Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience

Ximena Morales-Urrutia; David Vega-Camacho; Aracelly Núñez-Naranjo; Patricio Lara-Álvarez; Mery Ruiz-Guajala; Patricia Acosta-Vargas

doi:10.12688/f1000research.176930.2

Home Browse Environmental Degradation, Life Expectancy, and Institutional Quality:...

ALL Metrics

Views

Downloads

Get PDF

Get XML

Export

▬

✚

Research Article

Revised

Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience

[version 2; peer review: 1 approved with reservations]

Ximena Morales-Urrutia¹, David Vega-Camacho¹, Aracelly Núñez-Naranjo², Patricio Lara-Álvarez³, Mery Ruiz-Guajala¹, Patricia Acosta-Vargas ⁴

Ximena Morales-Urrutia¹, David Vega-Camacho¹, [...] Aracelly Núñez-Naranjo², Patricio Lara-Álvarez³, Mery Ruiz-Guajala¹, Patricia Acosta-Vargas ⁴

PUBLISHED 18 May 2026

Author details Author details

¹ Facultad de Contabilidad y Auditoría, Universidad Técnica de Ambato, Ambato, 180206, Ecuador
² Centro de Investigacion de Ciencias Humanas y de la Educación - CICHE, Facultad de Ciencias de la Educación, Universidad Tecnológica Indoamérica, Ambato, 180103, Ecuador
³ Facultad de Ingenierías, Universidad Tecnologica Indoamerica, Ambato, 180103, Ecuador
⁴ Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

Ximena Morales-Urrutia
Roles: Conceptualization, Data Curation, Funding Acquisition, Investigation, Methodology, Project Administration, Validation, Visualization, Writing – Original Draft Preparation

David Vega-Camacho
Roles: Investigation, Software, Writing – Review & Editing

Aracelly Núñez-Naranjo
Roles: Conceptualization, Project Administration, Supervision, Validation, Writing – Review & Editing

Patricio Lara-Álvarez
Roles: Conceptualization, Methodology

Mery Ruiz-Guajala
Roles: Data Curation, Software, Writing – Original Draft Preparation

Patricia Acosta-Vargas
Roles: Funding Acquisition, Resources, Writing – Original Draft Preparation, Writing – Review & Editing

OPEN PEER REVIEW

REVIEWER STATUS

This article is included in the Climate gateway.

This article is included in the Public Health and Environmental Health collection.

Abstract

Background

The relationship between the environment and public health has become an increasing concern on the international agenda, particularly in the European context, where the energy transition seeks to reconcile sustainable development with social well-being. This study examines how environmental degradation affects life expectancy in 20 European Union countries between 2000 and 2020, aiming to provide empirical evidence on the differentiated effects of pollutant emissions and energy sources across income levels.

Methods

A dynamic panel data model estimated using the Generalized Method of Moments (GMM) is employed, allowing for the control of endogeneity, unobserved heterogeneity, and autocorrelation.

Results

In middle-income countries, CO₂ emissions are negatively and significantly associated with life expectancy, reflecting potential institutional limitations in managing environmental risks Conversely, the use of renewable energy sources is positively associated with life expectancy in middle-income countries,, while the effect is not statistically significant in high-income countries, suggesting context-dependent health co-benefits derived from the energy transition.

Conclusions

In contrast, in high-income countries, a paradoxical relationship emerges between fossil fuel consumption and higher life expectancy, which can be attributed to the robustness of their welfare systems and environmental regulations.

Keywords

CO₂ emissions; fossil fuel consumption; renewable energy; ecological transitions; welfare systems; environmental governance

Corresponding author: Patricia Acosta-Vargas

Competing interests: No competing interests were disclosed.

Grant information: This research was funded by Universidad de Las Américas—Ecuador as part of the internal research project 518.A.XV.24.

Copyright: © 2026 Morales-Urrutia X et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

How to cite: Morales-Urrutia X, Vega-Camacho D, Núñez-Naranjo A et al. Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience [version 2; peer review: 1 approved with reservations]. F1000Research 2026, 15:435 (https://doi.org/10.12688/f1000research.176930.2) First published: 25 Mar 2026, 15:435 (https://doi.org/10.12688/f1000research.176930.1) Latest published: 18 May 2026, 15:435 (https://doi.org/10.12688/f1000research.176930.2)

Revised Amendments from Version 1

First, the theoretical framework was strengthened, moving from a general formulation of the institutionalist approach toward a more explicit integration of the concept of the “ecological welfare state.” This revision includes a more in-depth discussion of institutions as contextual factors rather than merely descriptive elements, thereby enhancing conceptual coherence and addressing previous concerns about limited theoretical articulation.
Second, the methodological section was expanded and clarified. The revised version incorporates key justifications, including the use of the Generalized Method of Moments (GMM), control of endogeneity, careful instrument selection to avoid overfitting, explanation of Variance Inflation Factor (VIF < 2), and discussion of stationarity using the Im–Pesaran–Shin (IPS) test. Additionally, limitations such as omitted variable bias and cross-sectional dependence are explicitly acknowledged, improving overall scientific transparency.
Third, the interpretation of variables was refined, particularly concerning CO₂ emissions. While the initial version implied more direct effects, the revised manuscript appropriately emphasizes their indirect impact through climate change, achieving better alignment with the existing literature.
Furthermore, the results were more carefully nuanced. Strong causal claims are avoided, and appropriate caveats are introduced regarding heterogeneity, average effects, and potential model limitations (e.g., non-significant AR(1)).
Finally, the academic writing and internal consistency were enhanced, with improved alignment between hypotheses, variables, and results, as well as greater precision in terminology.

See the authors' detailed response to the review by Matheus Koengkan

Introduction

Environmental degradation is a multidimensional and progressive process characterized by the deterioration of natural ecosystems, primarily driven by anthropogenic activities that exceed the resilience thresholds of ecological systems. As explained by Long and Steel,¹ this phenomenon weakens the structural and functional integrity of ecosystems, accelerates biodiversity loss, and disrupts essential ecological dynamics. The overexploitation of natural resources—exacerbated by agricultural expansion, unplanned urbanization, and pollution—has surpassed the carrying capacity of many ecosystems, diminishing their ability to provide essential services.^2,3

Climate change further intensifies this trend through rising temperatures, increased climatic variability, and sea-level rise, amplifying the negative socioeconomic impacts in vulnerable regions.⁴

From an institutionalist perspective, these environmental risks cannot be detached from the design and effectiveness of formal norms, regulatory agencies, and governance structures. As noted by,^5,6 institutions play a decisive role in shaping policy responses and determining how societies internalize environmental externalities. Thus, environmental degradation represents not only a biophysical process but also an institutional failure that perpetuates poverty, territorial exclusion, and socio-environmental vulnerability.⁷

In the 21st century, life expectancy must be interpreted as an outcome conditioned by both environmental and institutional determinants of well-being. Environmental degradation—through biodiversity loss, pollution, and resource depletion—directly undermines the conditions required for a healthy and dignified life, particularly among vulnerable populations.

As argued by,^8,9 sustainable well-being requires acknowledging the intrinsic link between quality of life and ecological integrity. Building on this, institutionalist theory posits that the design and strength of public institutions shape and condition the effects of environmental degradation on social outcomes, including health,^9,10 rather than acting as directly measured mediating mechanisms within the empirical model.

Accordingly, life expectancy should be linked to the institutional capacity to integrate environmental justice, territorial resilience, and intergenerational equity into public policy. However, in this study, such institutional dimensions are not directly measured but are considered as contextual factors shaping the observed relationships.^11,12

Consequently, the welfare state must transcend its traditional redistributive focus and incorporate ecological risks as structural components of long-term social policy.¹³

Empirical evidence confirms that air pollution, extreme weather events, and climate change exacerbate public health crises, disproportionately affecting low-income populations, older people, and children.^13–15

These environmental stressors not only reflect ecological imbalances but also reveal institutional weaknesses, especially when welfare systems lack adaptability and robustness to mitigate their effects.^16,17

In rural and agricultural regions, phenomena such as food and water insecurity—stemming from ecological degradation—undermine livelihoods and access to essential services.¹⁸

As highlighted in institutionalist studies, these outcomes are not inevitable but depend on the strength and coherence of governance systems. In this context, state intervention must evolve from reactive mechanisms toward proactive strategies, including climate-resilient infrastructure, integrated planning, and regulatory reforms that prioritize social and environmental justice.^19,20

This reorientation entails developing an “ecological welfare state” capable of addressing the overlapping demands of sustainability, equity, and resilience through institutional innovation and multilevel coordination.^9,10

Rethinking life expectancy within the ecological crisis framework requires reconfiguring its institutional foundations to embed the governance of environmental commons. This case includes adopting participatory governance tools, environmental impact assessments, and socio-ecological planning instruments that strengthen co-responsibility and legitimacy in public investment. Regulatory frameworks must reflect the interdependence between environmental integrity and social rights, moving beyond fragmented policy responses.

As emphasized by,⁵ institutions shape the incentives, constraints, and capacities of actors to respond effectively to complex challenges. Therefore, integrating scientific and local knowledge, ensuring equitable access to resources, and reinforcing institutional accountability can enhance the transformative potential of welfare regimes. In sum, reinforcing life expectancy as a central indicator of sustainable development requires embedding ecological dimensions as both preconditions and drivers of equitable and resilient societies.

Although previous studies have documented the effects of air pollution, CO₂ emissions, and fossil fuel use on life expectancy,^21,22 few have analyzed these relationships within a normative and institutional framework that incorporates the logic of the welfare state. This study seeks to bridge that gap by integrating environmental degradation variables—such as CO₂ emissions and fossil fuel dependency—into a dynamic empirical model of life expectancy, framed within the concept of an “ecological welfare state”.^9,10,23 In doing so, it aligns environmental outcomes not only with public health but also with the structural role of state intervention in providing collective goods, regulating externalities, and mitigating ecological risk through formal institutions.

Overview

The relationship between environmental degradation and life expectancy has attracted growing attention over recent decades, particularly in light of global climate change and its impact on public health. Foundational studies have linked ecological deterioration—including pollution, biodiversity loss, and climate variability—to adverse health effects, especially among vulnerable populations.^1–3

These findings highlight how environmental stressors disrupt ecosystem services essential to human well-being, such as access to clean air and water.

While numerous empirical studies have documented the adverse effects of CO₂ emissions and fossil fuel consumption on life expectancy,^22,24 and others have emphasized the importance of access to safe drinking water,^25,26 most have treated these variables in isolation, without situating them in an institutional context. This study addresses that gap by applying institutionalist theory, which emphasizes that development outcomes are shaped not only by economic or environmental factors but also by the quality, design, and responsiveness of institutions.^27,28

Under this framework, we argue that environmental degradation affects life expectancy through both direct and institutionally mediated channels. Strong institutions—such as comprehensive welfare states, effective environmental agencies, and inclusive infrastructure policies—can mitigate the harmful effects of pollution and ecological risks by providing essential services and regulatory safeguards.^9,10

Conversely, weak institutions exacerbate vulnerability by failing to internalize environmental costs and ensure equitable service provision.

This study proposes two key hypotheses:

(1) Higher CO₂ emissions and fossil fuel consumption reduce life expectancy by exposing populations to greater environmental risks. However, the magnitude and direction of these effects are expected to vary across countries with different institutional and socioeconomic contexts.
(2) The use of renewable energy sources improves life expectancy by strengthening resilience and reducing ecological pressures.

To empirically test these hypotheses, a dynamic panel data model is estimated using the Generalized Method of Moments (GMM) for 20 European Union countries over the period 2000–2020, capturing both temporal dynamics and unobserved heterogeneity.^21,29

Our contribution lies in integrating environmental and institutional variables within a theoretical model of life expectancy, offering a broader understanding of how public health outcomes are shaped by ecological risks and institutional capacity in a comparative European context.

Life expectancy

According to,²⁴ life expectancy refers to the average age an individual is expected to reach under specific social and health conditions. However, this measure goes beyond a mere demographic estimate.

As noted by,³⁰ life expectancy reflects not only the general health status of a population but also the cumulative effects of socioeconomic and environmental conditions, as well as effective access to essential services such as safe water, education, and healthcare.

In this sense,²² highlights that environmental degradation—measured through CO₂ emissions—has a significant negative impact on life expectancy. In other words, countries with higher levels of environmental pollution tend to experience a greater decline in population well-being.

Analyzing environmental degradation requires incorporating indicators that capture both direct environmental pressures and their collateral impacts on quality of life. In this study, five representative variables are selected: carbon dioxide (CO₂) emissions, fossil fuel consumption, and the share of renewable energy in the energy mix. Institutional structures play a critical role in mediating these relationships, as they determine the efficiency, equity, and responsiveness of public health and environmental policies influencing longevity.²⁷

CO₂ emissions

CO₂ emissions—recognized as a primary contributor to global greenhouse gases—originate mainly from the combustion of fossil fuels in electricity generation, transportation, and industrial production. However, CO₂ should not be considered a local air pollutant with direct effects on human health. Instead, its relevance lies in its contribution to climate change, which may indirectly affect health outcomes through mechanisms such as rising temperatures, increased climate variability, and broader environmental degradation. In Europe, despite progress in decarbonization, the correlation between economic prosperity and emissions remains evident, particularly in high-income countries where per capita energy consumption is high.

According to,³¹ although this relationship is more pronounced in Latin America, European countries are not exempt, as their industrial legacy, high mobility patterns, and energy-intensive lifestyles continue to pose environmental challenges.

Reference 32 further notes that in the European context, sectors such as transportation and manufacturing remain significant sources of emissions, driven by urbanization and the persistent dependence on fossil-based energy systems.

However, unlike many regions, Europe has made substantial progress in diversifying its energy matrix, integrating renewable sources such as wind, solar, and hydropower. These measures—alongside energy efficiency policies and clean technologies—are central to the European Green Deal’s ambition to achieve climate neutrality by 2050.^33,34

The effectiveness of these mitigation strategies largely depends on institutional governance, which influences both regulatory enforcement and technological diffusion.^25,35

Fossil fuel use

Fossil fuel use remains one of the main drivers of environmental degradation, as it constitutes the primary source of CO₂ emissions, one of the most significant greenhouse gases contributing to climate change.^26,36

Despite being a non-renewable, finite resource, its exploitation remains intensive globally.

According to,³⁶ even if humanity were to exhaust all existing reserves—a process that could take several centuries—the climatic consequences for the planet’s balance would be catastrophic. Therefore, relying on the natural depletion of fossil fuels cannot be considered a viable or responsible strategy in the face of the current climate crisis.

Today, key sectors such as industry, transportation, and electricity generation remain heavily dependent on coal, oil, and natural gas. This dependence intensifies the emission of pollutants that, once released into the atmosphere, can travel long distances, producing transboundary effects on air quality, public health, and global climate stability.^21,37

Institutional capacity, including the ability to regulate energy transitions and enforce environmental standards, remains a central variable in managing fossil fuel dependence.

Renewable energy use

The deployment of renewable energy is widely recognized as a key factor and strategic opportunity to mitigate the adverse effects of climate change and advance toward a sustainable development model.

According to,³⁸ urban areas account for more than 70% of global greenhouse gas (GHG) emissions, underscoring the urgency of transitioning to clean, sustainable energy sources.

Among the most viable alternatives are solar, wind, geothermal, and hydropower, which not only substantially reduce emissions but also promote energy diversification and decrease dependence on fossil fuels. This transition further generates co-benefits, including improved air quality, growth in local green economies, and enhanced energy resilience against future shocks.^33,38

Promoting these technologies—supported by coherent public policies and green financing mechanisms—is essential to meeting the climate targets set under the Paris Agreement. The institutional environment—through regulatory stability, investment incentives, and public–private partnerships—conditions the pace and inclusivity of this energy transformation.

When examining the relationship between life expectancy and environmental degradation, it is crucial to recognize that the issue transcends biophysical factors such as deforestation, rising global temperatures, or greenhouse gas emissions. What is ultimately at stake is human life, public health, and the quality of life of millions of people. Each environmental indicator reflects deeply rooted social and institutional dimensions, as ecological deterioration disproportionately affects vulnerable populations through systemic inequalities and governance gaps.

From an institutionalist perspective, these outcomes underscore the state’s central role as a regulator of environmental risks, an active provider of collective goods, and a redistributor of environmental protections. As emphasized by,⁹ integrating ecological sustainability into the design of life expectancy metrics requires acknowledging the interdependence between social systems and environmental governance.

In this context, well-being should not be defined solely in terms of income or service access, but also by the institutional capacity to ensure environmental conditions that sustain life. Understanding how ecological factors influence well-being thus becomes an urgent institutional challenge, not only for academic research but also for the design of resilient public policies grounded in equity, environmental justice, and long-term social protection frameworks.³⁸

Materials and methods

Gross Domestic Product (GDP) per capita was transformed using natural logarithms because its values were considerably higher than those of the other variables. This transformation helped stabilize the variance, reduce skewness, and ensure that the data could be processed and analyzed on an equal basis with the other indicators included in the econometric model.

The objective of this study is to analyze the behavior of the factors contributing to environmental degradation and their impact on life expectancy in European Union member countries. To this end, a quantitative research approach was employed, using panel-data regression models that capture both temporal and cross-sectional variability throughout the study period.

The data were obtained from the Our World in Data platform (https://ourworldindata.org/data), which compiles information from multiple international organizations — including the World Bank, OECD, and leading academic institutions. The data were downloaded in CSV format and subsequently filtered for analysis. The platform provides environmental, social, and public health statistics for EU countries.

The study sample comprises the following 20 countries: Austria, Belgium, Bulgaria, Croatia, Czechia, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, the Netherlands, Poland, Portugal, Romania, Spain, and Sweden.

To ensure a more representative analysis of Europe’s economic and environmental realities, countries were classified into two groups based on their average GDP per capita over the study period (2000–2020). Specifically, countries with higher average income levels were categorized as high-income, while those with relatively lower average income levels were classified as middle-income. This classification follows a relative grouping approach within the sample, allowing for the identification of structural differences in economic capacity, institutional context, and environmental dynamics.

Although this classification does not strictly follow external thresholds such as those defined by the World Bank, it provides a consistent and internally comparable framework for analyzing heterogeneity across countries. However, this approach may introduce some limitations related to sample-specific grouping, which should be considered when interpreting the results.

• The high-income group includes: Luxembourg, Ireland, Denmark, Sweden, the Netherlands, Austria, Finland, Germany, Belgium, and France.
• The middle-income group includes: Italy, Spain, Czechia, Portugal, Greece, Hungary, Poland, Croatia, Romania, and Bulgaria.

This classification enables the exploration of structural differences and distinct patterns in the relationship between environmental degradation and life expectancy. The analysis period covers 2000 to 2020.

Although institutional quality constitutes a central element of the theoretical framework, it is not explicitly included as a separate variable in the econometric specification. Instead, institutional differences are indirectly captured through the classification of countries into high- and middle-income groups, which reflect structural variations in governance capacity, regulatory frameworks, and welfare systems. This approach allows for the exploration of heterogeneous effects across institutional contexts, although it does not permit direct testing of institutional mediation.

In line with institutional theory, which emphasizes the role of governance structures, public policies, and institutional conditions in shaping social outcomes (6,27), this study integrates environmental and infrastructure indicators as reflections of the state’s capacity to regulate, provide, and adapt to ecological challenges ( Table 1). Further information is available in the Mendeley open repository.³⁹

Table 1. Variables in the model.

Type of variable	Variable	Conceptual definition	Unit of measurement	Source
Dependent	Life Expectancy	A summary measure of population health that estimates the average number of years a person is expected to live based on current mortality rates.⁴⁰	Years	Our World in Data
Independent	CO₂ Emissions	Amount of carbon dioxide released into the atmosphere, a process in which gases absorb part of the sun’s energy, thereby causing an increase in temperature.⁴¹	Metric tons of CO₂ per person per year	Our World in Data
Independent	Fossil Fuel Consumption	Primary source of air pollution and greenhouse gas emissions; there are three primary types of fossil fuels: coal, oil, and natural gas.⁴²	Percentage (%) of total energy consumption	Our World in Data
Independent	Use of Renewable Energy as a Primary Energy Source	Natural processes that can be repeated and allow the generation of clean and inexhaustible energy.⁴³	Percentage (%) of total energy consumption	Our World in Data
Independent	GDP per capita at constant prices (log-transformed)	Market value of the final goods and services produced in an economy during a given period of time.⁴⁴	U.S. dollars (log)	World Bank

1 Data sources: Our World in Data (Life Expectancy, CO₂ Emissions, Fossil Fuel Consumption, Renewable Energy Use); World Bank (GDP per capita at constant prices).

While the model includes key environmental and economic variables, it does not incorporate additional controls such as healthcare expenditure, education, urbanization, or demographic structure. This choice reflects a parsimonious specification based on data availability and comparability across countries and over time. However, the exclusion of these variables may introduce omitted variable bias, as life expectancy is influenced by multiple socio-economic and demographic factors. Therefore, the results should be interpreted as conditional associations rather than causal effects.

CO₂ emissions and fossil fuel consumption were included because they play complementary, non-redundant roles from both theoretical and empirical perspectives. Although fossil fuel combustion is the primary anthropogenic source of CO₂ emissions,⁴⁵ the impact of each variable on public health and life expectancy may operate through distinct mechanisms.

• Fossil fuel consumption reflects a country’s energy structure and dependency, as well as its energy intensity and the effectiveness of transition policies.³⁸
• CO₂ emissions, in contrast, represent a cumulative environmental outcome associated with climate change processes, which may indirectly affect human health through environmental and climatic pathways rather than direct exposure effects, particularly via mechanisms such as temperature increases and environmental degradation.^24,38

The logarithm of GDP per capita was used because its raw values were substantially higher than those of the other variables. This adjustment stabilizes variance, reduces skewness, and enables comparability across all indicators in the model.

The literature treats these two environmental dimensions—energy use and emissions—as distinct, justifying their simultaneous inclusion in multivariate econometric models. For example,^21,22 note that although emissions stem from fossil energy use, their behavior may vary depending on energy efficiency, fuel type (coal, gas, oil), and the adoption of mitigation technologies such as carbon capture and storage.

Empirically, this study confirms that collinearity between these variables does not compromise model robustness. The Variance Inflation Factor (VIF) values remain well below the critical threshold (VIF < 2), validating their simultaneous inclusion. Moreover, considering both variables enhances explanatory power by distinguishing between sources of environmental stress (energy use) and their direct environmental effects (emissions) on life expectancy—thus avoiding the omission of key analytical dimensions.

Econometric model specification

To formally specify the econometric structure, a dynamic panel data model estimated through the Generalized Method of Moments (GMM) was adopted, following.⁴⁶ This methodology controls for unobserved heterogeneity, potential endogeneity among explanatory variables, and autocorrelation in the error terms—issues commonly found in macro-level longitudinal data.

The basic form of the model is:

{Δ LE}_{it} = {α Δ LE}_{i (t - 1)} + β_{1} {Δ FF}_{i (t - 1)} + β_{2} ΔCO 2_{i (t - 1)} + β_{3} {ΔRE}_{i (t - 1)} + β_{4} {ΔlnGDP}_{i (t - 1)} + ε_{it}

Where:

• ${Δ LE}_{it}$ : Change in life expectancy for country iii at time ttt (dependent variable).
• ${Δ LE}_{i (t - 1)}$ : Lagged change in life expectancy, capturing the dynamic nature of the process.
• ${Δ FF}_{i (t - 1)}$ : Lagged change in fossil fuel consumption (% of energy use).
• $ΔCO 2_{i (t - 1)}$ : Lagged change in CO₂ emissions (metric tons per capita).
• ${ΔRE}_{i (t - 1)}$ : Lagged change in the share of renewable energy in total energy supply.
• $Δln {GDP}_{i (t - 1)}$ : Lagged change in natural logarithm GDP per capita (current USD), included as a control variable.
• ε it: Error term, assumed to be uncorrelated with the instruments used.

In the GMM estimation, instruments were constructed using lagged values of the endogenous variables, following the standard Arellano–Bond approach. To avoid overfitting and instrument proliferation, the number of instruments was deliberately restricted by limiting the lag depth and collapsing the instrument matrix. This procedure ensures a parsimonious specification while maintaining the validity of the instruments. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and helps improve the robustness and replicability of the estimates.

The decision to transform the variables into first differences is primarily guided by the IPS unit root test results, which suggest non-stationarity in levels and stationarity after differencing. However, it is important to recognize that this decision is not based solely on the IPS outcomes. Given the dynamic nature of the model and the need to avoid spurious regression results, differencing is also consistent with standard practices in GMM estimation when variables are integrated of order one.

At the same time, it is acknowledged that the IPS test assumes cross-sectional independence, which may not fully hold in the context of European Union countries due to their economic and environmental interconnections. Therefore, while the transformation provides a consistent estimation framework, the potential presence of cross-sectional dependence represents a limitation. Future research could address this issue by applying second-generation panel unit root tests that explicitly account for such dependencies.

From an institutionalist perspective, this model is particularly well-suited to capturing the cumulative effects of public policies over time, reflecting how institutional stability, regulatory frameworks, and welfare-state interventions shape the capacity to respond to environmental challenges. The use of a dynamic specification, incorporating lagged effects and infrastructure variables, enables testing for persistent features of institutional performance.

This specification aligns with established empirical practices in environmental economics and public health, where dynamic relationships and persistence effects are common. By integrating environmental, infrastructural, and economic indicators within a dynamic and institutionally informed framework, the model provides robust empirical evidence on the determinants of life expectancy in the European Union over the period 2000–2020.

Results

Figure 1 illustrates the temporal evolution between 2000 and 2020. The dependent variable in the analysis is life expectancy, while the independent variables include fossil fuel consumption, CO₂ emissions, renewable energy use, and GDP per capita.

Figure 1. Life expectancy vs. environmental degradation factors in high-income European Union countries.

The charts for high-income European countries present key trends in the relationships among environmental, economic, and health factors.

First, life expectancy remains relatively stable across all countries, with marginal but consistent increases during the 2000–2020 period. This study reflects the consolidation of strong health and welfare systems—characteristic of developed economies—where improvements in longevity tend to be sustained over time.

Second, fossil fuel use is declining in several countries, particularly Denmark, Sweden, and Germany, indicating progress in the energy transition. However, in countries such as Luxembourg and Ireland, fossil fuel consumption remains comparatively high, suggesting that energy dependence remains a structural challenge.

Similarly, renewable energy use is on the rise in most countries, particularly in Denmark, Sweden, and Germany, highlighting the impact of environmental policies on the energy matrix. Nonetheless, this growth has not yet been sufficient to replace dependence on fossil fuels entirely.

Finally, CO₂ emissions show relative stability or slight reductions, consistent with the gradual adoption of sustainability-oriented policies. Overall, the results indicate that although the energy transition is advancing in these high-income countries, reconciling economic growth (lnGDP per capita) with environmental sustainability remains a persistent challenge—particularly in nations with greater reliance on fossil energy.

The descriptive results presented in Table 2 provide an overview of the behavior of the variables analyzed in high-income European countries.

Table 2. Descriptive statistics — High-income European Union countries.

Variable	Mean	Std. Dev.	Min	Max
Life expectancy	80.26233	1.428743	76.54	83.05
Use of fossil fuels	74.07048	19.82745	27.14929	99.08259
CO2 emissions	9.825901	4.247067	3.544894	25.95329
Renewable energies	18.13581	14.85103	1.28	57.83
lnGDP per capita	10.77718	0.2969774	10.41196	11.62998

First, life expectancy averages 80.26 years, with low variability (standard deviation of 1.42) and a range of 76.54 to 83.05 years. These values indicate relatively high and homogeneous levels of longevity across the countries analyzed.

Regarding fossil fuel use, the mean is 74.07% of the energy matrix, with considerable dispersion (standard deviation of 19.82) and values ranging from 27.14% to 99.08%, suggesting substantial differences in energy dependence across countries.

CO₂ emissions average 9.82 metric tons per capita, ranging from 3.54 to 25.95, indicating heterogeneity in environmental pressure among countries.

Renewable energy use averages 18.13%, with high variability (standard deviation of 14.85%) and values ranging from 1.28% to 57.83%, reflecting uneven adoption of renewable energy sources.

Finally, GDP per capita (log-transformed) averages 10.77, with low dispersion (0.29) and a range of 10.41 to 11.62 indicating relatively similar income levels within this group.

The analysis of medium-income European Union countries reveals consistent patterns in the relationship between environmental and economic variables and life expectancy. In general, life expectancy maintains a relatively stable upward trajectory, though at a more moderate slope than in high-income countries.

Fossil fuel consumption shows a gradual decline in most cases, reflecting ongoing energy transition processes, although levels remain comparatively high in countries such as Romania, Bulgaria, and Greece.

In contrast, renewable energy use is growing notably, particularly in Portugal, Romania, and Bulgaria, where expansion is more pronounced. This increase reflects policy-driven transitions toward sustainable energy sources.

However, CO₂ emissions show slight but persistent increases in several countries, underscoring the challenges of decoupling economic growth from environmental pressures.

Finally, GDP per capita shows moderate growth, although disparities among countries remain evident, with southern and eastern European states showing slower progress.

These findings emphasize that institutional capacity and public policy frameworks play a decisive role in mediating the impact of environmental and economic changes on health outcomes ( Figure 2).

Figure 2. Life expectancy vs. explanatory variables in medium-income European Union countries.

The graphical evidence presented in Figures 1 and 2 suggests the presence of heterogeneous trajectories across countries, particularly in the evolution of environmental and economic variables. While the econometric model addresses part of this heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. This simplifying assumption facilitates estimation but may not fully capture country-specific dynamics. Therefore, the results should be interpreted as average effects within groups rather than country-specific relationships.

The results presented in Table 3 summarize the descriptive statistics for medium-income European Union countries.

Table 3. Descriptive statistics — Medium-income European Union countries.

Variable	Mean	Std. Dev.	Min	Max
Life expectancy	77.45319	3.344799	70.8189	83.4794
Use of fossil fuels	82.77991	8.220925	66.84081	99.32454
CO2 emissions	6.803062	2.164544	3.800979	12.50236
Renewable energies	15.70622	7.374068	4.63	33.61
lnGDP per capita	9.619094	0.527512	8.220922	10.43881

Life expectancy averages 77.45 years, with a standard deviation of 3.34 and a range of 70.81 to 83.47 years indicating moderate variability across countries.

CO₂ emissions average 6.80 metric tons per capita (standard deviation of 2.16), ranging from 3.80 to 12.50, indicating variation in environmental pressure levels.

Renewable energy use averages 15.70% (standard deviation of 7.37), with values ranging from 4.63% to 33.61, suggesting differences in the adoption of renewable energy sources.

GDP per capita (log) averages 9.62, with a standard deviation of 0.52 and a range of 8.22 to 10.43, reflecting variation in economic performance within the group.

The following sections present a comparative analysis of the relationship between environmental degradation and life expectancy, differentiating the results for high- and medium-income countries. This segmentation enables the identification of specific patterns and the contrast of how environmental variables affect population health across different institutional and socioeconomic contexts.

For high-income European countries Table 4 reports the results of the Im–Pesaran–Shin (IPS) unit root test used to examine the stationarity properties of the panel data. This test evaluates the null hypothesis that each series contains a unit root against the alternative of stationarity.

Table 4. Im–Pesaran–Shin unit root test — High-income European countries.

	Level	First difference
W-t-bar	P value	P value
Life expectancy	1.0000	0.0000
Use of fossil fuels	0.9577	0.0000
CO2 emissions	0.3164	0.0002
Use of renewable energy as a primary source of energy	0.8639	0.0024
lnGDP per capita	1.00	0.0083

The results indicate that, in levels, none of the variables exhibit statistical evidence of stationarity, as all p-values exceed the 0.05 threshold. For instance, life expectancy presents a p-value of 1.0000, while CO₂ emissions and fossil fuel consumption show p-values of 0.3164 and 0.9577, respectively, implying that the null hypothesis of non-stationarity cannot be rejected.

When the variables are transformed into first differences, the test results change substantially, with all series displaying statistically significant p-values (p < 0.05). This suggests that the variables become stationary after differencing, indicating that they are most likely integrated of order one. Such properties justify their inclusion in dynamic panel models such as GMM.

Nevertheless, it is important to consider that the IPS test relies on the assumption of cross-sectional independence and may be affected by heterogeneity across panel units. In the European context, where countries are economically and environmentally interconnected, cross-sectional dependence is a plausible concern. Consequently, the stationarity results should be interpreted with caution, as they may not fully reflect the complexity of the underlying panel structure.

The Arellano–Bond,⁴⁷ test results for high-income European countries indicate no statistically significant autocorrelation in the first-differenced residuals at either AR(1) or AR(2) levels. While the absence of AR(2) autocorrelation is consistent with the validity of the model, the lack of significant AR(1) autocorrelation is less typical in first-differenced specifications and should be interpreted with caution. This outcome may reflect sample-specific characteristics or limitations in the model specification and therefore does not provide strong confirmatory evidence regarding the dynamic structure of the model.

This result is relevant because it ensures that the residuals do not exhibit short- or medium-term serial correlation—a necessary condition for the validity of GMM estimators. Consequently, it confirms that the instruments used are appropriate and that the dynamic model is correctly specified, providing robustness and consistency to the results obtained ( Table 5).

Table 5. Arellano–Bond test for absence of autocorrelation in first-difference errors — High-income European countries.

Order	z	Prob > z
1	−0.88	0.378
2	0.29	0.768

Table 6 presents the Variance Inflation Factor (VIF) values for each explanatory variable included in the model. This test evaluates the degree of collinearity among independent variables, which is essential to ensure the validity of econometric estimates.

Table 6. Multicolinealidad — Países europeos de altos ingresos.

Variables	VIF
Life expectancy	1.95
Use of fossil fuels	1.82
CO2 emissions	1.40
Use of renewable energy as a primary source of energy	1.26
lnGDP per capita	1.24

As a general rule, VIF values below 5 indicate no serious multicollinearity, while values below 2 are considered optimal. The results show that all explanatory variables exhibit low VIF values, indicating the absence of problematic multicollinearity in the model.

Table 7 presents the results of the Hansen test for overidentifying restrictions in the GMM model. The chi²(2) statistic is 0.28, with an associated probability of 0.600.

Table 7. Hansen test of overidentification — High-income European countries.

restrictions: chi2(2) = 0.28

Prob > chi2 = 0.600

Since the p-value is greater than 0.05, the null hypothesis of valid instruments cannot be rejected. This case implies that the instruments used in the model are uncorrelated with the error term, thereby reinforcing the reliability of the estimated model.

Although the Hansen test results suggest that the instruments are valid, it is also important to consider the number of instruments relative to the sample size. In this study, the number of instruments (8) remains well below the number of cross-sectional units, reducing the risk of instrument proliferation and overfitting. This supports the reliability of the GMM estimates. Nevertheless, as in any dynamic panel model, the potential sensitivity to instrument specification should be acknowledged.

Before estimating the dynamic model using the Arellano–Bond Generalized Method of Moments (GMM), a diagnostic test was conducted to identify possible heteroscedasticity in the model variables. The results indicated that some variables exhibited this issue, potentially compromising estimator efficiency. To address this limitation and ensure robust results, the GMM model was estimated with robust standard errors, yielding consistent estimates even in the presence of heteroscedasticity.

The dynamic estimation following Arellano and Bond (1991) for high-income European countries yields several key findings. Table 8 presents the results of the dynamic panel-data estimation using the Arellano–Bond Generalized Method of Moments (GMM) for high-income European Union countries.

Table 8. Dynamic panel-data estimation using Arellano–Bond — High-income European countries.

Group variable: COUNTRY Time variable: YEAR		Number of obs	190
		Obs per group
		min	19
		Avg	19
		Max	19
Number of instruments	8	Prob > chi2	0.0000

D.Life expectancy	Coef.	Robust	P > \|t\|	[95% Conf. Interval]
D.Life expectancy	Coef.	Std. Err.	P > \|t\|	[95% Conf. Interval]
Life expectancy
LD.	−0.5915093	0.3489799	0.124	−1.380957 0.197938
Use of fossil fuels
LD.	0.0313014	0.0097929	0.011	0.0091483 0.0534545
CO2 emissions
LD.	−0.0062945	0.0306635	0.842	−0.0756601 0.063071
Use of renewable energy as a primary source of energy
LD.	0.0148982	0.0229679	0.533	−0.0370588 0.0668553
lnGDP per capita
LD.	−0.0950689	0.6232889	0.882	−1.505046 1.314909
cons	0.3090978	0.0732621	0.002	0.1433675 0.4748282

First, the dependent variable, life expectancy, shows a negative but statistically insignificant coefficient. Although the coefficient of the lagged dependent variable is not statistically significant, its magnitude suggests a moderate degree of persistence in life expectancy. However, this result should be interpreted with caution, as the lack of statistical significance prevents drawing strong conclusions about temporal dependence.

Second, a significant result is observed for fossil fuel use, with a positive, statistically significant coefficient. This study indicates that, in these countries, fossil fuel consumption is associated with higher life expectancy. A significant result is observed for fossil fuel use, with a positive and statistically significant coefficient. This finding indicates a positive association between fossil fuel consumption and life expectancy in high-income countries. However, this result should be interpreted with caution. Rather than representing a causal relationship, it may reflect alternative mechanisms such as omitted variable bias, reverse causality, or measurement issues. For example, higher-income economies with advanced healthcare systems and infrastructure may simultaneously exhibit higher energy consumption and better health outcomes.

In contrast, both CO₂ emissions and renewable energy use show statistically insignificant coefficients, suggesting that their marginal impact on life expectancy is weak in high-income countries—probably because public health outcomes are already supported by well-established welfare and environmental institutions.

Similarly, GDP per capita does not show a significant effect, suggesting that additional economic growth in advanced economies does not directly translate into measurable improvements in life expectancy.

For the group of medium-income countries, Table 9 reports the results of the Im–Pesaran–Shin (IPS) unit root test applied to the main variables to determine their integration orders. This test is used in panel-data models to verify the presence of a unit root (H₀) versus the alternative hypothesis of common stationarity (H₁).

Table 9. Im–Pesaran–Shin unit root test — Medium-income European countries.

	Level	First difference
W-t-bar	P value	P value
Life expectancy	1.0000	0.0025
Use of fossil fuels	0.0677	0.0330
CO2 emissions	0.2573	0.0023
Use of renewable energy as a primary source of energy	0.5805	0.0003
lnGDP per capita	0.4011	0.0439

According to statistical criteria, p-values below 0.05 indicate rejection of the null hypothesis and suggest that the series is stationary.

The results show that none of the variables is stationary in levels, since all exhibit p-values greater than 0.05. For example, the life expectancy variable has a p-value of 1.0000, and CO₂ emissions a p-value of 0.2573, indicating the presence of a unit root in their original form.

However, when the test is applied to first differences, several variables become stationary: life expectancy (p = 0.0025), CO₂ emissions (p = 0.0023), renewable energy use (p = 0.0003), and GDP per capita (p = 0.0439). These results suggest that the variables are integrated of order one, satisfying the conditions for their inclusion in dynamic models such as GMM.

However, it is important to note that the IPS test assumes cross-sectional independence and may be sensitive to panel heterogeneity. Given the integration of European economies and potential interdependencies among countries, the presence of cross-sectional dependence cannot be ruled out. Therefore, these results should be interpreted with caution, as they may not fully capture the underlying data structure.

Table 10 presents the results of the Arellano–Bond test applied to the medium-income European countries to assess the presence of autocorrelation in first-difference errors.

Table 10. Arellano–Bond test for absence of autocorrelation in first-difference errors — Medium-income European countries.

Order	z	Prob > z
1	−1.10	0.269
2	−0.85	0.393

The results show that for order 1, the z statistic is −1.10 with a probability of 0.269, while for order 2, the z statistic is −0.85 with a probability of 0.393.

Since in both cases the p-values exceed 0.05, the null hypothesis (H₀) of no autocorrelation in the residuals cannot be rejected. The results of the Arellano–Bond⁴⁸ test for medium-income European countries indicate that no statistically significant autocorrelation is detected in the first-differenced residuals at either the AR(1) or AR(2) levels. While the absence of second-order autocorrelation (AR(2)) is consistent with the assumptions of the model, the lack of evidence of first-order autocorrelation (AR(1)) is somewhat unusual in this type of specification and should therefore be interpreted with caution. This pattern may be associated with specific characteristics of the sample or potential limitations in the model specification and thus does not provide conclusive confirmation of the estimated dynamic structure.

Table 11 reports the Variance Inflation Factors (VIFs) for the explanatory variables in the model to assess multicollinearity. This indicator detects potential redundancies among independent variables that could distort econometric estimates.

Table 11. Multicollinearity — Medium-income European countries.

Variables	VIF
Use of fossil fuels	3.29
CO2 emissions	1.32
Use of renewable energy as a primary source of energy	1.27
lnGDP per capita	1.04

As a general rule, VIF values below 4 are considered acceptable.

In this case, the results indicate no serious multicollinearity among variables, since all values fall below the critical threshold. The variable use of fossil fuels (3.29) has the highest VIFs, though they remain within a manageable range. Other variables—CO₂ emissions (1.32), renewable energy use (1.27), and GDP per capita (1.04)—show low collinearity.

The Hansen test for overidentification yields χ²(2) = 1.85, with a p-value of 0.396, confirming the validity of the instruments used in the estimation. Since the p-value exceeds 0.05, there is no evidence of correlation between the instruments and the error term, avoiding overfitting issues and ensuring the consistency of the GMM model results ( Table 12).

Table 12. Hansen test of overidentification — Medium-income European countries.

restrictions: chi2(2) = 1.85

Prob > chi2 = 0.396

The dynamic panel-data estimation results for medium-income European countries are presented in Table 13. First, lagged life expectancy shows a positive coefficient (1.30), although not statistically significant (p = 0.603), suggesting weak temporal persistence of this variable within the group of countries analyzed.

Table 13. Dynamic panel-data estimation using Arellano–Bond — Medium-income European countries.

Group variable: COUNTRY Time variable: YEAR		Number of obs	189
		Obs per group
		min	18
		Avg	18.90
		Max	19
Number of instruments	8	Prob > chi2	0.0000

D.Life expectancy	Coef.	Robust	P > \|t\|	[95% Conf. Interval]
D.Life expectancy	Coef.	Std. Err.	P > \|t\|	[95% Conf. Interval]
Life expectancy
LD.	1.309437	2.430333	0.603	−4.188359 6.807233
Use of fossil fuels
LD.	0.0433404	0.0371055	0.273	−0.0405981 0.1272788
CO2 emissions
LD.	−0.7106007	0.3042238	0.044	−0.0223987−1.398803
Use of renewable energy as a primary source of energy
LD.	0.2687355	0.1170912	0.047	0.0038567 0.5336143
lnGDP per capita
LD.	0.1148107	0.9513675	0.258	−1.74 1.97
cons	−2.276037	1.801168	0.238	−63.50561 17.98487

Second, fossil fuel use shows a positive coefficient (0.043), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.

However, two relevant findings emerge:

• CO₂ emissions have a negative and statistically significant effect (−0.71; p = 0.044), consistent with the expected adverse impact of environmental degradation on population health in medium-income countries. The associated confidence interval is fully negative, reinforcing the robustness of this relationship. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, possibly reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.
• Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), indicating a positive association with life expectancy within this group of countries.

Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating a high degree of uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

The empirical findings provide partial but relevant support for the proposed hypotheses.

Regarding Hypothesis 1—that higher CO₂ emissions and higher fossil fuel consumption reduce life expectancy—the results show differentiated dynamics by income group. In high-income European countries, fossil fuel consumption has a positive, statistically significant effect on life expectancy—a counterintuitive result suggesting that, in advanced economies, fossil fuel use remains linked to economic activity and health infrastructure. By contrast, CO₂ emissions display a negative but not significant coefficient, indicating that their harmful impact is not yet statistically detectable in the short term. In medium-income countries, fossil fuel consumption is not significant, whereas CO₂ emissions show a negative and significant association with life expectancy.

With respect to Hypothesis 2—that renewable energy use improves life expectancy—the evidence is more substantial, albeit asymmetric. In medium-income countries, renewable energy use shows a positive, statistically significant effect, confirming that expanding sustainable energy sources contributes to resilience and yields measurable health benefits. In contrast, in high-income countries, renewables do not exhibit significant effects on life expectancy, which may be explained by the fact that, once countries have reached high levels of welfare and universal health coverage, additional improvements in environmental quality are less likely to produce immediate or detectable changes in longevity within the relatively short time horizon of the analysis.

Discussion

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcome.^23,49

A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would contribute to strengthening the empirical identification of institutional mechanisms.

From a methodological perspective, the results should be interpreted considering that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.

In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature,^21,37 phenomena such as transboundary pollution and interconnected energy systems suggest that part of the observed relationships may be influenced by cross-country interactions. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.

Similarly, the time span of the analysis (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics.^14,16 In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.

Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity.^21,22 Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.

Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship.^24,28 In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations.^10,25

These findings can be further interpreted through an institutional lens. Previous research highlights that development outcomes and well-being are shaped not only by economic or environmental conditions, but also by governance quality, social capital, and institutional capacity.⁴⁹ In this framework, institutions operate as mediating structures that influence the extent to which environmental risks translate into health outcomes. Accordingly, the stronger and statistically significant negative effect of CO₂ emissions observed in middle-income countries may reflect weaker regulatory enforcement, limited public health infrastructure, and lower adaptive capacity. In contrast, in high-income countries, more robust welfare systems and environmental governance frameworks may partially offset these adverse effects, reducing their observable impact on life expectancy.

In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings.^22,24 At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies.^21,26,36,50

Since the model does not include mediating variables—such as air quality indicators or environmental health measures the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.

From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that consider the institutional and structural characteristics of each country group. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Conclusions

This study contributes to the literature by examining the relationship between environmental degradation and life expectancy in the European Union, highlighting how these dynamics differ across countries with distinct economic and structural characteristics. The empirical results indicate that the effects of environmental factors on population health are not uniform, but vary depending on contextual conditions associated with income levels and broader institutional environments.

In high-income European countries, the findings suggest that environmental variables such as CO₂ emissions and renewable energy use do not exhibit statistically significant effects on life expectancy, while fossil fuel consumption shows a positive association. However, this result should be interpreted with caution, as it may reflect underlying structural factors—such as advanced healthcare systems, higher levels of economic development, or other omitted variables—rather than a direct causal relationship.

In contrast, medium-income European countries show a more pronounced sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively and significantly associated with life expectancy, indicating that environmental pressures may translate more directly into adverse health outcomes in these contexts. At the same time, renewable energy use is positively associated with life expectancy, although this relationship should be interpreted as an indirect association rather than a direct causal effect, given that the model does not explicitly capture the underlying transmission mechanisms.

Overall, the results suggest that environmental and economic factors interact with broader contextual conditions, which may include institutional, structural, and socio-economic elements, although these are not directly modeled in the empirical specification. Therefore, the findings should be interpreted as reduced-form relationships rather than structurally identified causal effects.

From a policy perspective, the results support the relevance of differentiated approaches that account for country-specific conditions when addressing environmental and public health challenges. However, these implications should be interpreted with caution. While the study adopts an institutional framework to contextualize the findings, the econometric model does not directly test specific institutional mechanisms or policy interventions.

Consequently, broader proposals—such as the transformation toward an “ecological welfare state”—should be understood as theoretical extensions consistent with the conceptual framework of the study, rather than as empirically validated conclusions. Future research could build on these findings by incorporating explicit institutional indicators, mediating variables (such as air quality), or alternative modeling approaches that allow for the identification of causal mechanisms and policy effects.

In addition, further studies could extend the analysis by accounting for spatial interactions, structural breaks, and heterogeneous country dynamics, which may play a significant role in shaping environmental and health outcomes in integrated regions such as the European Union. Addressing these dimensions would contribute to a more comprehensive understanding of the complex relationships between environmental degradation, institutional conditions, and population well-being.

Institutional review board statement

Not applicable.

Ethics and consent

Ethical approval and consent were not required.

Data availability statement

All data used in this study are publicly available from the Our World in Data database (https://ourworldindata.org) and the World Bank Open Data platform (https://data.worldbank.org). The dataset includes annual observations for 20 European Union countries covering the period 2000–2020. Data were accessed and downloaded in CSV format in January 2026. The variables employed in the analysis (life expectancy, CO₂ emissions, fossil fuel consumption, renewable energy use, and GDP per capita at constant prices) can be retrieved using the same country and period filters described in the Materials and Methods section. No proprietary or restricted-access data were used.

Acknowledgments

The authors would like to thank the Dirección de Investigación y Desarrollo-DIDE of the Universidad Técnica de Ambato. This article is derived from the research project entitled “Software for the integration of Industry 5.0 and the sustainable development of the Organizations of the Popular and Solidarity Economy of Zone 3 of Ecuador”, approved with Resolution No. UTA-CONIN-2025-0066-R by the Dirección de Investigación y Desarrollo-DIDE of the Universidad Técnica de Ambato, Ecuador.

References

1. Long JW, Steel EA: Shifting Perspectives in Assessing Socio-Environmental Vulnerability. Sustainability. 2020; 12(7): 2625. Publisher Full Text
2. Rockström J, Steffen W, Noone K, et al.: A safe operating space for humanity. Nature. Sep. 2009; 461(7263): 472–475. Publisher Full Text
3. Forster J: Summary for policymakers of the global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Díaz S, Settele J, Brondízio ES, et al., authors. 2019; vol. 45(3). Reference Source
4. Forster PM, Smith C, Walsh T, et al.: Indicators of Global Climate Change 2023: annual update of key indicators of the state of the climate system and human influence. Earth Syst. Sci. Data. Jun. 2024; 16(6): 2625–2658. Publisher Full Text
5. North DC: A Transaction Cost Theory of Politics. J. Theor. Polit. Oct. 1990; 2(4): 355–367. Publisher Full Text
6. Hall PA, Taylor RCR: Political Science and the Three New Institutionalisms. Polit. Stud. Dec. 1996; 44(5): 936–957. Publisher Full Text
7. Barbier EB, Hochard JP: Land degradation and poverty. Nat. Sustainability. Nov. 2018; 1(11): 623–631. Publisher Full Text
8. Kjell ONE: Sustainable Well-Being: A Potential Synergy between Sustainability and Well-Being Research. Rev. Gen. Psychol. Sep. 2011; 15(3): 255–266. Publisher Full Text
9. Gough I: Climate change and sustainable welfare: the centrality of human needs: Fig. 1. Camb. J. Econ. 2015; 39(5): 1191–1214. Publisher Full Text
10. Bailey D: The Environmental Paradox of the Welfare State: The Dynamics of Sustainability. New Polit. Econ. Nov. 2015; 20(6): 793–811. Publisher Full Text
11. Eisenmenger N, Pichler M, Krenmayr N, et al.: The Sustainable Development Goals prioritize economic growth over sustainable resource use: a critical reflection on the SDGs from a socio-ecological perspective. Sustain. Sci. Jul. 2020; 15(4): 1101–1110. Publisher Full Text
12. Singer-Brodowski M, von Seggern J , Duveneck A, et al.: Moving (Reflexively within) Structures. The Governance of Education for Sustainable Development in Germany. Sustainability. 2020; 12(7): 2778. Publisher Full Text
13. Easterlin: Munich Personal RePEc Archive Does Economic Growth Improve the Human Lot ? Some Empirical Evidence the Human Lot ?. Some Empirical Evidence. 1974; 111773. Reference Source
14. Martinez GS, Kendrovski V, Salazar MA, et al.: Heat-health action planning in the WHO European Region: Status and policy implications. Environ. Res. 2022; 214: 113709. Publisher Full Text
15. Ahoa E, Kassahun A, Verdouw C, et al.: Challenges and Solution Directions for the Integration of Smart Information Systems in the Agri-Food Sector. Sensors. Apr. 2025; 25(8): 2362. PubMed Abstract | Publisher Full Text | Free Full Text
16. Robben J, Antonio K, Kleinow T: The short-term association between environmental variables and mortality: evidence from Europe. J. R. Stat. Soc. Ser. A Stat. Soc. 2025. Publisher Full Text
17. Mouratidis K: Urban planning and quality of life: A review of pathways linking the built environment to subjective well-being. Cities. Aug. 2021; 115: 103229. Publisher Full Text
18. United Nations Children’s Fund: World Health Organization. Progress on household drinking water, sanitation and hygiene 2000–2022: Special focus on gender. WHO/UNICEF JMP; 2023. Reference Source
19. Otto A, Gugushvili D: Eco-Social Divides in Europe: Public Attitudes towards Welfare and Climate Change Policies. Sustainability. 2020; 12(1): 404. Publisher Full Text
20. O’Brien K, Barnett J: Global Environmental Change and Human Security. Annu. Rev. Environ. Resour. 2013; 38(1): 373–391. Publisher Full Text
21. Rehman A, Alam MM, Ozturk I, et al.: Globalization and renewable energy use: how are they contributing to upsurge the CO2 emissions? A global perspective. Environ. Sci. Pollut. Res. 2022; 30(4): 9699–9712. Publisher Full Text
22. Majeed MT, Ozturk I: Environmental degradation and population health outcomes: a global panel data analysis. Environ. Sci. Pollut. Res. May 2020; 27(13): 15901–15911. PubMed Abstract | Publisher Full Text
23. Gough I: Heat, Greed and Human Need. Edward Elgar Publishing; 2017. Publisher Full Text
24. Rahman MM, Rana R, Khanam R: Determinants of life expectancy in most polluted countries: Exploring the effect of environmental degradation. PLoS One. Jan. 2022; 17(1): e0262802. PubMed Abstract | Publisher Full Text | Free Full Text
25. UNICEF WHO: The State of Food Security and Nutrition in the World 2022. FAO; 2022. Publisher Full Text
26. Pu CJ, Patel P, Hornsby G, et al.: Necessary conditions for sustainable water and sanitation service delivery in schools: A systematic review. PLoS One. Jul. 2022; 17(7): e0270847. PubMed Abstract | Publisher Full Text | Free Full Text
27. North DC: Institutions, Institutional Change and Economic Performance. Cambridge University Press; 1990. Publisher Full Text
28. Rodrik D: Rethinking Economic Growth in Developing Countries. Centro Studi Luca d’Agliano; 2004. Reference Source
29. Morales-Urrutia X, Núñez-Naranjo A, Nogales-Portero R, et al.: The Natural Environment and Investment in Economic Growth: From the Perspective of the Prosperity of Developed and Developing Countries. Sustainability. 2025; 17(12): 5513. Publisher Full Text
30. Temporelli KL, Viego VN: Relación entre esperanza de vida e ingreso. Un análisis para América Latina y el Caribe. Lect. Econ. 2011; 74: 61–85. Publisher Full Text
31. Ortiz-Paniagua CF, Gómez M: Economic Growth and Environmental Quality in Latin America, Perspective from Kuznets, 1970-2016. Economía teoría y práctica. 2021; 29(55): 17–36. Publisher Full Text
32. Catalán H: Impacto de las energías renovables en las emisiones de gases efecto invernadero en México. Probl. del Desarro. Rev. Latinoam. Econ. Dec. 2020; 52(204). Publisher Full Text
33. Baste IA: Making Peace with Nature. United Nations; 2021. Publisher Full Text
34. Hassan Q, Viktor P, J. Al-Musawi T, et al.: The renewable energy role in the global energy Transformations. Renewable Energy Focus. 2024; 48: 100545. Publisher Full Text
35. Iñiguez Rojas L: Geografía y salud: temas y perspectivas en América Latina. Cad. Saude Publica. Oct. 1998; 14(4): 701–711. PubMed Abstract | Publisher Full Text
36. Ezbakhe F, Giné-Garriga R, Pérez-Foguet A: Leaving no one behind: Evaluating access to water, sanitation and hygiene for vulnerable and marginalized groups. Sci. Total Environ. Sep. 2019; 683: 537–546. PubMed Abstract | Publisher Full Text
37. Ortega-Ramirez AT, Beltrán-Rodríguez DG, Becerra-Ospina NL, et al.: Environmental aspects of natural resources and their relationship to the exploitation of fossil fuels: A reflection on sustainability. Rev. Fuentes el Reventón Energético. Dec. 2022; 20(2). Publisher Full Text
38. Barragán-Ocaña A, Cecilio-Ayala E, Silva-Borjas P, et al.: Policies and sustainable energy transition in the global environment: Challenges for Latin America. Heliyon. Mar. 2025; 11(6): e42295. Publisher Full Text
39. Nunez F: Dataset of References for the Article on Smart Irrigation with IoT. Mendeley Data. 2025; V1. Publisher Full Text
40. Hrzic R, Vogt T: The contribution of avoidable mortality to life expectancy differences and lifespan disparities in the European Union: a population-based study. The Lancet Regional Health - Europe. 2024; 46: 101042. Publisher Full Text
41. Javanshirova Z: The Impact of Carbon Emissions on Infant Mortality Rate in Azerbaijan. J. Sustain. Dev. Issues. 2024; 2(2): 104–114. Publisher Full Text
42. Perera F, Nadeau K: Climate Change, Fossil-Fuel Pollution, and Children’s Health. N. Engl. J. Med. Jun. 2022; 386(24): 2303–2314. Publisher Full Text
43. Sompolska-Rzechuła A, Bąk I, Becker A, et al.: The Use of Renewable Energy Sources and Environmental Degradation in EU Countries. Sustainability. 2024; 16(23): 10416. Publisher Full Text
44. Mankiw NG: Is government spending too easy an answer. The New York Times. Reference Source
45. Soeder DJ: Fossil Fuels and Climate Change. Fracking and the Environment. Cham: Springer International Publishing; 2021; 155–185. Publisher Full Text
46. Arellano M, Bond S: Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations. Rev. Econ. Stud. Apr. 1991; 58(2): 277. Publisher Full Text
47. Preston SH: Mortality Trends. Annu. Rev. Sociol. Aug. 1977; 3(1): 163–178. Publisher Full Text
48. Romanello M, Napoli C, Green C, et al.: The 2023 report of the Lancet Countdown on health and climate change: the imperative for a health-centred response in a world facing irreversible harms. Lancet. Dec. 2023; 402(10419): 2346–2394. PubMed Abstract | Publisher Full Text | Free Full Text
49. Núñez-Naranjo AF, Morales-Urrutia X: Social Capital and Economic Growth: An Analysis in South America [version 2; peer review: 2 approved]. F1000Research. 2025; 14: 797. Publisher Full Text
50. Icaza D, Vallejo-Ramirez D, Guerrero Granda C, et al.: Challenges, Roadmaps and Smart Energy Transition towards 100% Renewable Energy Markets in American Islands: A Review. Energies. 2024; 17(5): 1059. Publisher Full Text

Comments on this article Comments (0)

Version 2

VERSION 2 PUBLISHED 25 Mar 2026

Author details Author details

David Vega-Camacho
Roles: Investigation, Software, Writing – Review & Editing

Aracelly Núñez-Naranjo
Roles: Conceptualization, Project Administration, Supervision, Validation, Writing – Review & Editing

Patricio Lara-Álvarez
Roles: Conceptualization, Methodology

Mery Ruiz-Guajala
Roles: Data Curation, Software, Writing – Original Draft Preparation

Patricia Acosta-Vargas
Roles: Funding Acquisition, Resources, Writing – Original Draft Preparation, Writing – Review & Editing

Competing interests

No competing interests were disclosed.

Grant information

This research was funded by Universidad de Las Américas—Ecuador as part of the internal research project 518.A.XV.24.

Article Versions (2)

version 2

Revised

Published: 18 May 2026, 15:435

https://doi.org/10.12688/f1000research.176930.2

version 1

Published: 25 Mar 2026, 15:435

https://doi.org/10.12688/f1000research.176930.1

© 2026 Morales-Urrutia X et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Download

Export To

metrics

	Views	Downloads
F1000Research	-	-
PubMed Central Data from PMC are received and updated monthly.	-	-

Citations

SEE MORE DETAILS

CITE

how to cite this article

Morales-Urrutia X, Vega-Camacho D, Núñez-Naranjo A et al. Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience [version 2; peer review: 1 approved with reservations]. F1000Research 2026, 15:435 (https://doi.org/10.12688/f1000research.176930.2)

NOTE: If applicable, it is important to ensure the information in square brackets after the title is included in all citations of this article.

track

receive updates on this article

Track an article to receive email alerts on any updates to this article.

Open Peer Review

Version 1

VERSION 1

PUBLISHED 25 Mar 2026

Views

Reviewer Report 15 Apr 2026

Matheus Koengkan, University of Coimbra, Coimbra, Portugal

Approved with Reservations

https://doi.org/10.5256/f1000research.195058.r472715

Reviewer comment (R.C): The abstract presents strong causal language (e.g., “CO2 emissions significantly reduce life expectancy”) despite the empirical strategy relying on a difference GMM framework that primarily identifies associations rather than strict causality. The wording should be revised to reflect conditional correlations unless additional identification strategies are implemented.
Reviewer comment (R.C): The claim that institutional quality mediates the relationship is central to the paper, yet no explicit institutional quality variable is included in the econometric model (see model specification on page 7 ). This creates a disconnect between theory and empirics. A measurable proxy (e.g., governance indicators) should be incorporated or the claims softened.
Reviewer comment (R.C): The classification of countries into “high-income” and “middle-income” groups lacks a clear methodological justification. It is unclear whether this follows World Bank thresholds, sample medians, or arbitrary grouping, which raises concerns about reproducibility and potential sample selection bias.
Reviewer comment (R.C): The transformation of all variables into first differences (Δ) eliminates long-run information. Given that life expectancy evolves slowly, this approach may obscure meaningful long-term relationships. A cointegration framework or system GMM could better capture both short- and long-run dynamics.
Reviewer comment (R.C): The specification includes lagged differences of the dependent variable, but the interpretation of “limited persistence” is questionable, especially given the relatively high coefficient (0.59) in high-income countries, which is economically non-trivial despite statistical insignificance.
Reviewer comment (R.C): The interpretation of the positive and significant coefficient of fossil fuel consumption in high-income countries as a “paradox” is not sufficiently rigorous. Alternative explanations, such as omitted variable bias, reverse causality, or measurement issues, should be explored before attributing this to institutional buffering.
Reviewer comment (R.C): The paper states that CO2 emissions have “direct effects on air quality,” yet CO2 is not a local pollutant affecting health directly in the same way as particulate matter. This conceptual imprecision weakens the theoretical argument and should be corrected.
Reviewer comment (R.C): The absence of key control variables (e.g., healthcare expenditure, education, urbanization, demographic structure) raises serious concerns about omitted variable bias, especially given the complexity of determinants of life expectancy.
Reviewer comment (R.C): The VIF analysis includes “life expectancy” as an explanatory variable (Table 6), which is inappropriate since it is the dependent variable. This suggests either a reporting error or a misunderstanding of multicollinearity diagnostics.
Reviewer comment (R.C): The Hansen test p-values (0.600 and 0.396) are interpreted as evidence of valid instruments, but no discussion is provided regarding instrument proliferation or instrument count relative to sample size, which is critical in GMM estimation.
Reviewer comment (R.C): The number of instruments (8) appears low relative to the model structure, but the paper does not clarify how instruments were constructed or whether collapse or lag limits were applied. This lack of transparency limits replicability.
Reviewer comment (R.C): The Arellano–Bond test results show no AR(1) autocorrelation, which is unusual since first-differenced residuals typically exhibit AR(1) by construction. This raises concerns about either reporting accuracy or model implementation.
Reviewer comment (R.C): The IPS unit root test results are reported as showing non-stationarity at levels, yet p-values such as 0.3164 (CO2) suggest failure to reject the null, but the discussion does not consider panel heterogeneity or cross-sectional dependence.
Reviewer comment (R.C): The decision to differ all variables based solely on IPS results ignores the possibility of cross-sectional dependence, which is likely in EU countries. Second-generation panel unit root tests would be more appropriate.
Reviewer comment (R.C): The descriptive statistics are interpreted normatively (e.g., “this reflects strong welfare systems”), which goes beyond what summary statistics can support. These interpretations should be reserved for regression analysis.
Reviewer comment (R.C): The figures (pages 9 and 11) visually suggest heterogeneous country trajectories, yet the econometric model assumes homogeneous slope coefficients within groups. This assumption should be tested or relaxed (e.g., via heterogeneous panel models).
Reviewer comment (R.C): The interpretation of insignificant coefficients as evidence of “no effect” is misleading. Lack of statistical significance may reflect low power, measurement error, or model misspecification rather than the true absence of a relationship.
Reviewer comment (R.C): The sign inconsistency in Table 13 for CO2 emissions (reported as negative in the text but positive in the table before sign interpretation) suggests either a typographical error or misinterpretation that needs correction.
Reviewer comment (R.C): The confidence intervals reported in Table 13 appear inconsistent (e.g., extremely wide ranges for GDP per capita), indicating potential scaling or reporting errors that should be verified.
Reviewer comment (R.C): The discussion attributes differences between country groups to “institutional capacity,” but no robustness checks (e.g., interaction terms, subgroup regressions with institutional proxies) are provided to substantiate this claim.
Reviewer comment (R.C): The model does not address potential cross-country spillovers (e.g., pollution transport, EU-wide policies), which could bias estimates in a regional panel such as the European Union.
Reviewer comment (R.C): The time period (2000–2020) includes structural breaks such as the 2008 financial crisis and COVID-19 onset, yet no structural break tests or period controls are included, which may distort results.
Reviewer comment (R.C): The use of CO2 emissions per capita without considering total emissions or sectoral composition may mask important heterogeneity in environmental impact and policy relevance.
Reviewer comment (R.C): The argument that renewable energy improves life expectancy lacks a clear transmission mechanism in the empirical model. Mediating variables such as air quality improvements are not included, limiting interpretability.
Reviewer comment (R.C): The conclusions extend policy recommendations (e.g., ecological welfare state transformation) that are not directly tested in the empirical model, resulting in overgeneralization beyond the evidence presented.

Is the work clearly and accurately presented and does it cite the current literature?

Partly
Is the study design appropriate and is the work technically sound?

Partly
Are sufficient details of methods and analysis provided to allow replication by others?

Partly
If applicable, is the statistical analysis and its interpretation appropriate?

Partly
Are all the source data underlying the results available to ensure full reproducibility?

Partly
Are the conclusions drawn adequately supported by the results?

Partly

Competing Interests: No competing interests were disclosed.

Reviewer Expertise: Energy and Environmental Economics.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

CITE

Report a concern

Author Response 27 Apr 2026

Patricia Acosta-Vargas, Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

27 Apr 2026

Author Response

Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the ... Continue reading Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience.” We sincerely appreciate the time and effort the reviewers dedicated to providing us with valuable feedback.

In order to carefully address all comments and ensure a comprehensive and high-quality response, I request an extension to submit our replies until May 10, 2026.

This additional time will allow us to offer a comprehensive and well-structured review that fully addresses the reviewers’ concerns.

Thank you very much for your consideration. I look forward to your kind confirmation.

Sincerely,
Prof. Patricia Acosta-Vargas
Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience.” We sincerely appreciate the time and effort the reviewers dedicated to providing us with valuable feedback.

In order to carefully address all comments and ensure a comprehensive and high-quality response, I request an extension to submit our replies until May 10, 2026.

This additional time will allow us to offer a comprehensive and well-structured review that fully addresses the reviewers’ concerns.

Thank you very much for your consideration. I look forward to your kind confirmation.

Sincerely,
Prof. Patricia Acosta-Vargas
Competing Interests: No competing interests were disclosed. Close
Report a concern
Author Response 29 Apr 2026

Patricia Acosta-Vargas, Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

29 Apr 2026

Author Response
Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of ... Continue reading
Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of the study. Below, we address each comment in detail and describe the corresponding revisions implemented in the manuscript.

Reviewer Comment: The abstract presents strong causal language (e.g., “CO2 emissions significantly reduce life expectancy”) despite the empirical strategy relying on a difference GMM framework that primarily identifies associations rather than strict causality. The wording should be revised to reflect conditional correlations unless additional identification strategies are implemented.

Response:
We appreciate this important observation. The abstract has been revised to remove any causal language and to reflect the empirical strategy's associative nature. Expressions such as “reduce life expectancy” have been replaced with “are negatively associated with life expectancy.” Additionally, the wording throughout the manuscript has been carefully revised to ensure consistency with a conditional and non-causal interpretation of the results.

Reviewer comment (R.C): The claim that institutional quality mediates the relationship is central to the paper, yet no explicit institutional quality variable is included in the econometric model (see model specification on page 7 ). This case creates a disconnect between theory and empirics. A measurable proxy (e.g., governance indicators) should be incorporated, or the claims should be softened.

Response:

We thank the reviewer for pointing out this important issue. In the revised manuscript, we have clarified that institutional quality is not empirically modeled as a mediating variable. Instead, institutional theory is used strictly as an interpretive framework to contextualize the observed empirical patterns. We explicitly state that:

The study does not aim to identify or estimate institutional mediation mechanisms econometrically.

References to institutional capacity are not intended to provide causal explanations but rather to offer theoretically grounded interpretations consistent with the literature.

Furthermore, we have clarified that heterogeneity across countries is partially reflected in income-based groupings, which capture broad structural and institutional differences. However, we explicitly acknowledge that this approach does not allow for a direct assessment of institutional effects.

Building on this, institutionalist theory posits that the design and strength of public institutions shape and condition the effects of environmental degradation on social outcomes, including health [9], [10], rather than acting as directly measured mediating mechanisms within the empirical model.

Accordingly, life expectancy should be linked to the institutional capacity to integrate environmental justice, territorial resilience, and intergenerational equity into public policy. However, in this study, such institutional dimensions are not directly measured but are considered as contextual factors shaping the observed relationships [11], [12].

However, the magnitude and direction of these effects are expected to vary across countries with different institutional and socioeconomic contexts.

Reviewer comment (R.C): The classification of countries into “high-income” and “middle-income” groups lacks a clear methodological justification. It is unclear whether this follows World Bank thresholds, sample medians, or arbitrary grouping, which raises concerns about reproducibility and potential sample selection bias.

Response:
We have clarified that countries are classified based on their average GDP per capita over the study period (2000–2020), using a relative within-sample approach. This represents a deliberate methodological choice aimed at ensuring internal comparability and capturing structural heterogeneity within the dataset.
We also explicitly acknowledge that this classification does not follow external benchmarks (e.g., World Bank thresholds) and may introduce sample-specific limitations. This transparency enhances reproducibility and interpretability.

To ensure a more representative analysis of Europe’s economic and environmental realities, countries were classified into two groups based on their average GDP per capita over the study period (2000–2020). Specifically, countries with higher average incomes were categorized as high-income, while those with lower average incomes were classified as middle-income. This classification follows a relative-grouping approach within the sample, enabling the identification of structural differences in economic capacity, institutional context, and environmental dynamics.
Although this classification does not strictly adhere to external thresholds, such as those defined by the World Bank, it provides a consistent, internally comparable framework for analyzing heterogeneity across countries. However, this approach may introduce some limitations related to sample-specific grouping, which should be considered when interpreting the results.

Reviewer comment (R.C): The transformation of all variables into first differences (Δ) eliminates long-run information. Given that life expectancy evolves slowly, this approach may obscure meaningful long-term relationships. A cointegration framework or system GMM could better capture both short- and long-run dynamics.

Response:
We agree that transforming variables into first differences limits the ability to capture long-run relationships. In the revised manuscript, we clarify that this transformation reflects a deliberate econometric trade-off.
Specifically:

Differencing is used to ensure consistency and avoid spurious regression results in the presence of non-stationarity and potential endogeneity.

The dynamic GMM framework prioritizes short-run dynamics over long-run equilibrium relationships.

We explicitly acknowledge that long-run relationships are not identified in this specification and suggest that future research could explore cointegration approaches or system GMM to address this limitation.

The transformation of variables into first differences is consistent with the unit root test results, which indicate that the series are integrated of order one. This approach helps to avoid spurious regressions and ensures the validity of the GMM estimation. However, it is important to acknowledge that differencing may reduce the ability to capture long-run relationships, particularly for variables such as life expectancy that evolve gradually over time. Therefore, the results should be interpreted as reflecting short-term dynamics rather than long-term equilibrium relationships.

Reviewer Comment: The specification includes lagged differences of the dependent variable, but the interpretation of “limited persistence” is questionable, especially given the relatively high coefficient (0.59) in high-income countries, which is economically non-trivial despite statistical insignificance.

Response:
Thank you for this insightful observation. We agree that the original interpretation of “limited persistence” was not fully appropriate given the magnitude of the estimated coefficient.

In the revised manuscript, we have refined the interpretation to acknowledge that, although the coefficient is not statistically significant, its magnitude suggests a moderate degree of persistence in life expectancy. At the same time, we emphasize that the lack of statistical significance prevents drawing robust conclusions about temporal dependence. This revision improves the consistency between the statistical results and their economic interpretation.

Although the coefficient on the lagged dependent variable is not statistically significant, its magnitude suggests moderate persistence in life expectancy. However, this result should be interpreted with caution, as the lack of statistical significance prevents drawing strong conclusions about temporal dependence.

Reviewer comment (R.C): The interpretation of the positive and significant coefficient of fossil fuel consumption in high-income countries as a “paradox” is not sufficiently rigorous. Alternative explanations, such as omitted variable bias, reverse causality, or measurement issues, should be explored before attributing this to institutional buffering.

Response:
We have substantially revised the interpretation of this result. The positive and significant association between fossil fuel consumption and life expectancy in high-income countries is no longer presented as a paradox.
Instead, we explicitly state that:

The result should not be interpreted as evidence of a beneficial effect of fossil fuel consumption.

It likely reflects underlying structural correlations (e.g., higher income levels, advanced healthcare systems, or omitted variables).

Alternative explanations such as omitted variable bias, reverse causality, and measurement limitations are explicitly discussed.

This revision ensures a cautious and methodologically consistent interpretation.

This study indicates that, in these countries, fossil fuel consumption is associated with higher life expectancy. A significant result is observed for fossil fuel use, with a positive and statistically significant coefficient. This finding indicates a positive association between fossil fuel consumption and life expectancy in high-income countries. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may instead reflect omitted-variable bias, reverse causality, or measurement issues. For example, higher-income economies with advanced healthcare systems and infrastructure may simultaneously exhibit higher energy consumption and better health outcomes.

The most striking finding is a positive, statistically significant relationship between fossil fuel consumption and life expectancy, which appears counterintuitive given the expected negative environmental impacts. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may be driven by underlying structural factors, such as higher levels of economic development, more advanced healthcare systems, or omitted variables correlated with both energy consumption and population health outcomes. Additionally, potential issues such as reverse causality or measurement limitations cannot be ruled out. Therefore, attributing this pattern solely to institutional capacity would be premature without further empirical evidence.

Reviewer comment (R.C): The paper states that CO2 emissions have “direct effects on air quality,” yet CO2 is not a local pollutant affecting health directly in the same way as particulate matter. This conceptual imprecision weakens the theoretical argument and should be corrected.

Response:
We have corrected the conceptual treatment of CO₂ emissions. The manuscript now clearly states that CO₂ is not a local air pollutant with direct health effects, but rather a driver of climate change with indirect impacts on health through environmental and climatic pathways.

CO₂ emissions, in contrast, represent a cumulative environmental outcome associated with climate change processes, which may indirectly affect human health through environmental and climatic pathways rather than direct exposure effects, particularly via mechanisms such as temperature increases and environmental degradation [24], [38].

CO₂ emissions—recognized as a primary contributor to global greenhouse gases—originate mainly from the combustion of fossil fuels in electricity generation, transportation, and industrial production. However, CO₂ should not be considered a local air pollutant with direct effects on human health. Instead, its relevance lies in its contribution to climate change, which may indirectly affect health outcomes through mechanisms such as rising temperatures, increased climate variability, and broader environmental degradation.

The role of CO₂ emissions in this study should be interpreted within a broader environmental context. Unlike local air pollutants such as particulate matter, CO₂ does not directly affect human health through immediate exposure. Instead, its impact is primarily indirect, as it contributes to climate change, which in turn influences health outcomes through environmental and socio-economic pathways.

Reviewer comment (R.C): The absence of key control variables (e.g., healthcare expenditure, education, urbanization, demographic structure) raises serious concerns about omitted variable bias, especially given the complexity of determinants of life expectancy.

Response:
We acknowledge that excluding variables such as healthcare expenditure, education, urbanization, and demographic structure may introduce omitted-variable bias.
We clarify that the model adopts a parsimonious specification as a deliberate trade-off, balancing:

Data comparability across countries and time

Model tractability

Avoidance of instrument proliferation in the GMM framework

We explicitly state that the results should be interpreted as conditional associations rather than causal effects, and we identify the inclusion of additional controls as an avenue for future research.

While the model includes key environmental and economic variables, it does not incorporate additional controls such as healthcare expenditure, education, urbanization, or demographic structure. This choice reflects a parsimonious specification based on data availability and comparability across countries and over time. However, excluding these variables may introduce omitted-variable bias, as multiple socio-economic and demographic factors influence life expectancy. Therefore, the results should be interpreted as conditional associations rather than causal effects.

Reviewer comment (R.C): The VIF analysis includes “life expectancy” as an explanatory variable (Table 6), which is inappropriate since it is the dependent variable. This suggests either a reporting error or a misunderstanding of multicollinearity diagnostics.

Response:
Thank you for this important observation. We acknowledge that the inclusion of the dependent variable (life expectancy) in the VIF analysis was inappropriate.

In the revised manuscript, we have corrected Table 6 by removing the dependent variable and ensuring that the VIF analysis includes only explanatory variables. We have also clarified the corresponding text to reflect that multicollinearity diagnostics apply exclusively to independent variables.

These revisions correct the reporting error and improve the methodological accuracy of the manuscript.

Table 6 presents the Variance Inflation Factors (VIFs) for the explanatory variables in the model. This test evaluates the degree of collinearity among independent variables, which is essential to ensure the validity of econometric estimates. As a general rule, VIF values below 5 indicate no serious multicollinearity, while values below 2 are considered optimal. The results show that all explanatory variables have low VIF values, indicating no problematic multicollinearity in the model.
Table 6. Multicolinealidad — High-income European countries
Variables
VIF

Use of fossil fuels
1.82

CO2 emissions
1.40

Use of renewable energy as a primary source of energy
1.26

lnGDP per capita
1.24

In this case, the results indicate no serious multicollinearity among variables, since all values fall below the critical threshold. The variable use of fossil fuels (3.29) has the highest VIFs, though they remain within a manageable range. Other variables—CO₂ emissions (1.32), renewable energy use (1.27), and GDP per capita (1.04)—show low collinearity.
Table 11. Multicollinearity — Medium-income European countries
Variables
VIF

Use of fossil fuels
3.29

CO2 emissions
1.32

Use of renewable energy as a primary source of energy
1.27

lnGDP per capita
1.04

Reviewer comment (R.C): The Hansen test p-values (0.600 and 0.396) are interpreted as evidence of valid instruments, but no discussion is provided regarding instrument proliferation or instrument count relative to sample size, which is critical in GMM estimation.

Response
Although the Hansen test results suggest that the instruments are valid, we agree that this test alone is insufficient to assess instrument quality fully. Therefore, we have expanded the discussion to explicitly consider the number of instruments relative to the sample size, a critical aspect of GMM estimation.
In this study, the number of instruments (8) remains well below the number of cross-sectional units, following a conservative specification strategy to reduce the risk of instrument proliferation and overfitting. This approach enhances the reliability of the Hansen test and the robustness of the estimated coefficients.
Nevertheless, as in any dynamic panel model, we acknowledge that the results may be sensitive to instrument specification, and this limitation is now explicitly recognized in the manuscript.

Although the Hansen test results suggest that the instruments are valid, it is also important to consider the number of instruments relative to the sample size. In this study, the number of instruments (8) remains well below the number of cross-sectional units, reducing the risk of instrument proliferation and overfitting. This study supports the reliability of the GMM estimates. Nevertheless, as in any dynamic panel model, the potential sensitivity to instrument specification should be acknowledged.

Reviewer comment (R.C): The number of instruments (8) appears low relative to the model structure, but the paper does not clarify how instruments were constructed or whether collapse or lag limits were applied. This lack of transparency limits replicability.

Response:
In the GMM estimation, instruments were constructed using lagged values of the endogenous variables, following the standard Arellano–Bond difference GMM approach.
To address concerns related to instrument proliferation and overfitting, we implemented a deliberately conservative specification strategy. Specifically, the number of instruments was restricted by (i) limiting the lag depth used for instrument generation and (ii) collapsing the instrument matrix.
This procedure ensures a parsimonious instrument set while preserving the validity of the moment conditions. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and enhances both the robustness and replicability of the estimates.
These methodological details have now been explicitly incorporated into the revised manuscript to improve transparency.

In the GMM estimation, instruments were constructed from lagged endogenous variables, following the standard Arellano–Bond approach. To avoid overfitting and instrument proliferation, the number of instruments was deliberately restricted by limiting the lag depth and collapsing the instrument matrix. This procedure ensures a parsimonious specification while maintaining the validity of the instruments. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and helps improve the robustness and replicability of the estimates.

Reviewer comment (R.C): The Arellano–Bond test results show no AR(1) autocorrelation, which is unusual since first-differenced residuals typically exhibit AR(1) by construction. This case raises concerns about either reporting accuracy or model implementation.

Response:
We thank the reviewer for this important observation. The Arellano–Bond test results indicate no statistically significant autocorrelation in the first-differenced residuals at either the AR(1) or AR(2) levels for both high- and medium-income country groups.
We acknowledge that the absence of AR(1) autocorrelation is uncommon in first-differenced specifications, where negative first-order serial correlation is typically expected by construction. However, this result does not necessarily imply misspecification or implementation error.
In particular, the lack of significant AR(1) autocorrelation may reflect sample-specific characteristics, such as limited time variation after differencing, relatively smooth dynamics in life expectancy, or the restricted instrument set used in the estimation. These factors can weaken the detection of first-order serial correlation in finite samples.
Importantly, the absence of AR(2) autocorrelation— which is the key requirement for the validity of the moment conditions— is satisfied in all specifications. Therefore, the main consistency condition of the difference GMM estimator remains intact.
Nevertheless, we agree that this result should be interpreted with caution. In the revised manuscript, we explicitly acknowledge that the Arellano–Bond test does not provide strong confirmatory evidence for the model's dynamic structure and that the results should be interpreted accordingly.

The Arellano–Bond test results for high-income European countries indicate no statistically significant autocorrelation in the first-differenced residuals at either AR(1) or AR(2) levels. While the absence of AR(2) autocorrelation is consistent with the validity of the model, the lack of significant AR(1) autocorrelation is less typical in first-differenced specifications and should be interpreted with caution. This outcome may reflect sample-specific characteristics or limitations in the model specification and, therefore, does not provide strong confirmatory evidence of the model's dynamic structure.
The results of the Arellano–Bond test for medium-income European countries indicate that no statistically significant autocorrelation is detected in the first-differenced residuals at either the AR(1) or AR(2) levels. While the absence of second-order autocorrelation (AR(2)) is consistent with the model's assumptions, the lack of evidence for first-order autocorrelation (AR(1)) is somewhat unusual in this type of specification and should therefore be interpreted with caution. This pattern may be associated with specific characteristics of the sample or potential limitations in the model specification, and thus does not provide conclusive confirmation of the estimated dynamic structure.

Reviewer comment (R.C): The IPS unit root test results are reported as showing non-stationarity at levels, yet p-values such as 0.3164 (CO2) suggest failure to reject the null, but the discussion does not consider panel heterogeneity or cross-sectional dependence.

Response:
Thank you for this important observation. We agree that the Im–Pesaran–Shin (IPS) unit root test results should be interpreted with caution, particularly given its underlying assumptions of cross-sectional independence and potential sensitivity to panel heterogeneity.
In the revised manuscript, we have clarified the interpretation of the IPS results. Specifically, p-values such as 0.3164 for CO₂ emissions correctly indicate a failure to reject the null hypothesis of a unit root in levels, which is consistent with the conclusion that the variables are non-stationary in their original form. We have revised the wording to ensure that this interpretation is explicitly and accurately conveyed.
Furthermore, we now explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present in the context of European Union countries due to their economic and environmental interconnections. As a result, the stationarity findings may be affected by common shocks or interdependencies across panel units.
Accordingly, we emphasize in the revised manuscript that the unit root results should be interpreted as indicative rather than definitive. We also clarify that the decision to transform variables into first differences is not based solely on the IPS test, but is additionally supported by standard econometric practice in dynamic panel models when variables are likely integrated of order one.
Finally, we have explicitly identified this limitation and suggested that future research could apply second-generation panel unit root tests that account for cross-sectional dependence to validate the robustness of the results further.

For high-income European countries, Table 4 reports the results of the Im–Pesaran–Shin (IPS) unit root test for panel data, used to examine the stationarity of the panel data. This test evaluates the null hypothesis that each series contains a unit root against the alternative of stationarity.

The results indicate that, in levels, none of the variables exhibit statistical evidence of stationarity, as all p-values exceed the 0.05 threshold. For instance, life expectancy yields a p-value of 1.0000, while CO₂ emissions and fossil fuel consumption yield p-values of 0.3164 and 0.9577, respectively, suggesting that the null hypothesis of non-stationarity cannot be rejected.

When the variables are transformed into first differences, the test results change substantially, with all series displaying statistically significant p-values (p < 0.05). This case suggests that the variables become stationary after differencing, indicating that they are most likely integrated of order one. Such properties justify their inclusion in dynamic panel models such as GMM.

Nevertheless, it is important to consider that the IPS test relies on the assumption of cross-sectional independence and may be affected by heterogeneity across panel units. In the European context, where countries are economically and environmentally interconnected, cross-sectional dependence is a plausible concern. Consequently, the stationarity results should be interpreted with caution, as they may not fully reflect the complexity of the underlying panel structure.

For the group of medium-income countries, Table 9 reports the results of the Im–Pesaran–Shin (IPS) unit root test applied to the main variables to determine their integration orders. This test is used in panel-data models to verify the presence of a unit root (H₀) versus the alternative hypothesis of common stationarity (H₁). According to statistical criteria, p-values below 0.05 indicate rejection of the null hypothesis and suggest that the series is stationary.

The results show that none of the variables is stationary in levels, since all exhibit p-values greater than 0.05. For example, the life expectancy variable has a p-value of 1.0000, and CO₂ emissions a p-value of 0.2573, indicating the presence of a unit root in their original form. However, when the test is applied to first differences, several variables become stationary: life expectancy (p = 0.0025), CO₂ emissions (p = 0.0023), renewable energy use (p = 0.0003), and GDP per capita (p = 0.0439). These results suggest that the variables are likely integrated of order one, satisfying the conditions for their inclusion in dynamic models such as GMM.

However, it is important to note that the IPS test assumes cross-sectional independence and may be sensitive to panel heterogeneity. Given the integration of European economies and potential interdependencies among countries, the presence of cross-sectional dependence cannot be ruled out. Therefore, these results should be interpreted with caution, as they may not fully capture the underlying data structure.

Reviewer comment (R.C): The decision to differ all variables based solely on IPS results ignores the possibility of cross-sectional dependence, which is likely in EU countries. Second-generation panel unit root tests would be more appropriate.

Response:
Thank you for this insightful comment. We agree that the IPS test, as a first-generation panel unit root test, relies on the assumption of cross-sectional independence, which may not fully hold in the context of European Union countries.
In the revised manuscript, we clarify that the decision to transform variables into first differences is not based solely on the IPS test results. Rather, it reflects a standard and deliberate econometric choice within the dynamic panel GMM framework, where differencing is commonly employed to eliminate unobserved heterogeneity and to ensure estimator consistency in the presence of non-stationary variables.
While the IPS results provide indicative evidence that the variables are integrated of order one, the transformation is also motivated by the need to avoid spurious regression outcomes and to properly specify the dynamic structure of the model.
At the same time, we explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present among European countries due to shared economic and environmental shocks. As a result, the unit root findings should be interpreted with caution.
We have incorporated this discussion into the methodology section and explicitly identified this as a limitation of the study. In addition, we suggest that future research could employ second-generation panel unit root tests that explicitly account for cross-sectional dependence to further validate the robustness of the results.

The decision to transform the variables into first differences is primarily guided by the IPS unit root test results, which suggest non-stationarity in levels and stationarity after differencing. However, it is important to recognize that this decision is not based solely on the IPS outcomes. Given the model's dynamic nature and the need to avoid spurious regression results, differencing is also consistent with standard practices in GMM estimation when the variables are integrated of order one.

At the same time, it is acknowledged that the IPS test assumes cross-sectional independence, which may not fully hold in the context of European Union countries due to their economic and environmental interconnections. Therefore, while the transformation provides a consistent estimation framework, the potential presence of cross-sectional dependence represents a limitation. Future research could address this issue by applying second-generation panel unit root tests that explicitly account for such dependencies.

Reviewer comment (R.C): The descriptive statistics are interpreted normatively (e.g., “this reflects strong welfare systems”), which goes beyond what summary statistics can support. These interpretations should be reserved for regression analysis.

Response:
Thank you for this important observation. We agree that the interpretation of descriptive statistics should remain strictly descriptive and avoid normative or causal inferences.

In the revised manuscript, we have revised the wording of the descriptive analysis (Tables 2 and 3) to focus exclusively on the statistical properties of the data, such as means, dispersion, and ranges.

All interpretative and explanatory statements have been moved to the regression and discussion sections, where they are supported by econometric evidence. These revisions improve the clarity and methodological rigor of the manuscript.

The descriptive results presented in Table 2 provide an overview of the behavior of the variables analyzed in high-income European countries.
First, Life expectancy averages 80.26 years, with low variability (standard deviation of 1.42) and a range of 76.54 to 83.05 years. These values indicate relatively high and homogeneous longevity levels across the countries analyzed.
Regarding fossil fuel use, the mean is 74.07% of the energy matrix, with considerable dispersion (standard deviation of 19.82) and values ranging from 27.14% to 99.08%, suggesting substantial differences in energy dependence across countries.
CO₂ emissions average 9.82 metric tons per capita, ranging from 3.54 to 25.95, indicating heterogeneity in environmental pressure among countries.
Renewable energy use averages 18.13%, with high variability (standard deviation of 14.85%) and values ranging from 1.28% to 57.83, reflecting uneven adoption of renewable energy sources.
Finally, GDP per capita (log-transformed) averages 10.77, with low dispersion (0.29) and a range of 10.41 to 11.62, indicating relatively similar income levels within this group.

The results presented in Table 3 summarize the descriptive statistics for medium-income European Union countries. Life expectancy averages 77.45 years, with a standard deviation of 3.34 and a range of 70.81 to 83.47 years, indicating moderate variability across countries.
Fossil fuel use averages 82.78%, with a standard deviation of 8.22, reflecting differences in energy dependence among countries.
CO₂ emissions average 6.80 metric tons per capita (standard deviation of 2.16), ranging from 3.80 to 12.50, indicating variation in environmental pressure levels.
Renewable energy use averages 15.70% (standard deviation of 7.37), ranging from 4.63% to 33.61%, suggesting differences in the adoption of renewable energy sources.
GDP per capita (log) averages 9.62, with a standard deviation of 0.52 and a range of 8.22 to 10.43, reflecting variation in economic performance within the group.

Reviewer comment (R.C): The figures (pages 9 and 11) visually suggest heterogeneous country trajectories, yet the econometric model assumes homogeneous slope coefficients within groups. This assumption should be tested or relaxed (e.g., via heterogeneous panel models).

Response:
Thank you for this valuable observation. We agree that the graphical analysis highlights heterogeneous trajectories across countries.

In the revised manuscript, we have clarified that, while the econometric model accounts for heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. We now explicitly acknowledge this as a simplifying assumption and clarify that the estimated coefficients should be interpreted as average group effects rather than country-specific relationships.

The graphical evidence presented in Figures 1 and 2 suggests the presence of heterogeneous trajectories across countries, particularly in the evolution of environmental and economic variables. While the econometric model addresses part of this heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. This simplifying assumption facilitates estimation but may not fully capture country-specific dynamics. Therefore, the results should be interpreted as average effects within groups rather than country-specific relationships.

Another important methodological consideration concerns the assumption of homogeneous slope coefficients across country groups in the econometric model. While the graphical analysis (Figures 1 and 2) reveals heterogeneous trajectories across countries, the GMM specification estimates average effects within each income group. Although this grouping strategy captures broad structural differences, it may not fully reflect country-specific dynamics.
Therefore, the estimated coefficients should be interpreted as average relationships rather than uniform effects across all countries within each group. Future research could address this limitation by applying heterogeneous panel data models, such as mean-group (MG) or pooled mean-group (PMG) estimators, which offer greater flexibility for capturing country-level heterogeneity.

Reviewer comment (R.C): The sign inconsistency in Table 13 for CO2 emissions (reported as negative in the text but positive in the table before sign interpretation) suggests either a typographical error or misinterpretation that needs correction.

Response:
Thank you for this important observation. We have carefully reviewed the results and confirm that the coefficient for CO₂ emissions in Table 13 is negative and statistically significant.

The inconsistency identified was due to an imprecise interpretation in the text. In the revised manuscript, we have corrected the wording to ensure full consistency between the reported coefficient and its interpretation.

We now clearly state that CO₂ emissions have a negative and significant effect on life expectancy in medium-income countries, reflecting the direct adverse impact of environmental degradation in these contexts. These revisions improve the internal consistency and clarity of the results section.

Second, fossil fuel use shows a positive coefficient (0.43), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), confirming the expected adverse impact of environmental degradation on population health in medium-income countries. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, likely reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.      Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The confidence intervals reported in Table 13 appear inconsistent (e.g., extremely wide ranges for GDP per capita), indicating potential scaling or reporting errors that should be verified.

Response:
Thank you for this important observation. We have carefully reviewed the confidence intervals reported in Table 13 and identified a typographical error in the interval associated with GDP per capita.

In the revised manuscript, we have corrected the confidence interval to ensure consistency with the reported coefficient and standard error. The updated values now accurately reflect the estimation results.

This correction does not affect the statistical significance or interpretation of the variable, which remains non-significant. These revisions improve the accuracy and clarity of the reported results.

Group variable: COUNTRY Time variable: YEAR
Number of obs
189

Obs per group

min
18

Avg
18.90

Max
19

Number of instruments
8
Prob > chi2
0.0000

D.Life expectancy
Coef.
Robust
P>|t|
[95% Conf. Interval]

Std. Err.

Life expectancy

LD.
1.309437
2.430333
0.603
-4.188359    6.807233

Use of fossil fuels

LD.
0.0433404
0.0371055
0.273
-0.0405981    0.1272788

CO2 emissions

LD.
-0.7106007
0.3042238
0.044
-0.0223987   - 1.398803

Use of renewable energy as a primary source of energy

LD.
0.2687355
0.1170912
0.047
0.0038567    0.5336143

lnGDP per capita

LD.
0.1148107
0.9513675
0.258
-1.74              1.97

cons
-2.276037
1.801168
0.238
-63.50561    17.98487

The dynamic panel-data estimation results for medium-income European countries are presented in Table 13. First, lagged life expectancy shows a positive coefficient (1.30), although not statistically significant (p = 0.603), suggesting weak temporal persistence of this variable within the group of countries analyzed.
Second, fossil fuel use shows a positive coefficient (0.043), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), consistent with the expected adverse impact of environmental degradation on population health in medium-income countries. The associated confidence interval is fully negative, reinforcing the robustness of this relationship. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, possibly reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.

Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The discussion attributes differences between country groups to “institutional capacity,” but no robustness checks (e.g., interaction terms, subgroup regressions with institutional proxies) are provided to substantiate this claim

Response:
Thank you for this important observation. We agree that the role of institutional capacity is not directly tested within the econometric model.

In the revised manuscript, we have revised the discussion to avoid causal interpretations and instead present institutional capacity as a contextual and theoretical explanation for the observed differences between country groups. We now explicitly clarify that these interpretations are not empirically tested within the model.

Additionally, we have acknowledged this as a limitation and indicated that future research could incorporate robustness checks, such as interaction terms or institutional proxies, to more directly assess the role of governance factors.

These revisions improve the conceptual clarity and methodological rigor of the manuscript.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The model does not address potential cross-country spillovers (e.g., pollution transport, EU-wide policies), which could bias estimates in a regional panel such as the European Union.

Response:
Thank you for this insightful comment. We agree that, in the context of the European Union, cross-country spillovers—such as transboundary pollution and common policy frameworks—may influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not account for spatial dependence or cross-country interactions. We now emphasize that the estimated coefficients should be interpreted as average effects, potentially influenced by unobserved spillovers.

Additionally, we have highlighted that future research could benefit from applying spatial panel-data models or related approaches to better capture these dynamics. These revisions improve the study's transparency and methodological robustness.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The time period (2000–2020) includes structural breaks such as the 2008 financial crisis and COVID-19 onset, yet no structural break tests or period controls are included, which may distort results.

Response:
Thank you for this insightful observation. We agree that the study period includes major structural events that could influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not incorporate structural break tests or period-specific controls. We now emphasize that the estimated coefficients should be interpreted as average effects over the full period, which may mask short-term disruptions associated with these events.

Additionally, we highlight that future research could incorporate structural break tests or time-specific dummy variables to capture the impact of such shocks better. These revisions improve the transparency and robustness of the study.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted considering that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The argument that renewable energy improves life expectancy lacks a clear transmission mechanism in the empirical model. Mediating variables such as air quality improvements are not included, limiting interpretability.

Response:
Thank you for this insightful comment. We agree that the empirical model does not explicitly capture the mechanisms through which renewable energy may influence life expectancy.

In the revised manuscript, we have refined the discussion to clarify that the observed relationship should be interpreted as an indirect association rather than a direct causal effect. We now explicitly acknowledge that this relationship may operate through unobserved mechanisms, such as improvements in environmental quality or reductions in pollution exposure.

Additionally, we highlight that the absence of mediating variables—such as air quality indicators—limits the ability to identify specific transmission channels. We have included this as a limitation and suggested that future research incorporate such variables to better understand these pathways.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The conclusions extend policy recommendations (e.g., ecological welfare state transformation) that are not directly tested in the empirical model, resulting in overgeneralization beyond the evidence presented.

This study contributes to the literature by examining the relationship between environmental degradation and life expectancy in the European Union, highlighting how these dynamics differ across countries with distinct economic and structural characteristics. The empirical results indicate that the effects of environmental factors on population health are not uniform, but vary depending on contextual conditions associated with income levels and broader institutional environments.
In high-income European countries, the findings suggest that environmental variables, such as CO₂ emissions and renewable energy use, do not have statistically significant effects on life expectancy, whereas fossil fuel consumption shows a positive association. However, this result should be interpreted with caution, as it may reflect underlying structural factors—such as advanced healthcare systems, higher levels of economic development, or other omitted variables—rather than a direct causal relationship.
In contrast, medium-income European countries show a more pronounced sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively and significantly associated with life expectancy, indicating that environmental pressures may translate more directly into adverse health outcomes in these contexts. At the same time, renewable energy use is positively associated with life expectancy, though this relationship should be interpreted as an indirect effect rather than a direct causal effect, as the model does not explicitly capture the underlying mechanisms.
Overall, the results suggest that environmental and economic factors interact with broader contextual conditions, which may include institutional, structural, and socio-economic elements, although these are not directly modeled in the empirical specification. Therefore, the findings should be interpreted as reduced-form relationships rather than structurally identified causal effects.
From a policy perspective, the results support the relevance of differentiated approaches that account for country-specific conditions when addressing environmental and public health challenges. However, these implications should be interpreted with caution. While the study adopts an institutional framework to contextualize the findings, the econometric model does not directly test specific institutional mechanisms or policy interventions.
Consequently, broader proposals—such as the transformation toward an “ecological welfare state”—should be understood as theoretical extensions consistent with the study's conceptual framework, rather than as empirically validated conclusions. Future research could build on these findings by incorporating explicit institutional indicators, mediating variables (such as air quality), or alternative modeling approaches that allow for the identification of causal mechanisms and policy effects.
In addition, further studies could extend the analysis by accounting for spatial interactions, structural breaks, and heterogeneous country dynamics, which may play a significant role in shaping environmental and health outcomes in integrated regions such as the European Union. Addressing these dimensions would contribute to a more comprehensive understanding of the complex relationships between environmental degradation, institutional conditions, and population well-being.
Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of the study. Below, we address each comment in detail and describe the corresponding revisions implemented in the manuscript.

Reviewer Comment: The abstract presents strong causal language (e.g., “CO2 emissions significantly reduce life expectancy”) despite the empirical strategy relying on a difference GMM framework that primarily identifies associations rather than strict causality. The wording should be revised to reflect conditional correlations unless additional identification strategies are implemented.

Response:
We appreciate this important observation. The abstract has been revised to remove any causal language and to reflect the empirical strategy's associative nature. Expressions such as “reduce life expectancy” have been replaced with “are negatively associated with life expectancy.” Additionally, the wording throughout the manuscript has been carefully revised to ensure consistency with a conditional and non-causal interpretation of the results.

Reviewer comment (R.C): The claim that institutional quality mediates the relationship is central to the paper, yet no explicit institutional quality variable is included in the econometric model (see model specification on page 7 ). This case creates a disconnect between theory and empirics. A measurable proxy (e.g., governance indicators) should be incorporated, or the claims should be softened.

Response:

We thank the reviewer for pointing out this important issue. In the revised manuscript, we have clarified that institutional quality is not empirically modeled as a mediating variable. Instead, institutional theory is used strictly as an interpretive framework to contextualize the observed empirical patterns. We explicitly state that:

The study does not aim to identify or estimate institutional mediation mechanisms econometrically.

References to institutional capacity are not intended to provide causal explanations but rather to offer theoretically grounded interpretations consistent with the literature.

Furthermore, we have clarified that heterogeneity across countries is partially reflected in income-based groupings, which capture broad structural and institutional differences. However, we explicitly acknowledge that this approach does not allow for a direct assessment of institutional effects.

Building on this, institutionalist theory posits that the design and strength of public institutions shape and condition the effects of environmental degradation on social outcomes, including health [9], [10], rather than acting as directly measured mediating mechanisms within the empirical model.

Accordingly, life expectancy should be linked to the institutional capacity to integrate environmental justice, territorial resilience, and intergenerational equity into public policy. However, in this study, such institutional dimensions are not directly measured but are considered as contextual factors shaping the observed relationships [11], [12].

However, the magnitude and direction of these effects are expected to vary across countries with different institutional and socioeconomic contexts.

Reviewer comment (R.C): The classification of countries into “high-income” and “middle-income” groups lacks a clear methodological justification. It is unclear whether this follows World Bank thresholds, sample medians, or arbitrary grouping, which raises concerns about reproducibility and potential sample selection bias.

Response:
We have clarified that countries are classified based on their average GDP per capita over the study period (2000–2020), using a relative within-sample approach. This represents a deliberate methodological choice aimed at ensuring internal comparability and capturing structural heterogeneity within the dataset.
We also explicitly acknowledge that this classification does not follow external benchmarks (e.g., World Bank thresholds) and may introduce sample-specific limitations. This transparency enhances reproducibility and interpretability.

To ensure a more representative analysis of Europe’s economic and environmental realities, countries were classified into two groups based on their average GDP per capita over the study period (2000–2020). Specifically, countries with higher average incomes were categorized as high-income, while those with lower average incomes were classified as middle-income. This classification follows a relative-grouping approach within the sample, enabling the identification of structural differences in economic capacity, institutional context, and environmental dynamics.
Although this classification does not strictly adhere to external thresholds, such as those defined by the World Bank, it provides a consistent, internally comparable framework for analyzing heterogeneity across countries. However, this approach may introduce some limitations related to sample-specific grouping, which should be considered when interpreting the results.

Reviewer comment (R.C): The transformation of all variables into first differences (Δ) eliminates long-run information. Given that life expectancy evolves slowly, this approach may obscure meaningful long-term relationships. A cointegration framework or system GMM could better capture both short- and long-run dynamics.

Response:
We agree that transforming variables into first differences limits the ability to capture long-run relationships. In the revised manuscript, we clarify that this transformation reflects a deliberate econometric trade-off.
Specifically:

Differencing is used to ensure consistency and avoid spurious regression results in the presence of non-stationarity and potential endogeneity.

The dynamic GMM framework prioritizes short-run dynamics over long-run equilibrium relationships.

We explicitly acknowledge that long-run relationships are not identified in this specification and suggest that future research could explore cointegration approaches or system GMM to address this limitation.

The transformation of variables into first differences is consistent with the unit root test results, which indicate that the series are integrated of order one. This approach helps to avoid spurious regressions and ensures the validity of the GMM estimation. However, it is important to acknowledge that differencing may reduce the ability to capture long-run relationships, particularly for variables such as life expectancy that evolve gradually over time. Therefore, the results should be interpreted as reflecting short-term dynamics rather than long-term equilibrium relationships.

Reviewer Comment: The specification includes lagged differences of the dependent variable, but the interpretation of “limited persistence” is questionable, especially given the relatively high coefficient (0.59) in high-income countries, which is economically non-trivial despite statistical insignificance.

Response:
Thank you for this insightful observation. We agree that the original interpretation of “limited persistence” was not fully appropriate given the magnitude of the estimated coefficient.

In the revised manuscript, we have refined the interpretation to acknowledge that, although the coefficient is not statistically significant, its magnitude suggests a moderate degree of persistence in life expectancy. At the same time, we emphasize that the lack of statistical significance prevents drawing robust conclusions about temporal dependence. This revision improves the consistency between the statistical results and their economic interpretation.

Although the coefficient on the lagged dependent variable is not statistically significant, its magnitude suggests moderate persistence in life expectancy. However, this result should be interpreted with caution, as the lack of statistical significance prevents drawing strong conclusions about temporal dependence.

Reviewer comment (R.C): The interpretation of the positive and significant coefficient of fossil fuel consumption in high-income countries as a “paradox” is not sufficiently rigorous. Alternative explanations, such as omitted variable bias, reverse causality, or measurement issues, should be explored before attributing this to institutional buffering.

Response:
We have substantially revised the interpretation of this result. The positive and significant association between fossil fuel consumption and life expectancy in high-income countries is no longer presented as a paradox.
Instead, we explicitly state that:

The result should not be interpreted as evidence of a beneficial effect of fossil fuel consumption.

It likely reflects underlying structural correlations (e.g., higher income levels, advanced healthcare systems, or omitted variables).

Alternative explanations such as omitted variable bias, reverse causality, and measurement limitations are explicitly discussed.

This revision ensures a cautious and methodologically consistent interpretation.

This study indicates that, in these countries, fossil fuel consumption is associated with higher life expectancy. A significant result is observed for fossil fuel use, with a positive and statistically significant coefficient. This finding indicates a positive association between fossil fuel consumption and life expectancy in high-income countries. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may instead reflect omitted-variable bias, reverse causality, or measurement issues. For example, higher-income economies with advanced healthcare systems and infrastructure may simultaneously exhibit higher energy consumption and better health outcomes.

The most striking finding is a positive, statistically significant relationship between fossil fuel consumption and life expectancy, which appears counterintuitive given the expected negative environmental impacts. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may be driven by underlying structural factors, such as higher levels of economic development, more advanced healthcare systems, or omitted variables correlated with both energy consumption and population health outcomes. Additionally, potential issues such as reverse causality or measurement limitations cannot be ruled out. Therefore, attributing this pattern solely to institutional capacity would be premature without further empirical evidence.

Reviewer comment (R.C): The paper states that CO2 emissions have “direct effects on air quality,” yet CO2 is not a local pollutant affecting health directly in the same way as particulate matter. This conceptual imprecision weakens the theoretical argument and should be corrected.

Response:
We have corrected the conceptual treatment of CO₂ emissions. The manuscript now clearly states that CO₂ is not a local air pollutant with direct health effects, but rather a driver of climate change with indirect impacts on health through environmental and climatic pathways.

CO₂ emissions, in contrast, represent a cumulative environmental outcome associated with climate change processes, which may indirectly affect human health through environmental and climatic pathways rather than direct exposure effects, particularly via mechanisms such as temperature increases and environmental degradation [24], [38].

CO₂ emissions—recognized as a primary contributor to global greenhouse gases—originate mainly from the combustion of fossil fuels in electricity generation, transportation, and industrial production. However, CO₂ should not be considered a local air pollutant with direct effects on human health. Instead, its relevance lies in its contribution to climate change, which may indirectly affect health outcomes through mechanisms such as rising temperatures, increased climate variability, and broader environmental degradation.

The role of CO₂ emissions in this study should be interpreted within a broader environmental context. Unlike local air pollutants such as particulate matter, CO₂ does not directly affect human health through immediate exposure. Instead, its impact is primarily indirect, as it contributes to climate change, which in turn influences health outcomes through environmental and socio-economic pathways.

Reviewer comment (R.C): The absence of key control variables (e.g., healthcare expenditure, education, urbanization, demographic structure) raises serious concerns about omitted variable bias, especially given the complexity of determinants of life expectancy.

Response:
We acknowledge that excluding variables such as healthcare expenditure, education, urbanization, and demographic structure may introduce omitted-variable bias.
We clarify that the model adopts a parsimonious specification as a deliberate trade-off, balancing:

Data comparability across countries and time

Model tractability

Avoidance of instrument proliferation in the GMM framework

We explicitly state that the results should be interpreted as conditional associations rather than causal effects, and we identify the inclusion of additional controls as an avenue for future research.

While the model includes key environmental and economic variables, it does not incorporate additional controls such as healthcare expenditure, education, urbanization, or demographic structure. This choice reflects a parsimonious specification based on data availability and comparability across countries and over time. However, excluding these variables may introduce omitted-variable bias, as multiple socio-economic and demographic factors influence life expectancy. Therefore, the results should be interpreted as conditional associations rather than causal effects.

Reviewer comment (R.C): The VIF analysis includes “life expectancy” as an explanatory variable (Table 6), which is inappropriate since it is the dependent variable. This suggests either a reporting error or a misunderstanding of multicollinearity diagnostics.

Response:
Thank you for this important observation. We acknowledge that the inclusion of the dependent variable (life expectancy) in the VIF analysis was inappropriate.

In the revised manuscript, we have corrected Table 6 by removing the dependent variable and ensuring that the VIF analysis includes only explanatory variables. We have also clarified the corresponding text to reflect that multicollinearity diagnostics apply exclusively to independent variables.

These revisions correct the reporting error and improve the methodological accuracy of the manuscript.

Table 6 presents the Variance Inflation Factors (VIFs) for the explanatory variables in the model. This test evaluates the degree of collinearity among independent variables, which is essential to ensure the validity of econometric estimates. As a general rule, VIF values below 5 indicate no serious multicollinearity, while values below 2 are considered optimal. The results show that all explanatory variables have low VIF values, indicating no problematic multicollinearity in the model.
Table 6. Multicolinealidad — High-income European countries
Variables
VIF

Use of fossil fuels
1.82

CO2 emissions
1.40

Use of renewable energy as a primary source of energy
1.26

lnGDP per capita
1.24

In this case, the results indicate no serious multicollinearity among variables, since all values fall below the critical threshold. The variable use of fossil fuels (3.29) has the highest VIFs, though they remain within a manageable range. Other variables—CO₂ emissions (1.32), renewable energy use (1.27), and GDP per capita (1.04)—show low collinearity.
Table 11. Multicollinearity — Medium-income European countries
Variables
VIF

Use of fossil fuels
3.29

CO2 emissions
1.32

Use of renewable energy as a primary source of energy
1.27

lnGDP per capita
1.04

Reviewer comment (R.C): The Hansen test p-values (0.600 and 0.396) are interpreted as evidence of valid instruments, but no discussion is provided regarding instrument proliferation or instrument count relative to sample size, which is critical in GMM estimation.

Response
Although the Hansen test results suggest that the instruments are valid, we agree that this test alone is insufficient to assess instrument quality fully. Therefore, we have expanded the discussion to explicitly consider the number of instruments relative to the sample size, a critical aspect of GMM estimation.
In this study, the number of instruments (8) remains well below the number of cross-sectional units, following a conservative specification strategy to reduce the risk of instrument proliferation and overfitting. This approach enhances the reliability of the Hansen test and the robustness of the estimated coefficients.
Nevertheless, as in any dynamic panel model, we acknowledge that the results may be sensitive to instrument specification, and this limitation is now explicitly recognized in the manuscript.

Although the Hansen test results suggest that the instruments are valid, it is also important to consider the number of instruments relative to the sample size. In this study, the number of instruments (8) remains well below the number of cross-sectional units, reducing the risk of instrument proliferation and overfitting. This study supports the reliability of the GMM estimates. Nevertheless, as in any dynamic panel model, the potential sensitivity to instrument specification should be acknowledged.

Reviewer comment (R.C): The number of instruments (8) appears low relative to the model structure, but the paper does not clarify how instruments were constructed or whether collapse or lag limits were applied. This lack of transparency limits replicability.

Response:
In the GMM estimation, instruments were constructed using lagged values of the endogenous variables, following the standard Arellano–Bond difference GMM approach.
To address concerns related to instrument proliferation and overfitting, we implemented a deliberately conservative specification strategy. Specifically, the number of instruments was restricted by (i) limiting the lag depth used for instrument generation and (ii) collapsing the instrument matrix.
This procedure ensures a parsimonious instrument set while preserving the validity of the moment conditions. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and enhances both the robustness and replicability of the estimates.
These methodological details have now been explicitly incorporated into the revised manuscript to improve transparency.

In the GMM estimation, instruments were constructed from lagged endogenous variables, following the standard Arellano–Bond approach. To avoid overfitting and instrument proliferation, the number of instruments was deliberately restricted by limiting the lag depth and collapsing the instrument matrix. This procedure ensures a parsimonious specification while maintaining the validity of the instruments. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and helps improve the robustness and replicability of the estimates.

Reviewer comment (R.C): The Arellano–Bond test results show no AR(1) autocorrelation, which is unusual since first-differenced residuals typically exhibit AR(1) by construction. This case raises concerns about either reporting accuracy or model implementation.

Response:
We thank the reviewer for this important observation. The Arellano–Bond test results indicate no statistically significant autocorrelation in the first-differenced residuals at either the AR(1) or AR(2) levels for both high- and medium-income country groups.
We acknowledge that the absence of AR(1) autocorrelation is uncommon in first-differenced specifications, where negative first-order serial correlation is typically expected by construction. However, this result does not necessarily imply misspecification or implementation error.
In particular, the lack of significant AR(1) autocorrelation may reflect sample-specific characteristics, such as limited time variation after differencing, relatively smooth dynamics in life expectancy, or the restricted instrument set used in the estimation. These factors can weaken the detection of first-order serial correlation in finite samples.
Importantly, the absence of AR(2) autocorrelation— which is the key requirement for the validity of the moment conditions— is satisfied in all specifications. Therefore, the main consistency condition of the difference GMM estimator remains intact.
Nevertheless, we agree that this result should be interpreted with caution. In the revised manuscript, we explicitly acknowledge that the Arellano–Bond test does not provide strong confirmatory evidence for the model's dynamic structure and that the results should be interpreted accordingly.

The Arellano–Bond test results for high-income European countries indicate no statistically significant autocorrelation in the first-differenced residuals at either AR(1) or AR(2) levels. While the absence of AR(2) autocorrelation is consistent with the validity of the model, the lack of significant AR(1) autocorrelation is less typical in first-differenced specifications and should be interpreted with caution. This outcome may reflect sample-specific characteristics or limitations in the model specification and, therefore, does not provide strong confirmatory evidence of the model's dynamic structure.
The results of the Arellano–Bond test for medium-income European countries indicate that no statistically significant autocorrelation is detected in the first-differenced residuals at either the AR(1) or AR(2) levels. While the absence of second-order autocorrelation (AR(2)) is consistent with the model's assumptions, the lack of evidence for first-order autocorrelation (AR(1)) is somewhat unusual in this type of specification and should therefore be interpreted with caution. This pattern may be associated with specific characteristics of the sample or potential limitations in the model specification, and thus does not provide conclusive confirmation of the estimated dynamic structure.

Reviewer comment (R.C): The IPS unit root test results are reported as showing non-stationarity at levels, yet p-values such as 0.3164 (CO2) suggest failure to reject the null, but the discussion does not consider panel heterogeneity or cross-sectional dependence.

Response:
Thank you for this important observation. We agree that the Im–Pesaran–Shin (IPS) unit root test results should be interpreted with caution, particularly given its underlying assumptions of cross-sectional independence and potential sensitivity to panel heterogeneity.
In the revised manuscript, we have clarified the interpretation of the IPS results. Specifically, p-values such as 0.3164 for CO₂ emissions correctly indicate a failure to reject the null hypothesis of a unit root in levels, which is consistent with the conclusion that the variables are non-stationary in their original form. We have revised the wording to ensure that this interpretation is explicitly and accurately conveyed.
Furthermore, we now explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present in the context of European Union countries due to their economic and environmental interconnections. As a result, the stationarity findings may be affected by common shocks or interdependencies across panel units.
Accordingly, we emphasize in the revised manuscript that the unit root results should be interpreted as indicative rather than definitive. We also clarify that the decision to transform variables into first differences is not based solely on the IPS test, but is additionally supported by standard econometric practice in dynamic panel models when variables are likely integrated of order one.
Finally, we have explicitly identified this limitation and suggested that future research could apply second-generation panel unit root tests that account for cross-sectional dependence to validate the robustness of the results further.

For high-income European countries, Table 4 reports the results of the Im–Pesaran–Shin (IPS) unit root test for panel data, used to examine the stationarity of the panel data. This test evaluates the null hypothesis that each series contains a unit root against the alternative of stationarity.

The results indicate that, in levels, none of the variables exhibit statistical evidence of stationarity, as all p-values exceed the 0.05 threshold. For instance, life expectancy yields a p-value of 1.0000, while CO₂ emissions and fossil fuel consumption yield p-values of 0.3164 and 0.9577, respectively, suggesting that the null hypothesis of non-stationarity cannot be rejected.

When the variables are transformed into first differences, the test results change substantially, with all series displaying statistically significant p-values (p < 0.05). This case suggests that the variables become stationary after differencing, indicating that they are most likely integrated of order one. Such properties justify their inclusion in dynamic panel models such as GMM.

Nevertheless, it is important to consider that the IPS test relies on the assumption of cross-sectional independence and may be affected by heterogeneity across panel units. In the European context, where countries are economically and environmentally interconnected, cross-sectional dependence is a plausible concern. Consequently, the stationarity results should be interpreted with caution, as they may not fully reflect the complexity of the underlying panel structure.

For the group of medium-income countries, Table 9 reports the results of the Im–Pesaran–Shin (IPS) unit root test applied to the main variables to determine their integration orders. This test is used in panel-data models to verify the presence of a unit root (H₀) versus the alternative hypothesis of common stationarity (H₁). According to statistical criteria, p-values below 0.05 indicate rejection of the null hypothesis and suggest that the series is stationary.

The results show that none of the variables is stationary in levels, since all exhibit p-values greater than 0.05. For example, the life expectancy variable has a p-value of 1.0000, and CO₂ emissions a p-value of 0.2573, indicating the presence of a unit root in their original form. However, when the test is applied to first differences, several variables become stationary: life expectancy (p = 0.0025), CO₂ emissions (p = 0.0023), renewable energy use (p = 0.0003), and GDP per capita (p = 0.0439). These results suggest that the variables are likely integrated of order one, satisfying the conditions for their inclusion in dynamic models such as GMM.

However, it is important to note that the IPS test assumes cross-sectional independence and may be sensitive to panel heterogeneity. Given the integration of European economies and potential interdependencies among countries, the presence of cross-sectional dependence cannot be ruled out. Therefore, these results should be interpreted with caution, as they may not fully capture the underlying data structure.

Reviewer comment (R.C): The decision to differ all variables based solely on IPS results ignores the possibility of cross-sectional dependence, which is likely in EU countries. Second-generation panel unit root tests would be more appropriate.

Response:
Thank you for this insightful comment. We agree that the IPS test, as a first-generation panel unit root test, relies on the assumption of cross-sectional independence, which may not fully hold in the context of European Union countries.
In the revised manuscript, we clarify that the decision to transform variables into first differences is not based solely on the IPS test results. Rather, it reflects a standard and deliberate econometric choice within the dynamic panel GMM framework, where differencing is commonly employed to eliminate unobserved heterogeneity and to ensure estimator consistency in the presence of non-stationary variables.
While the IPS results provide indicative evidence that the variables are integrated of order one, the transformation is also motivated by the need to avoid spurious regression outcomes and to properly specify the dynamic structure of the model.
At the same time, we explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present among European countries due to shared economic and environmental shocks. As a result, the unit root findings should be interpreted with caution.
We have incorporated this discussion into the methodology section and explicitly identified this as a limitation of the study. In addition, we suggest that future research could employ second-generation panel unit root tests that explicitly account for cross-sectional dependence to further validate the robustness of the results.

The decision to transform the variables into first differences is primarily guided by the IPS unit root test results, which suggest non-stationarity in levels and stationarity after differencing. However, it is important to recognize that this decision is not based solely on the IPS outcomes. Given the model's dynamic nature and the need to avoid spurious regression results, differencing is also consistent with standard practices in GMM estimation when the variables are integrated of order one.

At the same time, it is acknowledged that the IPS test assumes cross-sectional independence, which may not fully hold in the context of European Union countries due to their economic and environmental interconnections. Therefore, while the transformation provides a consistent estimation framework, the potential presence of cross-sectional dependence represents a limitation. Future research could address this issue by applying second-generation panel unit root tests that explicitly account for such dependencies.

Reviewer comment (R.C): The descriptive statistics are interpreted normatively (e.g., “this reflects strong welfare systems”), which goes beyond what summary statistics can support. These interpretations should be reserved for regression analysis.

Response:
Thank you for this important observation. We agree that the interpretation of descriptive statistics should remain strictly descriptive and avoid normative or causal inferences.

In the revised manuscript, we have revised the wording of the descriptive analysis (Tables 2 and 3) to focus exclusively on the statistical properties of the data, such as means, dispersion, and ranges.

All interpretative and explanatory statements have been moved to the regression and discussion sections, where they are supported by econometric evidence. These revisions improve the clarity and methodological rigor of the manuscript.

The descriptive results presented in Table 2 provide an overview of the behavior of the variables analyzed in high-income European countries.
First, Life expectancy averages 80.26 years, with low variability (standard deviation of 1.42) and a range of 76.54 to 83.05 years. These values indicate relatively high and homogeneous longevity levels across the countries analyzed.
Regarding fossil fuel use, the mean is 74.07% of the energy matrix, with considerable dispersion (standard deviation of 19.82) and values ranging from 27.14% to 99.08%, suggesting substantial differences in energy dependence across countries.
CO₂ emissions average 9.82 metric tons per capita, ranging from 3.54 to 25.95, indicating heterogeneity in environmental pressure among countries.
Renewable energy use averages 18.13%, with high variability (standard deviation of 14.85%) and values ranging from 1.28% to 57.83, reflecting uneven adoption of renewable energy sources.
Finally, GDP per capita (log-transformed) averages 10.77, with low dispersion (0.29) and a range of 10.41 to 11.62, indicating relatively similar income levels within this group.

The results presented in Table 3 summarize the descriptive statistics for medium-income European Union countries. Life expectancy averages 77.45 years, with a standard deviation of 3.34 and a range of 70.81 to 83.47 years, indicating moderate variability across countries.
Fossil fuel use averages 82.78%, with a standard deviation of 8.22, reflecting differences in energy dependence among countries.
CO₂ emissions average 6.80 metric tons per capita (standard deviation of 2.16), ranging from 3.80 to 12.50, indicating variation in environmental pressure levels.
Renewable energy use averages 15.70% (standard deviation of 7.37), ranging from 4.63% to 33.61%, suggesting differences in the adoption of renewable energy sources.
GDP per capita (log) averages 9.62, with a standard deviation of 0.52 and a range of 8.22 to 10.43, reflecting variation in economic performance within the group.

Reviewer comment (R.C): The figures (pages 9 and 11) visually suggest heterogeneous country trajectories, yet the econometric model assumes homogeneous slope coefficients within groups. This assumption should be tested or relaxed (e.g., via heterogeneous panel models).

Response:
Thank you for this valuable observation. We agree that the graphical analysis highlights heterogeneous trajectories across countries.

In the revised manuscript, we have clarified that, while the econometric model accounts for heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. We now explicitly acknowledge this as a simplifying assumption and clarify that the estimated coefficients should be interpreted as average group effects rather than country-specific relationships.

The graphical evidence presented in Figures 1 and 2 suggests the presence of heterogeneous trajectories across countries, particularly in the evolution of environmental and economic variables. While the econometric model addresses part of this heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. This simplifying assumption facilitates estimation but may not fully capture country-specific dynamics. Therefore, the results should be interpreted as average effects within groups rather than country-specific relationships.

Another important methodological consideration concerns the assumption of homogeneous slope coefficients across country groups in the econometric model. While the graphical analysis (Figures 1 and 2) reveals heterogeneous trajectories across countries, the GMM specification estimates average effects within each income group. Although this grouping strategy captures broad structural differences, it may not fully reflect country-specific dynamics.
Therefore, the estimated coefficients should be interpreted as average relationships rather than uniform effects across all countries within each group. Future research could address this limitation by applying heterogeneous panel data models, such as mean-group (MG) or pooled mean-group (PMG) estimators, which offer greater flexibility for capturing country-level heterogeneity.

Reviewer comment (R.C): The sign inconsistency in Table 13 for CO2 emissions (reported as negative in the text but positive in the table before sign interpretation) suggests either a typographical error or misinterpretation that needs correction.

Response:
Thank you for this important observation. We have carefully reviewed the results and confirm that the coefficient for CO₂ emissions in Table 13 is negative and statistically significant.

The inconsistency identified was due to an imprecise interpretation in the text. In the revised manuscript, we have corrected the wording to ensure full consistency between the reported coefficient and its interpretation.

We now clearly state that CO₂ emissions have a negative and significant effect on life expectancy in medium-income countries, reflecting the direct adverse impact of environmental degradation in these contexts. These revisions improve the internal consistency and clarity of the results section.

Second, fossil fuel use shows a positive coefficient (0.43), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), confirming the expected adverse impact of environmental degradation on population health in medium-income countries. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, likely reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.      Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The confidence intervals reported in Table 13 appear inconsistent (e.g., extremely wide ranges for GDP per capita), indicating potential scaling or reporting errors that should be verified.

Response:
Thank you for this important observation. We have carefully reviewed the confidence intervals reported in Table 13 and identified a typographical error in the interval associated with GDP per capita.

In the revised manuscript, we have corrected the confidence interval to ensure consistency with the reported coefficient and standard error. The updated values now accurately reflect the estimation results.

This correction does not affect the statistical significance or interpretation of the variable, which remains non-significant. These revisions improve the accuracy and clarity of the reported results.

Group variable: COUNTRY Time variable: YEAR
Number of obs
189

Obs per group

min
18

Avg
18.90

Max
19

Number of instruments
8
Prob > chi2
0.0000

D.Life expectancy
Coef.
Robust
P>|t|
[95% Conf. Interval]

Std. Err.

Life expectancy

LD.
1.309437
2.430333
0.603
-4.188359    6.807233

Use of fossil fuels

LD.
0.0433404
0.0371055
0.273
-0.0405981    0.1272788

CO2 emissions

LD.
-0.7106007
0.3042238
0.044
-0.0223987   - 1.398803

Use of renewable energy as a primary source of energy

LD.
0.2687355
0.1170912
0.047
0.0038567    0.5336143

lnGDP per capita

LD.
0.1148107
0.9513675
0.258
-1.74              1.97

cons
-2.276037
1.801168
0.238
-63.50561    17.98487

The dynamic panel-data estimation results for medium-income European countries are presented in Table 13. First, lagged life expectancy shows a positive coefficient (1.30), although not statistically significant (p = 0.603), suggesting weak temporal persistence of this variable within the group of countries analyzed.
Second, fossil fuel use shows a positive coefficient (0.043), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), consistent with the expected adverse impact of environmental degradation on population health in medium-income countries. The associated confidence interval is fully negative, reinforcing the robustness of this relationship. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, possibly reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.

Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The discussion attributes differences between country groups to “institutional capacity,” but no robustness checks (e.g., interaction terms, subgroup regressions with institutional proxies) are provided to substantiate this claim

Response:
Thank you for this important observation. We agree that the role of institutional capacity is not directly tested within the econometric model.

In the revised manuscript, we have revised the discussion to avoid causal interpretations and instead present institutional capacity as a contextual and theoretical explanation for the observed differences between country groups. We now explicitly clarify that these interpretations are not empirically tested within the model.

Additionally, we have acknowledged this as a limitation and indicated that future research could incorporate robustness checks, such as interaction terms or institutional proxies, to more directly assess the role of governance factors.

These revisions improve the conceptual clarity and methodological rigor of the manuscript.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The model does not address potential cross-country spillovers (e.g., pollution transport, EU-wide policies), which could bias estimates in a regional panel such as the European Union.

Response:
Thank you for this insightful comment. We agree that, in the context of the European Union, cross-country spillovers—such as transboundary pollution and common policy frameworks—may influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not account for spatial dependence or cross-country interactions. We now emphasize that the estimated coefficients should be interpreted as average effects, potentially influenced by unobserved spillovers.

Additionally, we have highlighted that future research could benefit from applying spatial panel-data models or related approaches to better capture these dynamics. These revisions improve the study's transparency and methodological robustness.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The time period (2000–2020) includes structural breaks such as the 2008 financial crisis and COVID-19 onset, yet no structural break tests or period controls are included, which may distort results.

Response:
Thank you for this insightful observation. We agree that the study period includes major structural events that could influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not incorporate structural break tests or period-specific controls. We now emphasize that the estimated coefficients should be interpreted as average effects over the full period, which may mask short-term disruptions associated with these events.

Additionally, we highlight that future research could incorporate structural break tests or time-specific dummy variables to capture the impact of such shocks better. These revisions improve the transparency and robustness of the study.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted considering that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The argument that renewable energy improves life expectancy lacks a clear transmission mechanism in the empirical model. Mediating variables such as air quality improvements are not included, limiting interpretability.

Response:
Thank you for this insightful comment. We agree that the empirical model does not explicitly capture the mechanisms through which renewable energy may influence life expectancy.

In the revised manuscript, we have refined the discussion to clarify that the observed relationship should be interpreted as an indirect association rather than a direct causal effect. We now explicitly acknowledge that this relationship may operate through unobserved mechanisms, such as improvements in environmental quality or reductions in pollution exposure.

Additionally, we highlight that the absence of mediating variables—such as air quality indicators—limits the ability to identify specific transmission channels. We have included this as a limitation and suggested that future research incorporate such variables to better understand these pathways.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The conclusions extend policy recommendations (e.g., ecological welfare state transformation) that are not directly tested in the empirical model, resulting in overgeneralization beyond the evidence presented.

This study contributes to the literature by examining the relationship between environmental degradation and life expectancy in the European Union, highlighting how these dynamics differ across countries with distinct economic and structural characteristics. The empirical results indicate that the effects of environmental factors on population health are not uniform, but vary depending on contextual conditions associated with income levels and broader institutional environments.
In high-income European countries, the findings suggest that environmental variables, such as CO₂ emissions and renewable energy use, do not have statistically significant effects on life expectancy, whereas fossil fuel consumption shows a positive association. However, this result should be interpreted with caution, as it may reflect underlying structural factors—such as advanced healthcare systems, higher levels of economic development, or other omitted variables—rather than a direct causal relationship.
In contrast, medium-income European countries show a more pronounced sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively and significantly associated with life expectancy, indicating that environmental pressures may translate more directly into adverse health outcomes in these contexts. At the same time, renewable energy use is positively associated with life expectancy, though this relationship should be interpreted as an indirect effect rather than a direct causal effect, as the model does not explicitly capture the underlying mechanisms.
Overall, the results suggest that environmental and economic factors interact with broader contextual conditions, which may include institutional, structural, and socio-economic elements, although these are not directly modeled in the empirical specification. Therefore, the findings should be interpreted as reduced-form relationships rather than structurally identified causal effects.
From a policy perspective, the results support the relevance of differentiated approaches that account for country-specific conditions when addressing environmental and public health challenges. However, these implications should be interpreted with caution. While the study adopts an institutional framework to contextualize the findings, the econometric model does not directly test specific institutional mechanisms or policy interventions.
Consequently, broader proposals—such as the transformation toward an “ecological welfare state”—should be understood as theoretical extensions consistent with the study's conceptual framework, rather than as empirically validated conclusions. Future research could build on these findings by incorporating explicit institutional indicators, mediating variables (such as air quality), or alternative modeling approaches that allow for the identification of causal mechanisms and policy effects.
In addition, further studies could extend the analysis by accounting for spatial interactions, structural breaks, and heterogeneous country dynamics, which may play a significant role in shaping environmental and health outcomes in integrated regions such as the European Union. Addressing these dimensions would contribute to a more comprehensive understanding of the complex relationships between environmental degradation, institutional conditions, and population well-being.
Competing Interests: No competing interests were disclosed. Close
Report a concern
Respond or Comment

COMMENTS ON THIS REPORT

Author Response 27 Apr 2026

Patricia Acosta-Vargas, Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

27 Apr 2026

Author Response

Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the ... Continue reading Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience.” We sincerely appreciate the time and effort the reviewers dedicated to providing us with valuable feedback.

In order to carefully address all comments and ensure a comprehensive and high-quality response, I request an extension to submit our replies until May 10, 2026.

This additional time will allow us to offer a comprehensive and well-structured review that fully addresses the reviewers’ concerns.

Thank you very much for your consideration. I look forward to your kind confirmation.

Sincerely,
Prof. Patricia Acosta-Vargas
Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience.” We sincerely appreciate the time and effort the reviewers dedicated to providing us with valuable feedback.

In order to carefully address all comments and ensure a comprehensive and high-quality response, I request an extension to submit our replies until May 10, 2026.

This additional time will allow us to offer a comprehensive and well-structured review that fully addresses the reviewers’ concerns.

Thank you very much for your consideration. I look forward to your kind confirmation.

Sincerely,
Prof. Patricia Acosta-Vargas
Competing Interests: No competing interests were disclosed. Close
Report a concern
Author Response 29 Apr 2026

Patricia Acosta-Vargas, Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

29 Apr 2026

Author Response
Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of ... Continue reading
Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of the study. Below, we address each comment in detail and describe the corresponding revisions implemented in the manuscript.

Reviewer Comment: The abstract presents strong causal language (e.g., “CO2 emissions significantly reduce life expectancy”) despite the empirical strategy relying on a difference GMM framework that primarily identifies associations rather than strict causality. The wording should be revised to reflect conditional correlations unless additional identification strategies are implemented.

Response:
We appreciate this important observation. The abstract has been revised to remove any causal language and to reflect the empirical strategy's associative nature. Expressions such as “reduce life expectancy” have been replaced with “are negatively associated with life expectancy.” Additionally, the wording throughout the manuscript has been carefully revised to ensure consistency with a conditional and non-causal interpretation of the results.

Reviewer comment (R.C): The claim that institutional quality mediates the relationship is central to the paper, yet no explicit institutional quality variable is included in the econometric model (see model specification on page 7 ). This case creates a disconnect between theory and empirics. A measurable proxy (e.g., governance indicators) should be incorporated, or the claims should be softened.

Response:

We thank the reviewer for pointing out this important issue. In the revised manuscript, we have clarified that institutional quality is not empirically modeled as a mediating variable. Instead, institutional theory is used strictly as an interpretive framework to contextualize the observed empirical patterns. We explicitly state that:

The study does not aim to identify or estimate institutional mediation mechanisms econometrically.

References to institutional capacity are not intended to provide causal explanations but rather to offer theoretically grounded interpretations consistent with the literature.

Furthermore, we have clarified that heterogeneity across countries is partially reflected in income-based groupings, which capture broad structural and institutional differences. However, we explicitly acknowledge that this approach does not allow for a direct assessment of institutional effects.

Building on this, institutionalist theory posits that the design and strength of public institutions shape and condition the effects of environmental degradation on social outcomes, including health [9], [10], rather than acting as directly measured mediating mechanisms within the empirical model.

Accordingly, life expectancy should be linked to the institutional capacity to integrate environmental justice, territorial resilience, and intergenerational equity into public policy. However, in this study, such institutional dimensions are not directly measured but are considered as contextual factors shaping the observed relationships [11], [12].

However, the magnitude and direction of these effects are expected to vary across countries with different institutional and socioeconomic contexts.

Reviewer comment (R.C): The classification of countries into “high-income” and “middle-income” groups lacks a clear methodological justification. It is unclear whether this follows World Bank thresholds, sample medians, or arbitrary grouping, which raises concerns about reproducibility and potential sample selection bias.

Response:
We have clarified that countries are classified based on their average GDP per capita over the study period (2000–2020), using a relative within-sample approach. This represents a deliberate methodological choice aimed at ensuring internal comparability and capturing structural heterogeneity within the dataset.
We also explicitly acknowledge that this classification does not follow external benchmarks (e.g., World Bank thresholds) and may introduce sample-specific limitations. This transparency enhances reproducibility and interpretability.

To ensure a more representative analysis of Europe’s economic and environmental realities, countries were classified into two groups based on their average GDP per capita over the study period (2000–2020). Specifically, countries with higher average incomes were categorized as high-income, while those with lower average incomes were classified as middle-income. This classification follows a relative-grouping approach within the sample, enabling the identification of structural differences in economic capacity, institutional context, and environmental dynamics.
Although this classification does not strictly adhere to external thresholds, such as those defined by the World Bank, it provides a consistent, internally comparable framework for analyzing heterogeneity across countries. However, this approach may introduce some limitations related to sample-specific grouping, which should be considered when interpreting the results.

Reviewer comment (R.C): The transformation of all variables into first differences (Δ) eliminates long-run information. Given that life expectancy evolves slowly, this approach may obscure meaningful long-term relationships. A cointegration framework or system GMM could better capture both short- and long-run dynamics.

Response:
We agree that transforming variables into first differences limits the ability to capture long-run relationships. In the revised manuscript, we clarify that this transformation reflects a deliberate econometric trade-off.
Specifically:

Differencing is used to ensure consistency and avoid spurious regression results in the presence of non-stationarity and potential endogeneity.

The dynamic GMM framework prioritizes short-run dynamics over long-run equilibrium relationships.

We explicitly acknowledge that long-run relationships are not identified in this specification and suggest that future research could explore cointegration approaches or system GMM to address this limitation.

The transformation of variables into first differences is consistent with the unit root test results, which indicate that the series are integrated of order one. This approach helps to avoid spurious regressions and ensures the validity of the GMM estimation. However, it is important to acknowledge that differencing may reduce the ability to capture long-run relationships, particularly for variables such as life expectancy that evolve gradually over time. Therefore, the results should be interpreted as reflecting short-term dynamics rather than long-term equilibrium relationships.

Reviewer Comment: The specification includes lagged differences of the dependent variable, but the interpretation of “limited persistence” is questionable, especially given the relatively high coefficient (0.59) in high-income countries, which is economically non-trivial despite statistical insignificance.

Response:
Thank you for this insightful observation. We agree that the original interpretation of “limited persistence” was not fully appropriate given the magnitude of the estimated coefficient.

In the revised manuscript, we have refined the interpretation to acknowledge that, although the coefficient is not statistically significant, its magnitude suggests a moderate degree of persistence in life expectancy. At the same time, we emphasize that the lack of statistical significance prevents drawing robust conclusions about temporal dependence. This revision improves the consistency between the statistical results and their economic interpretation.

Although the coefficient on the lagged dependent variable is not statistically significant, its magnitude suggests moderate persistence in life expectancy. However, this result should be interpreted with caution, as the lack of statistical significance prevents drawing strong conclusions about temporal dependence.

Reviewer comment (R.C): The interpretation of the positive and significant coefficient of fossil fuel consumption in high-income countries as a “paradox” is not sufficiently rigorous. Alternative explanations, such as omitted variable bias, reverse causality, or measurement issues, should be explored before attributing this to institutional buffering.

Response:
We have substantially revised the interpretation of this result. The positive and significant association between fossil fuel consumption and life expectancy in high-income countries is no longer presented as a paradox.
Instead, we explicitly state that:

The result should not be interpreted as evidence of a beneficial effect of fossil fuel consumption.

It likely reflects underlying structural correlations (e.g., higher income levels, advanced healthcare systems, or omitted variables).

Alternative explanations such as omitted variable bias, reverse causality, and measurement limitations are explicitly discussed.

This revision ensures a cautious and methodologically consistent interpretation.

This study indicates that, in these countries, fossil fuel consumption is associated with higher life expectancy. A significant result is observed for fossil fuel use, with a positive and statistically significant coefficient. This finding indicates a positive association between fossil fuel consumption and life expectancy in high-income countries. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may instead reflect omitted-variable bias, reverse causality, or measurement issues. For example, higher-income economies with advanced healthcare systems and infrastructure may simultaneously exhibit higher energy consumption and better health outcomes.

The most striking finding is a positive, statistically significant relationship between fossil fuel consumption and life expectancy, which appears counterintuitive given the expected negative environmental impacts. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may be driven by underlying structural factors, such as higher levels of economic development, more advanced healthcare systems, or omitted variables correlated with both energy consumption and population health outcomes. Additionally, potential issues such as reverse causality or measurement limitations cannot be ruled out. Therefore, attributing this pattern solely to institutional capacity would be premature without further empirical evidence.

Reviewer comment (R.C): The paper states that CO2 emissions have “direct effects on air quality,” yet CO2 is not a local pollutant affecting health directly in the same way as particulate matter. This conceptual imprecision weakens the theoretical argument and should be corrected.

Response:
We have corrected the conceptual treatment of CO₂ emissions. The manuscript now clearly states that CO₂ is not a local air pollutant with direct health effects, but rather a driver of climate change with indirect impacts on health through environmental and climatic pathways.

CO₂ emissions, in contrast, represent a cumulative environmental outcome associated with climate change processes, which may indirectly affect human health through environmental and climatic pathways rather than direct exposure effects, particularly via mechanisms such as temperature increases and environmental degradation [24], [38].

CO₂ emissions—recognized as a primary contributor to global greenhouse gases—originate mainly from the combustion of fossil fuels in electricity generation, transportation, and industrial production. However, CO₂ should not be considered a local air pollutant with direct effects on human health. Instead, its relevance lies in its contribution to climate change, which may indirectly affect health outcomes through mechanisms such as rising temperatures, increased climate variability, and broader environmental degradation.

The role of CO₂ emissions in this study should be interpreted within a broader environmental context. Unlike local air pollutants such as particulate matter, CO₂ does not directly affect human health through immediate exposure. Instead, its impact is primarily indirect, as it contributes to climate change, which in turn influences health outcomes through environmental and socio-economic pathways.

Reviewer comment (R.C): The absence of key control variables (e.g., healthcare expenditure, education, urbanization, demographic structure) raises serious concerns about omitted variable bias, especially given the complexity of determinants of life expectancy.

Response:
We acknowledge that excluding variables such as healthcare expenditure, education, urbanization, and demographic structure may introduce omitted-variable bias.
We clarify that the model adopts a parsimonious specification as a deliberate trade-off, balancing:

Data comparability across countries and time

Model tractability

Avoidance of instrument proliferation in the GMM framework

We explicitly state that the results should be interpreted as conditional associations rather than causal effects, and we identify the inclusion of additional controls as an avenue for future research.

While the model includes key environmental and economic variables, it does not incorporate additional controls such as healthcare expenditure, education, urbanization, or demographic structure. This choice reflects a parsimonious specification based on data availability and comparability across countries and over time. However, excluding these variables may introduce omitted-variable bias, as multiple socio-economic and demographic factors influence life expectancy. Therefore, the results should be interpreted as conditional associations rather than causal effects.

Reviewer comment (R.C): The VIF analysis includes “life expectancy” as an explanatory variable (Table 6), which is inappropriate since it is the dependent variable. This suggests either a reporting error or a misunderstanding of multicollinearity diagnostics.

Response:
Thank you for this important observation. We acknowledge that the inclusion of the dependent variable (life expectancy) in the VIF analysis was inappropriate.

In the revised manuscript, we have corrected Table 6 by removing the dependent variable and ensuring that the VIF analysis includes only explanatory variables. We have also clarified the corresponding text to reflect that multicollinearity diagnostics apply exclusively to independent variables.

These revisions correct the reporting error and improve the methodological accuracy of the manuscript.

Table 6 presents the Variance Inflation Factors (VIFs) for the explanatory variables in the model. This test evaluates the degree of collinearity among independent variables, which is essential to ensure the validity of econometric estimates. As a general rule, VIF values below 5 indicate no serious multicollinearity, while values below 2 are considered optimal. The results show that all explanatory variables have low VIF values, indicating no problematic multicollinearity in the model.
Table 6. Multicolinealidad — High-income European countries
Variables
VIF

Use of fossil fuels
1.82

CO2 emissions
1.40

Use of renewable energy as a primary source of energy
1.26

lnGDP per capita
1.24

In this case, the results indicate no serious multicollinearity among variables, since all values fall below the critical threshold. The variable use of fossil fuels (3.29) has the highest VIFs, though they remain within a manageable range. Other variables—CO₂ emissions (1.32), renewable energy use (1.27), and GDP per capita (1.04)—show low collinearity.
Table 11. Multicollinearity — Medium-income European countries
Variables
VIF

Use of fossil fuels
3.29

CO2 emissions
1.32

Use of renewable energy as a primary source of energy
1.27

lnGDP per capita
1.04

Reviewer comment (R.C): The Hansen test p-values (0.600 and 0.396) are interpreted as evidence of valid instruments, but no discussion is provided regarding instrument proliferation or instrument count relative to sample size, which is critical in GMM estimation.

Response
Although the Hansen test results suggest that the instruments are valid, we agree that this test alone is insufficient to assess instrument quality fully. Therefore, we have expanded the discussion to explicitly consider the number of instruments relative to the sample size, a critical aspect of GMM estimation.
In this study, the number of instruments (8) remains well below the number of cross-sectional units, following a conservative specification strategy to reduce the risk of instrument proliferation and overfitting. This approach enhances the reliability of the Hansen test and the robustness of the estimated coefficients.
Nevertheless, as in any dynamic panel model, we acknowledge that the results may be sensitive to instrument specification, and this limitation is now explicitly recognized in the manuscript.

Although the Hansen test results suggest that the instruments are valid, it is also important to consider the number of instruments relative to the sample size. In this study, the number of instruments (8) remains well below the number of cross-sectional units, reducing the risk of instrument proliferation and overfitting. This study supports the reliability of the GMM estimates. Nevertheless, as in any dynamic panel model, the potential sensitivity to instrument specification should be acknowledged.

Reviewer comment (R.C): The number of instruments (8) appears low relative to the model structure, but the paper does not clarify how instruments were constructed or whether collapse or lag limits were applied. This lack of transparency limits replicability.

Response:
In the GMM estimation, instruments were constructed using lagged values of the endogenous variables, following the standard Arellano–Bond difference GMM approach.
To address concerns related to instrument proliferation and overfitting, we implemented a deliberately conservative specification strategy. Specifically, the number of instruments was restricted by (i) limiting the lag depth used for instrument generation and (ii) collapsing the instrument matrix.
This procedure ensures a parsimonious instrument set while preserving the validity of the moment conditions. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and enhances both the robustness and replicability of the estimates.
These methodological details have now been explicitly incorporated into the revised manuscript to improve transparency.

In the GMM estimation, instruments were constructed from lagged endogenous variables, following the standard Arellano–Bond approach. To avoid overfitting and instrument proliferation, the number of instruments was deliberately restricted by limiting the lag depth and collapsing the instrument matrix. This procedure ensures a parsimonious specification while maintaining the validity of the instruments. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and helps improve the robustness and replicability of the estimates.

Reviewer comment (R.C): The Arellano–Bond test results show no AR(1) autocorrelation, which is unusual since first-differenced residuals typically exhibit AR(1) by construction. This case raises concerns about either reporting accuracy or model implementation.

Response:
We thank the reviewer for this important observation. The Arellano–Bond test results indicate no statistically significant autocorrelation in the first-differenced residuals at either the AR(1) or AR(2) levels for both high- and medium-income country groups.
We acknowledge that the absence of AR(1) autocorrelation is uncommon in first-differenced specifications, where negative first-order serial correlation is typically expected by construction. However, this result does not necessarily imply misspecification or implementation error.
In particular, the lack of significant AR(1) autocorrelation may reflect sample-specific characteristics, such as limited time variation after differencing, relatively smooth dynamics in life expectancy, or the restricted instrument set used in the estimation. These factors can weaken the detection of first-order serial correlation in finite samples.
Importantly, the absence of AR(2) autocorrelation— which is the key requirement for the validity of the moment conditions— is satisfied in all specifications. Therefore, the main consistency condition of the difference GMM estimator remains intact.
Nevertheless, we agree that this result should be interpreted with caution. In the revised manuscript, we explicitly acknowledge that the Arellano–Bond test does not provide strong confirmatory evidence for the model's dynamic structure and that the results should be interpreted accordingly.

The Arellano–Bond test results for high-income European countries indicate no statistically significant autocorrelation in the first-differenced residuals at either AR(1) or AR(2) levels. While the absence of AR(2) autocorrelation is consistent with the validity of the model, the lack of significant AR(1) autocorrelation is less typical in first-differenced specifications and should be interpreted with caution. This outcome may reflect sample-specific characteristics or limitations in the model specification and, therefore, does not provide strong confirmatory evidence of the model's dynamic structure.
The results of the Arellano–Bond test for medium-income European countries indicate that no statistically significant autocorrelation is detected in the first-differenced residuals at either the AR(1) or AR(2) levels. While the absence of second-order autocorrelation (AR(2)) is consistent with the model's assumptions, the lack of evidence for first-order autocorrelation (AR(1)) is somewhat unusual in this type of specification and should therefore be interpreted with caution. This pattern may be associated with specific characteristics of the sample or potential limitations in the model specification, and thus does not provide conclusive confirmation of the estimated dynamic structure.

Reviewer comment (R.C): The IPS unit root test results are reported as showing non-stationarity at levels, yet p-values such as 0.3164 (CO2) suggest failure to reject the null, but the discussion does not consider panel heterogeneity or cross-sectional dependence.

Response:
Thank you for this important observation. We agree that the Im–Pesaran–Shin (IPS) unit root test results should be interpreted with caution, particularly given its underlying assumptions of cross-sectional independence and potential sensitivity to panel heterogeneity.
In the revised manuscript, we have clarified the interpretation of the IPS results. Specifically, p-values such as 0.3164 for CO₂ emissions correctly indicate a failure to reject the null hypothesis of a unit root in levels, which is consistent with the conclusion that the variables are non-stationary in their original form. We have revised the wording to ensure that this interpretation is explicitly and accurately conveyed.
Furthermore, we now explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present in the context of European Union countries due to their economic and environmental interconnections. As a result, the stationarity findings may be affected by common shocks or interdependencies across panel units.
Accordingly, we emphasize in the revised manuscript that the unit root results should be interpreted as indicative rather than definitive. We also clarify that the decision to transform variables into first differences is not based solely on the IPS test, but is additionally supported by standard econometric practice in dynamic panel models when variables are likely integrated of order one.
Finally, we have explicitly identified this limitation and suggested that future research could apply second-generation panel unit root tests that account for cross-sectional dependence to validate the robustness of the results further.

For high-income European countries, Table 4 reports the results of the Im–Pesaran–Shin (IPS) unit root test for panel data, used to examine the stationarity of the panel data. This test evaluates the null hypothesis that each series contains a unit root against the alternative of stationarity.

The results indicate that, in levels, none of the variables exhibit statistical evidence of stationarity, as all p-values exceed the 0.05 threshold. For instance, life expectancy yields a p-value of 1.0000, while CO₂ emissions and fossil fuel consumption yield p-values of 0.3164 and 0.9577, respectively, suggesting that the null hypothesis of non-stationarity cannot be rejected.

When the variables are transformed into first differences, the test results change substantially, with all series displaying statistically significant p-values (p < 0.05). This case suggests that the variables become stationary after differencing, indicating that they are most likely integrated of order one. Such properties justify their inclusion in dynamic panel models such as GMM.

Nevertheless, it is important to consider that the IPS test relies on the assumption of cross-sectional independence and may be affected by heterogeneity across panel units. In the European context, where countries are economically and environmentally interconnected, cross-sectional dependence is a plausible concern. Consequently, the stationarity results should be interpreted with caution, as they may not fully reflect the complexity of the underlying panel structure.

For the group of medium-income countries, Table 9 reports the results of the Im–Pesaran–Shin (IPS) unit root test applied to the main variables to determine their integration orders. This test is used in panel-data models to verify the presence of a unit root (H₀) versus the alternative hypothesis of common stationarity (H₁). According to statistical criteria, p-values below 0.05 indicate rejection of the null hypothesis and suggest that the series is stationary.

The results show that none of the variables is stationary in levels, since all exhibit p-values greater than 0.05. For example, the life expectancy variable has a p-value of 1.0000, and CO₂ emissions a p-value of 0.2573, indicating the presence of a unit root in their original form. However, when the test is applied to first differences, several variables become stationary: life expectancy (p = 0.0025), CO₂ emissions (p = 0.0023), renewable energy use (p = 0.0003), and GDP per capita (p = 0.0439). These results suggest that the variables are likely integrated of order one, satisfying the conditions for their inclusion in dynamic models such as GMM.

However, it is important to note that the IPS test assumes cross-sectional independence and may be sensitive to panel heterogeneity. Given the integration of European economies and potential interdependencies among countries, the presence of cross-sectional dependence cannot be ruled out. Therefore, these results should be interpreted with caution, as they may not fully capture the underlying data structure.

Reviewer comment (R.C): The decision to differ all variables based solely on IPS results ignores the possibility of cross-sectional dependence, which is likely in EU countries. Second-generation panel unit root tests would be more appropriate.

Response:
Thank you for this insightful comment. We agree that the IPS test, as a first-generation panel unit root test, relies on the assumption of cross-sectional independence, which may not fully hold in the context of European Union countries.
In the revised manuscript, we clarify that the decision to transform variables into first differences is not based solely on the IPS test results. Rather, it reflects a standard and deliberate econometric choice within the dynamic panel GMM framework, where differencing is commonly employed to eliminate unobserved heterogeneity and to ensure estimator consistency in the presence of non-stationary variables.
While the IPS results provide indicative evidence that the variables are integrated of order one, the transformation is also motivated by the need to avoid spurious regression outcomes and to properly specify the dynamic structure of the model.
At the same time, we explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present among European countries due to shared economic and environmental shocks. As a result, the unit root findings should be interpreted with caution.
We have incorporated this discussion into the methodology section and explicitly identified this as a limitation of the study. In addition, we suggest that future research could employ second-generation panel unit root tests that explicitly account for cross-sectional dependence to further validate the robustness of the results.

The decision to transform the variables into first differences is primarily guided by the IPS unit root test results, which suggest non-stationarity in levels and stationarity after differencing. However, it is important to recognize that this decision is not based solely on the IPS outcomes. Given the model's dynamic nature and the need to avoid spurious regression results, differencing is also consistent with standard practices in GMM estimation when the variables are integrated of order one.

At the same time, it is acknowledged that the IPS test assumes cross-sectional independence, which may not fully hold in the context of European Union countries due to their economic and environmental interconnections. Therefore, while the transformation provides a consistent estimation framework, the potential presence of cross-sectional dependence represents a limitation. Future research could address this issue by applying second-generation panel unit root tests that explicitly account for such dependencies.

Reviewer comment (R.C): The descriptive statistics are interpreted normatively (e.g., “this reflects strong welfare systems”), which goes beyond what summary statistics can support. These interpretations should be reserved for regression analysis.

Response:
Thank you for this important observation. We agree that the interpretation of descriptive statistics should remain strictly descriptive and avoid normative or causal inferences.

In the revised manuscript, we have revised the wording of the descriptive analysis (Tables 2 and 3) to focus exclusively on the statistical properties of the data, such as means, dispersion, and ranges.

All interpretative and explanatory statements have been moved to the regression and discussion sections, where they are supported by econometric evidence. These revisions improve the clarity and methodological rigor of the manuscript.

The descriptive results presented in Table 2 provide an overview of the behavior of the variables analyzed in high-income European countries.
First, Life expectancy averages 80.26 years, with low variability (standard deviation of 1.42) and a range of 76.54 to 83.05 years. These values indicate relatively high and homogeneous longevity levels across the countries analyzed.
Regarding fossil fuel use, the mean is 74.07% of the energy matrix, with considerable dispersion (standard deviation of 19.82) and values ranging from 27.14% to 99.08%, suggesting substantial differences in energy dependence across countries.
CO₂ emissions average 9.82 metric tons per capita, ranging from 3.54 to 25.95, indicating heterogeneity in environmental pressure among countries.
Renewable energy use averages 18.13%, with high variability (standard deviation of 14.85%) and values ranging from 1.28% to 57.83, reflecting uneven adoption of renewable energy sources.
Finally, GDP per capita (log-transformed) averages 10.77, with low dispersion (0.29) and a range of 10.41 to 11.62, indicating relatively similar income levels within this group.

The results presented in Table 3 summarize the descriptive statistics for medium-income European Union countries. Life expectancy averages 77.45 years, with a standard deviation of 3.34 and a range of 70.81 to 83.47 years, indicating moderate variability across countries.
Fossil fuel use averages 82.78%, with a standard deviation of 8.22, reflecting differences in energy dependence among countries.
CO₂ emissions average 6.80 metric tons per capita (standard deviation of 2.16), ranging from 3.80 to 12.50, indicating variation in environmental pressure levels.
Renewable energy use averages 15.70% (standard deviation of 7.37), ranging from 4.63% to 33.61%, suggesting differences in the adoption of renewable energy sources.
GDP per capita (log) averages 9.62, with a standard deviation of 0.52 and a range of 8.22 to 10.43, reflecting variation in economic performance within the group.

Reviewer comment (R.C): The figures (pages 9 and 11) visually suggest heterogeneous country trajectories, yet the econometric model assumes homogeneous slope coefficients within groups. This assumption should be tested or relaxed (e.g., via heterogeneous panel models).

Response:
Thank you for this valuable observation. We agree that the graphical analysis highlights heterogeneous trajectories across countries.

In the revised manuscript, we have clarified that, while the econometric model accounts for heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. We now explicitly acknowledge this as a simplifying assumption and clarify that the estimated coefficients should be interpreted as average group effects rather than country-specific relationships.

The graphical evidence presented in Figures 1 and 2 suggests the presence of heterogeneous trajectories across countries, particularly in the evolution of environmental and economic variables. While the econometric model addresses part of this heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. This simplifying assumption facilitates estimation but may not fully capture country-specific dynamics. Therefore, the results should be interpreted as average effects within groups rather than country-specific relationships.

Another important methodological consideration concerns the assumption of homogeneous slope coefficients across country groups in the econometric model. While the graphical analysis (Figures 1 and 2) reveals heterogeneous trajectories across countries, the GMM specification estimates average effects within each income group. Although this grouping strategy captures broad structural differences, it may not fully reflect country-specific dynamics.
Therefore, the estimated coefficients should be interpreted as average relationships rather than uniform effects across all countries within each group. Future research could address this limitation by applying heterogeneous panel data models, such as mean-group (MG) or pooled mean-group (PMG) estimators, which offer greater flexibility for capturing country-level heterogeneity.

Reviewer comment (R.C): The sign inconsistency in Table 13 for CO2 emissions (reported as negative in the text but positive in the table before sign interpretation) suggests either a typographical error or misinterpretation that needs correction.

Response:
Thank you for this important observation. We have carefully reviewed the results and confirm that the coefficient for CO₂ emissions in Table 13 is negative and statistically significant.

The inconsistency identified was due to an imprecise interpretation in the text. In the revised manuscript, we have corrected the wording to ensure full consistency between the reported coefficient and its interpretation.

We now clearly state that CO₂ emissions have a negative and significant effect on life expectancy in medium-income countries, reflecting the direct adverse impact of environmental degradation in these contexts. These revisions improve the internal consistency and clarity of the results section.

Second, fossil fuel use shows a positive coefficient (0.43), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), confirming the expected adverse impact of environmental degradation on population health in medium-income countries. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, likely reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.      Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The confidence intervals reported in Table 13 appear inconsistent (e.g., extremely wide ranges for GDP per capita), indicating potential scaling or reporting errors that should be verified.

Response:
Thank you for this important observation. We have carefully reviewed the confidence intervals reported in Table 13 and identified a typographical error in the interval associated with GDP per capita.

In the revised manuscript, we have corrected the confidence interval to ensure consistency with the reported coefficient and standard error. The updated values now accurately reflect the estimation results.

This correction does not affect the statistical significance or interpretation of the variable, which remains non-significant. These revisions improve the accuracy and clarity of the reported results.

Group variable: COUNTRY Time variable: YEAR
Number of obs
189

Obs per group

min
18

Avg
18.90

Max
19

Number of instruments
8
Prob > chi2
0.0000

D.Life expectancy
Coef.
Robust
P>|t|
[95% Conf. Interval]

Std. Err.

Life expectancy

LD.
1.309437
2.430333
0.603
-4.188359    6.807233

Use of fossil fuels

LD.
0.0433404
0.0371055
0.273
-0.0405981    0.1272788

CO2 emissions

LD.
-0.7106007
0.3042238
0.044
-0.0223987   - 1.398803

Use of renewable energy as a primary source of energy

LD.
0.2687355
0.1170912
0.047
0.0038567    0.5336143

lnGDP per capita

LD.
0.1148107
0.9513675
0.258
-1.74              1.97

cons
-2.276037
1.801168
0.238
-63.50561    17.98487

The dynamic panel-data estimation results for medium-income European countries are presented in Table 13. First, lagged life expectancy shows a positive coefficient (1.30), although not statistically significant (p = 0.603), suggesting weak temporal persistence of this variable within the group of countries analyzed.
Second, fossil fuel use shows a positive coefficient (0.043), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), consistent with the expected adverse impact of environmental degradation on population health in medium-income countries. The associated confidence interval is fully negative, reinforcing the robustness of this relationship. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, possibly reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.

Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The discussion attributes differences between country groups to “institutional capacity,” but no robustness checks (e.g., interaction terms, subgroup regressions with institutional proxies) are provided to substantiate this claim

Response:
Thank you for this important observation. We agree that the role of institutional capacity is not directly tested within the econometric model.

In the revised manuscript, we have revised the discussion to avoid causal interpretations and instead present institutional capacity as a contextual and theoretical explanation for the observed differences between country groups. We now explicitly clarify that these interpretations are not empirically tested within the model.

Additionally, we have acknowledged this as a limitation and indicated that future research could incorporate robustness checks, such as interaction terms or institutional proxies, to more directly assess the role of governance factors.

These revisions improve the conceptual clarity and methodological rigor of the manuscript.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The model does not address potential cross-country spillovers (e.g., pollution transport, EU-wide policies), which could bias estimates in a regional panel such as the European Union.

Response:
Thank you for this insightful comment. We agree that, in the context of the European Union, cross-country spillovers—such as transboundary pollution and common policy frameworks—may influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not account for spatial dependence or cross-country interactions. We now emphasize that the estimated coefficients should be interpreted as average effects, potentially influenced by unobserved spillovers.

Additionally, we have highlighted that future research could benefit from applying spatial panel-data models or related approaches to better capture these dynamics. These revisions improve the study's transparency and methodological robustness.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The time period (2000–2020) includes structural breaks such as the 2008 financial crisis and COVID-19 onset, yet no structural break tests or period controls are included, which may distort results.

Response:
Thank you for this insightful observation. We agree that the study period includes major structural events that could influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not incorporate structural break tests or period-specific controls. We now emphasize that the estimated coefficients should be interpreted as average effects over the full period, which may mask short-term disruptions associated with these events.

Additionally, we highlight that future research could incorporate structural break tests or time-specific dummy variables to capture the impact of such shocks better. These revisions improve the transparency and robustness of the study.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted considering that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The argument that renewable energy improves life expectancy lacks a clear transmission mechanism in the empirical model. Mediating variables such as air quality improvements are not included, limiting interpretability.

Response:
Thank you for this insightful comment. We agree that the empirical model does not explicitly capture the mechanisms through which renewable energy may influence life expectancy.

In the revised manuscript, we have refined the discussion to clarify that the observed relationship should be interpreted as an indirect association rather than a direct causal effect. We now explicitly acknowledge that this relationship may operate through unobserved mechanisms, such as improvements in environmental quality or reductions in pollution exposure.

Additionally, we highlight that the absence of mediating variables—such as air quality indicators—limits the ability to identify specific transmission channels. We have included this as a limitation and suggested that future research incorporate such variables to better understand these pathways.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The conclusions extend policy recommendations (e.g., ecological welfare state transformation) that are not directly tested in the empirical model, resulting in overgeneralization beyond the evidence presented.

This study contributes to the literature by examining the relationship between environmental degradation and life expectancy in the European Union, highlighting how these dynamics differ across countries with distinct economic and structural characteristics. The empirical results indicate that the effects of environmental factors on population health are not uniform, but vary depending on contextual conditions associated with income levels and broader institutional environments.
In high-income European countries, the findings suggest that environmental variables, such as CO₂ emissions and renewable energy use, do not have statistically significant effects on life expectancy, whereas fossil fuel consumption shows a positive association. However, this result should be interpreted with caution, as it may reflect underlying structural factors—such as advanced healthcare systems, higher levels of economic development, or other omitted variables—rather than a direct causal relationship.
In contrast, medium-income European countries show a more pronounced sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively and significantly associated with life expectancy, indicating that environmental pressures may translate more directly into adverse health outcomes in these contexts. At the same time, renewable energy use is positively associated with life expectancy, though this relationship should be interpreted as an indirect effect rather than a direct causal effect, as the model does not explicitly capture the underlying mechanisms.
Overall, the results suggest that environmental and economic factors interact with broader contextual conditions, which may include institutional, structural, and socio-economic elements, although these are not directly modeled in the empirical specification. Therefore, the findings should be interpreted as reduced-form relationships rather than structurally identified causal effects.
From a policy perspective, the results support the relevance of differentiated approaches that account for country-specific conditions when addressing environmental and public health challenges. However, these implications should be interpreted with caution. While the study adopts an institutional framework to contextualize the findings, the econometric model does not directly test specific institutional mechanisms or policy interventions.
Consequently, broader proposals—such as the transformation toward an “ecological welfare state”—should be understood as theoretical extensions consistent with the study's conceptual framework, rather than as empirically validated conclusions. Future research could build on these findings by incorporating explicit institutional indicators, mediating variables (such as air quality), or alternative modeling approaches that allow for the identification of causal mechanisms and policy effects.
In addition, further studies could extend the analysis by accounting for spatial interactions, structural breaks, and heterogeneous country dynamics, which may play a significant role in shaping environmental and health outcomes in integrated regions such as the European Union. Addressing these dimensions would contribute to a more comprehensive understanding of the complex relationships between environmental degradation, institutional conditions, and population well-being.
Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of the study. Below, we address each comment in detail and describe the corresponding revisions implemented in the manuscript.

Reviewer Comment: The abstract presents strong causal language (e.g., “CO2 emissions significantly reduce life expectancy”) despite the empirical strategy relying on a difference GMM framework that primarily identifies associations rather than strict causality. The wording should be revised to reflect conditional correlations unless additional identification strategies are implemented.

Response:
We appreciate this important observation. The abstract has been revised to remove any causal language and to reflect the empirical strategy's associative nature. Expressions such as “reduce life expectancy” have been replaced with “are negatively associated with life expectancy.” Additionally, the wording throughout the manuscript has been carefully revised to ensure consistency with a conditional and non-causal interpretation of the results.

Reviewer comment (R.C): The claim that institutional quality mediates the relationship is central to the paper, yet no explicit institutional quality variable is included in the econometric model (see model specification on page 7 ). This case creates a disconnect between theory and empirics. A measurable proxy (e.g., governance indicators) should be incorporated, or the claims should be softened.

Response:

We thank the reviewer for pointing out this important issue. In the revised manuscript, we have clarified that institutional quality is not empirically modeled as a mediating variable. Instead, institutional theory is used strictly as an interpretive framework to contextualize the observed empirical patterns. We explicitly state that:

The study does not aim to identify or estimate institutional mediation mechanisms econometrically.

References to institutional capacity are not intended to provide causal explanations but rather to offer theoretically grounded interpretations consistent with the literature.

Furthermore, we have clarified that heterogeneity across countries is partially reflected in income-based groupings, which capture broad structural and institutional differences. However, we explicitly acknowledge that this approach does not allow for a direct assessment of institutional effects.

Building on this, institutionalist theory posits that the design and strength of public institutions shape and condition the effects of environmental degradation on social outcomes, including health [9], [10], rather than acting as directly measured mediating mechanisms within the empirical model.

Accordingly, life expectancy should be linked to the institutional capacity to integrate environmental justice, territorial resilience, and intergenerational equity into public policy. However, in this study, such institutional dimensions are not directly measured but are considered as contextual factors shaping the observed relationships [11], [12].

However, the magnitude and direction of these effects are expected to vary across countries with different institutional and socioeconomic contexts.

Reviewer comment (R.C): The classification of countries into “high-income” and “middle-income” groups lacks a clear methodological justification. It is unclear whether this follows World Bank thresholds, sample medians, or arbitrary grouping, which raises concerns about reproducibility and potential sample selection bias.

Response:
We have clarified that countries are classified based on their average GDP per capita over the study period (2000–2020), using a relative within-sample approach. This represents a deliberate methodological choice aimed at ensuring internal comparability and capturing structural heterogeneity within the dataset.
We also explicitly acknowledge that this classification does not follow external benchmarks (e.g., World Bank thresholds) and may introduce sample-specific limitations. This transparency enhances reproducibility and interpretability.

To ensure a more representative analysis of Europe’s economic and environmental realities, countries were classified into two groups based on their average GDP per capita over the study period (2000–2020). Specifically, countries with higher average incomes were categorized as high-income, while those with lower average incomes were classified as middle-income. This classification follows a relative-grouping approach within the sample, enabling the identification of structural differences in economic capacity, institutional context, and environmental dynamics.
Although this classification does not strictly adhere to external thresholds, such as those defined by the World Bank, it provides a consistent, internally comparable framework for analyzing heterogeneity across countries. However, this approach may introduce some limitations related to sample-specific grouping, which should be considered when interpreting the results.

Reviewer comment (R.C): The transformation of all variables into first differences (Δ) eliminates long-run information. Given that life expectancy evolves slowly, this approach may obscure meaningful long-term relationships. A cointegration framework or system GMM could better capture both short- and long-run dynamics.

Response:
We agree that transforming variables into first differences limits the ability to capture long-run relationships. In the revised manuscript, we clarify that this transformation reflects a deliberate econometric trade-off.
Specifically:

Differencing is used to ensure consistency and avoid spurious regression results in the presence of non-stationarity and potential endogeneity.

The dynamic GMM framework prioritizes short-run dynamics over long-run equilibrium relationships.

We explicitly acknowledge that long-run relationships are not identified in this specification and suggest that future research could explore cointegration approaches or system GMM to address this limitation.

The transformation of variables into first differences is consistent with the unit root test results, which indicate that the series are integrated of order one. This approach helps to avoid spurious regressions and ensures the validity of the GMM estimation. However, it is important to acknowledge that differencing may reduce the ability to capture long-run relationships, particularly for variables such as life expectancy that evolve gradually over time. Therefore, the results should be interpreted as reflecting short-term dynamics rather than long-term equilibrium relationships.

Reviewer Comment: The specification includes lagged differences of the dependent variable, but the interpretation of “limited persistence” is questionable, especially given the relatively high coefficient (0.59) in high-income countries, which is economically non-trivial despite statistical insignificance.

Response:
Thank you for this insightful observation. We agree that the original interpretation of “limited persistence” was not fully appropriate given the magnitude of the estimated coefficient.

In the revised manuscript, we have refined the interpretation to acknowledge that, although the coefficient is not statistically significant, its magnitude suggests a moderate degree of persistence in life expectancy. At the same time, we emphasize that the lack of statistical significance prevents drawing robust conclusions about temporal dependence. This revision improves the consistency between the statistical results and their economic interpretation.

Although the coefficient on the lagged dependent variable is not statistically significant, its magnitude suggests moderate persistence in life expectancy. However, this result should be interpreted with caution, as the lack of statistical significance prevents drawing strong conclusions about temporal dependence.

Reviewer comment (R.C): The interpretation of the positive and significant coefficient of fossil fuel consumption in high-income countries as a “paradox” is not sufficiently rigorous. Alternative explanations, such as omitted variable bias, reverse causality, or measurement issues, should be explored before attributing this to institutional buffering.

Response:
We have substantially revised the interpretation of this result. The positive and significant association between fossil fuel consumption and life expectancy in high-income countries is no longer presented as a paradox.
Instead, we explicitly state that:

The result should not be interpreted as evidence of a beneficial effect of fossil fuel consumption.

It likely reflects underlying structural correlations (e.g., higher income levels, advanced healthcare systems, or omitted variables).

Alternative explanations such as omitted variable bias, reverse causality, and measurement limitations are explicitly discussed.

This revision ensures a cautious and methodologically consistent interpretation.

This study indicates that, in these countries, fossil fuel consumption is associated with higher life expectancy. A significant result is observed for fossil fuel use, with a positive and statistically significant coefficient. This finding indicates a positive association between fossil fuel consumption and life expectancy in high-income countries. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may instead reflect omitted-variable bias, reverse causality, or measurement issues. For example, higher-income economies with advanced healthcare systems and infrastructure may simultaneously exhibit higher energy consumption and better health outcomes.

The most striking finding is a positive, statistically significant relationship between fossil fuel consumption and life expectancy, which appears counterintuitive given the expected negative environmental impacts. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may be driven by underlying structural factors, such as higher levels of economic development, more advanced healthcare systems, or omitted variables correlated with both energy consumption and population health outcomes. Additionally, potential issues such as reverse causality or measurement limitations cannot be ruled out. Therefore, attributing this pattern solely to institutional capacity would be premature without further empirical evidence.

Reviewer comment (R.C): The paper states that CO2 emissions have “direct effects on air quality,” yet CO2 is not a local pollutant affecting health directly in the same way as particulate matter. This conceptual imprecision weakens the theoretical argument and should be corrected.

Response:
We have corrected the conceptual treatment of CO₂ emissions. The manuscript now clearly states that CO₂ is not a local air pollutant with direct health effects, but rather a driver of climate change with indirect impacts on health through environmental and climatic pathways.

CO₂ emissions, in contrast, represent a cumulative environmental outcome associated with climate change processes, which may indirectly affect human health through environmental and climatic pathways rather than direct exposure effects, particularly via mechanisms such as temperature increases and environmental degradation [24], [38].

CO₂ emissions—recognized as a primary contributor to global greenhouse gases—originate mainly from the combustion of fossil fuels in electricity generation, transportation, and industrial production. However, CO₂ should not be considered a local air pollutant with direct effects on human health. Instead, its relevance lies in its contribution to climate change, which may indirectly affect health outcomes through mechanisms such as rising temperatures, increased climate variability, and broader environmental degradation.

The role of CO₂ emissions in this study should be interpreted within a broader environmental context. Unlike local air pollutants such as particulate matter, CO₂ does not directly affect human health through immediate exposure. Instead, its impact is primarily indirect, as it contributes to climate change, which in turn influences health outcomes through environmental and socio-economic pathways.

Reviewer comment (R.C): The absence of key control variables (e.g., healthcare expenditure, education, urbanization, demographic structure) raises serious concerns about omitted variable bias, especially given the complexity of determinants of life expectancy.

Response:
We acknowledge that excluding variables such as healthcare expenditure, education, urbanization, and demographic structure may introduce omitted-variable bias.
We clarify that the model adopts a parsimonious specification as a deliberate trade-off, balancing:

Data comparability across countries and time

Model tractability

Avoidance of instrument proliferation in the GMM framework

We explicitly state that the results should be interpreted as conditional associations rather than causal effects, and we identify the inclusion of additional controls as an avenue for future research.

While the model includes key environmental and economic variables, it does not incorporate additional controls such as healthcare expenditure, education, urbanization, or demographic structure. This choice reflects a parsimonious specification based on data availability and comparability across countries and over time. However, excluding these variables may introduce omitted-variable bias, as multiple socio-economic and demographic factors influence life expectancy. Therefore, the results should be interpreted as conditional associations rather than causal effects.

Reviewer comment (R.C): The VIF analysis includes “life expectancy” as an explanatory variable (Table 6), which is inappropriate since it is the dependent variable. This suggests either a reporting error or a misunderstanding of multicollinearity diagnostics.

Response:
Thank you for this important observation. We acknowledge that the inclusion of the dependent variable (life expectancy) in the VIF analysis was inappropriate.

In the revised manuscript, we have corrected Table 6 by removing the dependent variable and ensuring that the VIF analysis includes only explanatory variables. We have also clarified the corresponding text to reflect that multicollinearity diagnostics apply exclusively to independent variables.

These revisions correct the reporting error and improve the methodological accuracy of the manuscript.

Table 6 presents the Variance Inflation Factors (VIFs) for the explanatory variables in the model. This test evaluates the degree of collinearity among independent variables, which is essential to ensure the validity of econometric estimates. As a general rule, VIF values below 5 indicate no serious multicollinearity, while values below 2 are considered optimal. The results show that all explanatory variables have low VIF values, indicating no problematic multicollinearity in the model.
Table 6. Multicolinealidad — High-income European countries
Variables
VIF

Use of fossil fuels
1.82

CO2 emissions
1.40

Use of renewable energy as a primary source of energy
1.26

lnGDP per capita
1.24

In this case, the results indicate no serious multicollinearity among variables, since all values fall below the critical threshold. The variable use of fossil fuels (3.29) has the highest VIFs, though they remain within a manageable range. Other variables—CO₂ emissions (1.32), renewable energy use (1.27), and GDP per capita (1.04)—show low collinearity.
Table 11. Multicollinearity — Medium-income European countries
Variables
VIF

Use of fossil fuels
3.29

CO2 emissions
1.32

Use of renewable energy as a primary source of energy
1.27

lnGDP per capita
1.04

Reviewer comment (R.C): The Hansen test p-values (0.600 and 0.396) are interpreted as evidence of valid instruments, but no discussion is provided regarding instrument proliferation or instrument count relative to sample size, which is critical in GMM estimation.

Response
Although the Hansen test results suggest that the instruments are valid, we agree that this test alone is insufficient to assess instrument quality fully. Therefore, we have expanded the discussion to explicitly consider the number of instruments relative to the sample size, a critical aspect of GMM estimation.
In this study, the number of instruments (8) remains well below the number of cross-sectional units, following a conservative specification strategy to reduce the risk of instrument proliferation and overfitting. This approach enhances the reliability of the Hansen test and the robustness of the estimated coefficients.
Nevertheless, as in any dynamic panel model, we acknowledge that the results may be sensitive to instrument specification, and this limitation is now explicitly recognized in the manuscript.

Although the Hansen test results suggest that the instruments are valid, it is also important to consider the number of instruments relative to the sample size. In this study, the number of instruments (8) remains well below the number of cross-sectional units, reducing the risk of instrument proliferation and overfitting. This study supports the reliability of the GMM estimates. Nevertheless, as in any dynamic panel model, the potential sensitivity to instrument specification should be acknowledged.

Reviewer comment (R.C): The number of instruments (8) appears low relative to the model structure, but the paper does not clarify how instruments were constructed or whether collapse or lag limits were applied. This lack of transparency limits replicability.

Response:
In the GMM estimation, instruments were constructed using lagged values of the endogenous variables, following the standard Arellano–Bond difference GMM approach.
To address concerns related to instrument proliferation and overfitting, we implemented a deliberately conservative specification strategy. Specifically, the number of instruments was restricted by (i) limiting the lag depth used for instrument generation and (ii) collapsing the instrument matrix.
This procedure ensures a parsimonious instrument set while preserving the validity of the moment conditions. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and enhances both the robustness and replicability of the estimates.
These methodological details have now been explicitly incorporated into the revised manuscript to improve transparency.

In the GMM estimation, instruments were constructed from lagged endogenous variables, following the standard Arellano–Bond approach. To avoid overfitting and instrument proliferation, the number of instruments was deliberately restricted by limiting the lag depth and collapsing the instrument matrix. This procedure ensures a parsimonious specification while maintaining the validity of the instruments. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and helps improve the robustness and replicability of the estimates.

Reviewer comment (R.C): The Arellano–Bond test results show no AR(1) autocorrelation, which is unusual since first-differenced residuals typically exhibit AR(1) by construction. This case raises concerns about either reporting accuracy or model implementation.

Response:
We thank the reviewer for this important observation. The Arellano–Bond test results indicate no statistically significant autocorrelation in the first-differenced residuals at either the AR(1) or AR(2) levels for both high- and medium-income country groups.
We acknowledge that the absence of AR(1) autocorrelation is uncommon in first-differenced specifications, where negative first-order serial correlation is typically expected by construction. However, this result does not necessarily imply misspecification or implementation error.
In particular, the lack of significant AR(1) autocorrelation may reflect sample-specific characteristics, such as limited time variation after differencing, relatively smooth dynamics in life expectancy, or the restricted instrument set used in the estimation. These factors can weaken the detection of first-order serial correlation in finite samples.
Importantly, the absence of AR(2) autocorrelation— which is the key requirement for the validity of the moment conditions— is satisfied in all specifications. Therefore, the main consistency condition of the difference GMM estimator remains intact.
Nevertheless, we agree that this result should be interpreted with caution. In the revised manuscript, we explicitly acknowledge that the Arellano–Bond test does not provide strong confirmatory evidence for the model's dynamic structure and that the results should be interpreted accordingly.

The Arellano–Bond test results for high-income European countries indicate no statistically significant autocorrelation in the first-differenced residuals at either AR(1) or AR(2) levels. While the absence of AR(2) autocorrelation is consistent with the validity of the model, the lack of significant AR(1) autocorrelation is less typical in first-differenced specifications and should be interpreted with caution. This outcome may reflect sample-specific characteristics or limitations in the model specification and, therefore, does not provide strong confirmatory evidence of the model's dynamic structure.
The results of the Arellano–Bond test for medium-income European countries indicate that no statistically significant autocorrelation is detected in the first-differenced residuals at either the AR(1) or AR(2) levels. While the absence of second-order autocorrelation (AR(2)) is consistent with the model's assumptions, the lack of evidence for first-order autocorrelation (AR(1)) is somewhat unusual in this type of specification and should therefore be interpreted with caution. This pattern may be associated with specific characteristics of the sample or potential limitations in the model specification, and thus does not provide conclusive confirmation of the estimated dynamic structure.

Reviewer comment (R.C): The IPS unit root test results are reported as showing non-stationarity at levels, yet p-values such as 0.3164 (CO2) suggest failure to reject the null, but the discussion does not consider panel heterogeneity or cross-sectional dependence.

Response:
Thank you for this important observation. We agree that the Im–Pesaran–Shin (IPS) unit root test results should be interpreted with caution, particularly given its underlying assumptions of cross-sectional independence and potential sensitivity to panel heterogeneity.
In the revised manuscript, we have clarified the interpretation of the IPS results. Specifically, p-values such as 0.3164 for CO₂ emissions correctly indicate a failure to reject the null hypothesis of a unit root in levels, which is consistent with the conclusion that the variables are non-stationary in their original form. We have revised the wording to ensure that this interpretation is explicitly and accurately conveyed.
Furthermore, we now explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present in the context of European Union countries due to their economic and environmental interconnections. As a result, the stationarity findings may be affected by common shocks or interdependencies across panel units.
Accordingly, we emphasize in the revised manuscript that the unit root results should be interpreted as indicative rather than definitive. We also clarify that the decision to transform variables into first differences is not based solely on the IPS test, but is additionally supported by standard econometric practice in dynamic panel models when variables are likely integrated of order one.
Finally, we have explicitly identified this limitation and suggested that future research could apply second-generation panel unit root tests that account for cross-sectional dependence to validate the robustness of the results further.

For high-income European countries, Table 4 reports the results of the Im–Pesaran–Shin (IPS) unit root test for panel data, used to examine the stationarity of the panel data. This test evaluates the null hypothesis that each series contains a unit root against the alternative of stationarity.

The results indicate that, in levels, none of the variables exhibit statistical evidence of stationarity, as all p-values exceed the 0.05 threshold. For instance, life expectancy yields a p-value of 1.0000, while CO₂ emissions and fossil fuel consumption yield p-values of 0.3164 and 0.9577, respectively, suggesting that the null hypothesis of non-stationarity cannot be rejected.

When the variables are transformed into first differences, the test results change substantially, with all series displaying statistically significant p-values (p < 0.05). This case suggests that the variables become stationary after differencing, indicating that they are most likely integrated of order one. Such properties justify their inclusion in dynamic panel models such as GMM.

Nevertheless, it is important to consider that the IPS test relies on the assumption of cross-sectional independence and may be affected by heterogeneity across panel units. In the European context, where countries are economically and environmentally interconnected, cross-sectional dependence is a plausible concern. Consequently, the stationarity results should be interpreted with caution, as they may not fully reflect the complexity of the underlying panel structure.

For the group of medium-income countries, Table 9 reports the results of the Im–Pesaran–Shin (IPS) unit root test applied to the main variables to determine their integration orders. This test is used in panel-data models to verify the presence of a unit root (H₀) versus the alternative hypothesis of common stationarity (H₁). According to statistical criteria, p-values below 0.05 indicate rejection of the null hypothesis and suggest that the series is stationary.

The results show that none of the variables is stationary in levels, since all exhibit p-values greater than 0.05. For example, the life expectancy variable has a p-value of 1.0000, and CO₂ emissions a p-value of 0.2573, indicating the presence of a unit root in their original form. However, when the test is applied to first differences, several variables become stationary: life expectancy (p = 0.0025), CO₂ emissions (p = 0.0023), renewable energy use (p = 0.0003), and GDP per capita (p = 0.0439). These results suggest that the variables are likely integrated of order one, satisfying the conditions for their inclusion in dynamic models such as GMM.

However, it is important to note that the IPS test assumes cross-sectional independence and may be sensitive to panel heterogeneity. Given the integration of European economies and potential interdependencies among countries, the presence of cross-sectional dependence cannot be ruled out. Therefore, these results should be interpreted with caution, as they may not fully capture the underlying data structure.

Reviewer comment (R.C): The decision to differ all variables based solely on IPS results ignores the possibility of cross-sectional dependence, which is likely in EU countries. Second-generation panel unit root tests would be more appropriate.

Response:
Thank you for this insightful comment. We agree that the IPS test, as a first-generation panel unit root test, relies on the assumption of cross-sectional independence, which may not fully hold in the context of European Union countries.
In the revised manuscript, we clarify that the decision to transform variables into first differences is not based solely on the IPS test results. Rather, it reflects a standard and deliberate econometric choice within the dynamic panel GMM framework, where differencing is commonly employed to eliminate unobserved heterogeneity and to ensure estimator consistency in the presence of non-stationary variables.
While the IPS results provide indicative evidence that the variables are integrated of order one, the transformation is also motivated by the need to avoid spurious regression outcomes and to properly specify the dynamic structure of the model.
At the same time, we explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present among European countries due to shared economic and environmental shocks. As a result, the unit root findings should be interpreted with caution.
We have incorporated this discussion into the methodology section and explicitly identified this as a limitation of the study. In addition, we suggest that future research could employ second-generation panel unit root tests that explicitly account for cross-sectional dependence to further validate the robustness of the results.

The decision to transform the variables into first differences is primarily guided by the IPS unit root test results, which suggest non-stationarity in levels and stationarity after differencing. However, it is important to recognize that this decision is not based solely on the IPS outcomes. Given the model's dynamic nature and the need to avoid spurious regression results, differencing is also consistent with standard practices in GMM estimation when the variables are integrated of order one.

At the same time, it is acknowledged that the IPS test assumes cross-sectional independence, which may not fully hold in the context of European Union countries due to their economic and environmental interconnections. Therefore, while the transformation provides a consistent estimation framework, the potential presence of cross-sectional dependence represents a limitation. Future research could address this issue by applying second-generation panel unit root tests that explicitly account for such dependencies.

Reviewer comment (R.C): The descriptive statistics are interpreted normatively (e.g., “this reflects strong welfare systems”), which goes beyond what summary statistics can support. These interpretations should be reserved for regression analysis.

Response:
Thank you for this important observation. We agree that the interpretation of descriptive statistics should remain strictly descriptive and avoid normative or causal inferences.

In the revised manuscript, we have revised the wording of the descriptive analysis (Tables 2 and 3) to focus exclusively on the statistical properties of the data, such as means, dispersion, and ranges.

All interpretative and explanatory statements have been moved to the regression and discussion sections, where they are supported by econometric evidence. These revisions improve the clarity and methodological rigor of the manuscript.

The descriptive results presented in Table 2 provide an overview of the behavior of the variables analyzed in high-income European countries.
First, Life expectancy averages 80.26 years, with low variability (standard deviation of 1.42) and a range of 76.54 to 83.05 years. These values indicate relatively high and homogeneous longevity levels across the countries analyzed.
Regarding fossil fuel use, the mean is 74.07% of the energy matrix, with considerable dispersion (standard deviation of 19.82) and values ranging from 27.14% to 99.08%, suggesting substantial differences in energy dependence across countries.
CO₂ emissions average 9.82 metric tons per capita, ranging from 3.54 to 25.95, indicating heterogeneity in environmental pressure among countries.
Renewable energy use averages 18.13%, with high variability (standard deviation of 14.85%) and values ranging from 1.28% to 57.83, reflecting uneven adoption of renewable energy sources.
Finally, GDP per capita (log-transformed) averages 10.77, with low dispersion (0.29) and a range of 10.41 to 11.62, indicating relatively similar income levels within this group.

The results presented in Table 3 summarize the descriptive statistics for medium-income European Union countries. Life expectancy averages 77.45 years, with a standard deviation of 3.34 and a range of 70.81 to 83.47 years, indicating moderate variability across countries.
Fossil fuel use averages 82.78%, with a standard deviation of 8.22, reflecting differences in energy dependence among countries.
CO₂ emissions average 6.80 metric tons per capita (standard deviation of 2.16), ranging from 3.80 to 12.50, indicating variation in environmental pressure levels.
Renewable energy use averages 15.70% (standard deviation of 7.37), ranging from 4.63% to 33.61%, suggesting differences in the adoption of renewable energy sources.
GDP per capita (log) averages 9.62, with a standard deviation of 0.52 and a range of 8.22 to 10.43, reflecting variation in economic performance within the group.

Reviewer comment (R.C): The figures (pages 9 and 11) visually suggest heterogeneous country trajectories, yet the econometric model assumes homogeneous slope coefficients within groups. This assumption should be tested or relaxed (e.g., via heterogeneous panel models).

Response:
Thank you for this valuable observation. We agree that the graphical analysis highlights heterogeneous trajectories across countries.

In the revised manuscript, we have clarified that, while the econometric model accounts for heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. We now explicitly acknowledge this as a simplifying assumption and clarify that the estimated coefficients should be interpreted as average group effects rather than country-specific relationships.

The graphical evidence presented in Figures 1 and 2 suggests the presence of heterogeneous trajectories across countries, particularly in the evolution of environmental and economic variables. While the econometric model addresses part of this heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. This simplifying assumption facilitates estimation but may not fully capture country-specific dynamics. Therefore, the results should be interpreted as average effects within groups rather than country-specific relationships.

Another important methodological consideration concerns the assumption of homogeneous slope coefficients across country groups in the econometric model. While the graphical analysis (Figures 1 and 2) reveals heterogeneous trajectories across countries, the GMM specification estimates average effects within each income group. Although this grouping strategy captures broad structural differences, it may not fully reflect country-specific dynamics.
Therefore, the estimated coefficients should be interpreted as average relationships rather than uniform effects across all countries within each group. Future research could address this limitation by applying heterogeneous panel data models, such as mean-group (MG) or pooled mean-group (PMG) estimators, which offer greater flexibility for capturing country-level heterogeneity.

Reviewer comment (R.C): The sign inconsistency in Table 13 for CO2 emissions (reported as negative in the text but positive in the table before sign interpretation) suggests either a typographical error or misinterpretation that needs correction.

Response:
Thank you for this important observation. We have carefully reviewed the results and confirm that the coefficient for CO₂ emissions in Table 13 is negative and statistically significant.

The inconsistency identified was due to an imprecise interpretation in the text. In the revised manuscript, we have corrected the wording to ensure full consistency between the reported coefficient and its interpretation.

We now clearly state that CO₂ emissions have a negative and significant effect on life expectancy in medium-income countries, reflecting the direct adverse impact of environmental degradation in these contexts. These revisions improve the internal consistency and clarity of the results section.

Second, fossil fuel use shows a positive coefficient (0.43), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), confirming the expected adverse impact of environmental degradation on population health in medium-income countries. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, likely reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.      Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The confidence intervals reported in Table 13 appear inconsistent (e.g., extremely wide ranges for GDP per capita), indicating potential scaling or reporting errors that should be verified.

Response:
Thank you for this important observation. We have carefully reviewed the confidence intervals reported in Table 13 and identified a typographical error in the interval associated with GDP per capita.

In the revised manuscript, we have corrected the confidence interval to ensure consistency with the reported coefficient and standard error. The updated values now accurately reflect the estimation results.

This correction does not affect the statistical significance or interpretation of the variable, which remains non-significant. These revisions improve the accuracy and clarity of the reported results.

Group variable: COUNTRY Time variable: YEAR
Number of obs
189

Obs per group

min
18

Avg
18.90

Max
19

Number of instruments
8
Prob > chi2
0.0000

D.Life expectancy
Coef.
Robust
P>|t|
[95% Conf. Interval]

Std. Err.

Life expectancy

LD.
1.309437
2.430333
0.603
-4.188359    6.807233

Use of fossil fuels

LD.
0.0433404
0.0371055
0.273
-0.0405981    0.1272788

CO2 emissions

LD.
-0.7106007
0.3042238
0.044
-0.0223987   - 1.398803

Use of renewable energy as a primary source of energy

LD.
0.2687355
0.1170912
0.047
0.0038567    0.5336143

lnGDP per capita

LD.
0.1148107
0.9513675
0.258
-1.74              1.97

cons
-2.276037
1.801168
0.238
-63.50561    17.98487

The dynamic panel-data estimation results for medium-income European countries are presented in Table 13. First, lagged life expectancy shows a positive coefficient (1.30), although not statistically significant (p = 0.603), suggesting weak temporal persistence of this variable within the group of countries analyzed.
Second, fossil fuel use shows a positive coefficient (0.043), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), consistent with the expected adverse impact of environmental degradation on population health in medium-income countries. The associated confidence interval is fully negative, reinforcing the robustness of this relationship. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, possibly reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.

Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The discussion attributes differences between country groups to “institutional capacity,” but no robustness checks (e.g., interaction terms, subgroup regressions with institutional proxies) are provided to substantiate this claim

Response:
Thank you for this important observation. We agree that the role of institutional capacity is not directly tested within the econometric model.

In the revised manuscript, we have revised the discussion to avoid causal interpretations and instead present institutional capacity as a contextual and theoretical explanation for the observed differences between country groups. We now explicitly clarify that these interpretations are not empirically tested within the model.

Additionally, we have acknowledged this as a limitation and indicated that future research could incorporate robustness checks, such as interaction terms or institutional proxies, to more directly assess the role of governance factors.

These revisions improve the conceptual clarity and methodological rigor of the manuscript.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The model does not address potential cross-country spillovers (e.g., pollution transport, EU-wide policies), which could bias estimates in a regional panel such as the European Union.

Response:
Thank you for this insightful comment. We agree that, in the context of the European Union, cross-country spillovers—such as transboundary pollution and common policy frameworks—may influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not account for spatial dependence or cross-country interactions. We now emphasize that the estimated coefficients should be interpreted as average effects, potentially influenced by unobserved spillovers.

Additionally, we have highlighted that future research could benefit from applying spatial panel-data models or related approaches to better capture these dynamics. These revisions improve the study's transparency and methodological robustness.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The time period (2000–2020) includes structural breaks such as the 2008 financial crisis and COVID-19 onset, yet no structural break tests or period controls are included, which may distort results.

Response:
Thank you for this insightful observation. We agree that the study period includes major structural events that could influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not incorporate structural break tests or period-specific controls. We now emphasize that the estimated coefficients should be interpreted as average effects over the full period, which may mask short-term disruptions associated with these events.

Additionally, we highlight that future research could incorporate structural break tests or time-specific dummy variables to capture the impact of such shocks better. These revisions improve the transparency and robustness of the study.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted considering that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The argument that renewable energy improves life expectancy lacks a clear transmission mechanism in the empirical model. Mediating variables such as air quality improvements are not included, limiting interpretability.

Response:
Thank you for this insightful comment. We agree that the empirical model does not explicitly capture the mechanisms through which renewable energy may influence life expectancy.

In the revised manuscript, we have refined the discussion to clarify that the observed relationship should be interpreted as an indirect association rather than a direct causal effect. We now explicitly acknowledge that this relationship may operate through unobserved mechanisms, such as improvements in environmental quality or reductions in pollution exposure.

Additionally, we highlight that the absence of mediating variables—such as air quality indicators—limits the ability to identify specific transmission channels. We have included this as a limitation and suggested that future research incorporate such variables to better understand these pathways.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The conclusions extend policy recommendations (e.g., ecological welfare state transformation) that are not directly tested in the empirical model, resulting in overgeneralization beyond the evidence presented.

This study contributes to the literature by examining the relationship between environmental degradation and life expectancy in the European Union, highlighting how these dynamics differ across countries with distinct economic and structural characteristics. The empirical results indicate that the effects of environmental factors on population health are not uniform, but vary depending on contextual conditions associated with income levels and broader institutional environments.
In high-income European countries, the findings suggest that environmental variables, such as CO₂ emissions and renewable energy use, do not have statistically significant effects on life expectancy, whereas fossil fuel consumption shows a positive association. However, this result should be interpreted with caution, as it may reflect underlying structural factors—such as advanced healthcare systems, higher levels of economic development, or other omitted variables—rather than a direct causal relationship.
In contrast, medium-income European countries show a more pronounced sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively and significantly associated with life expectancy, indicating that environmental pressures may translate more directly into adverse health outcomes in these contexts. At the same time, renewable energy use is positively associated with life expectancy, though this relationship should be interpreted as an indirect effect rather than a direct causal effect, as the model does not explicitly capture the underlying mechanisms.
Overall, the results suggest that environmental and economic factors interact with broader contextual conditions, which may include institutional, structural, and socio-economic elements, although these are not directly modeled in the empirical specification. Therefore, the findings should be interpreted as reduced-form relationships rather than structurally identified causal effects.
From a policy perspective, the results support the relevance of differentiated approaches that account for country-specific conditions when addressing environmental and public health challenges. However, these implications should be interpreted with caution. While the study adopts an institutional framework to contextualize the findings, the econometric model does not directly test specific institutional mechanisms or policy interventions.
Consequently, broader proposals—such as the transformation toward an “ecological welfare state”—should be understood as theoretical extensions consistent with the study's conceptual framework, rather than as empirically validated conclusions. Future research could build on these findings by incorporating explicit institutional indicators, mediating variables (such as air quality), or alternative modeling approaches that allow for the identification of causal mechanisms and policy effects.
In addition, further studies could extend the analysis by accounting for spatial interactions, structural breaks, and heterogeneous country dynamics, which may play a significant role in shaping environmental and health outcomes in integrated regions such as the European Union. Addressing these dimensions would contribute to a more comprehensive understanding of the complex relationships between environmental degradation, institutional conditions, and population well-being.
Competing Interests: No competing interests were disclosed. Close
Report a concern

Comments on this article Comments (0)

Version 2

VERSION 2 PUBLISHED 25 Mar 2026

Open Peer Review

Reviewer Status

Reviewer Reports

	Invited Reviewers
	1
Version 2 (revision) 18 May 26
Version 1 25 Mar 26	read

Matheus Koengkan, University of Coimbra, Coimbra, Portugal

Comments on this article

All Comments(0)

Add a comment

Browse by related subjects

Back to all reports

Reviewer Report

17 Views

15 Apr 2026 | for Version 1

Matheus Koengkan, University of Coimbra, Coimbra, Portugal

17 Views Cite this report Responses(2)

Approved With Reservations

Is the work clearly and accurately presented and does it cite the current literature?

Partly
Is the study design appropriate and is the work technically sound?

Partly
Are sufficient details of methods and analysis provided to allow replication by others?

Partly
If applicable, is the statistical analysis and its interpretation appropriate?

Partly
Are all the source data underlying the results available to ensure full reproducibility?

Partly
Are the conclusions drawn adequately supported by the results?

Partly

Competing Interests

No competing interests were disclosed.

Reviewer Expertise

Energy and Environmental Economics.

Respond to this report

Responses (2)

Author Response

27 Apr 2026

Patricia Acosta-Vargas, Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

Dear Reviewers,

I hope you are well.

Thank you for sharing the peer review report of our article, “Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience.” We sincerely appreciate the time and effort the reviewers dedicated to providing us with valuable feedback.

In order to carefully address all comments and ensure a comprehensive and high-quality response, I request an extension to submit our replies until May 10, 2026.

This additional time will allow us to offer a comprehensive and well-structured review that fully addresses the reviewers’ concerns.

Thank you very much for your consideration. I look forward to your kind confirmation.

Sincerely,
Prof. Patricia Acosta-Vargas

View more View less

Competing Interests

No competing interests were disclosed.

Author Response

29 Apr 2026

Patricia Acosta-Vargas, Intelligent and Interactive Systems Laboratory, Universidad de Las Americas, Quito, 170125, Ecuador

Response to Reviewer
We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. The comments have significantly improved the clarity, methodological rigor, and overall consistency of the study. Below, we address each comment in detail and describe the corresponding revisions implemented in the manuscript.

Reviewer Comment: The abstract presents strong causal language (e.g., “CO2 emissions significantly reduce life expectancy”) despite the empirical strategy relying on a difference GMM framework that primarily identifies associations rather than strict causality. The wording should be revised to reflect conditional correlations unless additional identification strategies are implemented.

Response:
We appreciate this important observation. The abstract has been revised to remove any causal language and to reflect the empirical strategy's associative nature. Expressions such as “reduce life expectancy” have been replaced with “are negatively associated with life expectancy.” Additionally, the wording throughout the manuscript has been carefully revised to ensure consistency with a conditional and non-causal interpretation of the results.

Reviewer comment (R.C): The claim that institutional quality mediates the relationship is central to the paper, yet no explicit institutional quality variable is included in the econometric model (see model specification on page 7 ). This case creates a disconnect between theory and empirics. A measurable proxy (e.g., governance indicators) should be incorporated, or the claims should be softened.

Response:

We thank the reviewer for pointing out this important issue. In the revised manuscript, we have clarified that institutional quality is not empirically modeled as a mediating variable. Instead, institutional theory is used strictly as an interpretive framework to contextualize the observed empirical patterns. We explicitly state that:

The study does not aim to identify or estimate institutional mediation mechanisms econometrically.
References to institutional capacity are not intended to provide causal explanations but rather to offer theoretically grounded interpretations consistent with the literature.

Furthermore, we have clarified that heterogeneity across countries is partially reflected in income-based groupings, which capture broad structural and institutional differences. However, we explicitly acknowledge that this approach does not allow for a direct assessment of institutional effects.

Building on this, institutionalist theory posits that the design and strength of public institutions shape and condition the effects of environmental degradation on social outcomes, including health [9], [10], rather than acting as directly measured mediating mechanisms within the empirical model.

Accordingly, life expectancy should be linked to the institutional capacity to integrate environmental justice, territorial resilience, and intergenerational equity into public policy. However, in this study, such institutional dimensions are not directly measured but are considered as contextual factors shaping the observed relationships [11], [12].

However, the magnitude and direction of these effects are expected to vary across countries with different institutional and socioeconomic contexts.

Reviewer comment (R.C): The classification of countries into “high-income” and “middle-income” groups lacks a clear methodological justification. It is unclear whether this follows World Bank thresholds, sample medians, or arbitrary grouping, which raises concerns about reproducibility and potential sample selection bias.

Response:
We have clarified that countries are classified based on their average GDP per capita over the study period (2000–2020), using a relative within-sample approach. This represents a deliberate methodological choice aimed at ensuring internal comparability and capturing structural heterogeneity within the dataset.
We also explicitly acknowledge that this classification does not follow external benchmarks (e.g., World Bank thresholds) and may introduce sample-specific limitations. This transparency enhances reproducibility and interpretability.

To ensure a more representative analysis of Europe’s economic and environmental realities, countries were classified into two groups based on their average GDP per capita over the study period (2000–2020). Specifically, countries with higher average incomes were categorized as high-income, while those with lower average incomes were classified as middle-income. This classification follows a relative-grouping approach within the sample, enabling the identification of structural differences in economic capacity, institutional context, and environmental dynamics.
Although this classification does not strictly adhere to external thresholds, such as those defined by the World Bank, it provides a consistent, internally comparable framework for analyzing heterogeneity across countries. However, this approach may introduce some limitations related to sample-specific grouping, which should be considered when interpreting the results.

Reviewer comment (R.C): The transformation of all variables into first differences (Δ) eliminates long-run information. Given that life expectancy evolves slowly, this approach may obscure meaningful long-term relationships. A cointegration framework or system GMM could better capture both short- and long-run dynamics.

Response:
We agree that transforming variables into first differences limits the ability to capture long-run relationships. In the revised manuscript, we clarify that this transformation reflects a deliberate econometric trade-off.
Specifically:

Differencing is used to ensure consistency and avoid spurious regression results in the presence of non-stationarity and potential endogeneity.
The dynamic GMM framework prioritizes short-run dynamics over long-run equilibrium relationships.

We explicitly acknowledge that long-run relationships are not identified in this specification and suggest that future research could explore cointegration approaches or system GMM to address this limitation.

The transformation of variables into first differences is consistent with the unit root test results, which indicate that the series are integrated of order one. This approach helps to avoid spurious regressions and ensures the validity of the GMM estimation. However, it is important to acknowledge that differencing may reduce the ability to capture long-run relationships, particularly for variables such as life expectancy that evolve gradually over time. Therefore, the results should be interpreted as reflecting short-term dynamics rather than long-term equilibrium relationships.

Reviewer Comment: The specification includes lagged differences of the dependent variable, but the interpretation of “limited persistence” is questionable, especially given the relatively high coefficient (0.59) in high-income countries, which is economically non-trivial despite statistical insignificance.

Response:
Thank you for this insightful observation. We agree that the original interpretation of “limited persistence” was not fully appropriate given the magnitude of the estimated coefficient.

In the revised manuscript, we have refined the interpretation to acknowledge that, although the coefficient is not statistically significant, its magnitude suggests a moderate degree of persistence in life expectancy. At the same time, we emphasize that the lack of statistical significance prevents drawing robust conclusions about temporal dependence. This revision improves the consistency between the statistical results and their economic interpretation.

Although the coefficient on the lagged dependent variable is not statistically significant, its magnitude suggests moderate persistence in life expectancy. However, this result should be interpreted with caution, as the lack of statistical significance prevents drawing strong conclusions about temporal dependence.

Reviewer comment (R.C): The interpretation of the positive and significant coefficient of fossil fuel consumption in high-income countries as a “paradox” is not sufficiently rigorous. Alternative explanations, such as omitted variable bias, reverse causality, or measurement issues, should be explored before attributing this to institutional buffering.

Response:
We have substantially revised the interpretation of this result. The positive and significant association between fossil fuel consumption and life expectancy in high-income countries is no longer presented as a paradox.
Instead, we explicitly state that:

The result should not be interpreted as evidence of a beneficial effect of fossil fuel consumption.
It likely reflects underlying structural correlations (e.g., higher income levels, advanced healthcare systems, or omitted variables).
Alternative explanations such as omitted variable bias, reverse causality, and measurement limitations are explicitly discussed.

This revision ensures a cautious and methodologically consistent interpretation.

This study indicates that, in these countries, fossil fuel consumption is associated with higher life expectancy. A significant result is observed for fossil fuel use, with a positive and statistically significant coefficient. This finding indicates a positive association between fossil fuel consumption and life expectancy in high-income countries. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may instead reflect omitted-variable bias, reverse causality, or measurement issues. For example, higher-income economies with advanced healthcare systems and infrastructure may simultaneously exhibit higher energy consumption and better health outcomes.

The most striking finding is a positive, statistically significant relationship between fossil fuel consumption and life expectancy, which appears counterintuitive given the expected negative environmental impacts. However, this result should be interpreted with caution. Rather than reflecting a causal relationship, it may be driven by underlying structural factors, such as higher levels of economic development, more advanced healthcare systems, or omitted variables correlated with both energy consumption and population health outcomes. Additionally, potential issues such as reverse causality or measurement limitations cannot be ruled out. Therefore, attributing this pattern solely to institutional capacity would be premature without further empirical evidence.

Reviewer comment (R.C): The paper states that CO2 emissions have “direct effects on air quality,” yet CO2 is not a local pollutant affecting health directly in the same way as particulate matter. This conceptual imprecision weakens the theoretical argument and should be corrected.

Response:
We have corrected the conceptual treatment of CO₂ emissions. The manuscript now clearly states that CO₂ is not a local air pollutant with direct health effects, but rather a driver of climate change with indirect impacts on health through environmental and climatic pathways.

CO₂ emissions, in contrast, represent a cumulative environmental outcome associated with climate change processes, which may indirectly affect human health through environmental and climatic pathways rather than direct exposure effects, particularly via mechanisms such as temperature increases and environmental degradation [24], [38].

CO₂ emissions—recognized as a primary contributor to global greenhouse gases—originate mainly from the combustion of fossil fuels in electricity generation, transportation, and industrial production. However, CO₂ should not be considered a local air pollutant with direct effects on human health. Instead, its relevance lies in its contribution to climate change, which may indirectly affect health outcomes through mechanisms such as rising temperatures, increased climate variability, and broader environmental degradation.

The role of CO₂ emissions in this study should be interpreted within a broader environmental context. Unlike local air pollutants such as particulate matter, CO₂ does not directly affect human health through immediate exposure. Instead, its impact is primarily indirect, as it contributes to climate change, which in turn influences health outcomes through environmental and socio-economic pathways.

Reviewer comment (R.C): The absence of key control variables (e.g., healthcare expenditure, education, urbanization, demographic structure) raises serious concerns about omitted variable bias, especially given the complexity of determinants of life expectancy.

Response:
We acknowledge that excluding variables such as healthcare expenditure, education, urbanization, and demographic structure may introduce omitted-variable bias.
We clarify that the model adopts a parsimonious specification as a deliberate trade-off, balancing:

Data comparability across countries and time
Model tractability
Avoidance of instrument proliferation in the GMM framework

We explicitly state that the results should be interpreted as conditional associations rather than causal effects, and we identify the inclusion of additional controls as an avenue for future research.

While the model includes key environmental and economic variables, it does not incorporate additional controls such as healthcare expenditure, education, urbanization, or demographic structure. This choice reflects a parsimonious specification based on data availability and comparability across countries and over time. However, excluding these variables may introduce omitted-variable bias, as multiple socio-economic and demographic factors influence life expectancy. Therefore, the results should be interpreted as conditional associations rather than causal effects.

Reviewer comment (R.C): The VIF analysis includes “life expectancy” as an explanatory variable (Table 6), which is inappropriate since it is the dependent variable. This suggests either a reporting error or a misunderstanding of multicollinearity diagnostics.

Response:
Thank you for this important observation. We acknowledge that the inclusion of the dependent variable (life expectancy) in the VIF analysis was inappropriate.

In the revised manuscript, we have corrected Table 6 by removing the dependent variable and ensuring that the VIF analysis includes only explanatory variables. We have also clarified the corresponding text to reflect that multicollinearity diagnostics apply exclusively to independent variables.

These revisions correct the reporting error and improve the methodological accuracy of the manuscript.

Table 6 presents the Variance Inflation Factors (VIFs) for the explanatory variables in the model. This test evaluates the degree of collinearity among independent variables, which is essential to ensure the validity of econometric estimates. As a general rule, VIF values below 5 indicate no serious multicollinearity, while values below 2 are considered optimal. The results show that all explanatory variables have low VIF values, indicating no problematic multicollinearity in the model.
Table 6. Multicolinealidad — High-income European countries
Variables
VIF

Use of fossil fuels
1.82

CO2 emissions
1.40

Use of renewable energy as a primary source of energy
1.26

lnGDP per capita
1.24

In this case, the results indicate no serious multicollinearity among variables, since all values fall below the critical threshold. The variable use of fossil fuels (3.29) has the highest VIFs, though they remain within a manageable range. Other variables—CO₂ emissions (1.32), renewable energy use (1.27), and GDP per capita (1.04)—show low collinearity.
Table 11. Multicollinearity — Medium-income European countries
Variables
VIF

Use of fossil fuels
3.29

CO2 emissions
1.32

Use of renewable energy as a primary source of energy
1.27

lnGDP per capita
1.04

Reviewer comment (R.C): The Hansen test p-values (0.600 and 0.396) are interpreted as evidence of valid instruments, but no discussion is provided regarding instrument proliferation or instrument count relative to sample size, which is critical in GMM estimation.

Response
Although the Hansen test results suggest that the instruments are valid, we agree that this test alone is insufficient to assess instrument quality fully. Therefore, we have expanded the discussion to explicitly consider the number of instruments relative to the sample size, a critical aspect of GMM estimation.
In this study, the number of instruments (8) remains well below the number of cross-sectional units, following a conservative specification strategy to reduce the risk of instrument proliferation and overfitting. This approach enhances the reliability of the Hansen test and the robustness of the estimated coefficients.
Nevertheless, as in any dynamic panel model, we acknowledge that the results may be sensitive to instrument specification, and this limitation is now explicitly recognized in the manuscript.

Although the Hansen test results suggest that the instruments are valid, it is also important to consider the number of instruments relative to the sample size. In this study, the number of instruments (8) remains well below the number of cross-sectional units, reducing the risk of instrument proliferation and overfitting. This study supports the reliability of the GMM estimates. Nevertheless, as in any dynamic panel model, the potential sensitivity to instrument specification should be acknowledged.

Reviewer comment (R.C): The number of instruments (8) appears low relative to the model structure, but the paper does not clarify how instruments were constructed or whether collapse or lag limits were applied. This lack of transparency limits replicability.

Response:
In the GMM estimation, instruments were constructed using lagged values of the endogenous variables, following the standard Arellano–Bond difference GMM approach.
To address concerns related to instrument proliferation and overfitting, we implemented a deliberately conservative specification strategy. Specifically, the number of instruments was restricted by (i) limiting the lag depth used for instrument generation and (ii) collapsing the instrument matrix.
This procedure ensures a parsimonious instrument set while preserving the validity of the moment conditions. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and enhances both the robustness and replicability of the estimates.
These methodological details have now been explicitly incorporated into the revised manuscript to improve transparency.

In the GMM estimation, instruments were constructed from lagged endogenous variables, following the standard Arellano–Bond approach. To avoid overfitting and instrument proliferation, the number of instruments was deliberately restricted by limiting the lag depth and collapsing the instrument matrix. This procedure ensures a parsimonious specification while maintaining the validity of the instruments. As a result, the total number of instruments remains relatively low (8), which is appropriate given the sample size and helps improve the robustness and replicability of the estimates.

Reviewer comment (R.C): The Arellano–Bond test results show no AR(1) autocorrelation, which is unusual since first-differenced residuals typically exhibit AR(1) by construction. This case raises concerns about either reporting accuracy or model implementation.

Response:
We thank the reviewer for this important observation. The Arellano–Bond test results indicate no statistically significant autocorrelation in the first-differenced residuals at either the AR(1) or AR(2) levels for both high- and medium-income country groups.
We acknowledge that the absence of AR(1) autocorrelation is uncommon in first-differenced specifications, where negative first-order serial correlation is typically expected by construction. However, this result does not necessarily imply misspecification or implementation error.
In particular, the lack of significant AR(1) autocorrelation may reflect sample-specific characteristics, such as limited time variation after differencing, relatively smooth dynamics in life expectancy, or the restricted instrument set used in the estimation. These factors can weaken the detection of first-order serial correlation in finite samples.
Importantly, the absence of AR(2) autocorrelation— which is the key requirement for the validity of the moment conditions— is satisfied in all specifications. Therefore, the main consistency condition of the difference GMM estimator remains intact.
Nevertheless, we agree that this result should be interpreted with caution. In the revised manuscript, we explicitly acknowledge that the Arellano–Bond test does not provide strong confirmatory evidence for the model's dynamic structure and that the results should be interpreted accordingly.

The Arellano–Bond test results for high-income European countries indicate no statistically significant autocorrelation in the first-differenced residuals at either AR(1) or AR(2) levels. While the absence of AR(2) autocorrelation is consistent with the validity of the model, the lack of significant AR(1) autocorrelation is less typical in first-differenced specifications and should be interpreted with caution. This outcome may reflect sample-specific characteristics or limitations in the model specification and, therefore, does not provide strong confirmatory evidence of the model's dynamic structure.
The results of the Arellano–Bond test for medium-income European countries indicate that no statistically significant autocorrelation is detected in the first-differenced residuals at either the AR(1) or AR(2) levels. While the absence of second-order autocorrelation (AR(2)) is consistent with the model's assumptions, the lack of evidence for first-order autocorrelation (AR(1)) is somewhat unusual in this type of specification and should therefore be interpreted with caution. This pattern may be associated with specific characteristics of the sample or potential limitations in the model specification, and thus does not provide conclusive confirmation of the estimated dynamic structure.

Reviewer comment (R.C): The IPS unit root test results are reported as showing non-stationarity at levels, yet p-values such as 0.3164 (CO2) suggest failure to reject the null, but the discussion does not consider panel heterogeneity or cross-sectional dependence.

Response:
Thank you for this important observation. We agree that the Im–Pesaran–Shin (IPS) unit root test results should be interpreted with caution, particularly given its underlying assumptions of cross-sectional independence and potential sensitivity to panel heterogeneity.
In the revised manuscript, we have clarified the interpretation of the IPS results. Specifically, p-values such as 0.3164 for CO₂ emissions correctly indicate a failure to reject the null hypothesis of a unit root in levels, which is consistent with the conclusion that the variables are non-stationary in their original form. We have revised the wording to ensure that this interpretation is explicitly and accurately conveyed.
Furthermore, we now explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present in the context of European Union countries due to their economic and environmental interconnections. As a result, the stationarity findings may be affected by common shocks or interdependencies across panel units.
Accordingly, we emphasize in the revised manuscript that the unit root results should be interpreted as indicative rather than definitive. We also clarify that the decision to transform variables into first differences is not based solely on the IPS test, but is additionally supported by standard econometric practice in dynamic panel models when variables are likely integrated of order one.
Finally, we have explicitly identified this limitation and suggested that future research could apply second-generation panel unit root tests that account for cross-sectional dependence to validate the robustness of the results further.

For high-income European countries, Table 4 reports the results of the Im–Pesaran–Shin (IPS) unit root test for panel data, used to examine the stationarity of the panel data. This test evaluates the null hypothesis that each series contains a unit root against the alternative of stationarity.

The results indicate that, in levels, none of the variables exhibit statistical evidence of stationarity, as all p-values exceed the 0.05 threshold. For instance, life expectancy yields a p-value of 1.0000, while CO₂ emissions and fossil fuel consumption yield p-values of 0.3164 and 0.9577, respectively, suggesting that the null hypothesis of non-stationarity cannot be rejected.

When the variables are transformed into first differences, the test results change substantially, with all series displaying statistically significant p-values (p < 0.05). This case suggests that the variables become stationary after differencing, indicating that they are most likely integrated of order one. Such properties justify their inclusion in dynamic panel models such as GMM.

Nevertheless, it is important to consider that the IPS test relies on the assumption of cross-sectional independence and may be affected by heterogeneity across panel units. In the European context, where countries are economically and environmentally interconnected, cross-sectional dependence is a plausible concern. Consequently, the stationarity results should be interpreted with caution, as they may not fully reflect the complexity of the underlying panel structure.

For the group of medium-income countries, Table 9 reports the results of the Im–Pesaran–Shin (IPS) unit root test applied to the main variables to determine their integration orders. This test is used in panel-data models to verify the presence of a unit root (H₀) versus the alternative hypothesis of common stationarity (H₁). According to statistical criteria, p-values below 0.05 indicate rejection of the null hypothesis and suggest that the series is stationary.

The results show that none of the variables is stationary in levels, since all exhibit p-values greater than 0.05. For example, the life expectancy variable has a p-value of 1.0000, and CO₂ emissions a p-value of 0.2573, indicating the presence of a unit root in their original form. However, when the test is applied to first differences, several variables become stationary: life expectancy (p = 0.0025), CO₂ emissions (p = 0.0023), renewable energy use (p = 0.0003), and GDP per capita (p = 0.0439). These results suggest that the variables are likely integrated of order one, satisfying the conditions for their inclusion in dynamic models such as GMM.

However, it is important to note that the IPS test assumes cross-sectional independence and may be sensitive to panel heterogeneity. Given the integration of European economies and potential interdependencies among countries, the presence of cross-sectional dependence cannot be ruled out. Therefore, these results should be interpreted with caution, as they may not fully capture the underlying data structure.

Reviewer comment (R.C): The decision to differ all variables based solely on IPS results ignores the possibility of cross-sectional dependence, which is likely in EU countries. Second-generation panel unit root tests would be more appropriate.

Response:
Thank you for this insightful comment. We agree that the IPS test, as a first-generation panel unit root test, relies on the assumption of cross-sectional independence, which may not fully hold in the context of European Union countries.
In the revised manuscript, we clarify that the decision to transform variables into first differences is not based solely on the IPS test results. Rather, it reflects a standard and deliberate econometric choice within the dynamic panel GMM framework, where differencing is commonly employed to eliminate unobserved heterogeneity and to ensure estimator consistency in the presence of non-stationary variables.
While the IPS results provide indicative evidence that the variables are integrated of order one, the transformation is also motivated by the need to avoid spurious regression outcomes and to properly specify the dynamic structure of the model.
At the same time, we explicitly acknowledge that the IPS test does not account for cross-sectional dependence, which is likely to be present among European countries due to shared economic and environmental shocks. As a result, the unit root findings should be interpreted with caution.
We have incorporated this discussion into the methodology section and explicitly identified this as a limitation of the study. In addition, we suggest that future research could employ second-generation panel unit root tests that explicitly account for cross-sectional dependence to further validate the robustness of the results.

The decision to transform the variables into first differences is primarily guided by the IPS unit root test results, which suggest non-stationarity in levels and stationarity after differencing. However, it is important to recognize that this decision is not based solely on the IPS outcomes. Given the model's dynamic nature and the need to avoid spurious regression results, differencing is also consistent with standard practices in GMM estimation when the variables are integrated of order one.

At the same time, it is acknowledged that the IPS test assumes cross-sectional independence, which may not fully hold in the context of European Union countries due to their economic and environmental interconnections. Therefore, while the transformation provides a consistent estimation framework, the potential presence of cross-sectional dependence represents a limitation. Future research could address this issue by applying second-generation panel unit root tests that explicitly account for such dependencies.

Reviewer comment (R.C): The descriptive statistics are interpreted normatively (e.g., “this reflects strong welfare systems”), which goes beyond what summary statistics can support. These interpretations should be reserved for regression analysis.

Response:
Thank you for this important observation. We agree that the interpretation of descriptive statistics should remain strictly descriptive and avoid normative or causal inferences.

In the revised manuscript, we have revised the wording of the descriptive analysis (Tables 2 and 3) to focus exclusively on the statistical properties of the data, such as means, dispersion, and ranges.

All interpretative and explanatory statements have been moved to the regression and discussion sections, where they are supported by econometric evidence. These revisions improve the clarity and methodological rigor of the manuscript.

The descriptive results presented in Table 2 provide an overview of the behavior of the variables analyzed in high-income European countries.
First, Life expectancy averages 80.26 years, with low variability (standard deviation of 1.42) and a range of 76.54 to 83.05 years. These values indicate relatively high and homogeneous longevity levels across the countries analyzed.
Regarding fossil fuel use, the mean is 74.07% of the energy matrix, with considerable dispersion (standard deviation of 19.82) and values ranging from 27.14% to 99.08%, suggesting substantial differences in energy dependence across countries.
CO₂ emissions average 9.82 metric tons per capita, ranging from 3.54 to 25.95, indicating heterogeneity in environmental pressure among countries.
Renewable energy use averages 18.13%, with high variability (standard deviation of 14.85%) and values ranging from 1.28% to 57.83, reflecting uneven adoption of renewable energy sources.
Finally, GDP per capita (log-transformed) averages 10.77, with low dispersion (0.29) and a range of 10.41 to 11.62, indicating relatively similar income levels within this group.

The results presented in Table 3 summarize the descriptive statistics for medium-income European Union countries. Life expectancy averages 77.45 years, with a standard deviation of 3.34 and a range of 70.81 to 83.47 years, indicating moderate variability across countries.
Fossil fuel use averages 82.78%, with a standard deviation of 8.22, reflecting differences in energy dependence among countries.
CO₂ emissions average 6.80 metric tons per capita (standard deviation of 2.16), ranging from 3.80 to 12.50, indicating variation in environmental pressure levels.
Renewable energy use averages 15.70% (standard deviation of 7.37), ranging from 4.63% to 33.61%, suggesting differences in the adoption of renewable energy sources.
GDP per capita (log) averages 9.62, with a standard deviation of 0.52 and a range of 8.22 to 10.43, reflecting variation in economic performance within the group.

Reviewer comment (R.C): The figures (pages 9 and 11) visually suggest heterogeneous country trajectories, yet the econometric model assumes homogeneous slope coefficients within groups. This assumption should be tested or relaxed (e.g., via heterogeneous panel models).

Response:
Thank you for this valuable observation. We agree that the graphical analysis highlights heterogeneous trajectories across countries.

In the revised manuscript, we have clarified that, while the econometric model accounts for heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. We now explicitly acknowledge this as a simplifying assumption and clarify that the estimated coefficients should be interpreted as average group effects rather than country-specific relationships.

The graphical evidence presented in Figures 1 and 2 suggests the presence of heterogeneous trajectories across countries, particularly in the evolution of environmental and economic variables. While the econometric model addresses part of this heterogeneity by grouping countries into high- and medium-income categories, it assumes homogeneous slope coefficients within each group. This simplifying assumption facilitates estimation but may not fully capture country-specific dynamics. Therefore, the results should be interpreted as average effects within groups rather than country-specific relationships.

Another important methodological consideration concerns the assumption of homogeneous slope coefficients across country groups in the econometric model. While the graphical analysis (Figures 1 and 2) reveals heterogeneous trajectories across countries, the GMM specification estimates average effects within each income group. Although this grouping strategy captures broad structural differences, it may not fully reflect country-specific dynamics.
Therefore, the estimated coefficients should be interpreted as average relationships rather than uniform effects across all countries within each group. Future research could address this limitation by applying heterogeneous panel data models, such as mean-group (MG) or pooled mean-group (PMG) estimators, which offer greater flexibility for capturing country-level heterogeneity.

Reviewer comment (R.C): The sign inconsistency in Table 13 for CO2 emissions (reported as negative in the text but positive in the table before sign interpretation) suggests either a typographical error or misinterpretation that needs correction.

Response:
Thank you for this important observation. We have carefully reviewed the results and confirm that the coefficient for CO₂ emissions in Table 13 is negative and statistically significant.

The inconsistency identified was due to an imprecise interpretation in the text. In the revised manuscript, we have corrected the wording to ensure full consistency between the reported coefficient and its interpretation.

We now clearly state that CO₂ emissions have a negative and significant effect on life expectancy in medium-income countries, reflecting the direct adverse impact of environmental degradation in these contexts. These revisions improve the internal consistency and clarity of the results section.

Second, fossil fuel use shows a positive coefficient (0.43), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), confirming the expected adverse impact of environmental degradation on population health in medium-income countries. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, likely reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.      Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The confidence intervals reported in Table 13 appear inconsistent (e.g., extremely wide ranges for GDP per capita), indicating potential scaling or reporting errors that should be verified.

Response:
Thank you for this important observation. We have carefully reviewed the confidence intervals reported in Table 13 and identified a typographical error in the interval associated with GDP per capita.

In the revised manuscript, we have corrected the confidence interval to ensure consistency with the reported coefficient and standard error. The updated values now accurately reflect the estimation results.

This correction does not affect the statistical significance or interpretation of the variable, which remains non-significant. These revisions improve the accuracy and clarity of the reported results.

Group variable: COUNTRY Time variable: YEAR
Number of obs
189

Obs per group

min
18

Avg
18.90

Max
19

Number of instruments
8
Prob > chi2
0.0000

D.Life expectancy
Coef.
Robust
P>|t|
[95% Conf. Interval]

Std. Err.

Life expectancy

LD.
1.309437
2.430333
0.603
-4.188359    6.807233

Use of fossil fuels

LD.
0.0433404
0.0371055
0.273
-0.0405981    0.1272788

CO2 emissions

LD.
-0.7106007
0.3042238
0.044
-0.0223987   - 1.398803

Use of renewable energy as a primary source of energy

LD.
0.2687355
0.1170912
0.047
0.0038567    0.5336143

lnGDP per capita

LD.
0.1148107
0.9513675
0.258
-1.74              1.97

cons
-2.276037
1.801168
0.238
-63.50561    17.98487

The dynamic panel-data estimation results for medium-income European countries are presented in Table 13. First, lagged life expectancy shows a positive coefficient (1.30), although not statistically significant (p = 0.603), suggesting weak temporal persistence of this variable within the group of countries analyzed.
Second, fossil fuel use shows a positive coefficient (0.043), but it is not statistically significant (p = 0.273), indicating a relationship without sufficient statistical evidence.
However, two relevant findings emerge. CO₂ emissions have a negative and statistically significant effect (–0.71; p = 0.044), consistent with the expected adverse impact of environmental degradation on population health in medium-income countries. The associated confidence interval is fully negative, reinforcing the robustness of this relationship. This result suggests that, in these contexts, environmental pressures translate more directly into health outcomes, possibly reflecting limitations in regulatory capacity, environmental governance, or health system resilience. Therefore, higher levels of emissions are associated with reductions in life expectancy.

Renewable energy use shows a positive and statistically significant association (0.26; p = 0.047), supporting its role in improving population health and promoting sustainability in these economies.
Finally, GDP per capita does not exhibit a statistically significant effect on life expectancy (p = 0.258). This result is accompanied by a relatively large standard error and a wide confidence interval, indicating substantial uncertainty in the estimated coefficient. Consequently, no robust conclusions can be drawn regarding the role of income in explaining variations in life expectancy within this group of countries.

Reviewer comment (R.C): The discussion attributes differences between country groups to “institutional capacity,” but no robustness checks (e.g., interaction terms, subgroup regressions with institutional proxies) are provided to substantiate this claim

Response:
Thank you for this important observation. We agree that the role of institutional capacity is not directly tested within the econometric model.

In the revised manuscript, we have revised the discussion to avoid causal interpretations and instead present institutional capacity as a contextual and theoretical explanation for the observed differences between country groups. We now explicitly clarify that these interpretations are not empirically tested within the model.

Additionally, we have acknowledged this as a limitation and indicated that future research could incorporate robustness checks, such as interaction terms or institutional proxies, to more directly assess the role of governance factors.

These revisions improve the conceptual clarity and methodological rigor of the manuscript.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The model does not address potential cross-country spillovers (e.g., pollution transport, EU-wide policies), which could bias estimates in a regional panel such as the European Union.

Response:
Thank you for this insightful comment. We agree that, in the context of the European Union, cross-country spillovers—such as transboundary pollution and common policy frameworks—may influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not account for spatial dependence or cross-country interactions. We now emphasize that the estimated coefficients should be interpreted as average effects, potentially influenced by unobserved spillovers.

Additionally, we have highlighted that future research could benefit from applying spatial panel-data models or related approaches to better capture these dynamics. These revisions improve the study's transparency and methodological robustness.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The time period (2000–2020) includes structural breaks such as the 2008 financial crisis and COVID-19 onset, yet no structural break tests or period controls are included, which may distort results.

Response:
Thank you for this insightful observation. We agree that the study period includes major structural events that could influence the relationships analyzed.

In the revised manuscript, we have explicitly acknowledged this limitation and clarified that the econometric model does not incorporate structural break tests or period-specific controls. We now emphasize that the estimated coefficients should be interpreted as average effects over the full period, which may mask short-term disruptions associated with these events.

Additionally, we highlight that future research could incorporate structural break tests or time-specific dummy variables to capture the impact of such shocks better. These revisions improve the transparency and robustness of the study.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted considering that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The argument that renewable energy improves life expectancy lacks a clear transmission mechanism in the empirical model. Mediating variables such as air quality improvements are not included, limiting interpretability.

Response:
Thank you for this insightful comment. We agree that the empirical model does not explicitly capture the mechanisms through which renewable energy may influence life expectancy.

In the revised manuscript, we have refined the discussion to clarify that the observed relationship should be interpreted as an indirect association rather than a direct causal effect. We now explicitly acknowledge that this relationship may operate through unobserved mechanisms, such as improvements in environmental quality or reductions in pollution exposure.

Additionally, we highlight that the absence of mediating variables—such as air quality indicators—limits the ability to identify specific transmission channels. We have included this as a limitation and suggested that future research incorporate such variables to better understand these pathways.

This study provides relevant evidence by showing that the relationship between environmental factors and life expectancy differs across country groups with distinct economic and structural characteristics. While these differences may be associated with variations in institutional capacity and governance frameworks, the empirical model does not directly test these mechanisms. Therefore, institutional factors should be interpreted as contextual elements rather than as empirically verified causal effects. In this sense, the observed differences between high- and medium-income countries may reflect a combination of institutional, economic, and structural conditions that shape how environmental pressures translate into health outcomes.
A more direct assessment of the role of institutional capacity would require additional robustness checks, such as the inclusion of interaction terms, subgroup regressions using institutional proxies (e.g., governance indicators), or alternative model specifications. Although these extensions are beyond the scope of the present study, they represent a relevant avenue for future research and would help strengthen the empirical identification of institutional mechanisms.
From a methodological perspective, the results should be interpreted with the understanding that the model focuses on a parsimonious set of environmental and economic variables. While this allows for identifying clear associations with life expectancy, it does not explicitly incorporate additional socio-economic determinants—such as healthcare expenditure, education, urbanization, or demographic structure—which are known to influence health outcomes. As a result, the estimated relationships may capture broader dynamics that combine environmental and socio-economic effects.
In addition, the European context is characterized by a high degree of regional integration, where environmental and economic processes frequently extend beyond national borders. As noted in the literature [21], [37], phenomena such as transboundary pollution and interconnected energy systems suggest that cross-country interactions may influence part of the observed relationships. While the model captures average effects within country groups, these should be interpreted as aggregate patterns that may partially reflect such interdependencies.
Similarly, the analysis period (2000–2020) includes major global disruptions, such as the global financial crisis and the onset of the COVID-19 pandemic, which have been shown to affect economic and health dynamics [14], [16]. In this context, the results should be understood as average relationships over time, rather than as responses to specific short-term shocks.
Furthermore, the empirical specification assumes homogeneous effects within country groups. While this approach facilitates comparative analysis, it may not fully capture country-specific trajectories, as highlighted in previous studies on panel heterogeneity [21], [22]. Therefore, the estimated coefficients should be interpreted as general patterns rather than uniform effects across all countries within each group.
Within this framework, the results reveal differentiated dynamics across income levels. In high-income European countries, life expectancy exhibits limited temporal variation, suggesting that institutional maturity contributes to stable health outcomes. The positive association observed between fossil fuel consumption and life expectancy should be interpreted with caution, as it may reflect underlying structural factors—such as higher levels of economic development and more advanced healthcare systems—rather than a direct causal relationship [24], [48]. In this context, the absence of significant effects from CO₂ emissions, renewable energy use, and GDP per capita reinforces the idea that well-established welfare and environmental systems may buffer marginal environmental variations [10], [25].
In contrast, medium-income European countries show greater sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively associated with life expectancy, indicating that environmental pressures may translate more directly into health outcomes in these contexts, consistent with previous findings [22], [24]. At the same time, renewable energy use shows a positive association, which may reflect broader processes of environmental and socio-economic transition. However, this relationship likely operates through indirect mechanisms—such as improvements in environmental quality or reductions in pollution exposure—that are not explicitly captured in the empirical model, as suggested in related studies [21], [26], [36], [49].
Since the model does not include mediating variables—such as air quality indicators or environmental health measures—the underlying transmission channels remain unobserved. Therefore, the results should be interpreted as reduced-form associations rather than structurally identified causal relationships. This interpretation is consistent with the broader literature, which emphasizes the complexity of linking environmental factors to health outcomes through multiple indirect pathways.
From a policy-oriented perspective, the results are consistent with the relevance of differentiated strategies that account for each country group's institutional and structural characteristics. In high-income countries, the challenge lies in advancing regulatory innovation and integrating environmental sustainability into welfare systems. In medium-income countries, strengthening institutional capacity and promoting cleaner energy transitions appear particularly relevant. However, these implications should be interpreted as indicative rather than prescriptive, given that the empirical analysis does not directly evaluate specific policy interventions.

Reviewer comment (R.C): The conclusions extend policy recommendations (e.g., ecological welfare state transformation) that are not directly tested in the empirical model, resulting in overgeneralization beyond the evidence presented.

This study contributes to the literature by examining the relationship between environmental degradation and life expectancy in the European Union, highlighting how these dynamics differ across countries with distinct economic and structural characteristics. The empirical results indicate that the effects of environmental factors on population health are not uniform, but vary depending on contextual conditions associated with income levels and broader institutional environments.
In high-income European countries, the findings suggest that environmental variables, such as CO₂ emissions and renewable energy use, do not have statistically significant effects on life expectancy, whereas fossil fuel consumption shows a positive association. However, this result should be interpreted with caution, as it may reflect underlying structural factors—such as advanced healthcare systems, higher levels of economic development, or other omitted variables—rather than a direct causal relationship.
In contrast, medium-income European countries show a more pronounced sensitivity of health outcomes to environmental conditions. CO₂ emissions are negatively and significantly associated with life expectancy, indicating that environmental pressures may translate more directly into adverse health outcomes in these contexts. At the same time, renewable energy use is positively associated with life expectancy, though this relationship should be interpreted as an indirect effect rather than a direct causal effect, as the model does not explicitly capture the underlying mechanisms.
Overall, the results suggest that environmental and economic factors interact with broader contextual conditions, which may include institutional, structural, and socio-economic elements, although these are not directly modeled in the empirical specification. Therefore, the findings should be interpreted as reduced-form relationships rather than structurally identified causal effects.
From a policy perspective, the results support the relevance of differentiated approaches that account for country-specific conditions when addressing environmental and public health challenges. However, these implications should be interpreted with caution. While the study adopts an institutional framework to contextualize the findings, the econometric model does not directly test specific institutional mechanisms or policy interventions.
Consequently, broader proposals—such as the transformation toward an “ecological welfare state”—should be understood as theoretical extensions consistent with the study's conceptual framework, rather than as empirically validated conclusions. Future research could build on these findings by incorporating explicit institutional indicators, mediating variables (such as air quality), or alternative modeling approaches that allow for the identification of causal mechanisms and policy effects.
In addition, further studies could extend the analysis by accounting for spatial interactions, structural breaks, and heterogeneous country dynamics, which may play a significant role in shaping environmental and health outcomes in integrated regions such as the European Union. Addressing these dimensions would contribute to a more comprehensive understanding of the complex relationships between environmental degradation, institutional conditions, and population well-being.

View more View less

Competing Interests

No competing interests were disclosed.

Alongside their report, reviewers assign a status to the article:

Approved - the paper is scientifically sound in its current form and only minor, if any, improvements are suggested

Approved with reservations - A number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.

Not approved - fundamental flaws in the paper seriously undermine the findings and conclusions

[1] 1. Long JW, Steel EA: Shifting Perspectives in Assessing Socio-Environmental Vulnerability. Sustainability. 2020; 12(7): 2625. Publisher Full Text

[2] 2. Rockström J, Steffen W, Noone K, et al.: A safe operating space for humanity. Nature. Sep. 2009; 461(7263): 472–475. Publisher Full Text

[3] 3. Forster J: Summary for policymakers of the global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Díaz S, Settele J, Brondízio ES, et al., authors. 2019; vol. 45(3). Reference Source

[4] 4. Forster PM, Smith C, Walsh T, et al.: Indicators of Global Climate Change 2023: annual update of key indicators of the state of the climate system and human influence. Earth Syst. Sci. Data. Jun. 2024; 16(6): 2625–2658. Publisher Full Text

[5] 5. North DC: A Transaction Cost Theory of Politics. J. Theor. Polit. Oct. 1990; 2(4): 355–367. Publisher Full Text

[6] 6. Hall PA, Taylor RCR: Political Science and the Three New Institutionalisms. Polit. Stud. Dec. 1996; 44(5): 936–957. Publisher Full Text

[7] 7. Barbier EB, Hochard JP: Land degradation and poverty. Nat. Sustainability. Nov. 2018; 1(11): 623–631. Publisher Full Text

[8] 8. Kjell ONE: Sustainable Well-Being: A Potential Synergy between Sustainability and Well-Being Research. Rev. Gen. Psychol. Sep. 2011; 15(3): 255–266. Publisher Full Text

[9] 9. Gough I: Climate change and sustainable welfare: the centrality of human needs: Fig. 1. Camb. J. Econ. 2015; 39(5): 1191–1214. Publisher Full Text

[10] 10. Bailey D: The Environmental Paradox of the Welfare State: The Dynamics of Sustainability. New Polit. Econ. Nov. 2015; 20(6): 793–811. Publisher Full Text

[11] 11. Eisenmenger N, Pichler M, Krenmayr N, et al.: The Sustainable Development Goals prioritize economic growth over sustainable resource use: a critical reflection on the SDGs from a socio-ecological perspective. Sustain. Sci. Jul. 2020; 15(4): 1101–1110. Publisher Full Text

[12] 12. Singer-Brodowski M, von Seggern J , Duveneck A, et al.: Moving (Reflexively within) Structures. The Governance of Education for Sustainable Development in Germany. Sustainability. 2020; 12(7): 2778. Publisher Full Text

[13] 13. Easterlin: Munich Personal RePEc Archive Does Economic Growth Improve the Human Lot ? Some Empirical Evidence the Human Lot ?. Some Empirical Evidence. 1974; 111773. Reference Source

[14] 14. Martinez GS, Kendrovski V, Salazar MA, et al.: Heat-health action planning in the WHO European Region: Status and policy implications. Environ. Res. 2022; 214: 113709. Publisher Full Text

[15] 15. Ahoa E, Kassahun A, Verdouw C, et al.: Challenges and Solution Directions for the Integration of Smart Information Systems in the Agri-Food Sector. Sensors. Apr. 2025; 25(8): 2362. PubMed Abstract | Publisher Full Text | Free Full Text

[16] 16. Robben J, Antonio K, Kleinow T: The short-term association between environmental variables and mortality: evidence from Europe. J. R. Stat. Soc. Ser. A Stat. Soc. 2025. Publisher Full Text

[17] 17. Mouratidis K: Urban planning and quality of life: A review of pathways linking the built environment to subjective well-being. Cities. Aug. 2021; 115: 103229. Publisher Full Text

[18] 18. United Nations Children’s Fund: World Health Organization. Progress on household drinking water, sanitation and hygiene 2000–2022: Special focus on gender. WHO/UNICEF JMP; 2023. Reference Source

[19] 19. Otto A, Gugushvili D: Eco-Social Divides in Europe: Public Attitudes towards Welfare and Climate Change Policies. Sustainability. 2020; 12(1): 404. Publisher Full Text

[20] 20. O’Brien K, Barnett J: Global Environmental Change and Human Security. Annu. Rev. Environ. Resour. 2013; 38(1): 373–391. Publisher Full Text

[21] 21. Rehman A, Alam MM, Ozturk I, et al.: Globalization and renewable energy use: how are they contributing to upsurge the CO2 emissions? A global perspective. Environ. Sci. Pollut. Res. 2022; 30(4): 9699–9712. Publisher Full Text

[22] 22. Majeed MT, Ozturk I: Environmental degradation and population health outcomes: a global panel data analysis. Environ. Sci. Pollut. Res. May 2020; 27(13): 15901–15911. PubMed Abstract | Publisher Full Text

[23] 23. Gough I: Heat, Greed and Human Need. Edward Elgar Publishing; 2017. Publisher Full Text

[24] 24. Rahman MM, Rana R, Khanam R: Determinants of life expectancy in most polluted countries: Exploring the effect of environmental degradation. PLoS One. Jan. 2022; 17(1): e0262802. PubMed Abstract | Publisher Full Text | Free Full Text

[25] 25. UNICEF WHO: The State of Food Security and Nutrition in the World 2022. FAO; 2022. Publisher Full Text

[26] 26. Pu CJ, Patel P, Hornsby G, et al.: Necessary conditions for sustainable water and sanitation service delivery in schools: A systematic review. PLoS One. Jul. 2022; 17(7): e0270847. PubMed Abstract | Publisher Full Text | Free Full Text

[27] 27. North DC: Institutions, Institutional Change and Economic Performance. Cambridge University Press; 1990. Publisher Full Text

[28] 28. Rodrik D: Rethinking Economic Growth in Developing Countries. Centro Studi Luca d’Agliano; 2004. Reference Source

[29] 29. Morales-Urrutia X, Núñez-Naranjo A, Nogales-Portero R, et al.: The Natural Environment and Investment in Economic Growth: From the Perspective of the Prosperity of Developed and Developing Countries. Sustainability. 2025; 17(12): 5513. Publisher Full Text

[30] 30. Temporelli KL, Viego VN: Relación entre esperanza de vida e ingreso. Un análisis para América Latina y el Caribe. Lect. Econ. 2011; 74: 61–85. Publisher Full Text

[31] 31. Ortiz-Paniagua CF, Gómez M: Economic Growth and Environmental Quality in Latin America, Perspective from Kuznets, 1970-2016. Economía teoría y práctica. 2021; 29(55): 17–36. Publisher Full Text

[32] 32. Catalán H: Impacto de las energías renovables en las emisiones de gases efecto invernadero en México. Probl. del Desarro. Rev. Latinoam. Econ. Dec. 2020; 52(204). Publisher Full Text

[33] 33. Baste IA: Making Peace with Nature. United Nations; 2021. Publisher Full Text

[34] 34. Hassan Q, Viktor P, J. Al-Musawi T, et al.: The renewable energy role in the global energy Transformations. Renewable Energy Focus. 2024; 48: 100545. Publisher Full Text

[35] 35. Iñiguez Rojas L: Geografía y salud: temas y perspectivas en América Latina. Cad. Saude Publica. Oct. 1998; 14(4): 701–711. PubMed Abstract | Publisher Full Text

[36] 36. Ezbakhe F, Giné-Garriga R, Pérez-Foguet A: Leaving no one behind: Evaluating access to water, sanitation and hygiene for vulnerable and marginalized groups. Sci. Total Environ. Sep. 2019; 683: 537–546. PubMed Abstract | Publisher Full Text

[37] 37. Ortega-Ramirez AT, Beltrán-Rodríguez DG, Becerra-Ospina NL, et al.: Environmental aspects of natural resources and their relationship to the exploitation of fossil fuels: A reflection on sustainability. Rev. Fuentes el Reventón Energético. Dec. 2022; 20(2). Publisher Full Text

[38] 38. Barragán-Ocaña A, Cecilio-Ayala E, Silva-Borjas P, et al.: Policies and sustainable energy transition in the global environment: Challenges for Latin America. Heliyon. Mar. 2025; 11(6): e42295. Publisher Full Text

[39] 39. Nunez F: Dataset of References for the Article on Smart Irrigation with IoT. Mendeley Data. 2025; V1. Publisher Full Text

[40] 40. Hrzic R, Vogt T: The contribution of avoidable mortality to life expectancy differences and lifespan disparities in the European Union: a population-based study. The Lancet Regional Health - Europe. 2024; 46: 101042. Publisher Full Text

[41] 41. Javanshirova Z: The Impact of Carbon Emissions on Infant Mortality Rate in Azerbaijan. J. Sustain. Dev. Issues. 2024; 2(2): 104–114. Publisher Full Text

[42] 42. Perera F, Nadeau K: Climate Change, Fossil-Fuel Pollution, and Children’s Health. N. Engl. J. Med. Jun. 2022; 386(24): 2303–2314. Publisher Full Text

[43] 43. Sompolska-Rzechuła A, Bąk I, Becker A, et al.: The Use of Renewable Energy Sources and Environmental Degradation in EU Countries. Sustainability. 2024; 16(23): 10416. Publisher Full Text

[44] 44. Mankiw NG: Is government spending too easy an answer. The New York Times. Reference Source

[45] 45. Soeder DJ: Fossil Fuels and Climate Change. Fracking and the Environment. Cham: Springer International Publishing; 2021; 155–185. Publisher Full Text

[46] 46. Arellano M, Bond S: Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations. Rev. Econ. Stud. Apr. 1991; 58(2): 277. Publisher Full Text

[47] 47. Preston SH: Mortality Trends. Annu. Rev. Sociol. Aug. 1977; 3(1): 163–178. Publisher Full Text

[48] 48. Romanello M, Napoli C, Green C, et al.: The 2023 report of the Lancet Countdown on health and climate change: the imperative for a health-centred response in a world facing irreversible harms. Lancet. Dec. 2023; 402(10419): 2346–2394. PubMed Abstract | Publisher Full Text | Free Full Text

[49] 49. Núñez-Naranjo AF, Morales-Urrutia X: Social Capital and Economic Growth: An Analysis in South America [version 2; peer review: 2 approved]. F1000Research. 2025; 14: 797. Publisher Full Text

[50] 50. Icaza D, Vallejo-Ramirez D, Guerrero Granda C, et al.: Challenges, Roadmaps and Smart Energy Transition towards 100% Renewable Energy Markets in American Islands: A Review. Energies. 2024; 17(5): 1059. Publisher Full Text

Environmental Degradation, Life Expectancy, and Institutional Quality: Evidence from the European Experience

Abstract

Background

Methods

Results

Conclusions

Keywords

Revised Amendments from Version 1

Introduction

Overview

Life expectancy

CO2 emissions

Fossil fuel use

Renewable energy use

Materials and methods

Table 1. Variables in the model.

Econometric model specification

Results

Figure 1. Life expectancy vs. environmental degradation factors in high-income European Union countries.

Table 2. Descriptive statistics — High-income European Union countries.

Figure 2. Life expectancy vs. explanatory variables in medium-income European Union countries.

Table 3. Descriptive statistics — Medium-income European Union countries.

Table 4. Im–Pesaran–Shin unit root test — High-income European countries.

Table 5. Arellano–Bond test for absence of autocorrelation in first-difference errors — High-income European countries.

Table 6. Multicolinealidad — Países europeos de altos ingresos.

Table 7. Hansen test of overidentification — High-income European countries.

Table 8. Dynamic panel-data estimation using Arellano–Bond — High-income European countries.

Table 9. Im–Pesaran–Shin unit root test — Medium-income European countries.

Table 10. Arellano–Bond test for absence of autocorrelation in first-difference errors — Medium-income European countries.

Table 11. Multicollinearity — Medium-income European countries.

Table 12. Hansen test of overidentification — Medium-income European countries.

Table 13. Dynamic panel-data estimation using Arellano–Bond — Medium-income European countries.

Discussion

Conclusions

Institutional review board statement

Ethics and consent

Data availability statement

Acknowledgments

References

Comments on this article Comments (0)

Open Peer Review

Comments on this article Comments (0)

Open Peer Review

Reviewer Status

Reviewer Reports

Comments on this article

Browse by related subjects

Competing Interests Policy

Stay Updated

CO₂ emissions