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

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

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.