Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study is to explore the relationship between the performance of logistics and Environmental, Social, and Governance (ESG) performance using the multi-methodological framework of combining econometric with state-of-the-art machine learning approaches. The topic is timely. The literature is properly reviewed and the study model is well addressed. The study analysis and results are properly proceeded to get valid findings.  The following are some comments to improve the quality of the study.   

- It is focused on comparative analysis among different methods to explore the relationship between the performance and ESG while the topic seems to focus on exploring the relationship. Focus on one aspect. Otherwise, readers may get lost.      

- In line 14, IV panel ... => instrumental variable (IV) panel   

- In line 54, address it with a normal sentence, not a questioning sentence.   

- Address overall study purpose and study questions in the introduction section.     

- In line 65, This article => This study   

- In line 68,  the paper ... => the study or previous study ...   

- Throughout the manuscript, authors use the study with several different words such as article, paper, study, and research. Make it consistent.  Study or research is recommended to use.   

- Provide proper study hypotheses and examine them  

- In Tables 1, 7 and 12, all numbers with three decimal points are good enough. Modify them properly.   

- In Tables 1, 7 and 12, the meaning of ***  should be denoted below the table.    

- In line 1342, add theoretical implications and practical implications.   

- Enhance the overall study contents. The study title, study motivation, study purpose and study model  and study analysis, and implications must be congruent with each other.    

- Enhance the overall contents.     

Author Response

Point to Point Answers to Reviewer 1

Q1. This study is to explore the relationship between the performance of logistics and Environmental, Social, and Governance (ESG) performance using the multi-methodological framework of combining econometric with state-of-the-art machine learning approaches. The topic is timely. The literature is properly reviewed and the study model is well addressed. The study analysis and results are properly proceeded to get valid findings.  The following are some comments to improve the quality of the study.   

A1.  Thanks, dear reviewer.

Q2. It is focused on comparative analysis among different methods to explore the relationship between the performance and ESG while the topic seems to focus on exploring the relationship. Focus on one aspect. Otherwise, readers may get lost.      

A2. We would like to thank this comment. Indeed, as the purpose of the present work, the focus on the relationship between the logistics performance and the dimensions of ESG is clearly restated. In this regard, the role of the multiple methodological approach isn’t seen as its own comparative task, but as a complement set of tools, designed with the purpose of adding depth, richness, and insights to the analysis of that unique relationship. With this regard, this manuscript specifies that the econometric approach, as well as the machine-learning approach, do both share exactly the same objective: that of delivering a comprehensive, thorough, and robust analysis with regard to the relationship between the specific logistics performance, as well as that with every dimension of ESG.

Q3. In line 14, IV panel ... => instrumental variable (IV) panel   

A3. The expression on line 14 has been changed from “IV panel” to “Instrumental Variable (IV) Panel Data”.

Q4. In line 54, address it with a normal sentence, not a questioning sentence.   

A4. The sentence was written without a question mark as follows:

• How the interactions between the quality of logistics performance and each of the ESG pillars vary by country.

Q5. Address overall study purpose and study questions in the introduction section.   

A5. The following propositions have been added within the introduction

Study Purpose and Research Themes. This proposed research aims to develop a theoretical framework that explains the relationship between Logistics Capabilities Metrics and Environmental, Social, and Governance (ESG) factors within Sustainable Development. The emerging literature shows a strong correlation between national logistics capabilities metrics and social or environmental issues (Larson, 2021). Although the correlation between global logistics capabilities metrics and Sustainable Development remains uncharted, the general implication is that logistics capabilities significantly affect economic competitiveness, trade, and Sustainable Development. However, there seems to be limited insight into the independent or cumulative aspects of such logistics capabilities, particularly as a Sustainable Development factor. With regard to the literature established in the precedent of existing literature on logistics capabilities metrics/sustainability, more recent literature asserts that the Sustainability path within G20 nations is largely dependent on their capabilities, thus establishing that policy as collectively decisive (Harsono, 2023). By implication, that this literature fills a much-needed aspect, within existing Sustainable literature, that might explore this correlation between determinants of ESG factors, through measures of Logistics Efficiency established within pertinent literature, as emergent empirical notions demonstrate that ESG Logistics, defined through necessary environmental, social, or governance protocols, remain collectively independent, with negative, positive, notions within the macroeconomic paradigm (Nenavani et al., 2024). Improving ESG Capabilities, particularly within the realm of small-to-large-scale businesses, demonstrates a push towards more Sustainable, People-centric paradigms (Tsang et al., 2023). Simultaneously, technology such as Industry 4.0 remains effective within the strategic paradigm of transforming urban, corporate, or logistics infrastructure, as defined through the resultant ESG paradigm (Barykin et al., 2023). With this background, the proposed analysis will conduct a meticulous, sequential, multivariate analysis of Logistics Performance Index variables, cross-checked against more refined ESG variables, focusing on a dataset comprising 163 governance units for the years 2007–2023. To increase the validity of the results, posterior predictive checks and robustness analyses will be used. This will allow for issues of model specification parsimony and sensitivity issues that might qualify the conclusions. More specifically, this analysis will focus on the interaction between improved logistics capabilities under environmental stress and environmental efficiency, as captured by variables such as customs simplicity, customs clearance, and lead time reliability. Social-influence variables, such as education, working conditions, demographics, accessibility, ease of arranging shipments, and customer satisfaction, will be considered antecedents of logistics capabilities. More specifically, the knowledge generated by this analysis of education and working conditions might serve as a basis for formulating personnel management policies through train-and-develop programs or by establishing benchmark standards for laboratory practices, thus further reinforcing the social component of ESG. Governance quality will be analyzed through its constituent parts, including governance structures, governance frameworks, scientific productivity, and facilitative governance frameworks, which comprise tracking and tracing capabilities and transparency. By combining instrumental-variable panel regression with more sophisticated machine-learning analytics, this analysis will explore the two-way interaction between logistics capabilities and sustainable development.

 

Q6. In line 65, This article => This study   

A6. A change was made from "This article" to "This study" on line 65.

Q7. In line 68,  the paper ... => the study or previous study ...   

A7. A change was made from "This article" to "This study" on line 65.

Q8. Throughout the manuscript, authors use the study with several different words such as article, paper, study, and research. Make it consistent.  Study or research is recommended to use. 

A8. We have replaced the word article with study and the word paper with research throughout the text.

Q9. Provide proper study hypotheses and examine them  

A9. The following propositions have been added within the introduction

Study Hypotheses. With the aforementioned research questions as the background, this study formulates three inclusive hypotheses that provide direction for the analysis, thereby aligning the conceptual framework with the methodology. These hypotheses assume that the correla-tion between logistics performance metrics and sustainability outcomes is complex, inter-acting with environmental, social, and governance factors within the ESG framework (Zhang et al., 2025).

H1. Logistics performance shows a systematic relationship with mixed environmen-tal effects, reflecting trade-offs between development and the environment. This hypothesis argues that improvements in logistics infrastructure can minimize resource use and some types of pollutants, but simultaneously increase other pollutants, such as GHG emissions. The existing literature suggests that ESG innovations focused on logistics and transporta-tion can improve environmental efficiency while addressing new environmental pres-sures, such as increased energy use and GHG emissions (Wang et al., 2024; Rodionova et al., 2022). Using disaggregated measures of environmental effects, such as air, GHG emis-sions, heat stress, and land use, this research will examine the impact of environmental stresses and efficiencies as forces behind changes in the Logistics Performance Index (Zhang et al., 2025).

H2. Socioeconomic variables significantly and diversely affect logistics performance. This hypothesis assesses the influence of education, basic service accessibility, de-mographics, working conditions, and income distribution on logistics performance. Evi-dence confirms that socioeconomic variables, such as employee education, fair working conditions, and service accessibility, affect logistics efficiency (Rodionova et al., 2022). So-cial determinants, such as education, access to basic services, demographics, working conditions, and income distribution, create inequality in human capital, working condi-tions, or both, affecting the efficiency of global logistics.

H3. Improving governance quality promotes a positive outcome on logistics perfor-mance. This hypothesis assumes that a high-quality institution, with attributes of proper regulation, the rule of law, efficient administration, and scientific strength, fosters a sup-portive environment that facilitates the establishment of a sound, modern, and trustworthy logistics infrastructure. Empirical evidence shows that sound governance principles or regulations can enhance ESG practices and sustainable development across nations (Jílková & Kotěšovcová, 2023; Wang et al., 2024). These hypotheses collectively form the focal point of this analysis, through which the rest of this report will explore the relation-ships that exist between logistical performance and the environmental, social, and gov-ernance aspects of sustainable development.

 

Q10. In Tables 1, 7 and 12, all numbers with three decimal points are good enough. Modify them properly.   

A10. The numbers in Tables 1, 7 and 12 have been rounded to three decimal places.

Q11. In Tables 1, 7 and 12, the meaning of ***  should be denoted below the table.    

A11. A note explaining the meaning of the asterisks has been added to Tables 1, 7 and 12.

Q12. In line 1342, add theoretical implications and practical implications “Policy Implications”.   

A12. The following subsessions have been added with the “Policy Implications”:

Theoretical implications. .  Several theoretical implications of this analysis emerge as a consequence of its results. Among them, the most central implication lies within the assumption of ‘two-sided logistics’ that not only captures its enabling dimensions but is also sensitive to any negative externality that might emerge as a consequence of deregulation (Rodrigues, Fiorini, & Piato, 2022). From a theoretical perspective, this report clearly shows that, through the use of appropriate causal analysis, logistics, as well as sustainability, move beyond traditional linear or unidirectional notions. Rather, they assume multiple-layered, structurally defined trade-off forms that transform and evolve in specifically dynamic ways (Yontar, 2022). The application of instrumental variables with machine learning tools in logistics and sustainability pushes beyond conventional descriptive or correlation-based literature, opening theoretically innovative dimensions in defining logistics and sustainability as theoretically robust. Logistics, on the one hand, can be regarded as enabling infrastructure for achieving logistics sustainability, whereas, on the other hand, negative externalities may emerge with unregulated logistics (Barykin et al., 2023). Empirical confirmations of the Logistics Performance Index (LPI) as well as its improvements that contribute, along with their effects, towards achieving logistics sustainability exist. Improvements in LPI will enhance social sustainability, such as better education or the absence of child labor in some industries and countries. However, LPI, along with logistics, may cause negative social imbalances or inequalities that may call for structural and meaningful corrective measures (Mutambik, 2024). Moreover, environmental sustainability may emerge due to the promotion of clean technologies along with more effective resource management. However, negative environmental imbalances, such as pollution, may emerge from such LPI, along with their effects that may call for structural and meaningful remedies (Yontar, 2022). Thus, theories recognize the application of "two-sided logistics sustainability" that relies heavily on, as well as being dependent on, some extremely specific assumptions: that logistics may enhance social sustainability in multiple aspects (improving educations or eliminating child labor in some industries) or may generate social inequities requiring attention, along with contributing towards environmental sustainability in some ways, however, with the possibility of environmental pollutions (Rodrigues et al., 2022; Barykin et al., 2023). Further, this analysis refutes the existing theoretical consideration that takes a marginal stance concerning the role of environmental and demographic variables in the field of logistics. However, the causal dynamics of air pollutants, GHG, heat stress, and logistics efficiency identify the importance of theoretically formulating logistics systems as units that respond to environmental dynamics (Yoo, 2025). This goes beyond the existing theoretical focus that considers infrastructure, cost, and trading volume. An extra theoretical implication relates to the formulation of ESG variables. By employing a disaggregated framework, there would be the possibility of proving that the interaction of the three variables of ESG with logistics system dynamics differs in ways that would not be recognized via the aggregate metric (Mutambik, 2024; Yoo, 2025). Finally, in light of the above, the interaction of machine learning with econometric insights within the methodology is a material theoretical component, clearly showing that large-scale sustainability dynamics are, in fact, nonlinear, such that country-level profiles must be factored in for their effective probing. In other words, this theoretical exercise again vindicates the need for hybrid studies in the theory of supply chain dynamics and sustainability (Rodrigues et al., 2022; Yontar, 2022). Moving ahead, a practical policy implication would be that carbon pricing policies must be adopted with a focus on keeping corporations on their toes in terms of carbon emissions, along with providing incentives for more investments in green technology. At the same time, green freight routes may significantly enhance the efficiency of logistics along with its positive effects on the environment (Barykin et al., 2023).

Practical implications. A number of the results have implications that are important for policymakers, decision-making bodies in international forums, investment circles, and managers responsible for logistics. Recent studies suggest that improvements in logistics do not always yield positive outcomes for environmental, social, and governance (ESG) factors, requiring policy measures to reconcile efficiency with sustainability (Karountzos et al., 2025). However, it seems entirely valid that improvements in logistics will increasingly necessitate more focused strategies, as greater efficiency may increase costs in terms of sustainability. With regard to environmental aspects, stronger links in the logistics system exhibit a strong positive correlation with NOx emissions, whereas air pollutants, specifically particulate matter (PM2.5), negatively affect efficiency. These findings align with existing empirical insights that link logistics practices with material environmental spillovers, such as higher NOx levels, when effective sustainability-mitigating strategies are not considered (Kim et al., 2024; Truant et al., 2024). This further explains that any project linked to logistics practices must effectively address comprehensive environmental protection, including decarbonization, environmentally friendly transportation, ecologically safe warehouses, and climate-proof infrastructure, as mentioned by Kim et al. (2024). Socially, the topic highlights the importance of incorporating human development considerations into logistics planning. Specifically, the positive causal link between Child Labor (CET) and the Logistics Performance Index (LPI) across different societies highlights that some logistics improvements may have stemmed from problematic social behaviors. These results confirm that logistics improvement must operate within the parameters of responsibility, so that the development of the value chain occurs in a fair, humane manner (Truant et al., 2024). Also, the negative impacts of demographics, such as aging societies or increasing school enrollment, on logistics efficiency indicate that the adoption of effective active labor policies, as well as the promotion of automated technology, would be necessary as a buffer against changes in the unavailability of the workforce (Popescu et al.2024). Thus, social logistics and sustainable logistics practices must be seen as structural aspects that ensure the supply chain's resilience, rather than mere afterthoughts. With regards to governance, nations with better quality governance, defined as better regulation, rule of law, accountability, as well as scientific capabilities, show better logistics capabilities (Karountzos et al. 2025). This explicitly indicates that betterment in the Logistics Performance Index requires improvements in governance quality (Kim et al. 2024). Machine learning analyses provide additional helpful tools for decision-makers. Prediction algorithms such as Random Forests or K-Nearest Neighbors (KNN) may help governments and companies predict logistics efficiency. In addition, cluster analysis helps identify different national profiles. This will allow global organizations and development finance institutions to develop strategies grounded in concrete environmental or structural issues (Truant et al., 2024). In conclusion, the implications of the mentioned dimensions are that Sustainable Logistics requires that the processes of infrastructure development, environmental issues, social aspects, and the quality of institutions must move forward together. Thus, logistics transformation must be incorporated into broader sustainable development paradigms so that efficiency for both the economy and the environment can be achieved (Karountzos et al., 2025; Popescu et al., 2024).

 

Q13. Enhance the overall study contents. The study title, study motivation, study purpose and study model  and study analysis, and implications must be congruent with each other.  

A13.  

• Study title: the title has been changed has follows “Logistics Performance and the Three Pillars of ESG: A Detailed Causal and Predictive Investigation”
• Study motivation: the following sentences have been added within the introduction “The current work proposes and seeks to explore answers to the question of whether and to what extent the sustainability implications of logistics performance can be captured and quantified in relation to the three sets of ESG factors: environmental pressures, social change, and governance change, over the period between 2007 and 2023, for the total of 163 nations. In this process, the current work seeks to fill a gap in the literature that relies solely on economic grounds to interpret the LPI, whereas the ESG literature and scholarship interpret this domain on purely organizational or financial grounds, in parallel or in tandem, yet independently of each other. This is particularly important in this case because logistics itself is reflected in its infrastructure and activities, which have direct implications for greenhouse gas emissions, the use of natural resources, social labor patterns, service delivery, and governance quality and strength. In this case, it can be seen that changes in supply chain and logistics technologies are spreading quickly around the world, and, in this context, this current work examines the practical applications of ESG frameworks in logistics for validation.
• Study purpose: This research endeavors to identify a comprehensive paradigm grounded in empirical re-lationships between logistics performance metrics and the Environmental, Social, and Governance (ESG) dimensions of Sustainable Development. Although the importance of global logistics competencies as primary determinants of economic competitiveness, trade efficiency, and development is well established, their interrelationship with Sustainable Development remains largely uncharted. There is still a significant knowledge gap re-garding the independent and cumulative effects of different logistics competencies on ESG metrics, especially in light of recent controversies over their impacts on sustainable de-velopment. The importance of this research effort lies in its attempt to fill this knowledge gap by recognizing the determinants of ESG with regard to logistics competencies, as well as the interaction between logistics efficiency and ESG dimensions, as generally estab-lished in the existing literature. With this objective, the proposed research will conduct a rigorous, sequential analysis of the Logistics Performance Index (LPI) variables cross-nationally, together with detailed ESG variables, for the jurisdictions of 163 national governance units over the years 2007 through 2023. Specifically, this inquiry will explore: (1) whether enhanced logistics capacities correspond with environmental stress, as op-posed to environmental efficiency. (2) Social conditioning as the antecedent of logistics capacities, focusing on aspects such as education, working conditions, demographics, and accessibility. (3) Governance quality focusing on its constitutive elements, such as enabling institutions, regulatory frameworks, scientific productivity, as well as facilitative governance structures. Utilizing instrumental variables (IV) panel regression analyses, together with sophisticated machine learning models, this scientific investigation will seek to identify the two-way dynamics between logistics capacities and sustainable de-velopment.
• Study model: This work adopts a multi-method design to analyze the link between logistics performance and Environmental, Social, and Governance (ESG) dimensions at the macro level. At the heart of this design is the Logistics Performance Index (LPI) as the dependent variable, with its Environmental, Social, and Governance variables as its determinants. In exploring the form of this link, this work will use Instrumental Variable (IV) panel fixed-effect regres-sion models, including Two-Stage Least Squares (2SLS) and Generalized (2SLS) models. Such designs continually address issues of endogeneity, missing data, and reverse causal-ity. Such economic specifications measure the differential roles of environmental, social, and governance factors in influencing logistics performance across a broad list of 163 na-tions from 2007 to 2023. Apart from this established model for determining causation, this work will use more intricate machine learning models to explore nonlinear associations, thereby providing models with enhanced precision and the ability to discover hidden dy-namics across different nations. Regression models (RF, SVM, kNN, Tree, Boost, as well as L) will evaluate precision, while cluster models (DBSCAN, FCM, Hierarchical, Mod-el-based, Neighborhood, as well as RF) will capture structural differences, particularly across models associated with different nations. By coming together, any assessment as-sociated with the use of ESG variables as discrete variables will seek to measure their net effects on the entire logistics industry, providing this work with a complex perspective on the contents of any linkages between sustainability studies. Such models will further posi-tion the logistics industry as a key ecological lever, providing this critical perspective on the different ways in which this industry’s better practices seek to demonstrably lessen resource intensity and environmental impacts. Effective ways through which freight con-solidation, as well as routes, offer effective solutions in terms of optimizing processes, as well as lowering environmental impact. Such will clearly demonstrate the applicability of any improvement in the logistics aspect.
• Study analysis: This paper combines econometric identification with machine learning predictions to examine the interplay between environmental, social, and governance (ESG) variables’ impacts on the logistics performance of 163 nations from 2007 through 2023. Using IV panel regression models, the empirical results suggest that environmental variables exert a two-faced, double-edged, but otherwise negative impact on the Logistics Performance Index (LPI). Specifically, GHG emissions, agricultural value added, air pollutants, and extensive agricultural land usage positively/negatively contribute, respectively, towards better logistics efficiency. Social variables, in contrast, demonstrate equally complex correlations. Water, sanitation, aging, education, as well as increased elementary education enrollment rates, deliver mild negative adjustments, whereas child labor increases LPI indices. Income inequality strongly dampens logistics activity, clearly establishing that sound social development with equitable economic distribution facilitates efficient value chain activity. By contrast, machine learning predictions indicate that IV estimates of environmental stress, education, and demographics remain relevant, valid, proper, robust, and serviceable. Clustering-based approaches establish unique national group profiles, with air pollutants, extreme temperatures, and agricultural intensity defining such national loci. Thus, this empirical assessment proves that multifaceted logistics sustainability prevails, implying that successful logistics evolution requires complementary, not sequential, attention to environmental protection, societal advancement, and sound governance.
• Study implications: The implications of the results of this research are important to policymakers interested in linking logistics performance improvements to more general Environmental, Social, and Governance (ESG) goals. The results show how improvements in Logistics Performance Index (LPI) are interwoven with the core elements of the ESG framework. This finding indicates how goods and services traditionally associated with a sector or operations domain are a much more integral part of pursuing sustainable and fair development objectives (Sharawi et al., 2025). From a governance standpoint, the empirical results highlight the importance of good institutions capable of supporting regulatory quality as well as government effectiveness. Enhanced performance in logistics is positively related to better regulatory practices and governmental institutions, as evidenced by the striking correlation between LPI and Government Effectiveness (GEE) and Regulatory Quality Estimate (RQE) variables (Göçer et al., 2022). Policymakers are thus implored to pursue strengthening transparency, efficiency, and accountability in public sector organizations and to recognize how such improvements are likely to have spillover effects on the efficiency of logistic networks as well as on overall national competitive power. Social considerations are addressed by the research in showing how greater efficiency in logistic networks contributes significantly to a better realization of broader social rights outcomes like Economic and Social Rights Performance Score (ESRPS). More efficient logistic networks are likely to promote greater accessibility to essential goods and services and to promote fairness in social development. National development policies are thus urged to address the role of logistic infrastructure as more than an economic imperative but as a social imperative. Logistics investments are to be planned with clear social objectives to ensure the benefits from improvements in efficiency in the supply chain are shared equitably across various social classes and among different regions. Environmental consequences also appear as essential from the research. While the LPI does not directly capture environmental outcomes as a measurable variable, decomposition of ESG elements by the research shows how the impact of efficiency improvements in the logistic networks has to be accompanied by proactive environmental regulation and incentive arrangements motivated by a reduction in the environmental footprint of logistic chains. Governments are thus urged to contemplate adopting convergence to green logistic standards and promoting green transport practices as well as providing incentives to adopt low-emission technologies in the logistic industries. Beyond borders, the interconnection of global supply chains necessitates global cooperation. The countries with high LPI scores promote trade both domestically and in a broader region. This emphasizes the functioning and efficiency of regional organizations and trade agreements in aligning the standards of logistics and sustainability policies (Sharawi et al., 2025). Policymakers thus need to pursue diplomatic efforts enshrining ESG issues in trade and transport agreements to avoid making efficiency in logistics a price paid at the expense of causing harm to the environment and social exclusion. Furthermore, the cluster analysis in this research exhibits heterogeneous patterns in countries' performance regarding both logistics and ESG outcomes to support research on the spatial heterogeneity of supply chain efficiency and governance performance (Yıldırım, 2023). This result highlights the necessity of differentiated policy responses. One-size-fits-all policies are unlikely to succeed with the differentenciing institutional, economic, and infrastructural circumstances across countries. Countries with low LPI scores and poor ESG indicators are best focused on core governance and infrastructure reforms, while countries in higher-performing groups might fine-tune their logistics ecosystems towards even higher levels of environmental and social sustainability (Lee, 2024). Further, the established causual dynamics connecting scientific output, as a proxy by the scientific and technical journal articles count (STJA), and logistics performance imply that policy on innovation has to accompany any policy on logistics. Governments need to promote research and development work focused on improving the features of logistics technologies, promoting the digitalization of the supply chains and the design of sustainable transport options. Government investment in advanced research and education on logistic issues will enhance LPI scores but also support the overall ESG agenda—an increasingly identified dynamic in recent research on ESG and logistics (Nenavani et al., 2024). The component of political stability as captured by the Political Stability and Absence of Violence/Terrorism (PSAOV) estimate also has a pertinent impact on the performance of logistics. Stable politics facilitates efficient and stable systems of logistics which in turn facilitate trade, economic progress, and social welfare. This shows that policy focused on improving political stability, dampening corruption and conflict are part and parcel of policy on logistics as well. In reality, transport and trade and also social and environmental affairs ministries have to collaborate much more cooperatively and across disciplines. Sectoral boundaries are likely to interfere with the type of across-the-board policymaking the results of this research call for. Cross-sectoral data analysis and evidence-supported assessment informed policy framework plans of action have to become the rule and not the exception. Lastly, also international development agencies and multilaterals and financial institutions should realign part of their investments in logistic infrastructures by tying ESG appraisal parameters to their evaluation methodology. Financing development has the power to become a potent driver in improving the performance of logistics and reaching ESG targets as long as it is aligned to the required standards of sustainability (Rodionova et al., 2022). In summary, the findings of this research locate the role of logistics performance as a key lever of sustainable development. Policy design has to acknowledge the multifarious nature of the impact of logistics on governance quality, social entitlements, green sustainability, and economic dynamism. Policymaking in the future has to be holistic, strategic, and responsive to context to release the full power of logistics as a source of ESG-compatible development (Nenavani et al., 2024; Lee, 2024).
• Theoretical implications. The findings of this analysis include a series of important theoretical implications that contribute meaningfully to the literature on the performance of logistics and the implications of ESG for nations. At its core, this research introduces the premise of 'two-sided logistics,' capturing both the enabling aspects of logistics as well as the externality-producing potential when not carefully regulated. In theoretical terms, this analysis shows that, through technically valid methods of causal analysis, logistics and sustainability are much more than just linear or unidirectional concepts. Rather, logistics and sustainability are identified as multi-layered and structurally trade-off relationships that are shaped and reshaped in highly characteristic and correlated ways. The combination of IV analysis and machine learning analysis of logistics and sustainability goes well beyond any literature on description or correlational relationships between concepts. The literature extends beyond mere description or correlations and defines logistics and sustainability in logical, innovative ways. Logistics can be seen as 'enabling infrastructures' for sustainability on one side of the ledger and as a contributor to 'negative externalities' on the other side of the ledger when logistics is not carefully and meaningfully regulated. Evidence of LPI and improvements in logistics and its effect on sustainability can be seen clearly: LPI in logistics performance leads to improvements in social sustainability, such as enhancements in education or the elimination of child labor in certain sectors of the economy and in certain nations. Conversely, LPI and logistics can lead to social imbalance or inequalities that need to be structurally and meaningfully addressed. In addition, LPI and logistics promote environmental sustainability through the generation of clean technologies and better, more effective use of resources. However, LPI and logistics can also generate pollution and environmental problems that need to be structurally and meaningfully addressed. Thus, theory supports an understanding of 'two-sided logistics sustainability' driven and shaped in very valid and specific ways: logistics can lead to or drive social sustainability in various ways (for example, improvement of education or elimination of child labor); yet logistics can also create social imbalance or inequalities that need to be addressed; and logistics that are valid can generate environmental sustainability while also potentially leading to pollution. Further, this analysis challenges the prevailing theoretical tenet that environmental and demographic factors play only a marginal role in logistics. The causal relationships among air pollution, greenhouse gas emissions, heat exposure, and logistics performance underscore the need to conceptualize logistics systems as units sensitive to their environments. This extends the domain of theory beyond logistics’ current preoccupation with infrastructure, cost, and trade volumes to include a wider spectrum of factors, including climate risk, land use, and environmental damage as vital logistics efficiency factors. Another important implication of this work for theory pertains to the conceptualization of ESG factors. Through this disaggregated approach, it is possible to illustrate that the three ESG factors each operate in relation to logistics systems in ways that cannot be captured in an aggregated approach and, in turn, contribute in different ways to theory on sustainability itself. In conclusion, the integration of machine learning and econometrics into the methodology is itself an important theoretical insight, as it illustrates that large-scale sustainability phenomena are nonlinear and that country profiles must be considered to explore them in novel ways. In other words, this work reconfirms the argument that hybrid methodologies are required in the theory of supply chain and sustainability. Building on these insights, a concrete recommendation for immediate action could be to implement emissions pricing mechanisms. These would hold companies accountable for their carbon footprints and encourage investment in green technologies. Additionally, establishing green freight corridors could significantly enhance logistical efficiency while minimizing environmental impact. By leveraging these regulatory levers, decision-makers can operationalize the theoretical insights from this research into effective sustainability policies.
• Practical implications. Some of the findings have several implications that are of great interest to policymakers and decision-making authorities in international bodies, investors, and logistics managers. First, it can be seen that better logistics performance is not necessarily expected to promote favorable environmental, social, or governance factors. Nonetheless, it can be observed that logistics improvement is expected to require more targeted interventions to counter the problem of efficiency translating into unintended sustainability costs. Turning to environmental considerations, it is clear that logistics improvements are closely linked to increased nitrous oxide (NOx) emissions, and that pollution from particulate matter (PM2.5) reduces efficiency. Unfortunately, this indicates that any logistics-related initiative needs to incorporate comprehensive environmental protection measures, such as decarbonization, green transport, energy-efficient storage facilities, and climate-proof infrastructure. Socially, the issue at hand underscores the need to integrate human development factors into logistics planning. For example, the positive causal effect of child labor (CET) on the Logistics Performance Index (LPI) in some areas highlights one of the most important social considerations: that some logistical advancements can be derived from child labor. In addition, the negative effects of some demographic factors (ageing or rising school attendance rates) on logistics performance underscore the need for active labor market programs, such as training workers to be better equipped in logistics and employing technological and automated means to offset changes in logistics workforce availability. Thus, it is important that social and sustainable logistics planning be seen as structural elements that can sustain supply chain resilience rather than afterthoughts. In terms of governance, it emerges that nations with superior institutional qualities, including better regulation, the rule of law, accountability, and scientific strength, consistently demonstrate better logistics performance. Here, it can be noted that in order for improvements in LPI to be achieved, changes must be made in administrative strength, and developments in the area of governance must be initiated. In practical terms, this means that logistics investments must be aligned with governance improvement and corruption eradication. Machine learning analysis also provides additional practical tools for decision makers. Predictive analysis methods such as Random Forests or K-Nearest Neighbors (KNN) can help governments and companies predict logistics efficiency across different environmental and social contexts. In addition, cluster analysis reveals different country profiles that can help international organizations and development banks develop interventions tailored to specific types of environmental stress, agricultural systems, or risk factors. In conclusion, it is clear that the ultimate implication of these areas is this: sustainable logistics requires that infrastructure development, environmental protection, social considerations, and institutional quality be developed in tandem. Logistics modernization must thus be situated within the context of comprehensive sustainable development to achieve efficiency that benefits either or both of these areas.

Q14. Enhance the overall contents.     

A14. We have answered to this question in the previous Q13.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

 

1. General Assessment of the Issues and Originality In the article, the authors examine the relationship between the Logistics Performance Index and various dimensions of sustainable development. They accomplish this task using advanced research tools, a large sample, and a long time period. The intention itself is attractive, original, and valuable. Research in this area is being conducted, but it is characterized by less complexity and geographical scope. I positively evaluate the scientific value of the article.
2. Abstract The abstract is properly constructed. It contains the objective, methodology, and main conclusions. I have no comments on this section of the article.
3. Introduction – Justification In the introduction, the authors formulate the research questions quite quickly. A section justifying the undertaking of the research and indicating why it is important would be useful before the scope of the research is presented.
4. Literature Review The literature review is based on 103 publications. This is an extensive and current review. In the body of the article, the authors refer to specific findings from previous research, which effectively illustrates their line of reasoning. No excessive self-citation was found. The literature is appropriately selected.
5. Research Methodology In the article, the authors utilize a wide range of modern research techniques. The machine learning models—models based upon regression, viz. Random Forest, k-Nearest Neighbors, Support Vector Machines, Boosting Regression, Decision Tree Regression, and Linear Regressions, and clustering, viz. Density-Based, Neighborhood-Based, and Hierarchical clustering, Fuzzy c-Means, Model Based, and Random Forest—were applied to uncover unknown structures and predict the behaviour of LPI. This demonstrates very strong research expertise. Nevertheless, for one publication, it could be reduced, as the text exceeds the standard length of an article. These methods are properly applied. The authors consistently repeat the research pattern for each of the examined groups of LPI determinants. This increases the clarity of the text and allows the reader to follow the reasoning. The analysis of 163 countries has great value, although with such a large number, detailed conclusions are lost. It would be worthwhile to supplement the research process diagram before the studies.
6. Results The results are described consistently for each dimension: environmental, social, logistical, and managerial. This is a good structure that allows tracking the authors' line of reasoning. I have no comments on this section.
7. Discussion and Conclusions Within the discussion, the authors include policy implications. However, I consider the statement "In summary, the findings of this research locate the role of logistics performance as a key lever of sustainable development" to be decidedly exaggerated. Sustainable development is determined by many different factors. I also suggest that the guidelines based on such extensive research be grouped, for example, in a table for each of the analyzed areas. The summary of such extensive research is too brief.
8. Graphic Material The graphic material in the article includes numerous charts and tables. They are carefully prepared, and I have no comments on them.
9. Suggestions for Improvement a) supplement the introduction with research justification and research gap; b) citations should be corrected to comply with the journal's requirements; c) supplement the research process diagram before the detailed description of the methodology; d) expand the policy implications to include guidelines for each of the analyzed areas; e) emphasize the research contribution to sustainable development.

Author Response

Point to Point Answers to Reviewer 2

 

Q1. General Assessment of the Issues and Originality In the article, the authors examine the relationship between the Logistics Performance Index and various dimensions of sustainable development. They accomplish this task using advanced research tools, a large sample, and a long time period. The intention itself is attractive, original, and valuable. Research in this area is being conducted, but it is characterized by less complexity and geographical scope. I positively evaluate the scientific value of the article.

 

A1. Thanks dear reviewer.

 

Q2. Abstract The abstract is properly constructed. It contains the objective, methodology, and main conclusions. I have no comments on this section of the article.

 

A2. Thanks dear reviewer.

 

Q3. Introduction – Justification In the introduction, the authors formulate the research questions quite quickly. A section justifying the undertaking of the research and indicating why it is important would be useful before the scope of the research is presented.

 

A3. The following sentences have been added before the research questions:

 

Justification for the study. Despite the increasing importance of both the logistics performance perspective and the environmental, social, and governance (ESG) framework in the design of sustainable economic systems, the intersection between the two areas remains largely uncharted (Larson, 2021; Nenavani et al., 2024). Currently, existing literature focuses predominantly on logistics performance as economic infrastructure, with most economic studies being carried out on a firm level, whereas, conversely, most existing studies on ESG frameworks focus mostly on a firm level paradigm, thereby largely ignoring their systemic dynamics within the country’s economic infrastructure (Radu et al., 2023). Currently, this systemic divide also leaves a huge knowledge gap, particularly with increasing recognition being accorded to the roles of logistics systems, as either facilitators or barriers, within environmental sustainability, as well as social well-being, and within governance dynamics (Larson, 2021). Knowledge within this intersection is also highly required, particularly since logistics infrastructure designs significantly impact issues such as energy consumption, carbon emissions, resource use efficiency, working conditions, inclusive supply chains, and transparency in governance (Nenavani et al., 2024). In addition, most global commitments, such as the United Nations Sustainable Development Goals, significantly depend on the assurance of sustainable logistics systems, so empirical studies within this field, particularly within the intersection of systemic dynamics within logistics infrastructure design, within ESG frameworks within different countries, remain largely uncharted (Radu et al., 2023).

 

 

Q4. Literature Review The literature review is based on 103 publications. This is an extensive and current review. In the body of the article, the authors refer to specific findings from previous research, which effectively illustrates their line of reasoning. No excessive self-citation was found. The literature is appropriately selected.

 

A4. Thanks dear reviewer.

 

Q5. Research Methodology. In the article, the authors utilize a wide range of modern research techniques. The machine learning models—models based upon regression, viz. Random Forest, k-Nearest Neighbors, Support Vector Machines, Boosting Regression, Decision Tree Regression, and Linear Regressions, and clustering, viz. Density-Based, Neighborhood-Based, and Hierarchical clustering, Fuzzy c-Means, Model Based, and Random Forest—were applied to uncover unknown structures and predict the behaviour of LPI. This demonstrates very strong research expertise. Nevertheless, for one publication, it could be reduced, as the text exceeds the standard length of an article.

 

A5.  Regarding the remark on diminishing returns in machine learning models, we would like to acknowledge the reviewer’s comment. At the same time, we would like to clarify the reasoning behind this methodological decision. Multiple predictive models, as compared with other studies, remain one of the essential strengths of this work, particularly given the existing literature's relative absence of guidance on selecting the best-suited models to approximate the non-linear correlates that may link LPI variables with ESG variables. Thus, rather than rendering this task redundant, it remains necessary. The diversity of models helps provide:

• uncover non-linear patterns that would not appear if traditional econometric models alone were being analyzed.
• comparisons of different levels of heterogeneities to prevent comparison with only one, maybe inefficient, method.
• Evaluate the robustness of your results with complementary methodological approaches. Therefore, the extensive comparison of machine learning algorithms is considered a necessary component of the research.

 

 

Q6. Research Methodology. These methods are properly applied. The authors consistently repeat the research pattern for each of the examined groups of LPI determinants. This increases the clarity of the text and allows the reader to follow the reasoning. The analysis of 163 countries has great value, although with such a large number, detailed conclusions are lost.

 

A6. The following  sentences have been added within the methodological section.

 

Limitations. Across all analyses, the dataset includes 163 countries. Although such a large cross-country dataset helps derive corresponding cross-country correlations with relative ease, this inevitably affects the level of granularity available to inspect individual national settings. To address this tension, the necessity of accounting for national cross-country variability is carefully explained in this manuscript through complementary cross-country analysis strategies, including cluster and machine-learning algorithms.

 

Q7. Research Methodology. It would be worthwhile to supplement the research process diagram before the studies.

 

A7. The following image has been added in the methodological section as a graphical abstract:

 

Figure xxx. Overview of the Research Design and Analytical Framework. Note: the diagram illustrates the full research workflow, including the reconstruction of missing LPI values through polynomial interpolation, the integration of Environmental, Social, and Governance indicators, and the application of both econometric and machine learning approaches for the analysis.

 

 

Q8. Results The results are described consistently for each dimension: environmental, social, logistical, and managerial. This is a good structure that allows tracking the authors' line of reasoning. I have no comments on this section.

 

A8. Thanks dear reviewer.

 

Q9.Discussion and Conclusions Within the discussion, the authors include policy implications. However, I consider the statement "In summary, the findings of this research locate the role of logistics performance as a key lever of sustainable development" to be decidedly exaggerated. Sustainable development is determined by many different factors. I also suggest that the guidelines based on such extensive research be grouped, for example, in a table for each of the analyzed areas. The summary of such extensive research is too brief.

 

 

A9. A new section entitled “Discussion” has been added:

 

 

8. Discussion

 

In this regard, the analysis illustrates the complex but intricate link between logistics performance and the trilogy of E(S)G issues with sufficient evidence that shows that improvements in the Logistics Performance Index (LPI) positively correlate with environmental, social, as well as governance issues, although with different degrees of intensity (Nenavani et al., 2024; Yoo, 2025). Regardless of the models, be it economics, AI, or cluster models, the fact remains that logistics efficiency is both driven as well as a driver of sustainability, with the LPI being the mediator that facilitates the link between environmental sustainability, social issues, and good governance (Mutambik, 2024).

Environmentally, the findings indicate a strong trade-off among economic development, resource use, and environmental conditions. There is a positive correlation between higher LPI rankings and greater GHG emissions, such as nitrogen oxides, indicating that as economic development occurs, associated carbon-intensive practices often increase. This observation aligns with the traditional diseconomies observed in early-stage industrialization, where increased transport infrastructure, warehousing, and freight activity lead to higher associated GHG emissions (Juvvla et al., 2025). At the same time, environmental degradation, particularly air pollutants such as PM2.5, significantly costs the logistics system. Air pollutants contribute to poor logistics performance, leading to reduced productivity and logistical infrastructure disruptions (Mutambik, 2024).

It is further found that there is a positive correlation between extreme heat exposure and LPI, which is explained by the efforts of hot-region countries that invested significantly in resilient logistics structures to overcome inefficiencies caused by weather conditions (Yoo, 2025). Land-use factors further interact with this aspect, such that a higher percentage of agricultural land is associated with poorly developed logistics infrastructure, but higher value addition in agriculture is associated with better logistics efficiency. This difference indicates that subsistence agriculture does not enhance logistics efficiency, but a more commercialized agricultural sector promotes investment in the Cold Chain, exports, and transport (Nenavani et al., 2024). References confirm the importance of environmental variables, pinpointing agricultural land, nitrous oxide, PM2.5 air quality, and heat stress as the top determinants of LPI (Mutambik, 2024). Taken together, this literature collectively suggests that, contrary to the topic’s periphery, environmental issues play a fundamental, defining role within the efficiency of logistics, finding that any improvement within this realm will necessarily require addressing issues of environmental degradation as well as those of climate change (Juvvala et al., 2025). Socially, the implication of this finding is that there is a complex interaction of positive and negative effects, pointing towards a convergence of logistics performance with human development. Higher LPI scores indicate improved education, as evident through higher education enrollment and fewer children in the workforce, as the workforce, processes, and technology required in logistics systems depend on human knowledge, competencies, and technology-enabled human resource capabilities, thus their development in societies with positive education dynamics (Nenavani et al. in press, 2024). However, income inequality appears to be a robust negative factor for logistics performance, suggesting that societies with income inequality will suffer from inefficient human resource allocation, unconsolidated service value, and inadequate infrastructure accessibility. Simultaneously, other dimensions, such as an aging population and imbalanced access to basic services, appear as weak negative determinants of logistics efficiency. Such results indicate that, despite human development, logistics development may increase inequalities in societies, which may further strain them if proper social policies do not channel appropriate attention (Juvvala et al., in press, 2025). The results confirm that logistics performance extends beyond techno-infrastructural aspects, encompassing social cohesion, human capital development, and equal opportunities for economic participation (Yoo, 2025). In terms of governance, the results are more definitive, being strongly positive. There is a positive correlation between higher governance, as indicated by more effective governments, the rule of law, the quality of regulation, scientific productivity, and logistics performance. This occurs because the enabling effects of governance shape proper regulations, customs policies, the enforceability of contracts, and stable environments that attract investment in logistics infrastructure (Mutambik, 2024). "Voice and accountability, combined with active scientific production, further enhance the enabling effects of innovation-modified logistics systems" (Yoo, 2025). Similarly, machine learning models identify governance variables as robust predictors of LPI, with cluster analyses affirming that nations with strong governance, as identified via systematic clustering, position closely with strong logistics practices (Nenavani et al., 2024). These indicate that governance as a consideration is essential, as it promotes efforts toward logistics modernization, allowing better alignment between the scale of logistics development and ESG principles (Juvvala et al., 2025). In any case, the conclusions of this research indicate that logistics performance is firmly positioned at the center of the dynamics of ESG. Environmental considerations call for proper management via greener innovation as well as more climate-resilient infrastructure, with the social factor pinpointing that logistics systems always remain strong within more equitable, more educated societies, contrary to effects that generate reduced efficiency within more inequitable, poorly developed societies, with governance proving as a cornerstone much more closely associated with improved logistics performance (Mutambik,2024; Yoo,2025).

 

ESG Component	Main Relationships Identified	Direction of Effects	Interpretation
Environmental (E)	Nitrous oxide emissions rise alongside improvements in logistics performance; PM2.5 air pollution reduces logistics efficiency; exposure to extreme heat correlates positively with LPI; larger agricultural land share is associated with weaker logistics systems; greater agricultural value added improves logistics performance.	Mixed effects: both positive and negative depending on the indicator.	Logistics modernization is linked to higher emissions, showing development–pollution trade-offs; environmental degradation harms logistics, while commercialized agriculture and climate-adapted systems support efficiency.
Social (S)	Higher education levels and increased school enrollment correspond with stronger logistics performance; reductions in child labor correlate with higher LPI; income inequality negatively influences logistics performance; aging populations and gaps in basic services exert moderate negative effects.	Generally positive for education-related variables and negative for inequality or demographic strain.	Social conditions shape logistics capabilities through human capital quality, workforce stability, and access to services, but unequal or aging societies face structural barriers to logistics efficiency.
Governance (G)	Government effectiveness, rule of law, regulatory quality, and scientific research productivity all show strong positive correlations with logistics performance; better governance systems support modern customs procedures and transparent institutional environments.	Strongly positive.	Governance quality is a foundational enabler of logistics performance, reinforcing institutional stability, innovation, and efficient regulatory environments that support logistics modernization.

 

 

 

Q10. Graphic Material The graphic material in the article includes numerous charts and tables. They are carefully prepared, and I have no comments on them.

 

A10. Thanks, dear editor.

 

Q11. Suggestions for Improvement a) supplement the introduction with research justification and research gap; b) citations should be corrected to comply with the journal's requirements; c) supplement the research process diagram before the detailed description of the methodology; d) expand the policy implications to include guidelines for each of the analyzed areas; e) emphasize the research contribution to sustainable development.

 

A11. All these suggestions have been added.

 

 

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The reviewed article examines relationship between logistics performance (LPI) and disaggregated pillars of sustainable development (ESG: environmental, social, governance) using a large sample of 163 countries. Authors employed advanced research and statistical methods, including econometric instrumental variables (TSLS/G2SLS), machine learning algorithms (Random Forest, k-Nearest Neighbors) and clustering techniques. They demonstrate that although logistics development is crucial for the economy its current progress is associated with critical trade-offs such as increased pollution and the risk of deepening inequalities. They also emphasize that Governance pillar is essential for achieving synergies between logistics performance and ESG objectives.

The main strengths of the article are its high thematic originality and methodological approach. The literature review is comprehensive and up to date effectively justifying the research gap and leading to clearly defined objectives. The key contribution is the accurate identification of critical trade-offs in the results and the combined use of causal econometrics (IV-TSLS) with modern machine learning (ML) algorithms and SHAP analysis.

Identified limitations concern primarily the performance of statistical models and inconsistencies in key findings.

At first, the main causal model (TSLS regression) performs very poorly as it explains less than 1% of the variance in the Logistics Performance Index (R² < 0.0093). Therefore, I recommend to present conclusions as cautious suggestions regarding the potential direction of the effect rather than as definitive statements.

What is more, the clustering analysis (country grouping) in the environmental pillar was unsuccessful, as 91% of the countries were assigned to one very large cluster. I suggest to adjust clustering parameters to obtain meaningful and well-balanced groups.

Additionally, I recommend shortening the article to improve readability and enhance the clarity of the argumentation. In its current form it is overly lengthy and certain sections repeat the same information.

Author Response

Point to Point Answers to Reviewer 3

Q1. The reviewed article examines relationship between logistics performance (LPI) and disaggregated pillars of sustainable development (ESG: environmental, social, governance) using a large sample of 163 countries. Authors employed advanced research and statistical methods, including econometric instrumental variables (TSLS/G2SLS), machine learning algorithms (Random Forest, k-Nearest Neighbors) and clustering techniques. They demonstrate that although logistics development is crucial for the economy its current progress is associated with critical trade-offs such as increased pollution and the risk of deepening inequalities. They also emphasize that Governance pillar is essential for achieving synergies between logistics performance and ESG objectives.

A1. Thanks dear reviewer.

Q2. The main strengths of the article are its high thematic originality and methodological approach. The literature review is comprehensive and up to date effectively justifying the research gap and leading to clearly defined objectives. The key contribution is the accurate identification of critical trade-offs in the results and the combined use of causal econometrics (IV-TSLS) with modern machine learning (ML) algorithms and SHAP analysis.

A2. Thanks dear reviewer.

Q3. Identified limitations concern primarily the performance of statistical models and inconsistencies in key findings. At first, the main causal model (TSLS regression) performs very poorly as it explains less than 1% of the variance in the Logistics Performance Index (R² < 0.0093). Therefore, I recommend to present conclusions as cautious suggestions regarding the potential direction of the effect rather than as definitive statements.

A3. Also, the reported R² statistic in the TSLS and G2SLS panel models is not a valid measure of goodness of fit. In IV estimation, conventional measures of fit such as R² are not applicable because the estimated models use instrumented variables, and the within transformation changes the structure of the variance. Therefore, the most authoritative literature on econometric analysis recommends that IV models be assessed on the basis of the robustness of their coefficients, their significance, and specification testing, not through the use of R² statistics. In this case, the coefficients of the estimated models are highly robust, significantly different from zero, and their joint explanatory abilities were supported with Wald chi-squared statistics.

References: Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data. MIT press.

Q4. What is more, the clustering analysis (country grouping) in the environmental pillar was unsuccessful, as 91% of the countries were assigned to one very large cluster. I suggest to adjust clustering parameters to obtain meaningful and well-balanced groups.

A4. The clustering analysis was redone as follows:

 

 

The analysis shows that the Density-Based approach is the best technique based on metrics that are either maximized or minimized. The overall ranking score of this approach is the lowest in the comparison. This shows that this approach generally yields good results when the correct sign interpretation of each measure is considered. The approach performs best in terms of clustering separation and cohesion, and it scores highest on the Minimum Separation and Dunn Index measures. These measures indicate larger inter-cluster separation and stronger intra-cluster cohesion. The best possible results are also obtained in terms of AIC and BIC measures; both are minimized and measure model efficiency and simplicity. The results confirm that this approach obtains clustering configurations that are sound and less noisy. Other merits include the best Silhouette Score and Pearson’s correlation measure, which validate stronger intra-cluster cohesion. Despite the approach’s relatively mediocre favorableness in the Maximum Diameter measure, contributing factors do not cause significant impairment in overall effectiveness. The hierarchical algorithm ranks second overall because of excellent results in Pearson’s correlation coefficient, Silhouette measure, and R²; however, its accuracy in AIC and BIC is not as good due to relatively lower clustering separation. The Neighborhood algorithm ranks third overall in terms of the best possible values of Calinski’s index and R². The analysis of the size distribution in the produced clusters provides additional information. Density-Based clustering provides three clusters with sizes 2517, 238, and 8, together with eight noise samples. This indicates excellent performance but lacks balance in the structural composition, given the presence of the small cluster. The results are more balanced in fuzzy c-means clustering but highly unbalanced in hierarchical clustering. The outcomes of model-based clustering and k-means clustering are the most balanced and stable. This additional analysis further strengthens the preceding result that Density-Based clustering has the best structure quality, while model-based clustering and k-means clustering are recommended if balanced clustering sizes are considered.

However, since clustering is an unsupervised machine learning model, we decided to verify the composition and size of the clusters produced by the various algorithms used. The idea is to choose a machine learning algorithm capable of producing a balanced clustering. Therefore, we propose the following table:

	Size
Clustering Algorithms	Noisepoints	1	2	3	4	5	6	7	8	9	10
Density Based	8	2517	8	238
Fuzzy C-Means		206	131	215	159	29	587	519	218	636	71
Hierarchical		2319	90	61	18	11	4	8	216	22	22
Model Based		694	550	238	287	255	266	271	210
K-Means		111	290	88	217	468	60	978	222	158	179
Random Forest		1432	225	114	70	95	173	130	196	239	97

 

From the analysis carried out we must therefore reject Density Based clustering as it presents a very polarized type of clustering in which approximately 90% of the observations are present within the same cluster. In this sense the Model-Based clustering outcomes provide relevant information on the inter-structuring of the environmental (E) factors considered in this study, namely nitrous oxide emissions, PM2.5 exposure levels, heat stress levels, agricultural land use, and value added in agriculture. With regard to the objective of this clustering process—that is, assessing the levels of environmental pressures impacting logistics capacity through an ESG framework—the use of the clustering mean enables the detection of various levels of environmental and logistical profiles of differing intensities. Given that there are eight clusters overall in this analysis, there are LPI levels that are functionally close to the overall standard mean. The greatest difference arises from the factors of the Environment. This indicates a pattern in which LPI values are not indifferent or primarily or fundamentally defined by a single Environmental factor. For example, in Cluster 1, there’s a moderate emission level and PM2.5 concentration, along with a lower-than-average LPI, indicating situations in which environmental pressures may thwart logistical efficiency. For another example, in Cluster 2, there’s nearly the reverse pattern in terms of either relatively clean or more pristine environmental circumstances coupled with slightly negative LPI, emphasizing the point that in situations of relatively clean or pristine environmental circumstances, simply being environmentally clean or pristine would not be sufficient in bringing about systematic improvements in logistical efficiency. The first, or most environmentally stressful, situation would be in Cluster 3, with high exposure to PM2.5 and hot temperatures, coupled with the highest LPI scores among all other clusters. This would point to situations in which the circumstances of an industrially developed or advanced nation are reflected in substantial simultaneous exposure to environmentally stressful pressures. The remaining clusters from 4 through 8 are characterized by minimal levels of PM2.5 exposure and negligible or near-zero levels of agricultural land use, although they vary by levels of emissions and the value added by agriculture. The relatively balanced or equivalent LPI standard would tend to support the point that, in situations of minimal exposure to environmental parameters in Clusters 4 through 8, variations in logistical capacity are not appreciably affected by incremental improvements in either factor. On the whole, the Model-Based clusters tend to confirm the article's overall conclusion that the environmental factor in the ESG has a heterogeneous, nonlinear effect on the LPI. Highly developed logistics infrastructure may coexist with significant environmental degradation, whereas favorable environmental factors do not necessarily yield the best possible LPI outcomes. This tends to support the study's overall message that, under the ESG framework, not only improvements but also environmental degradation in the LPI are best tackled through intersectoral policies that address both environmental pressures and LPI infrastructure.

Cluster	LPI	NOE	PM2.5AE	HI35	ALPA	AFFVA
1	-0.045	0.621	0.304	-0.298	-0.141	-0.412
2	-0.011	-0.559	-0.646	-0.312	0.193	0.576
3	0.169	0.684	0.606	3.250	-0.033	-0.125
4	-0.002	-1.709	5.289×10-5	-0.323	-1.735×10-7	0.442
5	-0.002	-5.890×10-4	5.289×10-5	-0.303	-1.735×10-7	6.433×10-4
6	-0.002	-0.760	5.289×10-5	-0.305	-1.735×10-7	-0.119
7	-0.002	0.952	5.289×10-5	-0.291	-1.735×10-7	-0.483
8	-0.002	0.706	5.289×10-5	-0.311	-1.735×10-7	0.166

The shape of the mixing probabilities distribution sheds further light on the significance of each component in capturing the correlation between Logistics Performance Index (LPI) and the environmental (E) component of the ESG structure. The probabilities indicate the number of observations associated with each component and allow us to identify which forms of environmental-logistics correlation the majority of observations in the data are distributed across. Component 1 has a probability of 0.247 and is the most common correlation type. This corresponds to about one out of every five observations in the dataset and asserts that the environmentally-related factors and LPIs pertaining to this component capture the most common correlation pattern in the overall relationship between environmental pressures and LPI. Component 2, with probability 0.202, also shows a relatively abundant data representation and indicates the presence of another common correlation pattern in the data, through which environmental factors affect LPI. The remaining components, with probabilities of 0.075-0.104, capture less abundant data patterns and highlight more specific correlations between environmental factors and LPIs in smaller country groups. However, the existence of all components underlines that the correlation pattern between LPI’s and environmental factors is not homogenous in various contexts but depends in each particular case upon levels of release of polluting gases into the atmosphere from industry and transport, levels of air pollution caused after those releases by both gases and solid matter wastes from those releases through various meteorological factors like temperature and wetness levels, and structure parameters of agriculture in each country. The entire mixture probability distribution shows, in essence, that there are, in the overall correlation structure of the LPI-ESG-Environment model, not only the dominant but also other relatively rare patterns that are essential for covering the variability of this correlation phenomenon worldwide. This pattern further underscores the multidimensional nature of LPI-ESG-Environment correlation patterns, highlighting the overall interrelationships between environmental sustainability and LPIs.

Components	Mixing probability
Component 1	0.247
Component 2	0.202
Component 3	0.086
Component 4	0.104
Component 5	0.092
Component 6	0.096
Component 7	0.098
Component 8	0.075

The analysis of the standardised means for each component sheds further light on how different components of the environmental variables are linked to the values of the Logistics Performance Index (LPI). The most interesting result is Component 3, with the LPI well above the average. This particular component has moderate GHG emissions and PM2.5 exposure levels, high heat stress, and high weights for agricultural factors. This particular combination shows that quite high levels of logistical strength can coexist with strong levels of environmental pressure, further establishing the nonlinear relationship between environmental sustainability and logistical effectiveness. The remaining components are negative in terms of the LPI values. This shows that each component has underperformed in terms of average logistical capability. Component 1 shows relatively lower levels of GHG emissions and air pollution, along with reduced LPI levels and strong levels of agricultural factors. This further suggests that there may be restrictions on overall logistical efficiency in the particular economy, due more to structural factors than to environmental pressures. Component 2 further provides evidence that the values resulting from environmental factors are not linear. This component shows relatively higher GHG emissions alongside relatively higher levels of both temperature and humidity. The next components, 4 through 8, appear negligible in terms of both PM2.5 levels and Heat Index 35 levels, in combination with considerable levels of agricultural land use. This further indicates overall particular levels of logistical capacity in each of these economies, with relatively more homogeneous particular levels of environmental factors. This further suggests that there may not be substantial nonlinear variations in particular levels of LPI, in combination with relatively marginal variations in particular levels of individual factors.

Means	LPI	NOE	PM2.5AE	HI35	ALPA	AFFVA
Component 1	-0.298	-0.150	-0.410	0.296	0.609	-0.063
Component 2	-0.311	0.197	0.556	-0.620	-0.524	0.010
Component 3	3.250	-0.033	-0.125	0.606	0.684	0.169
Component 4	-0.323	-1.735×10-7	0.449	5.289×10-5	-1.709	-0.002
Component 5	-0.303	-1.735×10-7	6.433×10-4	5.289×10-5	-5.890×10-4	-0.002
Component 6	-0.305	-1.735×10-7	-0.126	5.289×10-5	-0.748	-0.002
Component 7	-0.291	-1.735×10-7	-0.483	5.289×10-5	0.952	-0.002
Component 8	-0.311	-1.735×10-7	0.163	5.289×10-5	0.710	-0.002

The figure provides a summary of the model-based clustering outcomes with respect to both identifying the number of clusters and visualizing the data structure in the data space. The figure - Panel A illustrates the change in Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) scores and Within-Cluster Sum of Squares (WSS) values with respect to the increase in the number of clusters. The red spot in the figure marks the point with the lowest BIC score; this indicates that the best compromise between data fit and model complexity is achieved with the model including eight clusters. This indicates that this number of factors provides sufficient information about the data structure pattern without sacrificing accuracy through overfitting or underfitting. The plot of the WSS indicates a gradual reduction in values until reaching the point corresponding to the medium number of clusters, beyond which the values vary, though in a smooth pattern. This pattern corresponds to data with a complex structure that cannot be adequately represented by a few clusters. The figure - Panel B provides information regarding the pattern of the eight clusters in the projected feature space. The clusters are well differentiated in this representation and express complex configurations in most cases. This color representation indicates that each component differs in predetermined regions of the data point space. This indicator also shows regions of differentiation in most components, with some appearing sparser than others. This provides information regarding the complexity of the patterns of various environmental and logistical factors represented through the model. The structure that emerges from this representation provides information on the effectiveness of the clustering analysis through the model's grouping of profiles in conformity with the analysis's objectives.

The figure depicts the standardised means for each variable across the eight clusters formed using the model-based approach, thereby clarifying the distinct environmental-logistical configurations within each grouping. The LPI values show small variation, indicating minimal differences among the clusters in LPI. However, environmental factors exhibit significant variation. The nitrous oxide emissions (NOE) and PM2.5 exposure values show both positive and negative standardisations, indicating clusters with high levels of pollution and others that are relatively clean, based on environmental conditions. The Heat Index (HI35) showed the highest discrimination value, with one cluster recording a substantially high value, indicating higher heat stress in this grouping than in others. The other clusters recorded values close to the overall average. The Agricultural land share (ALPA) and Agricultural value added (AFFVA) recorded moderate discrimination values. The figure illustrates that overall group formation is driven by environmental factors rather than LPI values, indicating that the majority of the model's variation is attributable to these factors.

The figure summarizes a pairwise scatter plot matrix of six standardized variables across eight clusters derived from a model-based clustering analysis. Each subplot presents the relationship between two variables using colored ellipses, noting the probability distribution of each component. The combined results indicate that environmental factors such as NOE, PM2.5 exposure levels, and HI35 are the key determinants in distinguishing the various clusters. Clusters are arranged in well-defined regions based on these factors, especially in the NOE and PM2.5 exposure level plots, where clear separations occur between high and low emission factors. However, the most significant determinant in this analysis is the HI35 variable, which shows one component with abnormally high levels of heat stress compared to other factors. Conversely, LPI shows minimal variation between factors and groups, yet remains clustered around the standard mean in terms of standardization. The result indicates that variations in logistical performance do not significantly account for the observed pattern in the data and supports the primary idea of relying on the environmental factors outlined in this analysis. The agricultural factors of ALPA and AFFVA provide more information, though they exhibit less-defined edges in the figure, suggesting minute variations in land use or in the overall economic role of agriculture. This analysis indicates that the figure presents a complex clustering pattern across multiple dimensions, in which environmental factors account for overall variation, except for LPI, which is significant yet supplementary in the overall context of Environmental Social Governance.

Q5. Additionally, I recommend shortening the article to improve readability and enhance the clarity of the argumentation. In its current form it is overly lengthy and certain sections repeat the same information.

A5. The text was reduced from 31,986 words to 28,106 words, or -3,880, equivalent to -12.13%.  However, it should also be considered that there are approximately 25 pages of appendices which are not part of the main text. Furthermore, there are also 23 tables and 9 figures, as well as 127 bibliographical references.

 

 

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Dear Authors, thank you for your detailed and comprehensive response to the review, as well as for the revisions you've implemented, which significantly enhance the methodological quality of the article. I accept your explanation regarding the inadequacy of the standard R2 in TSLS/IV models as entirely valid and professional. Furthermore, the corrected cluster analysis, employing the Model-Based Clustering method, has successfully yielded satisfactory, well-balanced, and interpretable environmental-logistical profiles. I also appreciate the substantial shortening of the main text. All critical comments have been addressed reliably and correctly.