Global Dataset of the Energy Transition Vulnerability Index 2000-2023 Authors: Yifan Shen, Yuxuan Wang, Jia He, Yichen Zhang, Xunpeng Shi, Xuanye Cai, Jian Yu, Yingzhu Li Published: 2026 ABSTRACT AbstractCountries differ in their energy infrastructures, economic structures, and socioeconomic contexts, resulting in uneven vulnerability to the energy transition. Advancing a just global transition requires that the most vulnerable are not left behind and that all countries make measurable progress. To support this effort, we introduce the Energy Transition Vulnerability Index (ETVI), a global metric assessing national-level vulnerability. The framework draws on the Vulnerability Scoping Diagram and evaluates vulnerability across three dimensions—exposure, sensitivity, and adaptive capacity—in accordance with the Intergovernmental Panel on Climate Change (IPCC) definition. The dataset provides ETVI scores and sub-indices for 142 countries and regions from 2000 to 2023. Developed using a consistent methodology, the spatiotemporal dataset supports both cross-country comparison and within-country tracking over time, and serves as a foundation for further research. Insights derived from such research can inform the design of inclusive energy transition strategies and support international policy coordination.Subject terms: Sustainability, Climate-change mitigation Background & Summary Fossil fuel combustion remains the dominant source of global greenhouse gas emissions, making the transition to low- or zero-carbon energy systems central to limiting global warming and meeting the targets of the Paris Agreement1. However, countries vary widely in their energy infrastructures and socioeconomic contexts, leading to uneven exposure to transition-related challenges such as the scale of investment required for renewable deployment, risks of job displacement, and mounting cost-of-living pressures2. Advancing a just, orderly and inclusive global energy transition requires a clear understanding of each nation’s vulnerability and the structural factors that shape it3. Although energy transition vulnerability has been examined at scales ranging from national and municipal levels to communities and households, most studies have relied on qualitative methods, such as interviews, surveys, and participatory assessments4–6. These approaches yield valuable, locally grounded insights but are limited in their capacity to support cross-country comparisons or monitor progress over time. To address this gap, Carley et al.7 adapted the vulnerability scoping diagram (VSD)—originally developed for climate adaptation—to assess energy transition vulnerability across U.S. counties. The framework has since been applied to fossil fuel-dependent areas in the United States and extended to subnational contexts in Europe, supporting more equity-oriented policy design8,9. Energy transition—particularly the phase-out of fossil fuels—can be understood as a system-wide structural transformation that generates uneven socioeconomic impacts across regions, including changes in energy costs, labor markets, and local economic structures. These impacts map directly onto the core dimensions of vulnerability, namely exposure (to transition pressures), sensitivity (of affected systems), and adaptive capacity (to respond and adjust). Building on this foundation, our previous work10 developed a national-level Energy Transition Vulnerability Index (ETVI) covering 135 countries over 2010–2019, providing the first globally comparable assessment of energy transition vulnerability. In this study, we extend our previous work10 by providing an updated and substantially expanded dataset of energy transition vulnerability. The new release covers 142 countries and regions from 2000 to 2023, representing over 97% of global GDP, 93% of the world’s population, 98% of global energy consumption, and 96% of global CO₂ emissions as of 202311,12. The dataset is publicly available via Figshare13 at 10.6084/m9.figshare.29818313.v3. Compared to our earlier dataset, this version extends the temporal coverage, improves data consistency, and provides an updated panel dataset comprising the composite ETVI and its dimensional indices. A more detailed conceptual and methodological description of the framework is provided in our previous work10. The globally consistent ETVI dataset provides a robust basis for both cross-country comparison and temporal tracking of national vulnerabilities. Spatially, it enables the identification of countries and regions requiring policy attention, as well as those demonstrating resilience that may serve as benchmarks. Temporally, it facilitates analysis of evolving vulnerability patterns. In addition, the ETVI can be readily incorporated into econometric analyses as either a dependent or explanatory variable, enabling investigation of the drivers and consequences of energy transition vulnerability. As shown in our previous work10, the index can identify distinct transition profiles across countries, distinguishing between economies facing high exposure but strong adaptive capacity and those with compounded vulnerabilities. The ETVI can also be linked to broader development outcomes, such as the Sustainable Development Goals (SDGs), to examine how reductions in transition vulnerability align with sustainable development pathways. These applications support more targeted policy design and contribute to a more inclusive and well-managed energy transition. Methods The national-level ETVI framework (Fig. 1) adapts the county-level VSD proposed by Carley et al.7. Following the Intergovernmental Panel on Climate Change (IPCC)’s original definition of vulnerability, the ETVI evaluates three core dimensions—exposure, sensitivity, and adaptive capacity. The components within each dimension have been adjusted to reflect the study’s emphasis on fossil fuel phase-out, diverse socioeconomic contexts, and constraints in data availability. Each dimension includes at least two components, which are in turn represented by multiple indicators rigorously validated in the literature. Conceptual framework for the Energy Transition Vulnerability Index (ETVI). Building on the work of Carley et al.7 and the IPCC definition of vulnerability43, the ETVI conceptualizes vulnerability as a function: (i) the magnitude of transformation required to decarbonize a country’s energy system (exposure); (ii) the degree of impact such changes have on the country (sensitivity); and (iii) the ability to manage or adapt to resulting disruptions (adaptive capacity). Each dimension comprises distinct conceptual components and quantitative indicators, with fossil fuel phase-out as a central pillar of the low-carbon transition44. The national-level ETVI framework Our three-dimensional framework posits that a country’s vulnerability to energy transition depends not only on its exposure but also on its sensitivity to that exposure and its capacity to mitigate the associated risks. The framework follows a hierarchical structure in which dimensions are subdivided into components and indicators, resulting in 20 indicators in total (Table 1). Indicators within a given component are assigned equal weight. Table 1Further information on selected indicators.Sub IndexComponentsIndicators (unit)Upper boundLower boundData sourcesExposureFossil fuels in energy mixFossil fuel in energy supply mix (%)99.990.00IEA-EB42Fossil fuel in power generation mix (%)100.000.03IEA-EB42Fossil fuels in national incomeFossil fuel exports, share of GDP (%)49.460.00EI-SRWE11Fossil fuel rents, share of GDP (%)42.080.00EI-SRWE11SensitivityEnergy dependenceEnergy consumption (toe per capita)0.940.07IEA-EB42Energy intensity (toe per thousand 2015 USD)0.650.05IEA-EB42WealthPoverty rate ($3.65 a day, % of population)86.390.00WB-WDI12InequalityGini index55.8024.80WB-WDI12Bottom 20% income share (%)9.802.90WB-WDI12Susceptible demographicsUnemployment rate (%)22.080.75WB-WDI12Age dependency ratio (%)98.2232.32WB-WDI12Adaptive capacityEconomic capabilityGDP per capita (PPP, constant 2015 international $)64,882.61476.15WB-WDI12Science & TechnologyR&D expenditure, share of GDP (%)3.250.04WB-WDI12Researchers in R&D (per million people)6375.5517.08WB-WDI12EducationMean years of schooling (years)13.401.95UNDP-HDRDC15Education expenditure, share of GDP (%)7.591.78WB-WDI12Primary completion rate (% of relevant age group)109.1444.69WB-WDI12School enrollment rate, tertiary (% gross)94.782.70WB-WDI12Fiscal & Governance capacityGovernment spending, share of GDP (%)25.746.27WB-WDI12Government revenue, share of GDP (%)44.799.76WB-WDI12Notes: IEA-EB = IEA, Energy Balances; EI-SRWE = Energy Institute, Statistical Review of World Energy; WB-WDI = World Bank, World Development Indicators; UNDP-HDRDC = United Nations Development Programme, Human Development Report Data Center. toe denotes tons of oil equivalent, a standard unit of energy used to compare different energy sources. The upper and lower bounds reported for each indicator correspond to the 97.5th and 2.5th percentiles of the global distribution, which are used in the normalization procedure. Further information on selected indicators. Notes: IEA-EB = IEA, Energy Balances; EI-SRWE = Energy Institute, Statistical Review of World Energy; WB-WDI = World Bank, World Development Indicators; UNDP-HDRDC = United Nations Development Programme, Human Development Report Data Center. toe denotes tons of oil equivalent, a standard unit of energy used to compare different energy sources. The upper and lower bounds reported for each indicator correspond to the 97.5th and 2.5th percentiles of the global distribution, which are used in the normalization procedure. All indicators are selected within a unified conceptual framework based on the vulnerability scoping diagram (VSD). Specifically, each variable is explicitly mapped to one of the three core dimensions—exposure, sensitivity, or adaptive capacity—ensuring internal consistency in index construction. While the indicators draw on multiple strands of literature, their inclusion is determined by conceptual relevance within the framework rather than by individual data sources. To ensure cross-country comparability, all indicators are sourced from internationally harmonized datasets, including major global databases such as the World Bank (WB-WDI), the International Energy Agency (IEA-EB), and other widely used international data providers. The exposure dimension quantifies the extent of transformation required for energy systems to achieve decarbonization. Countries facing larger transformations are more exposed and therefore more vulnerable. While decarbonizing energy systems involves numerous facets, transitioning away from fossil fuels remains the most widespread challenge globally, as emphasized at COP2814. Within the exposure dimension, two components—fossil fuels in the energy mix and national income—capture the magnitude of potential energy system transformation and the associated socioeconomic challenges15–17. The first component is measured by the proportion of fossil fuels in total energy supply and in power generation, while the second is quantified by the share of fossil fuel exports and rents in GDP. The sensitivity dimension captures the extent to which a country is affected by energy system transformation. Given equal exposure, more sensitive countries are likely to experience greater socioeconomic disruption and thus face heightened vulnerability. Drawing on prior research and policy discourse, we evaluate sensitivity through four components: energy dependence, wealth, inequality, and demographic vulnerability. Economies with higher per capita energy consumption16,17, widespread poverty18–21, entrenched inequality22,23, and a greater share of socially or economically marginalized populations20,24–26 tend to be more acutely impacted. The indicators selected for each component are widely applied in assessments of social vulnerability27,28. The adaptive capacity dimension reflects a country’s ability to mitigate or respond to the socioeconomic disruptions of the energy transition. Given equal exposure and sensitivity, countries with greater socioeconomic resources and institutional capabilities are better positioned to manage such disruptions and are therefore less vulnerable. Based on existing literature and policy discussions, we assess adaptive capacity through four elements: economic capability, science and technology, education, and fiscal and governance capacity. Countries with lower levels of economic development2,7,29,30, weaker technological capability31–33, lower educational attainment34–37, and limited fiscal or financial resources are generally less equipped to alleviate or adapt to transition-related challenges. Scope and data sources This study evaluates the Energy Transition Vulnerability Index (ETVI) for 142 countries and regions from 2000 to 2023, selected on the basis of data availability. Together, these economies accounted in 2023 for over 97% of global GDP, 93% of the world’s population, 98% of global energy consumption, and 96% of global CO2 emissions11,12. Specifically, energy data were sourced from the International Energy Agency’s World Energy Balance Table and the Energy Institute’s Statistical Review of World Energy, socioeconomic and environmental data from the World Bank’s World Development Indicators, and education data from United Nations Development Programme’s Human Development Report. Summary statistics and data sources for all indicators used to construct the ETVI are presented in Table 1. Calculation of national-level ETVI scores The ETVI is calculated in two steps: normalization of indicators, followed by aggregation into dimensional and composite scores. Step 1: Normalization of indicator values Indicator values vary substantially across dimensions, necessitating normalization to a 0–100 scale for comparability over time and between countries. This approach allows both relative positioning and progress tracking. For instance, if a country ranks below the global average on an indicator in both 2000 and 2023 but improves over time, its normalized score in 2023 will still be higher than in 2000. Normalization follows the conventional min-max normalization method: where x denotes the original value of the indicator, x* the normalized value on a 0–100 scale, and max and min the upper and lower bounds used for normalization, respectively. To reduce the influence of extreme values and skewed distributions, the 97.5th and 2.5th percentiles are used as upper and lower bounds for all indicators—a common approach in index construction38,39. Equation (1) is applied to indicators for which higher values require direct scaling (e.g., GDP per capita), and Eq. (2) to those for which higher values require inverse scaling (e.g., Bottom 20% income share). Values above the upper threshold are set to 100 and those below the lower threshold to 0, with intermediate values linearly scaled between 0 and 100. Following normalization, exposure and sensitivity indicators are oriented such that higher values correspond to higher vulnerability. In contrast, adaptive capacity indicators retain their original interpretation, with higher values indicating stronger capacity and thus lower vulnerability. To ensure consistency in the composite index, adaptive capacity is incorporated inversely in the aggregation step as (100 − A), such that higher values reflect lower capacity and thus greater vulnerability. Accordingly, exposure and sensitivity increase vulnerability, whereas adaptive capacity mitigates it. The ETVI is primarily intended for comparison across countries and over time. Although normalized to a 0–100 scale, its absolute magnitude should be interpreted with caution, as it reflects a country’s position within the global distribution of indicators rather than a fixed threshold of vulnerability. Changes over time therefore capture both country-specific dynamics and broader shifts in the global distribution. Step 2: Aggregation into ETVI scores To operationalize the VSD framework, the composite ETVI is computed as the geometric mean of the three dimensional indices, each calculated as the arithmetic mean of its components: where V is the vulnerability score, Ei the ith component of exposure, Sj the jth component of sensitivity, and Ak the kth component of adaptive capacity. Using the geometric mean ensures that zero exposure to a given transition dimension results in zero overall vulnerability. The index follows a hierarchical aggregation structure, in which indicators are grouped into components and dimensions, with equal weights assigned within each component7,38. This weighting scheme is consistent with standard practice in composite index construction, as exemplified by the Human Development Index (HDI)40 and the Sustainable Development Goals (SDG) Index41. It is appropriate in the absence of a clear theoretical or empirical basis for differential weighting and reduces subjectivity, as each indicator captures a distinct aspect of vulnerability within its conceptual category. The robustness of the results to alternative index construction approaches is assessed in the Technical Validation section. Data Records The dataset “Global ETVI Data” is publicly available via Figshare13 at 10.6084/m9.figshare.29818313.v3. It contains 13,632 records, including the Energy Transition Vulnerability Index and its three-dimensional Sub-indices: Exposure, Sensitivity, and Adaptive Capacity. Specifically, 3408 records are national-level Energy Transition Vulnerability Index (142 countries and regions, from 2000 to 2023) [File tab ‘Vulnerability Index’] 3408 records are national-level Energy Transition Vulnerability Index (142 countries and regions, from 2000 to 2023) [File tab ‘Vulnerability Index’] 3408 records are national-level Exposure Sub-index (142 countries and regions, from 2000 to 2023) [File tab ‘Exposure Sub-index’] 3408 records are national-level Exposure Sub-index (142 countries and regions, from 2000 to 2023) [File tab ‘Exposure Sub-index’] 3408 records are national-level Sensitivity Sub-index (142 countries and regions, from 2000 to 2023) [File tab ‘Sensitivity Sub-index’] 3408 records are national-level Sensitivity Sub-index (142 countries and regions, from 2000 to 2023) [File tab ‘Sensitivity Sub-index’] 3408 records are national-level Adaptive Capacity Sub-index (142 countries and regions, from 2000 to 2023) [File tab ‘Adaptive Capacity Sub-index’] 3408 records are national-level Adaptive Capacity Sub-index (142 countries and regions, from 2000 to 2023) [File tab ‘Adaptive Capacity Sub-index’] Figure 2 presents the spatial distribution of the ETVI and its dimensional indices in 2023. The results reveal substantial regional disparities in energy transition vulnerability. Developed countries are generally the least vulnerable, although some remain highly exposed to fossil fuel phase-out. By contrast, many low- and lower-middle-income countries exhibit higher vulnerability, with weaker performance across all three dimensions. This global heterogeneity highlights the need for targeted national policies, where appropriate, coordinated international action—particularly to strengthen adaptive capacity and ensure a just, inclusive transition. Energy Transition Vulnerability Index (ETVI) and dimensional sub-indices, 2023. Map (a) presents overall vulnerability, where lower values indicate reduced susceptibility to adverse effects of the energy transition. Maps (b–d) display the three constituent dimensions: exposure (b), sensitivity (c), and adaptive capacity (d). All maps use a quantile classification to ensure an equal number of countries in each category. Technical Validation To assess the robustness of the ETVI, we introduced uncertainty into key assumptions through a series of sensitivity tests across three alternative configurations: (1) varying the upper and lower normalization bounds; (2) sequentially omitting indicators; and (3) applying alternative aggregation methods. For normalization, we tested alternative percentile bounds—specifically, the 99th and 1st percentiles, as well as the 95th and 5th percentiles. In the indicator-omission tests, we sequentially removed each indicator to examine whether the results were unduly influenced by any single input. For aggregation, we replaced the arithmetic mean with the geometric mean in calculating the dimensional sub-indices. Overall, the results indicate that the ETVI is robust to reasonable modifications in index construction. Figure 3 presents the variations in national ETVI scores and rankings across sensitivity tests using 2023 data. Countries and regions are ordered by their median score or rank, with boxplots showing the distribution of outcomes under alternative configurations. Blue and green markers denote results derived from alternative upper and lower quantile bounds of 0.01–0.99 and 0.05–0.95, respectively. Rankings for economies in the top and bottom quintiles are generally more robust, exhibiting smaller interquartile ranges. While mid-ranked economies exhibit greater variability, their baseline rankings remain close to the median, with over half deviating by no more than one position. Removing an indicator that captures a country’s comparative strength or weakness in energy transition vulnerability can shift its score or ranking. However, the overall distribution of outcomes across sensitivity tests indicates that, for most economies, ETVI rankings are not unduly influenced by any single indicator or dimension. Sensitivity analysis for energy transition vulnerability index (ETVI). The robustness of the ETVI is assessed by applying alternative upper and lower normalization bounds, sequentially excluding individual indicators, and testing different aggregation methods. Countries are ranked based on their median score or rank, with the original value of ETVI is indicated by orange markers. Results based on alternative upper and lower bounds of 0.01–0.99 and 0.05–0.95 are shown in blue and green, respectively. In each boxplot, the central rectangle spans from the first quartile (Q1) to the third quartile (Q3), representing the interquartile range (IQR = Q3 – Q1). If the maximum observed ETVI score exceeds Q3 + 1.5 × IQR, the upper whisker is fixed at that threshold; otherwise, it extends to the observed maximum. Similarly, if the minimum score falls below Q1 – 1.5 × IQR, the lower whisker is set at that threshold; otherwise, it extends to the observed minimum. Figure 4 shows the standard deviations in the scores and rankings of the ETVI and its three sub-indices across the sensitivity tests. The ETVI exhibits minimal fluctuations, with all countries and regions having a standard deviation below 6 in their scores (Fig. 4a) and 93% showing a standard deviation below 12 in their rankings (Fig. 4b), indicating that relative global standings in energy transition vulnerability are largely robust to the alternative configurations tested. For the sub-indices, 77% of countries and regions recorded a score standard deviation below 6 for Exposure (Fig. 4c), 96% for Sensitivity (Fig. 4e), and 100% for Adaptive Capacity (Fig. 4g). In terms of rankings, 79% recorded a standard deviation under 12 for Exposure (Fig. 4d), 62% for Sensitivity (Fig. 4f), and 100% for Adaptive Capacity (Fig. 4h). Although sub-index rankings exhibited slightly more variability than the ETVI ranking, global relative positions within these sub-indices have not changed substantially. Standard deviations of index scores and rankings across sensitivity tests. Panels on the left (a,c,e,g) show the standard deviations of ETVI and sub-index scores across sensitivity tests, including applying alternative upper and lower normalization bounds, sequentially excluding individual indicators, and testing different aggregation methods. Panels on the right (b,d,f,h) present the standard deviations of ETVI and sub-index rankings across sensitivity tests. Limitations While this study represents one of the first attempts to quantify national-level vulnerability in the transition to a low-carbon future, the dataset has several limitations. First, although grounded in a consistent conceptual framework that captures key dimensions of energy transition vulnerability and enables meaningful cross-country comparisons, it may overlook country-specific factors that are critical in particular contexts. For countries identified as highly vulnerable and in need of targeted policy interventions, the framework could be further refined to reflect their unique characteristics and thereby produce more accurate assessments. Second, the sensitivity and adaptive capacity dimensions do not explicitly account for the potential socioeconomic co-benefits of the energy transition, such as those associated with renewable energy expansion and green industrial development. The ETVI is designed to measure vulnerability to transition risks, rather than the net balance of risks and opportunities. As such, countries may simultaneously exhibit high vulnerability to transition shocks while benefiting from emerging low-carbon opportunities. Third, the dataset focuses primarily on fossil fuel phase-out, without explicitly addressing other elements of the low-carbon transition—such as advances in renewable technologies and carbon capture and storage. Future research could extend the framework to encompass these aspects. Acknowledgements This study was supported by National Natural Science Foundation of China (72474156, 72574197, 72104184, and 72373172), Major Project of the National Social Science Foundation of China (25&ZD197), the STU Scientific Research Initiation Grant (STF25018T), and the Program for Innovation Research in Central University of Finance and Economics. Author contributions Yifan Shen: Conceptualization, Formal analysis, Writing. Xuanye Cai: Formal analysis, Writing – original draft. Jian Yu: Formal analysis, Writing – original draft. Yingzhu Li: Conceptualization, Formal analysis, Writing – review and editing. Yuxuan Wang, Jia He, and Yichen Zhang: Formal analysis, Data curation. Xunpeng Shi: Writing – review and editing. Data availability All indicators are derived from internationally recognized and harmonized databases. The energy-related data were obtained from the International Energy Agency (IEA-EB, https://www.iea.org/data-and-statistics/data-product/world-energy-balances) and the Energy Institute (EI-SRWE, https://www.energyinst.org/statistical-review). The socioeconomic data were sourced from the World Bank (WB-WDI, https://databank.worldbank.org/source/world-development-indicators) and the United Nations Development Programme (UNDP-HDRDC, https://hdr.undp.org/data-center). The dataset generated in this study, entitled “Global ETVI Data”, is publicly available in the Figshare13 at 10.6084/m9.figshare.29818313.v3. Code availability The code used to generate the main results has been deposited on GitHub and is available on the Zenodo repository at 10.5281/zenodo.19256028. Additional information necessary to reproduce the analyses reported in this paper is available from the authors upon request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Xuanye Cai, Email: xuanye_cai@stu.edu.cn. Jian Yu, Email: jianyu@cufe.edu.cn. Yingzhu Li, Email: li_yingzhu@zju.edu.cn. References Associated Data Data Availability Statement All indicators are derived from internationally recognized and harmonized databases. The energy-related data were obtained from the International Energy Agency (IEA-EB, https://www.iea.org/data-and-statistics/data-product/world-energy-balances) and the Energy Institute (EI-SRWE, https://www.energyinst.org/statistical-review). The socioeconomic data were sourced from the World Bank (WB-WDI, https://databank.worldbank.org/source/world-development-indicators) and the United Nations Development Programme (UNDP-HDRDC, https://hdr.undp.org/data-center). The dataset generated in this study, entitled “Global ETVI Data”, is publicly available in the Figshare13 at 10.6084/m9.figshare.29818313.v3. The code used to generate the main results has been deposited on GitHub and is available on the Zenodo repository at 10.5281/zenodo.19256028. Additional information necessary to reproduce the analyses reported in this paper is available from the authors upon request.