated using input–output analysis, which accounts for spillovers and outsourcing of upstream environmental burdens across countries enabled by international trade by allocating them to final consumers, no matter where in the world such burdens occur. Such spillovers and outsourcing are critical to account for inequalities in consumption across societies but they are not relevant in the global Doughnut, as production and consumption are equal at the global scale.

Ecological overshoot was calculated for each per capita indicator–boundary pair using the same method described above for the global Doughnut (see ‘Calculating social shortfall and ecological overshoot’ subsection). The planetary boundaries for nutrient pollution and biodiversity breakdown are both represented by two separate indicators (phosphorus and nitrogen, and species loss and HANPP, respectively). We note that this approach to measuring ecological overshoot on the basis of country-cluster per capita consumption with respect to downscaled planetary boundaries should be seen as complementary to assessments of locally relevant ecological pressures and thresholds. Although increasing consumption and affluence, including a growing middle class, are widely held to be the primary drivers of global impacts on Earth-system stability³³, local ecological concerns may be more strongly affected by other factors, such as overexploitation, urban and agricultural encroachment, pollution and population growth⁶³. Further details on the individual per capita indicator–boundary pairs are provided in Supplementary Discussion 2 and Supplementary Table 2.

Calculating proportions of global totals

In our Analysis, we present the share of global population in social shortfall held by each country cluster in 2017 alongside the contribution of each country cluster to global ecological overshoot in the same year (see Extended Data Fig. 10). In both cases, the results are expressed as proportions of total shortfall and total overshoot in 2017 in percentage terms.

For each social indicator, we calculated the number of people in deprivation in each country cluster by multiplying its proportion of population in deprivation by its total population. These country-cluster populations in deprivation were summed to give total global population in shortfall and the share of global social shortfall held by each country cluster was defined as its respective proportion of this total population, or in mathematical terms for each country cluster n: share of global shortfall_n = (share deprived_n × population_n)/population deprived_total.

For each ecological indicator, we calculated the absolute level of environmental footprint beyond equality-based population shares in each country cluster by subtracting the global per capita boundary from its per capita environmental footprint and then multiplying this excess environmental footprint per capita by the country cluster’s total population. These excess environmental footprints per country cluster were summed to give total environmental footprint overshoot and the contribution of each country cluster n was defined as its respective proportion of this total overshoot, or in mathematical terms: share of global excess EF_n = ((EF per cap_n − boundary per cap_n) × population_n)/excess EF_total, in which EF is the environmental footprint indicator.

We note that this single-year approach does not account for the historical contributions of wealthy countries to global ecological overshoot, which is not ideal but compiling comparable national time-series data on environmental footprints and per capita boundaries from disparate sources is a non-trivial exercise that was beyond the scope of the present study. Building on previous research showing that wealthy countries contribute approximately 90% of excess cumulative CO₂ emissions beyond population-based fair shares of safe global carbon budgets (compared with 50% using our single-year approach)^64,65, a cumulative assessment of trends in ecological overshoot over time for several footprint indicators across country clusters is an important avenue for future research, in our view.

Limitations

Our Analysis is necessarily limited by the quality and availability of global time-series data (detailed descriptions of each social and ecological indicator are provided in Supplementary Information). Furthermore, although the Doughnut monitoring framework tracks each social and ecological indicator separately, we acknowledge the interdependence of many indicators. Although we do not analyse such complex interdependencies formally here, the representation of indicators in the Doughnut diagram conveys a visual sense of holistic interconnection that could frame and support future research in this area, which is beginning to emerge^26,48,49. That being said, we note that the radial representation of indicators, such as the Doughnut plots in Figs. 1 and 3 (and the planetary boundaries framework), has been criticized, notably because quantitative values scaled in terms of wedge radius leads to a quadratic increase in wedge area⁶⁶. To mitigate the risk that some readers may perceive small changes as more significant than they are, we also represent comparable results using bar charts (Figs. 2 and 4).

We acknowledge that our focus on the global and country-cluster scales masks wide inequalities in levels of social shortfall and ecological overshoot between countries and within them. We recognize the importance of accounting for such inequalities, particularly to envision equitable trajectories towards the Doughnut, that we see as complementary to our approach monitoring high-level social and ecological trends on a changing planet, which also need to be taken into account for considering Earth-system trajectories. We note that the environmental footprint data from ref. ¹⁵ is also available by expenditure decile within each country, but we analysed national averages to enable comparison with the social indicators, which are not generally available with such within-country disaggregation. A useful step for future research could therefore be to account for within-country social deprivations by income groups and/or other characteristics in a cross-country comparable framework (see ref. ⁶⁷ for a single-country application in Norway).

Finally, the Doughnut of social and planetary boundaries will continue to evolve. Its social foundation—including the dimensions, indicators, boundaries and data—will continue to be revised as internationally agreed social norms and standards continue to evolve and as improved international data become available. We acknowledge that some of the indicator thresholds that we have selected in this iteration lack officially recognized thresholds of minimum acceptable standards at present, such as a homicide rate of 5 or more per 100,000, whereas other indicators lack data altogether, such as racial equality and several disaggregated ecological boundaries. However, in these relatively early days of devising metrics fit for monitoring progress with respect to social and ecological goals, our view is that such data gaps are to be expected and one of the best ways to improve them is by making them visible. Future iterations of the social foundation could, for example, include dimensions concerning cultural rights and community resilience. Likewise, the ecological ceiling’s dimensions, indicators, boundaries and data will continue to be revised and refined as scientific research and understanding of Earth-system processes proceeds and gets translated for application at smaller scales. Future iterations could, for example, include more specific forms of chemical pollution, such as plastics, and improved metrics for air pollution and biosphere breakdown.