How to Distinguish Income Indicators of Energy and Transport Vulnerability—A Case Study of Greece
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe article begins by situating the study within the broader context of climate change, energy transitions, and their societal impacts. It draws on existing literature examining how climate change disproportionately affects vulnerable populations and emphasizes the importance of integrating social considerations into policy design and research. The authors argue that, despite notable progress in reducing greenhouse gas emissions, meaningful disparities persist among low-income groups due to inadequate financial mechanisms for decarbonization.
The paper also references key frameworks from previous studies, such as the Social Climate Fund (SCF) proposed by the European Commission. While the SCF aims to support vulnerable citizens through targeted energy policies, the absence of a robust method for identifying and addressing specific social groups raises questions about its equitable application.
Despite the potential significance of the findings, the review process was hindered by the lack of curated data. In the interest of transparency, the authors should upload their processed or curated datasets—specifically, the version used directly in their computations—to a public repository (e.g., Figshare). Although the Data Availability Statement indicates where the data can be obtained, the primary value of this work lies in data curation; thus, reviewers must access the curated data to properly assess its quality. The reliability of the study’s conclusions fundamentally depends on accurate data management.
In addition, comparisons across different surveys must be verified for logical consistency before any further analysis. For example, Table 4 appears to contain contradictory information regarding RI.1 (19.4% of households unable to keep cool or warm) and RI.2 (31.4% of households unable to pay utility bills). One might expect that the proportion of households unable to pay utility bills would be equal to or lower than the proportion of those unable to maintain adequate heating or cooling. This discrepancy requires further clarification or resolution.
Another technical issue is the presence of broken hyperlinks—likely artifacts of conversion from text documents to PDF (e.g., lines 234, 451, 495, 503, and so on)—which should be corrected before publication.
The authors are encouraged to analyze potential correlations between the various indicators, assuming the data allow. For instance, is there a high probability that a person who answers “yes” to TI.2 will also answer “yes” to TI.6? Likewise, are RI.1 and RI.2 strongly correlated, given that households unable to pay utility bills are also likely to struggle with heating or cooling?
Moreover, since most indicators are based on self-assessment, it remains unclear whether a “yes” response (e.g., TI.6: “No—cannot afford”) reflects a genuine financial constraint or a personal preference. For example, higher-income households might choose properties with easy access to public transportation, eliminating their need for a private car, or they may simply prefer a more expensive private car, complicating the interpretation of such responses.
The authors estimate energy poverty by averaging four indicators (page 12, line 434). Clarification is needed on why these indicators carry equal weight. The study should explore whether conclusions would change if different weights were assigned, and it should justify the final weighting scheme.
Another issue arises in Table 7, which lists categories as “<3500” and “>3501” without indicating where 3501 itself is categorized. This omission should be addressed.
Given the array of indicators and the general lack of consensus on their use, the authors might also consider how combining uncorrelated or weakly correlated indicators would affect their results. A more thorough analysis is needed to substantiate the study’s conclusions before it can warrant publication in Sustainability.
Author Response
Despite the potential significance of the findings, the review process was hindered by the lack of curated data. In the interest of transparency, the authors should upload their processed or curated datasets—specifically, the version used directly in their computations—to a public repository (e.g., Figshare). Although the Data Availability Statement indicates where the data can be obtained, the primary value of this work lies in data curation; thus, reviewers must access the curated data to properly assess its quality. The reliability of the study’s conclusions fundamentally depends on accurate data management.
Response: thank you for making this important point. We have now added the link to the curated and processed dataset used in our analysis in a respective footnote (Also the data availability statement) . The dataset includes the harmonized and calculated indicator values, household-level extrapolations, and income thresholds used in constructing vulnerability groups.
In addition, comparisons across different surveys must be verified for logical consistency before any further analysis. For example, Table 4 appears to contain contradictory information regarding RI.1 (19.4% of households unable to keep cool or warm) and RI.2 (31.4% of households unable to pay utility bills). One might expect that the proportion of households unable to pay utility bills would be equal to or lower than the proportion of those unable to maintain adequate heating or cooling. This discrepancy requires further clarification or resolution.
Response: thank you for bringing this to our attention. While it may appear counterintuitive, this discrepancy stems from the nature of the two indicators as RI.1 is a subjective perception of thermal discomfort, while RI.2 refers to objective arrears on any utility bills, including electricity, heating, and water. As such, many households may prioritize energy payments by reducing consumption or skipping other utility bills—thus reporting arrears without necessarily reporting discomfort. This clarification has now been added in Section 3.2.1 of the revised manuscript.
Another technical issue is the presence of broken hyperlinks—likely artifacts of conversion from text documents to PDF (e.g., lines 234, 451, 495, 503, and so on)—which should be corrected before publication.
Response: thanks for catching that. Most Hyperlinks were removed where unnecessary or fixed where needed.
Response: The manuscript now acknowledges this limitation in the Discussion (Section 4.1), noting that self-assessed affordability indicators may reflect both material deprivation and subjective preferences, particularly among middle- or higher-income groups. Where possible, we cross-validated these responses with income and expenditure data to ensure more robust interpretations.
The authors estimate energy poverty by averaging four indicators (page 12, line 434). Clarification is needed on why these indicators carry equal weight. The study should explore whether conclusions would change if different weights were assigned, and it should justify the final weighting scheme.
Response: We now clarify that the equal weighting follows the European Commission’s guidance under Article 8.3 of the Energy Efficiency Directive (2023/1791), which recommends an arithmetic average across RI.1, RI.2, RI.10, and RI.13.
Another issue arises in Table 7, which lists categories as “<3500” and “>3501” without indicating where 3501 itself is categorized. This omission should be addressed.
Response: This has been corrected. The final category in the previous Table 7 (which is now Table 2) now reads “≥3501€” to clearly reflect inclusion of the €3,501 threshold within the last category.
Given the array of indicators and the general lack of consensus on their use, the authors might also consider how combining uncorrelated or weakly correlated indicators would affect their results. A more thorough analysis is needed to substantiate the study’s conclusions before it can warrant publication in Sustainability.
Response: We have included a paragraph in the Discussion (Section 4.1) acknowledging that combining weakly correlated indicators may dilute signal strength or mask specific types of vulnerability. However, our new thematic grouping to analyse results and propose a discussion also help mitigate this issue by preserving internal consistency within each vulnerability dimension.
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper present a case study to find out the income indicator for infrastructure vulnerability. My major comments are as follows.
- The writing and the organization of the paper need to be improved. And conduct the proof reading to eliminate the grammar error.
- Before discussing the terminology, I think you need to define the term like "transport vulnerability", "energy poverty".
- There is a lot of reference error, which doesn't show up in the paper. Probably because of missing or incomplete references
- how data from different sources like EU SILC, EU LFS were combined.
- Table 5 shows some indicators (TI.2, TI.3, TI.4, TI.5) with very little or no data for certain years.
- How do you choose these indicators?
- I suggest to split the first section in to introduction section and related work section. Consider to include the recent work "Optimizing seismic retrofit of bridges: integrating efficient graph neural network surrogates and transportation equity" and "Graph neural network surrogate for seismic reliability analysis of highway bridge systems" in the related work.
- Why and how the indicator "inability to purchase a zero-emissions vehicle" is included, more explanation is needed. And why it is so important?
- Some columns in the tables don't have proper explanation headings.
- The writing and the organization of the paper need to be improved. And conduct the proof reading to eliminate the grammar error.
Author Response
- The writing and the organization of the paper need to be improved.
Response: We have substantially revised the paper to improve overall structure, logical flow, and clarity. Now, in the revised version we followed a clearer organization, including a more robust introduction, a distinct section on theoretical and policy frameworks, and separate chapters on methodology, results, and discussion. Transitions between sections have also been improved, and the argumentation has been clarified throughout.
- Conduct the proof reading to eliminate the grammar errors.
Response: We conducted a thorough proofreading of the entire manuscript to eliminate grammar errors, sentence-level inconsistencies, and awkward phrasing.
- Before discussing the terminology, I think you need to define the terms like "transport vulnerability", "energy poverty".
Response: thank you for this comment, and we fully agree and have addressed this in the revised manuscript. Now, the terms are defined early in Section 1.1 and 1.3, referencing the latest EU legislation (e.g., EED 2023/1791 and SCF Regulation) as well as peer-reviewed literature. Thus, the distinction between poverty and vulnerability is now made explicit to avoid ambiguity throughout the paper. - There is a lot of reference error, which doesn't show up in the paper. Probably because of missing or incomplete references
response: All references have been reviewed, corrected, and cross-checked to ensure they are properly cited and included in the bibliography. Also, Broken reference links have been removed, and inaccurate citation codes from the draft stage have been replaced with complete bibliographic entries. - how data from different sources like EU SILC, EU LFS were combined.
response: We added a new subsection in the Methodology chapter (Section 2.2.1) to clarify this process. Indicators were extracted independently from each source (EU-SILC, HBS, LFS, EQLS), and only harmonized when common definitions existed, or as indicated by EED. When combining data, we ensured that (i) the same years were compared, (ii) household-level data was extrapolated carefully. Limitations of this process are discussed transparently in the revised Discussion (Section 4.3). - Table 5 shows some indicators (TI.2, TI.3, TI.4, TI.5) with very little or no data for certain years; How do you choose these indicators?
Response: Thank you for pointing this out. We clarified in Section 2.2.1 and 3.2.2 that these indicators were selected based on their inclusion in EU-level guidance and studies (e.g., OEKO Institute for DG EMPL), despite data limitations. In our research, they were not used in composite estimates but discussed qualitatively as critical gaps in national transport poverty assessments.
- I suggest to split the first section into introduction section and related work section. Consider to include the recent work "Optimizing seismic retrofit of bridges: integrating efficient graph neural network surrogates and transportation equity" and "Graph neural network surrogate for seismic reliability analysis of highway bridge systems" in the related work.
Response: We thank the reviewer for this very useful suggestion. We have restructured the early part of the paper, now splitting Section 1 into a general Introduction (Section 1) and a contextual framing with a review of energy justice, policy, and transition literature (Section 1.1). However, as the paper focuses on household energy and transport vulnerability rather than infrastructure resilience or structural engineering, the suggested papers on seismic retrofitting and graph neural networks, while interesting, were deemed outside the scope of this study. We instead expanded the related work with sources specifically addressing transport equity, accessibility, and mobility poverty in the European context.
- Why and how the indicator "inability to purchase a zero-emissions vehicle" is included, more explanation is needed. And why it is so important?
Response: We have now provided a detailed justification in Section 2.2.4 and again in the results discussion (Section 3.2.2 and 4.1) as this indicator is aligned with the SCF Regulation, which specifically includes it as a criterion for defining vulnerable transport users. In the Greek context, this indicator shows a structural constraint: households reliant on older, fossil-fuel-based vehicles are disproportionately affected by ETS2 costs and might lack the means to transition to more sustainable (fossil-free) alternatives. - Some columns in the tables don't have proper explanation headings.
Response: We revised tables, and changed some to figures, for full clarity and readability.
Reviewer 3 Report
Comments and Suggestions for AuthorsThis manuscript addresses the challenge of defining and measuring vulnerability to energy and transport poverty, using Greece as a case study. It takes a structured approach to integrating EU-level indicators with national data, particularly within the framework of the Social Climate Fund. By analyzing income-based and expenditure-based measures, the paper highlights how different indicators capture distinct but complementary dimensions of energy and transport poverty.
While the manuscript offers a practical perspective for policymakers, its clarity and consistency should be improved. Refining the theoretical framework, clarifying indicator definitions, and ensuring uniform citation of references would enhance the manuscript for a future submission. Given these concerns, I do not recommend acceptance in its current form. Below, you will find my comments and recommendations.
The introduction provides a good overview of the policy context. However, it might help to more clearly state the research gap. Therefore for a future submission I recommend to strength the argument around why current approaches are insufficient and how this paper’s proposed indicators improve on them would enhance the clarity of the research gap.
The manuscript references recent EU directives and prior academic debates, but does not strongly anchor these indicators in a broader theoretical framework. Consider to include for example a short reflection on conceptual frameworks of energy justice, capabilities approach, and/or social vulnerability studies.
The discussion of the greek household budget survey is helpful, though a bit more detail on sampling would reassure readers about data robustness. In this same way, your composite approach could benefit from a short explanation of why these specific composite rules are methodologically justified.
This discussion section is comparatively succinct. You can consider further elaboration on policy implications: how might Greek authorities update eligibility criteria for subsidies or the social climate plan to address the findings? Or a brief reflection on how these indicators might be adapted to other member states with different datasets or contexts.
Comments on the Quality of English Language
Overall, the paper is readable, but some editorial issues (spelling, grammar, referencing) need to be polished.
Author Response
This manuscript addresses the challenge of defining and measuring vulnerability to energy and transport poverty, using Greece as a case study. It takes a structured approach to integrating EU-level indicators with national data, particularly within the framework of the Social Climate Fund. By analyzing income-based and expenditure-based measures, the paper highlights how different indicators capture distinct but complementary dimensions of energy and transport poverty.
While the manuscript offers a practical perspective for policymakers, its clarity and consistency should be improved. Refining the theoretical framework, clarifying indicator definitions, and ensuring uniform citation of references would enhance the manuscript for a future submission. Given these concerns, I do not recommend acceptance in its current form. Below, you will find my comments and recommendations.
The introduction provides a good overview of the policy context. However, it might help to more clearly state the research gap. Therefore for a future submission I recommend to strength the argument around why current approaches are insufficient and how this paper’s proposed indicators improve on them would enhance the clarity of the research gap.
Response: Agreed. In the revised manuscript, Section 1.4 has been expanded into a new dedicated subsection titled “Research Gap Detection”, which now explicitly articulates the limitations of existing EU and national-level approaches. It explores how current indicators rely heavily on income-expenditure thresholds and fail to capture structural, behavioral, and chronic vulnerabilities—especially those relevant for ETS2 impacts.
The manuscript references recent EU directives and prior academic debates, but does not strongly anchor these indicators in a broader theoretical framework. Consider to include for example a short reflection on conceptual frameworks of energy justice, capabilities approach, and/or social vulnerability studies.
Response: We have significantly improved the theoretical grounding in the revised version. In Section 1.1, we added a dedicated framing using energy justice theory, referencing the three tenets—distributional, procedural, and recognitional justice—as an analytical lens. Additionally, we reflect on sustainable transition theory to contextualize the need for paradigm shifts in measurement and policymaking. Although we did not include the full capabilities approach due to scope constraints, we reference social vulnerability literature and link our work to systemic and structural inequities (e.g., housing, mobility, and health).
The discussion of the greek household budget survey is helpful, though a bit more detail on sampling would reassure readers about data robustness. In this same way, your composite approach could benefit from a short explanation of why these specific composite rules are methodologically justified.
response: We added clarification in Section 2.3 on the robustness and relevance of the HBS dataset and now it describes the sampling structure of the HBS (nationally representative, annual updates, stratified by region and household size), and explain why it is suited among EU Member States for vulnerability assessments.
This discussion section is comparatively succinct. You can consider further elaboration on policy implications: how might Greek authorities update eligibility criteria for subsidies or the social climate plan to address the findings? Or a brief reflection on how these indicators might be adapted to other member states with different datasets or contexts.
Response: In the Recommendations section, we elaborate on the need for eligibility criteria updates and adjustments via a two-fold process. As described, “first, the proposed criteria for optimum vulnerability detection must align with the terms defined by the SCF frameworks. Next, these criteria must be populated by additional parameters already existing in regulation that allow for a more meticulous approach and detection of vulnerable groups”.
Then, we reflect on the potential of the methodological approach's adoption by other EU MSs, suggesting, that such adaptation process would require " high flexibility by the researchers’ and policy-making authorities, updated datasets, as well as collaboration between the states and civil society", since each national context varies, and simple replication actions could lead to incorrect results, obstructing the promotion of adequate measures.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors have made significant changes to the manuscript. However, two issues that were highlighted in the previous review are not resolved. A new format issue is present in Table 1 of the revised manuscript.
Despite the potential significance of the findings, the review process was hindered by the lack of curated data. In the interest of transparency, the authors should upload their processed or curated datasets—specifically, the version used directly in their computations—to a public repository (e.g., Figshare). Although the Data Availability Statement indicates where the data can be obtained, the primary value of this work lies in data curation; thus, reviewers must access the curated data to properly assess its quality. The reliability of the study’s conclusions fundamentally depends on accurate data management.
Response: thank you for making this important point. We have now added the link to the curated and processed dataset used in our analysis in a respective footnote (Also the data availability statement) . The dataset includes the harmonized and calculated indicator values, household-level extrapolations, and income thresholds used in constructing vulnerability groups.
Response of reviewer: We were referring to the curated dataset that can be manipulated directly to generate all the figures in the manuscript. It seems that the authors have used excel to generate the Figures. If that is the case, we are referring to the xlsx file, not the link to where the authors get their data. In addition, data availability should be mentioned in the Data Availability Statement, not as foot note in other sections.
In addition, comparisons across different surveys must be verified for logical consistency before any further analysis. For example, Table 4 appears to contain contradictory information regarding RI.1 (19.4% of households unable to keep cool or warm) and RI.2 (31.4% of households unable to pay utility bills). One might expect that the proportion of households unable to pay utility bills would be equal to or lower than the proportion of those unable to maintain adequate heating or cooling. This discrepancy requires further clarification or resolution.
Another issue arises in Table 7, which lists categories as “<3500” and “>3501” without indicating where 3501 itself is categorized. This omission should be addressed.
Response: This has been corrected. The final category in the previous Table 7 (which is now Table 2) now reads “≥3501€” to clearly reflect inclusion of the €3,501 threshold within the last category.
Response of reviewer: 3500 is still missing in the revised header. In table 2, the last headers of the last two columns are “<3500” and “>=3501”. The former exclude 3500, the later only include 3501, so 3500 is still missing.
Author Response
Comment: We were referring to the curated dataset that can be manipulated directly to generate all the figures in the manuscript. It seems that the authors have used excel to generate the Figures. If that is the case, we are referring to the xlsx file, not the link to where the authors get their data. In addition, data availability should be mentioned in the Data Availability Statement, not as foot note in other sections.
Response: The curated dataset has been uploaded together with the indicators in figshare as instructed (https://doi.org/10.6084/m9.figshare.28892639.v1)
Comment: In addition, comparisons across different surveys must be verified for logical consistency before any further analysis. For example, Table 4 appears to contain contradictory information regarding RI.1 (19.4% of households unable to keep cool or warm) and RI.2 (31.4% of households unable to pay utility bills). One might expect that the proportion of households unable to pay utility bills would be equal to or lower than the proportion of those unable to maintain adequate heating or cooling. This discrepancy requires further clarification or resolution.
Response:
We agree that at first glance, the higher proportion of households in arrears on utility bills (RI.2) compared to those unable to keep adequately warm or cool (RI.1) may appear contradictory. However, these two indicators capture distinct aspects of energy vulnerability.
RI.1 is a subjective, perception-based indicator reflecting households' self-reported discomfort in maintaining thermal adequacy. This perception can vary depending on individual tolerance, behavioural responses (e.g., reducing heating or cooling to control costs), or housing conditions. Conversely, RI.2 is an objective financial indicator representing households that report arrears on utility bills.
In practice, many households may prioritise paying energy bills (even through borrowing or cutting other essential spending) to avoid disconnection, while still falling behind on other utility payments or entering debt. This means a household can report arrears without reporting discomfort, or vice versa.
To address this and clarify for readers, we have added the following sentence to Section 3.2.1:
“It is important to note that RI.1 and RI.2 reflect distinct aspects of vulnerability: while RI.1 (19.4%) captures a household’s subjective experience of thermal discomfort, RI.2 (31.4%) refers to reported arrears on utility bills, which may encompass various types of services and signal broader financial stress. Many households may prioritise energy payments while cutting back on other needs or accumulating debt elsewhere, resulting in high arrears without necessarily experiencing inadequate comfort, therefore, the two indicators should not be expected to align directly.”
Comment: 3500 is still missing in the revised header. In table 2, the last headers of the last two columns are “<3500” and “>=3501”. The former exclude 3500, the later only include 3501, so 3500 is still missing.
Response: Adapted as suggested
Reviewer 2 Report
Comments and Suggestions for Authors no further question.Author Response
Comment: No comment received
Round 3
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors have addressed our concerns.