Prospects for global sustainable development through integrating the environmental impacts of economic activities

Abstract

Human economic activities drive the production and consumption of goods and services, contribute to the achievement of the United Nations Sustainable Development Goals (SDGs). However, the extent of economic growth’s influence on the SDGs remains unclear. To fill this knowledge gap, here, we quantified the environmental effects of economic activities and explored correlations between environmental effect and achieving SDGs. We developed six Environmental Footprint Indices, with a higher score indicating better efficiency or lower burden. Here we show that the various Environmental Footprint Indices had synergistic and trade-off effects on most SDG targets indices, but the synergistic effects prevailed. As income increased, the correlation between Environmental Footprint Indices and SDG target indices gradually strengthened. improved production efficiency and consumption changes notably advance SDGs, especially in low-income group countries. Our work provides scientific insights into the impact and prospects of environmental regulation required for achieving the SDGs by 2030.

Introduction

The 17 global Sustainable Development Goals (SDGs) proposed by the United Nations provide a strategic blueprint for developing healthy economies, societies and environment in a way that does not compromise the wellbeing of future generations^1,2,3,4. Human economic activities influence how these goals are achieved, for example by lifting incomes (SDG 1—no poverty), providing livelihoods (SDG 8—economic growth) and supporting human needs (SDG 9—infrastructure innovation)⁵. However, the pursuit of these positive goals also results in unintended adverse environmental impacts that in turn adversely affect sustainable development itself^1,2,6,7. This is because human consumption drives production that can cause resource depletion, ecosystem degradation, and climate change⁸. These impacts then provide feedback on human society, for example because deteriorating resource and environmental conditions may affect citizen health and social equity^9,10, and stringent environmental protection policies can hamper economic development¹¹. For example, economically critical industries in China, such as coal, steel and cement manufacturing, have had to increase their investment in technology upgrades to undergo a green transformation, which has reduced profitability¹².

Since the proposal of the SDGs, previous work has focused on evaluating the performance of achieving goals at different scales^13,14,15, the relationships between goals and their changes over time and space^1,2,6,16, as well as qualitatively analyzing the affecting factors of achieving the SDGs^9,17,18,19. Finding quantitative measures describing the progress of countries in achieving SDGs remains a research gap. Moreover, the synergies of environmental conditions and progress towards SDG targets have so far not been quantified. It is known that production efficiency and consumption patterns influence resource depletion and environmental burdens^20,21. Further, global trade has shifted economic output and associated environmental damage from importing towards exporting countries²². However, the details of how economic—production and consumption—activities determine and alter environmental conditions and in turn change countries’ SDG progress are not well understood. Assessing these connections is a prerequisite for supporting countries’ decision-makers in steering their societies towards the SDG targets.

Environmental footprints (EFPs) are an approach for quantifying the effect of humans’ economic activities on environmental and resource aspects (Fig. 1a)^23,24. Assessing EFPs is the basis for formulating environment-related sustainable-development policies and measures from a demand-side perspective^{25,26,27,28,29}. EFPs come in different forms: water footprints (WF)^25,30, energy footprints (EF)^31,32, carbon footprints (CF)^33,34, material footprints (MF)^27,28, nitrogen footprints (NF)^29,35, and land footprints (LF)^36,37. However, only the MF has been adopted as an official evaluation index in the SDGs^27,38. In this work, we establish links between environmental footprint practice^23,39 in terms of the above six indicators and SDGs.

Fig. 1: The environmental footprint of economic activities.

a The environmental footprint of economic activities, including six environmental footprints, WF water footprints, EF energy footprints, CF carbon footprints, MF material footprints, NF nitrogen footprints, and LF land footprints, displays the global average levels of unit USD and per capita environmental footprint (refer to the “Method” section). b Comparison of the environmental footprint by per USD and per capita among countries; the coordinate value is the percentage difference from the global average, where symbols represent countries with four income categories. The size of the circles represents the total population of the countries. The country names corresponding to the ISO-3 codes in the figure are provided in Table S1 in the supplementary information. c environmental footprints trends in four income categories, income H high-income, income UM upper-middle-income, income LM low-middle-income, income L low-income. d Fitting the relationship between the FI footprint intensity score, FP footprint pressure score, and the SDG index separately.

EFPs can be viewed against a country’s GDP and its population. This is because, generally, EFPs increase with increasing economic activity, and with increasing population⁴⁰. EFPs per unit of monetary total output characterizes the environmental intensity of a country’s domestic production, in the sense that the generation of economic value requires, ideally as little as possible, environmental damage. This Footprint Intensity (FI) index is essentially a measure of the efficiency with which a country’s domestic purchases and imports are being produced. As we show in our results, there exists large disparities between those countries with a low FI that can act as role models for the future technological development of those with a high FI. EFPs per capita characterize the environmental pressure that every citizen of a country exerts through the volume of their consumption. This Footprint Pressure (FP) index is a reflection of a society’s affluence. Again, there exists large disparities between countries, and those that have a low FP typically aspire and strive for the living standards of those that have a high FP.

Improving production efficiency and stemming overconsumption in affluent countries are the most direct ways to address environmental impacts²¹. The latter is now deemed to be necessary because technological measures alone are unlikely to achieve environmental goals⁴¹. In addition, focusing solely on efficiency may lead to rebound effects, that is, an overall increase in consumption levels⁴². Truly alleviating environmental pressure requires simultaneous changes in both consumption patterns and production efficiency, and hence we examine both indicators in this work⁴³.

This work proceeds in three stages. First, we use multi-region input-output (MRIO) analysis to compute both FI and FP indices from primary data at the individual country level and EFPs index was constructed based on these two indices. Second, we establish relationships between these indices and country’s performance on SDG targets in four income group defined by UN, which are low (income L), lower-middle (income LM), upper-middle (income UM) and high (income H) income group⁴⁴. Finally, we employ Computable General Equilibrium (CGE) analysis to project our MRIO data under different scenarios to 2030, establish future FI and FP indices, and using our regression model, project future country’s performance on SDG targets. A particular innovation of this research is that we capture interactions occurring in this three-step chain, facilitated by production and consumption simultaneously affecting Environmental Footprint Indices to various extents, and these in turn determining countries’ performance on multiple SDGs at once.

Results

The EFPs indices for economic activities

The characteristics of the environmental footprint differ among countries. Fig.1a, b shows the average level of global environmental FI and pressure and the percentage difference in the EFPs of countries from the global average. The FP of the United States, Australia, and Canada are generally higher than the global average, but their footprint intensities (FI) are lower. Factors such as population density, advanced production technologies, and resource abundance may contribute to this. As a country with a higher population density, China’s FP is close to the global average, but the per-capita NF is more than double the global average. Comparing FI and FP across income groups, we observe that countries with higher incomes tend to feature lower FI and higher FP (Fig. S1). During the past 16 years, the EFPs-to-GDP ratio has significantly decreased, while the EFPs per-capita has increased (Fig. S2), which means that the improvement of environmental efficiency may not bring relief to environmental pressure.

We use the mean of FI and FP for different environmental indicators as the environmental footprint index to assess the corresponding environmental performance of the various countries (Fig. 1c). Most EFPs Indices have experienced fluctuations over the period. The low-income group’s indices of WF, EF, and LF have all risen by nearly 10%, while its WF is still at a relatively low level. The CF index of the middle-income group has recorded the highest relative increase. For the high-income group, in addition to CF, the NF index has also increased significantly. Comparing the four income groups, the WF index shows a gradually improving trend with the increase of income level, while the materials footprint (MF) index shows no significant difference. The most significant gap lies in the WF index between the high-income and low-income countries, with the high-income group leading by ~65–105%. Over time however, this gap has slightly narrowed. The overall performance of high-income and upper-middle-income groups is higher than that of low-income and lower-middle-income groups. The specific scores of the EFPs index for countries worldwide can be found in Tables S2–S7.

The mean values of FI and FP for the six EFPs were calculated as the overall performance of the environmental footprint FI and FP, respectively. FI and FP were fitted to the SDGs index, respectively (Fig. 1d). In low-income group, the relationship between FI, FP, and SDG is not significant; With the increase in income level, starting from the lower-middle-income group, there is gradually a negative correlation, that is, with the increase of FI and FP indices, SDGs slightly decrease; After entering the upper-middle-income group, this relationship reversed and became positively correlated, showing a synergistic growth of FI, FP, and SDGs; In high-income countries, this positive correlation is more significant, with a slope exceeding that of the upper-middle-income group.

Correlations between the EFPs index and SDG targets

Figure 2 delineates correlation between the EFPs Indices and the attainment of SDG targets indices across four income groups, highlighting both positive and negative effects. The environmental footprint, as quantified by the EFPs Index, have a noteworthy correlation with most of SDG targets indices among the four income groups. Among high-income, lower-middle-income, and low-income groups, the EFPs Index tends to exert more pronounced positive correlations with these targets. However, in upper-middle-income countries, negative correlation slightly dominates. As income levels increase, the average level of correlation strength between environmental footprint and various SDG targets indices gradually increases (Fig. S3). At the same time, the strength of the positive correlation between the CF index, the NF index and SDG target indices gradually increases with the increase of income level, while the strength of the negative correlation between LF and WF index gradually increases.

Fig. 2: Positive and negative effects between targets and environmental footprints (EFPs) Index across four income categories.

Multiple regression approach was adopted to explore the positive or negative effects of the environmental footprint on targets (using the Target Index as the dependent variable and the Environmental Footprint Index as the explanatory variable, with all data ranging from 0 to 100). In all models, statistically significant factors (p < 0.05) were selected. The coefficient of the independent variable serves as a positive or negative effect. In the graphs (a–d) respectively depict the positive or negative effects between the environmental footprint (WF water footprint, EF energy footprint, CF carbon footprint, MF material footprint, NF nitrogen footprint, LF land footprint) and targets for four income categories. On the left side of the graphs, blue lines represent positive effects, while red lines represent negative effects, the sum of the coefficients of each environmental footprint index is displayed on the left, representing the positive or negative effects of each footprint index on all targets. On the right side, the corresponding targets are displayed, with the thickness of the lines indicating the coefficients of the as an explanatory variable for each target. The length of the environmental footprint represents the sum of their coefficients for the target, indicating the positive and negative strength. For the specific meaning of “target”, it is provided in the PDF which is available on the website as follow. (https://unstats.un.org/sdgs/indicators/Global%20Indicator%20Framework%20after%202022%20refinement_Eng.pdf) or the Table S8 in the Supplementary Information.

Specifically, the LF index demonstrates a relatively robust positive correlation in low, lower-middle and upper-middle-income groups, whereas negative impacts are more pronounced in high-income groups. It is worth noting that CF shows a prominent negative correlation in low-income and middle-income countries, while this phenomenon is completely reversed in the middle to high-income and high-income groups, showing a strong positive correlation. For instance, the CF index has a significant positive correlation with promoting an increase in renewable energy consumption (Target 7.2) in the high-income group and promoting equal employment opportunities (Target 8.5) in upper-middle-income group, while in lower-middle-income and low-income groups, the CF index is negatively correlated with ensuring universal access to affordable, reliable and modern energy services (Target 7.1) and ensuring sustainable utilization of ecosystems (Target 15.1), respectively. Individually, there is a strong synergistic relationship between the WF of low-income group and the goals related to equitable education (Target 4.1, 4.5).

Key sectors associated with the EFPs index

The connections between sectors in economic activities are complex and intricate, we use the environmental footprint index as the dependent variable, the environmental footprint FI and FP of each sector as the independent variables, and the random forest as the prediction model. We evaluate the importance of the FI and FP of different sectors to the corresponding environmental footprint index through the Accuracy Increase method. The heatmap (Fig. 3) in the figure illustrates the relative importance of the FI and FP of the sectors to the EFPs index. The vertical direction represents the different types of FI and FP. For example, EFI (EF intensity) and EFP (EF pressure) represent the EF intensity and the EF per capita, respectively.

Fig. 3: Identification of importance of sectors impacting the environmental footprint index.

The horizontal variables represent sectors differentiated in the calculation of environmental footprint index, and the vertical variables represent the different types of footprint intensity (FI) and footprint pressure (FP) in each sector, for example, CFI represent carbon footprint intensity, CFP represent carbon footprint pressure. A greater importance (darker color blocks) index indicates greater significance of the FP/FI of the sector on the environmental footprint index. Methods for measuring feature importance in machine learning are applied, where the Accuracy Decrease method, within Random Forest, assesses feature importance by removing each feature individually and measuring the resultant decrease in model accuracy, indicating how critical the feature is to the model’s performance.

Overall, except for the lower-middle-income group, the agriculture and food manufacturing sectors have shown significant importance, especially in terms of Land footprint per capita (LFP) in agriculture, material footprint per capita  (MFP) in food manufacturing. This also indicates that agriculture and food production, the fundamental industries for human survival, are still the main sectors in terms of their environmental impact, often accompanied by high resource consumption and environmental loads. In addition, changes in the construction sector are more likely to affect the Environmental Footprint Index among the lower-middle-income group, which may indicate that lower-middle-income countries are traditionally in a stage of rapid urbanization and infrastructure construction. The large demand in these sectors leads to resource use, material consumption, and emissions, which determine the overall performance of the country’s environmental footprint.

Simulating future EFPs index and the synergistic effects of SDG targets

Based on a CGE model, the global input‒output matrix for 2030 was simulated under the assumption of business as usual, with population growth, economic development, and EFPs projections. Compared with that in 2015, the growth rate of the environmental footprint in 2030 was more than 54.3%, of which the CF was expected to increase by 85.4%, and the total MF was more than twice the original value (Figs. S4, S5). Moreover, three scenarios were set: production efficiency improvement, a decrease in trade amount between developed and developing countries, and an increase in the service sector’s share of the consumption structure. Each scenario has three gradients, ranging from low to high (10%, 20%, 30%). In all the scenarios, the EFPs indices show the most significant improvement under the SP scenario, and low-income group show a greater improvement rate than others (Fig. 4a–d).

Fig. 4: Synergistic effects of SDG targets under the 2030 scenario simulation.

a–d Shows the changes in the average environmental footprint index of four income categories in the different scenarios and (e–h) shows the changes of SDG targets in 2030 compared to those in the business-as-usual (BAU) scenario. For the specific meaning of “target”, refer to the PDF at (https://unstats.un.org/sdgs/indicators/Global%20Indicator%20Framework%20after%202022%20refinement_Eng.pdf).

From the simulation of the synergistic effect of environmental footprint on the progress of SDG targets (Fig. 4e–h), there is still a significant trade-off between improving environmental footprint and hunger-related goals, and this does not hold true in high-income group. The balance between improving in environmental protection and food security is still a major concern in some countries. Among low-income groups, the progress of most targets improves with the increase of the environmental footprint index and is the largest among the four income categories. Improvement in the production scenario is relatively the most significant, while the increase in high-income countries is relatively the lowest. This is also because the performance of high-income groups on SDG targets is already at a relatively high level.

Discussion

Since the 2030 Agenda for Sustainable Development was proposed by the United Nations, both UN officials and researchers have conducted numerous studies to evaluate the spatiotemporal changes in the progress of achieving SDGs of different scales, aiming to identify targeted policy actions and track hot spots in achieving SDGs. The synergies and trade-offs between many goals and targets have also been thoroughly analyzed and dissected^45,46. However, these studies focus on the network analysis within the SDGs system, and there is relatively little research on the role of the “off-site factors” of SDGs. Understanding the interactions between complex systems can provide more comprehensive insights, help optimize the overall performance of countries on SDGs, and address challenges in complex situations. Meanwhile, qualitative research accounts for the vast majority of previous research, and quantitative changes in the synergy and trade-offs between goals and targets, as well as between goals and other factors, need to be analyzed to provide more precise information for reference and bias when making trade-off decisions. This article uses multiple models to analyze the correlation between the environmental impact of economic activities and progress of SDG targets in different income groups, simulating the synergistic effect of each target index with environmental footprint under different measures by 2030, achieving quantitative analysis for sustainable development in future scenarios.

Most economic activities drive industrialization, infrastructure, and innovation (SDG 9) and economic productivity (SDG 8), as expected. Simultaneously, many economic activities fulfill key needs (SDGs 2, 3, 4, 6, 7, and 11)⁵. This paper quantifies the environmental effects of economic activities and explores correlations between environmentally sustainable transformations and achieving various SDG targets. Based on the application of the comprehensive model, we analyzed the synergistic effects of SDG target indices and EFPs. National and international economic policy usually overlooks environmental impacts. In areas where the environment is beginning to influence policy, as in the “General Agreement on Tariffs”, it remains a tangential concern⁴⁷. Research suggests that with increasing income, environmental degradation increases until a certain point, after which environmental quality improves (this relationship follows an inverted U-shape)^48,49.

In the high-income group, the CF index shows the highest positive effect, however, the CF index has a negative effect with achieving Target 12.4 (Reduce waste into water and soil). This is similar to Lusseau’s research, where there are barriers between achieving climate action and responsible consumption goals in high-income countries¹. This study provides some explanations on typical synergies and trade-offs to enhance understanding and communication. For example, as an alternative energy source, biofuels may reduce greenhouse gas emissions, but their production process may require a large amount of water, land resources and generate waste. In addition, due to the high complexity of their industries and manufacturing, some industries may benefit from reducing greenhouse gas emissions, but face difficulties in reducing waste emissions. For example, in the heavy industry and chemical industry, waste disposal may become more difficult when reducing greenhouse gases, leading to conflicts. In low-income countries, there is a significant negative correlation between the material footprint index and the performance on Target 2.1 (access to safe, nutritious, and sufficient food) and Target 2.3 (access to productive resources and enhancing productivity). Low-income countries have outdated agricultural production technologies and are highly dependent on a large amount of resource inputs, including land, water, and fertilizers⁵⁰. Reducing the use of these resources may directly affect agricultural production, leading to insufficient food supply. For the lower-middle-income group, the CF index is significantly negatively correlated with Target 7.1 (access to affordable, reliable, and modern energy services), indicating that there is a trade-off between widespread energy access and a country’s greenhouse gas emission reduction measures. Stricter emission reduction policies may lead to rising energy prices, resulting in inequities in energy services⁵¹.

The environmental burden caused by exports in developing countries is mostly attributed to the demandof developed countries (Figs. S6, S7). In instances where developed nations curtail their importation from developing countries, there is a marginal decline in the EFPs index. This phenomenon is principally attributed to the necessity for these developed countries to augment domestic production to satisfy internal demand, consequently resulting in an elevated per capita environmental footprint. Furthermore, reducing imports may result in economic development in terms of imports for developed countries, fairer allocation of societal resources, and impacts on the domestic resource environment, leading to a slight decrease in related targets. However, the increase in demand and rising environmental cost prompt developed countries to respond more quickly to technological advancements, thereby enhancing agricultural production efficiency (Target 2.3)^52,53.

With improvements in environmental efficiency, the low-income group shows a more pronounced response to health-related SDG 3 targets than other income groups. This is likely due to the reduction in environmental pollution (caused by CF and NF), which results in a decrease in pollution-related diseases⁵⁴. Additionally, goals related to water resources and energy in low-income countries exhibit a more significant response to improvements in the environmental footprint⁵⁵. In the trade scenario aiming to reduce developed countries’ imports from developing countries, the results indicate a general slight decrease in most target indices for upper-middle- and high-income countries and a slight increase in most target indices for lower-middle and low-income countries, consistent with previous research findings⁵⁶. Overall, improving the environmental footprint generally results in a synergistic effect with the SDGs. Some countries have also taken multiple measures to improve resources and environment, while promoting the progress of multiple SDGs.

Denmark has made progress on multiple SDGs through a series of comprehensive measures to decrease carbon emissions. By investing heavily in the development of renewable energy, encouraging energy-saving equipment, advocating green lifestyles, etc., the country has achieved SDG3, 4, 7, 12, 15, and other goals in synergy⁵⁷. As the largest middle- and high-income country, China has achieved multiple SDGs through improving land-use efficiency, such as implementing a large-scale project to convert farmland into forests and grasslands, reducing excessive land use and soil erosion, improving the quality of soil and water resources (SDG 6), thereby reducing air and water pollution, and improving the health level of residents (SDG 3). In addition, China promotes water-saving irrigation technology and organic agriculture, reduces the use of fertilizers and pesticides, improves land use efficiency, ensures food security (SDG 2), and reduces the pollution of pesticides and fertilizer residues to water sources and food, improving public health conditions (SDG 3)^58,59.

This research explored the potential effects of changes in EFPs on progress towards the SDG targets, however, both the data and methodology used were limited. First, the identified positive and negative relationships are based on data from the “Sustainable Development Report 2023”, which employs a much smaller set of indicators than the 231 United Nations indicators. The limited data availability is the primary reason behind this constraint⁶⁰. Warchold⁷ also proposed that the selection of data for evaluating SDGs has impact on the result of correlations of the goals. In the future, authoritative official releases of highly available data systems are needed to ensure the consistency of research methods. A robustness analysis of the EFPs index calculation results is discussed in the supplementary materials^61,62,63. Richer data and more rigorous analytical methods could complement our conclusions by further explaining causal relationships through which SDGs are impacted by economic activities.

In conclusion, this study determines the response of environmental SDGs in four income categories to changes in the EFPs of economic activity, aiming to achieve as many SDGs as possible by 2030. Improve land use efficiency can synergistically promote progress towards more goals in all income groups. In high-income and middle-income groups there is a more pronounced synergy between optimizing carbon emissions and achieving SDG among high-income and middle-income groups. Efforts related to water resource utilization in low-income countries can help progress towards more goals. However, the trade-off effect between different EFPs and progress of SDGs should not be ignored, such as the CF of low-income countries and the NF of middle- and low-income countries. Future research could explore the intricate mechanisms underlying the negative and positive aspects of specific actions or events about sustainability.

Methods

First, we collected foundational data on EFPs, including WF, EF, CF, material footprint, NF, and LF. We then applied the input-output method to calculate the global EFPs of various countries from both production and consumption perspectives. By considering the environmental footprint per USD and per capita, we constructed the EFPs Index to represent the overall performance of different countries in various environmental aspects. Second, we conducted a multivariate regression analysis to examine the impact of EFPs on SDG targets. We introduced income categories as interaction effects to clarify the differences among four groups of countries. Finally, we combined CGE modeling with regression modeling to predict the potential changes in the EFPs Index under different scenarios for SDG targets in the future (Fig. 5).

Fig. 5: Research method framework.

The research framework is divided into four parts, with the analysis of the environmental footprint accounting and the assessment of the SDGs target index as the data foundation of this study. The Computable General Equilibrium (CGE) model and regression analysis are used to predict future environmental footprints and the SDGs target index.

Environmental footprints calculation based on input-output analysis

The input-output method has been widely used in the study of EFPs, which can conduct consistent accounting and analysis of water, energy, greenhouse gases, land, and materials. Six EFPs were selected for research, involving water consumption, energy utilization, material consumption, greenhouse gas emissions, active nitrogen emissions, and land use. These six footprints are amongst the most concern resource and environmental issues globally^23,38,64, and are also directly related to the agenda of the SDGs. Many specific indicators are used as evaluation indicators for achieving the goals in the framework⁶⁵. We utilize the Eora Global Multi-Regional Input-Output (MRIO) database⁶⁶, including its intermediate demand matrix T, final demand y, and value-added v. The Eora MRIO database provides a fully balanced world MRIO table from major sources such as the United Nations System of National Accounts, United Nations Commodity Trade Organization, Eurostat, Institute of Developing Economies and many national inputs and outputs database. It is freely open for research use (www.worldmrio.com). Based on the environmental footprint theory, we have established direct consumption (emission) accounts of resources (environment) in various sectors by combining the existing global environmental account database with the input-output table and calculating the environmental footprint of each country. The data of Intergovernmental Panel on Climate Change, I-PRIMAP, International Energy Agency, Food and Agriculture Organization of the United Nations, and Emissions Database for Global Atmospheric Research were combined with the IO table in the Eora database, which generated the Satellite Q of water resources, energy, greenhouse gases, reactive nitrogen, materials, and land use direct consumption and emissions data for 24 sectors in 136 countries (agriculture sector and fishing sector were merged as well as wholesale trade sector and retail trade sector). The total environmental footprint, per capita environmental footprint, and unit GDP environmental footprint are obtained from the perspectives of producing and consuming countries. All GDP data are calculated and analyzed with comparable prices in 2015. The global population data and GDP information were adopted from the World Bank database.

Environmental footprint framework

Water footprint

WF is an “indicator of water uses behind all the goods and services consumed by one individual or the individuals of a country”²⁵. We considered the actual consumption of water resources in the process of human activities. Therefore, the green and the blue WF are considered in the scope of the WF in this paper. We did not consider the calculation of the greywater footprint, which will duplicate the NF included in the analysis. The agricultural blue and green water consumption is calculated by combining the agricultural commodities production ($C\left[p\right]$) of each country and the WF coefficient (${WF}[p]$) each country proposed by method of Hoekstra²⁵. The water consumption of industrial and service sectors mainly belongs to blue water consumption. For industrial water consumption, we use the value added ($V{A}_{{industry}/{service}}$) of each sector by the water use efficiency (${WUE}$), which is from FAO AQUASTAT. 5% of industrial water is used as the actual consumption of blue water, and the number of the service sector is 10%^25,67. The calculation process is shown in the formula (1), (2).

$$W{U}_{{Agri}}={\sum}_{p}(C\left[p\right]\times {WF}[p])$$

(1)

$$W{U}_{{industry}/{service}}=V{A}_{{industry}(service)} \times {WUE}\times 0.05(0.1)$$

(2)

Energy footprint

The EF is the embodied energy content of producing or consuming sectors, which mainly refers to the national energy supply in Eora based on the IEA database. IEA collects and publishes the most extensive selection of energy statistics with charts and tables on 16 energy topics for over 170 countries and regions, which contain coal, natural gas, oil, biofuels and waste, hydro, nuclear and wind, solar, etc. The unit of EF is measured in joules⁶⁸.

Carbon footprint

CF is usually expressed in mass units (kilograms) of CO₂ equivalent resulting from Global Warming Potential (GWP). The CF in this research mainly refers to the greenhouse gas emissions in Eora. Gas categories use GWPs from the IPCC’s (Intergovernmental Panel on Climate Change) Assessment Report 4⁶⁹. Therefore, this indicator can also be called “GHG footprint.” The primary data is included in the PRIMAPHIST database, which consists of three main types of greenhouse gases: CO₂, N₂O, and CH₄.

Nitrogen footprint

NF refers to the reactive nitrogen emitted during production and consumption. Four types of reactive nitrogen emissions, N₂O, NH₃, NO_x, and N_wp, were mainly considered in the research⁷⁰. The N₂O, NH₃, and NO_x data are used in the EDGAR database, and N_wp is the leaching of agricultural fertilizers and manure excretion to the water of livestock, which is used in the FAO database. The calculation process is shown in the formula (3).

$$N{E}_{{sectors}}={{NE}}_{{wp}}+{{NE}}_{{NH}3}+N{E}_{{NOx}}+N{E}_{N2O}$$

(3)

Material footprint

Material footprint in this research mainly focuses on uncovering the global patterns of material extraction and use embedded in goods or services producing and consuming²⁷; The material footprint adopts the domestic material consumption in the Technical Annex for the Global Material Flows Database, which mainly includes 13 sub-categories of material input in four categories: biomass, fossil fuels, metal ores and non-metallic minerals (Table 1).

Table 1 Material categories used in the calculation

Land footprint

LF considerations are based on the amount of land used to produce biological resources in the FAO database, including agricultural land, forest land, and inland water land^38,71.

We use the economic input-output analysis method to calculate the EFPs of all countries. MRIO in the Eora database was used in this research⁶⁶. It contains goods and services within and between countries in the middle of the demand of data exchange, final demand matrix y(N × K) and v(M × N) value matrix, where N is the number of the production sector in all countries, K is the number of final demand category, M is the number of value-added categories. The original MRIO database covers 188 countries and 26 economic sectors. The basic formula for calculating EFPs is ${{\bf{W}}}=\widehat{{{\bf{w}}}}{{\bf{L}}}\widehat{{{\bf{y}}}}$, here $\widehat{{{\bf{w}}}}$ is the direct consumption (emission) of unit output, L is Leontief inverse matrix, representing the total output of each industry required by unit demand of an industry, and y is the final demand matrix⁷⁰. The specific calculation formula is as follows:

$${{\bf{X}}}={({{\bf{I}}}-{{\bf{A}}})}^{-{{\bf{1}}}}{{\bf{y}}}$$

(4)

$${{\bf{X}}}=[\begin{array}{ccc}{{{\bf{X}}}}^{{{\bf{1}}}} & \begin{array}{cc}{{{\bf{X}}}}^{{{\bf{2}}}} & \cdots \end{array} & {{{\bf{X}}}}^{{{\bf{n}}}}\end{array}],{{\bf{A}}}=\left[\begin{array}{ccc}\begin{array}{cc}{{{\bf{A}}}}^{{{\bf{11}}}} & {{{\bf{A}}}}^{{{\bf{12}}}}\\ {{{\bf{A}}}}^{{{\bf{21}}}} & {{{\bf{A}}}}^{{{\bf{22}}}}\end{array} & \cdots & \begin{array}{c}{{{\bf{A}}}}^{{{\bf{1}}}{{\bf{n}}}}\\ {{{\bf{A}}}}^{{{\bf{2}}}{{\bf{n}}}}\end{array}\\ \vdots & \ddots & \vdots \\ \begin{array}{cc}{{{\bf{A}}}}^{{{\bf{n}}}{{\bf{1}}}} & {{{\bf{A}}}}^{{{\bf{n}}}{{\bf{2}}}}\end{array} & \cdots & {{{\bf{A}}}}^{{{\bf{nn}}}}\end{array}\right],{{\bf{y}}}=\left[\begin{array}{ccc}\begin{array}{cc}{{{\bf{y}}}}^{{{\bf{11}}}} & {{{\bf{y}}}}^{{{\bf{12}}}}\\ {{{\bf{y}}}}^{{{\bf{21}}}} & {{{\bf{y}}}}^{{{\bf{22}}}}\end{array} & \cdots & \begin{array}{c}{{{\bf{y}}}}^{{{\bf{1}}}{{\bf{n}}}}\\ {{{\bf{y}}}}^{{{\bf{2}}}{{\bf{n}}}}\end{array}\\ \vdots & \ddots & \vdots \\ \begin{array}{cc}{{{\bf{y}}}}^{{{\bf{n}}}{{\bf{1}}}} & {{{\bf{y}}}}^{{{\bf{n}}}{{\bf{2}}}}\end{array} & \cdots & {{{\bf{y}}}}^{{{\bf{nn}}}}\end{array}\right]$$

(5)

$$\left[\begin{array}{ccc}\begin{array}{cc}{{{\bf{w}}}}_{{{\bf{11}}}} & {{{\bf{w}}}}_{{{\bf{12}}}}\\ {{{\bf{w}}}}_{{{\bf{21}}}} & {{{\bf{w}}}}_{{{\bf{22}}}}\end{array} & \cdots & \begin{array}{c}{{{\bf{w}}}}_{{{\bf{1}}}{{\bf{n}}}}\\ {{{\bf{w}}}}_{{{\bf{2}}}{{\bf{n}}}}\end{array}\\ \vdots & \ddots & \vdots \\ \begin{array}{cc}{{{\bf{w}}}}_{{{\bf{n}}}{{\bf{1}}}} & {{{\bf{w}}}}_{{{\bf{n}}}{{\bf{2}}}}\end{array} & \cdots & {{{\bf{w}}}}_{{{\bf{nn}}}}\end{array}\right]=\left[\begin{array}{ccc}\begin{array}{cc}{\hat{{{\bf{w}}}}}_{{{\bf{11}}}} & {{\bf{0}}}\\ {{\bf{0}}} & {\hat{{{\bf{w}}}}}_{{{\bf{22}}}}\end{array} & \cdots & \begin{array}{c}{{\bf{0}}}\\ {{\bf{0}}}\end{array}\\ \vdots & \ddots & \vdots \\ \begin{array}{cc}{{\bf{0}}} & {{\bf{0}}}\end{array} & \cdots & {\hat{{{\bf{w}}}}}_{{{\bf{nn}}}}\end{array}\right] \\ \left[\begin{array}{ccc}\begin{array}{cc}{{{\bf{L}}}}_{{{\bf{11}}}} & {{{\bf{L}}}}_{{{\bf{12}}}}\\ {{{\bf{L}}}}_{{{\bf{21}}}} & {{{\bf{L}}}}_{{{\bf{22}}}}\end{array} & \cdots & \begin{array}{c}{{{\bf{L}}}}_{{{\bf{1}}}{{\bf{n}}}}\\ {{{\bf{L}}}}_{{{\bf{2}}}{{\bf{n}}}}\end{array}\\ \vdots & \ddots & \vdots \\ \begin{array}{cc}{{{\bf{L}}}}_{{{\bf{n}}}{{\bf{1}}}} & {{{\bf{L}}}}_{{{\bf{n}}}{{\bf{2}}}}\end{array} & \cdots & {{{\bf{L}}}}_{{{\bf{nn}}}}\end{array}\right]\left[\begin{array}{ccc}\begin{array}{cc}{\hat{{{\bf{y}}}}}_{{{\bf{11}}}} & {\hat{{{\bf{y}}}}}_{{{\bf{12}}}}\\ {\hat{{{\bf{y}}}}}_{{{\bf{21}}}} & {\hat{{{\bf{y}}}}}_{{{\bf{22}}}}\end{array} & \cdots & \begin{array}{c}{\hat{{{\bf{y}}}}}_{{{\bf{1}}}{{\bf{n}}}}\\ {\hat{{{\bf{y}}}}}_{{{\bf{2}}}{{\bf{n}}}}\end{array}\\ \vdots & \ddots & \vdots \\ \begin{array}{cc}{\hat{{{\bf{y}}}}}_{{{\bf{n}}}{{\bf{1}}}} & {\hat{{{\bf{y}}}}}_{{{\bf{n}}}{{\bf{2}}}}\end{array} & \cdots & {\hat{{{\bf{y}}}}}_{{{\bf{nn}}}}\end{array}\right]$$

(6)

$${{{\bf{f}}}}_{{{\bf{c}}}}={\sum}_{{{\bf{r}}}}{{{\bf{e}}}}^{{\prime} }{{{\bf{W}}}}_{{{\bf{rs}}}}{{\bf{e}}}={{{\bf{e}}}}^{{\prime} }{{{\bf{W}}}}_{{{\bf{ss}}}}{{\bf{e}}}+{\sum}_{{{\bf{r}}}\ne {{\bf{s}}}}{{{\bf{e}}}}^{{\prime} }{{{\bf{W}}}}_{{{\bf{rs}}}}{{\bf{e}}}$$

(7)

$${{{\bf{f}}}}_{{{\bf{p}}}}={\sum}_{{{\bf{r}}}}{{{\bf{e}}}}^{{\prime} }{{{\bf{W}}}}_{{{\bf{sr}}}}{{\bf{e}}}={{{\bf{e}}}}^{{\prime} }{{{\bf{W}}}}_{{{\bf{ss}}}}{{\bf{e}}}+{\sum}_{{{\bf{r}}}\ne {{\bf{s}}}}{{{\bf{e}}}}^{{\prime} }{{{\bf{W}}}}_{{{\bf{sr}}}}{{\bf{e}}}$$

(8)

Where ${{{\bf{f}}}}_{{{\bf{c}}}}$ and ${{{\bf{f}}}}_{{{\bf{p}}}}$ are the EFPs based on consumption and production perspectives, respectively, we divide the environmental footprint by the value added and the total population separately to obtain the EFPs per unit added value and per capita.

Data resources

The data sources for environmental footprint calculation are provided in Table 2:

Table 2 List and explanations of sources for calculating environmental footprints

Calculation of EFPs index and SDG scores

The indicators used to assess the SDGs on a global scale, as well as the specific SDG Index for individual countries, are derived from the “Sustainable Development Report 2023”⁶⁰. This report has been consistently published for 9 years, covering 2015–2023. For our study, we adopted the indicators, principles of attribution, and calculation methods outlined in the report to evaluate the progress of each country’s SDGs from 2000 to 2015. This approach allowed us to create a comparable dataset from 2000 to 2015, which was employed for our SDG analysis.

We calculated the SDG Index for 166 countries based on 125 indicators (detailed information about indicator selection and sources is available in the supplementary materials Table S8). We employed sensitivity analysis^15,72 to examine the sensitivity of country’s SDGs score to data from alternative years and countries. We used the sensitivity index described in ref. ¹⁵. and decreased and increased (separately) the value for each indicator by 10% for six countries at the national level, then recalculated their SDGs score and obtained the sensitivity index for each. These countries were Australia, Brazil, China, Germany, Nigeria and USA, and were selected based on their large population sizes on each continent. The results show that the sensitivity of to changes in indicators’ original data values was less than 0.2 (Fig. S8). For assessing the EFPs index, we employed a combined calculation method using two footprint indicators for each category. These indicators are the per capita footprint and the per unit output footprint. As an example, the WF score was obtained by arithmetic averaging the scores of a country’s WFP Index and WFI Index. At the same time, we analyzed the correlation between the results after changing the weights and the average value to demonstrate the impact of weights on the EFPs index results in supplementary information. Similarly, the correlation analysis of the results obtained using geometric and arithmetic averages is also presented in the supplementary information (Figs. S9, S10). The calculation of WFP and WFI followed the same methodology indicated in the “Sustainable Development Report 2023”.

Data standardization

After establishing the upper and lower bounds for the indicators⁶⁰, we transformed all indicators into a range of 0–100 using the following formula (9):

$$x^{\prime}=\frac{x-\min (x)}{\max (x)-\min (x)}\times 100$$

(9)

Where $x$ is the raw data value; max/min denote the upper and lower bounds, respectively; and $x^{\prime}$ is the normalized value after rescaling.

Weighting and aggregation

Several rounds of expert consultations on earlier drafts of the SDG Index made it clear that there is little consensus across different epistemic communities on assigning higher weights to some SDGs over others. As a normative assumption, we therefore opted to transfer a fixed, equal weight to every SDG to reflect policymakers’ commitment to treating all SDGs equally and as an integrated and indivisible set of goals. This implies that countries need to pay attention to all goals to improve their SDG Index score but focus mainly on those where they are furthest from achieving the SDGs and where incremental progress might be expected to be fastest. The EFPs composite index is obtained by calculating the average score of the six footprint indices considered in this study.

To compute SDG Index scores, we first estimate scores on each goal using the arithmetic mean of indicators for that goal. These goal scores are averaged across all 17 SDGs to obtain the final Index score. Differing from the approach in the “Sustainable Development Report 2023”, this paper computes scores for each target within the SDGs. Similarly, the score for a target is derived by calculating the arithmetic mean of the indicators falling under that target. The matching of indicators to targets follows the methodology outlined in the “Sustainable Development Report 2023”. In the “Sustainable Development Report 2023”, the information of “Related SDG Target” for each indicator was included in Table A.4 “Indicators included in the Sustainable Development Report 2023” for the first time (Table S8).

Interaction between EFPs index and targets

Six environmental footprint indices were used as explanatory variables and the target variable as the dependent variable to construct a regression model. We further introduced the income categories as a fixed effect. We examined the positive or negative effects of the environmental footprint indices on the target variable among four income groups by analyzing their relationship with the dependent variable. The coefficients of factors in the regression model are used as the correlation between environmental footprint and SDG target. The absolute value of the correlation coefficient quantifies the strength of the interaction. The p value of each factor with the target variable was calculated to determine significance (p < 0.05). The results of significance are presented in Supplementary Table S9, along with the R² values for each model. Models with R² > 0.5 are used to simulate SDG targets and environmental synergistic effects.

The mean decrease accuracy method for analyzing the importance of environmental footprint of sectors

Mean Decreasing Accuracy (MDA) is a method used to evaluate the importance of each feature in a random forest model by measuring the impact of randomly arranging feature values on model accuracy^73,74. In this research, we used the FI and FP (Footprint Production) of all sectors in the country as independent variables, and the country’s environmental footprint index as the dependent variable. We used the MDA method to analyze the importance of each variable in the dependent variable prediction model. Specifically, the random forest model is first trained using environmental footprint data from all sectors of the country. Then, using the model on unused samples (i.e., out-of-bag samples) for training, Evaluate its accuracy on out of bag (OOB) samples as the baseline accuracy. Next, for each independent variable, its value in the OOB sample is randomly substituted to break the relationship between the feature and the dependent variable, and the accuracy of the model is reevaluated using the substituted OOB sample. The importance of a feature is measured by the difference in accuracy before and after permutation, and the decrease in accuracy is the importance score of that feature. This process is repeated for each independent variable and averaged across all decision trees to obtain the final feature importance. This method directly evaluates the impact of each independent variable on the predictive performance of the model, effectively identifying the variable that contributes the most to the prediction of environmental footprint index, while considering the linear and nonlinear interactions between variables.

General equilibrium quantitative simulation methods and data processing

To analyze the impact of global cross-regional commodity trade on the implicit resource use of the world and various economies, quantify the changes in the CF under different policy scenarios and mitigation measures, and identify feasible paths for optimizing carbon emissions, this study constructs a global general equilibrium economic system comprising 36 countries or regions, three commodity sectors, and two production factors. The model system depicts the main conditions and relationships in the world economy. To match the Eora data used in previous studies, this paper establishes a benchmark dataset based on accurate data from 2015. Real economic data, such as output data and production factor input data from the World Bank’s World Development Indicators (WDI) database, trade data from the United Nation’s commodity trade database, and tariff data from the World Trade Organization’s statistical database, are used to calibrate and estimate the parameters required by the model. As a result, a numerical economic system that simulates actual economic operations is obtained. Using this model, the study sets fifteen specific simulation scenarios from the perspectives of production, consumption, and trade, quantifying the impact of different perspectives and varying degrees of policy measures on changes in industry inputs and demands.

Basic framework and structure of the general equilibrium quantitative (CGE) model

Due to the involvement of international trade between nations and the cross-regional allocation and utilization of goods with implicit resources, it is necessary to establish a global trade model system with a general equilibrium framework as its basis involving multiple countries. The objective is to analyze the impact of international cross-regional commodity trade on global resources and individual economies, quantify the changes in EFPs under different policy scenarios and response measures, and identify feasible pathways for optimizing resource and environmental sustainability.

The production structure of the model assumes constant elasticity of substitution (CES) between factors of production. A standard CES production function is set, and the factor endowments of each country are predetermined. By minimizing costs under the budget constraint, an optimal plan can determine the input quantities of individual factors and the output quantities of products.

The demand structure of the model assumes that each country’s consumers have a CES utility function. In this case, consumers can choose between domestically produced or imported tradable goods (agriculture, manufacturing) and domestically non-tradable goods (services). It is assumed that there is imperfect substitution between domestic and imported products, following the Armington assumption, meaning that products from different countries are heterogeneous.

The equilibrium in product and factor markets determines the prices of products and factors. Trade costs exist for imports and exports between countries, with tariff barriers being an essential source of trade costs. Specifically, this is reflected in import tariffs (referred to as tariffs) imposed by individual countries. Tariffs affect consumer consumption decisions and firm production decisions through price transmission mechanisms. The import tariff rate for country $i$ importing product $l$ from countries is denoted as ${t}_{{ij}}^{l}$. The price transmission mechanism is defined as follows:

$$p{c}_{{ij}}^{l}=\left(1+{t}_{{ij}}^{l}\right){p}_{j}^{l}$$

(10)

In the above equation, $p{c}_{{ij}}^{l}$ represents the consumer price of product $l$ produced by country $j$ and consumed by country $i$. ${p}_{j}^{l}$ represents the production price of product $l$ produced by country $j$.

The model equilibrium is determined by the simultaneous clearing of product markets, factor markets, global trade, and the condition of zero profits in a perfectly competitive market environment. Specifically, under product market clearing, all output is fully consumed, meaning that all produced goods are consumed. Under factor market clearing, the supply of factors equals the demand for factors, ensuring that the quantity of factors supplied matches the quantity demanded. Under trade clearing, global trade is balanced, with the sum of trade imbalances across countries equaling zero. Under the zero-profit condition, the value of production equals the cost of the input factors, ensuring no economic profits in the model. Table 3 provides specific mathematical expressions for market clearing.

Table 3 The mathematical for the framework and structure of the model

Once the theoretical model is established, real economic data must be collected to calibrate and estimate the parameters and construct a numerical model system. To match the Eora database used in the previous sections, this study establishes a benchmark dataset based on real data from 2015 to calibrate and construct the numerical model system, considering the complexity and feasibility of model system construction and the research requirements. The numerical simulation includes 36 countries or regions, namely, Argentina, Australia, Austria, Belgium, Brazil, Canada, China, Denmark, France, Germany, India, Indonesia, Ireland, Israel, Italy, Japan, South Korea, Malaysia, Mexico, the Netherlands, Nigeria, Norway, the Philippines, Poland, Russia, Saudi Arabia, Singapore, South Africa, Spain, Sweden, Switzerland, Thailand, Turkey, the United Kingdom, the United States, and the rest of the world (ROW).

Regarding data processing, the data for ROW are obtained by subtracting the data of all the other 35 countries and regions in the model from the world total. The base-year data for 2030 are obtained by assuming that future economic growth rates will remain at the average annual growth rates of the past 20 years, based on the base-year data of 2015, i.e., the economic scale growth is maintained at the average yearly GDP growth rate from 2001 to 2021, simulated using the quantitative model. For factor inputs in production, labor input is represented by total labor income rather than the number of laborers, and capital input is represented by the actual amount of capital input. The units of the data are all billions of U.S. dollars.

Based on the need for parameter calibration and numerical model equilibrium, adjustments are made to address anomalies or inconsistencies in the data. The production-level data are sourced from the World Bank’s WDI database. The sectoral output data for agriculture, manufacturing, and services are calculated based on the respective sector’s share of GDP and annual GDP data. For factor input data in production, the individual sector’s share of GDP is multiplied by the total capital formation to calculate the capital factor input, and wage income data are used to represent the labor factor input. The factor endowment is obtained by summing capital and labor.

For the demand, data are indirectly calculated based on the product market clearing conditions of production and consumption, as well as the trade volumes between countries. The trade data are sourced from the United Nations Comtrade trade database, and the import and export data between countries and the ROW are obtained by subtracting the imports and exports of all countries in the model.

Trade costs, specifically import tariffs, are sourced from the World Trade Organization’s Statistics Database (WTO Statistics Database) MFN (Most-Favored-Nation Treatment) tariff levels. The import tariff levels for the ROW are determined using the world average import tariff rate.

The elasticity of factor substitution in production and product demand in consumption cannot be calculated from parameter calibration. There are generally two methods for determining these values: one is to estimate them using historical statistical data, but estimating these elasticities for all 36 countries individually would be labor intensive, and the results of historical data estimation represent the past, while the elasticities for the current year are still uncertain; the other method is to find existing estimates from other relevant literature, which is relatively more straightforward but introduces a degree of subjectivity. To facilitate processing and ensure consistency, we adopted the common practice used in other studies, such as ref. ⁷⁵ and ref. ⁷⁶, in which these elasticity values were set to 2. There are two primary methods for determining parameters in CGE models to address the issue of subjectivity sensitivity.

One method involves exogenously given parameters, such as estimating substitution elasticities in production and consumption functions using econometric/statistical methods, estimating trade costs, or obtaining data from other sources, including random values. The other method involves endogenously calibrating parameters, the most common approach for determining parameter values. Calibration refers to a method within the general equilibrium framework where a benchmark equilibrium dataset is used to satisfy the equilibrium conditions of the model and determine the parameter values. The basic principle of calibration is to treat variables in the model system as parameters and consider parameters as variables. Using the benchmark equilibrium data in the model system, the optimal equilibrium can be solved backward to determine the parameter values.

The parameters in the model mainly focus on the production function and utility function, including the scale parameter and factor input share parameters in the production function, as well as the product consumption share parameters in the utility function. By incorporating the benchmark data framework into the mathematical general equilibrium system and solving for equilibrium in reverse, the values of the parameters can be calibrated. Once the parameter values are determined, a simulation numerical model system that replicates the operation of the real-world economy is obtained.

Once the simulated numerical model system of the CGE is established, it is necessary to validate its effectiveness and test whether the system can fit real-world economic conditions. We employ three feasible methods for validation. First, we simulate real-world economic operation using a numerical model system and compare the simulated values with actual financial data to assess the fitting effect and degree of the model. Second, in a general equilibrium model system, changes in nominal prices do not affect the equilibrium results, only relative prices impact the equilibrium. Therefore, we can test the reliability of the numerical model system by varying nominal prices and examining whether equilibrium changes occur. Finally, we examine individual variables by increasing the import tariff rates of certain countries, observe the changes in import values and import prices in the simulation results, analyze whether they align with logical expectations, and thereby test the model’s effectiveness. The specific results of the validation are not listed individually.

The results of the three validation methods indicate that the large-scale numerical model system we constructed for the 36 countries and regions can effectively simulate real-world economic operations and closely replicate actual financial data. Based on this foundation, we can further utilize this large-scale numerical model system to simulate the potential impacts of different policies and evaluate the effects of policy choices.

To analyze the effects of different policy measures from the perspectives of production, consumption, and trade, this study established nine specific simulation scenarios. These scenarios quantify the effects of different policy measures on changes in industry inputs and demands from different perspectives and to varying degrees.

Three scenario assumptions for production, trade, and consumption were provided in Table 4, each of which represents the radical degree of our pursuit of resource and environmental sustainability with conservative (10%), suboptimal (20%), and ambitious (30%) scenarios. In terms of scenario production (SP), there is no doubt that countries need to narrow the gap with countries with high resource and environmental efficiency through technological progress to reduce the demand for resources and the environment. Between 2000 and 2015, the global environmental footprint per unit of USD experienced growth rates ranging from 9.1% to 25%. For the scenario trade (ST), all developing countries reduce their exports to all developed countries. To analyze the changes in their EFPs and the impacts on the economic and social aspects of all countries, our analysis revealed that more than 70% of the environmental burden generated by developing countries due to trade is attributable to imports from developed countries (SI). For the scenario consumption (SC), gradual progress toward a more modern consumption structure is the mainstream trend in the future, and structural changes will also change the sustainability of resources and the environment⁷⁷, 10%, 20%, and 30% were calculated to represent the three scenarios and assess the reduction in total global resources and the environment based on the BAU in the future and the sustainability performance of the national environmental footprint targets.

Table 4 Scenario settings

To correspond to the Eora database used in the previous section (2015), it is necessary to process the general equilibrium numerical simulation results to obtain input-output data for different policy simulation scenarios. The most crucial aspect is constructing 15 input-output tables with the same structure as that of Eora (2015) using the consumption and output data from the 15 simulation scenarios. The specific construction method is as follows:

Step 1: Establish the correspondence between the original countries in Eora (35 + 155) and the model countries (35 + 1). The Eora database includes 190 countries or regions. Nevertheless, due to the difficulty in collecting complete data and the relatively small economic size of some countries, 35 major world economies and resource-intensive countries were selected for the quantitative simulation. In contrast, the other countries or regions are grouped as ROW in the model. According to Eora (2015), in addition to the first 35 countries in the model, the other 155 countries correspond to the ROW in the model.

Step 2: Establish the correspondence between the original sectors in Eora (26) and the model sectors (3). Match each industry in Eora (2015) with the corresponding sectors in the model: agriculture, manufacturing, and services.

Step 3: Establish the correspondence between the data structure of Eora (2015) and the data structure of the model under the 2015 base period scenario. The example structure of the three countries and two sectors in Eora (2015) is shown in Table 5, which represents the input‒output structure of 190 countries and 26 sectors.

Table 5 Example of Eora’s (2015) data structure

Step 4: Calculate the growth rates of the output data under the different policy simulation scenarios compared to the 2015 base period scenario.

Step 5: Multiply the input-output data from Eora (2015) by the corresponding growth rates determined in Step 4, using the correspondence established in Step 3, to obtain the input-output tables under different policy simulation scenarios.

Finally, use the calculated input-output data for further calculation and analysis of the implied EFPs.