Ecological Footprint and Population Health Outcomes: Evidence from E7 Countries

Abstract

This study investigates the relationship between ecological footprint and health outcomes in E7 countries from 1990 to 2017. This study makes use of panel fully modified ordinary least square (FMOLS) and dynamic ordinary least square (DOLS) models to assess the relationship between the ecological footprint and health outcomes. Although the findings show that ecological footprint has a positive effect on life expectancy, implying that the current levels of ecological footprints support life expectancy, failure to strictly observe the level of ecological footprint, in the long run, may result in a negative impact on life expectancy. Therefore, more serious efforts and strategies are needed to keep the size of ecological footprints to be favorable to human life.

1. Introduction

A healthy population is key to the country’s productivity and economic growth. Economic development, living environment, and social welfare systems are crucial factors that can have a direct link with life expectancy [1]. Improved life expectancy is attributable to improvements in education, and healthcare, among others. An improved living environment or environmental quality is also expected to cause an improvement in life expectancy. Under environmental factors, ecological footprint could serve as one crucial indicator. Rees [2] defines ecological footprint as the ‘[total] area of productive land and water ecosystems required to produce the resources that the population consumes and assimilate the wastes that the population produces, wherever on Earth that land and water may be located’. The ecological footprint is used as a measure of the sustainable use of resources. An ecological footprint exceeding the level of control implies overutilization or unsustainable utilization of resources [3]. In an area or country that suffers from ecological deficits, a situation the ecological footprint exceeds the area’s biocapacity, over-suppressing its own ecological assets (e.g., overfishing), and emitting carbon dioxide could be inevitable (see https://www.footprintnetwork.org/our-work/ecological-footprint, (accessed on 15 March 2023). The critique on ecological footprint has been added by Fiala [4] that most measurements of footprint put a strong assumption of zero greenhouse emissions (Fiala [4] also reminds us of the weak representation of sustainability that ecological footprint can serve since the correlation between ecological footprint and land degradation is weak. Another interesting debate is the uselessness of ecological footprint for future prediction in the presence of technology that could turn future production beyond expectation [5]).

Figure 1 below shows the trends in ecological footprint for the E7 member countries for the period from 1990 to 2015. The y-axis and x-axis represent the ecological footprint per capita and the years, respectively. On a positive note, the level of ecological footprint in all E7 countries is generally at an acceptable point. The average ecological footprint in Brazil is merely 1.04, while Mexico is around 1.1 and Russia is around 1.7. These figures are far below Qatar (14.72), Luxembourg (12.79), and the United Arab Emirates (8.95) in 2017 and could be ideal in supporting the life expectancy of the population (See https://worldpopulationreview.com/country-rankings/ecological-footprint-by-country, (accessed on 1 February 2023)). Similarly, with the exception of Russia, whose ecological footprint is above 1.4, India, Indonesia, China, and Turkey are currently under control, with the highest level still below 1.4.

Nonetheless, the uprising trend in these countries could induce health problems in the future if the level is not strictly monitored and controlled. For instance, the quality of air has deteriorated over the years and poses a serious threat to human existence. It is estimated that about 3.4 million child mortality each year is caused by air pollution [6]. Reduction in air quality has led to an increase in diseases such as cancer, heart disease, stroke, chronic obstructive pulmonary disease (COPD), pneumonia, allergy, asthma, etc. It is estimated that 3.8 million and 4.4 million people die from indoor air pollution and outdoor air pollution each year, respectively [6]. In the developing world, most respiratory diseases are caused by indoor air pollution [7]. The impact of air pollution on human health is much of a challenge with the rapid population growth coupled with industrialization that has led to the reduction in the quality of air attributable to an increase in emissions and a high ecological footprint [8]. Interestingly, people with low socioeconomic backgrounds are more susceptible to the harmful effect of air pollution [9,10]. With the industrial sector becoming the primary driver of economies across the globe, sustainability erosion is inevitable, which reflects a high ecological footprint. Given the mixed results of past studies regarding the effect of the ecological footprint on life expectancy, this study hypothesizes that it could be due to the size of the ecological footprint. In other words, low ecological footprints are expected to positively support life expectancy and vice versa. Efforts to establish the link between life expectancy against ecological footprints and life expectancy against emissions, particularly when the ecological footprint is getting weaker to fully support human health, its finding will hint at the future threats to the ecological footprint that urge for the formulation of strategic environmental policies to maintain optimal ecological footprint [11,12,13].

Therefore, it is the objective of this study to examine the potential impact of a high ecological footprint on health outcomes in the E7 countries, namely China, Brazil, Turkey, India, Russia, Indonesia, and Mexico. Expecting that the high ecological footprint may induce more damage to environmental quality, this study also examines the impact of higher emissions (i.e., carbon dioxide, nitrogen, and methane) on life expectancy due to the failure to monitor the size of the ecological footprint properly. These countries have moved into the phase of rapid industrialization, and there is a possibility of high carbon emissions, which has the potential to negatively impact their respective health outcomes. It is against this backdrop that this study seeks to examine the interactions between ecological footprints and health outcomes. The outcome of this study will enable policymakers to put in place measures to balance the relationship between ecological footprint and health outcomes.

The researchers used FMOLS and DOLS approaches, which cater to the long-run effects of an ecological footprint on health outcomes. The life expectancy at birth, total (years) and mortality rate, and infant (per 1000 live births) are used as the measure of health outcomes.

The organization of this study is as follows. The next section reviews relevant past studies, followed by the methodology section. Results are displayed and discussed in the fourth section. The fifth section concludes this article.

2. Literature Review

Life expectancy is widely accepted as a good measurement of the health status of any country’s population as well as comparative national development [14,15]. According to Frenk [16], a 10 percent improvement in life expectancy can generate 3–4 percent economic growth. Life expectancy was constantly rising about two decades ago, but discrepancy exists between developing and developed countries, which is rooted in differences in socioeconomic and environmental conditions in each country [17]. Since improvement in life expectancy should also mean improvement in socioeconomic and environmental factors, they form the foundation for the life expectancy model.

The critique on ecological footprint by Fiala [4] that most measurements of footprint put a strong assumption of zero greenhouse emissions has led Long et al. [18] to propose a new ecological well-being performance (EWP) index, as opposed to ecological footprint by combining ecological footprint (EFP) and human development index (HDI) as $EWP=\frac{HDI}{EFP}$ in the four islands, namely Chongming, Zhoushan, Hainan, and Taiwan. Prior to the calculation of EWP, the early correlation between HDI and EFP suggests that while most of the four islands are enjoying a high HDI, the main contribution comes from high life expectancy; only in Taiwan does that ‘acceptable’ ecological footprint lead to a high HDI. Hainan Island suffers from a high ecological footprint that likely explains the low HDI. Some of the arguments could suggest a positive connection between ecological footprint and health, especially life expectancy. First, viewing and connecting to the natural green environment, such as trees, can lead to faster recovery among surgical patients [19,20,21] Second, physical activities in forests and outdoor parks could improve health conditions [22]. Third, health issues should not be addressed by referring to medicine alone as the ever-growing stress-related diseases may suggest that environmental elements could be useful to mediate the situation [23,24]. Fourth, indirectly, natural resources can maintain clean water and air that are suitable to naturally cure the physical and mental conditions of people [25]. Fifth, environmental amenities could also offer physically challenging jobs and be capable of enhancing health and life longevity [26]. Nonetheless, there are several studies on life expectancy; however, studies that focus on the effect of ecological footprint on health outcomes are relatively scarce (with the exception of Dietz et al. [27]). This could be because the relationship is only felt in the later stage, at which point the situation has become too difficult to handle and disastrous to human life. Hence, this study could serve as an important finding to stress the possibility that health outcomes might worsen over time.

Barlow and Vissandjee [14] observed that income has a positive effect on life expectancy in several countries. Obviously, income allows individuals to afford healthy food and water supply. Moreover, income, which is also one of the indicators of economic development, determines improvement in social conditions and is expected to improve life expectancy [17]. Among the classic sources of income leading to better life expectancy could be healthcare spending and transfer payment (e.g., unemployment compensation, disability pay, and maternity pay), which is called welfare effort by Crepaz and Crepaz [28]. Daniels et al. [29] and Crepaz and Crepaz [28] highlight that the issue of life expectancy could stem from the issue of health inequalities, or the unhealthy is generally among the poor and could be addressed by overcoming the issue of income inequality. Crepaz and Crepaz [28] particularly show that income inequality has a non-linear relationship with life expectancy. At the low-income inequality level, it does affect life expectancy too much that life expectancy keeps on increasing. Nonetheless, after income inequality reaches a certain level of high, life expectancy starts to decline. Regardless of whether income or income inequality, the key to the results on life expectancy lies in the psycho-social stress, which tends to be low when income is high or more equally distributed and vice versa (Nonetheless, Mellor and Milyo [30], Deaton [31] and Mackenbach [32] cannot find any systematic or robust correlation between income inequality and life expectancy. In line with the suggestion by Wilkinson [33], a level of chronic anxiety is the key to mediating the relationship between income inequality and life expectancy). Interestingly, although life expectancy could be higher among the rich rather than the poor [33], some diseases are more pertinent to high-class people, such as coronary heart attack [34]. Finally, a detailed analysis of workers with a similar type of work suggests a higher incidence of heart disease among the lower class or income earners of civil servants [35].

Taking into account another perspective, GDP, which can represent political tension, crimes, and internal conflict, may have an unfavorable consequence on life expectancy if low GDP is primarily due to conflict and war, either domestic or across the border [36,37]. As estimated by WHO [37], 90 million people are living in critical situations due to conflicts combined with disasters and sanctions. Hence, if high GDP may mean strong security enforcement, and well-designed laws, leading to a minimum or zero social unrest and violence, then life expectancy can be predicted to be positively improved [36]. Sabri [38] offers another fascinating idea that the imports and exports in the era of globalization not only affect GDP but could also have a positive and negative impact on life expectancy. The positive aspect of globalization could be due to the installation of new medical devices, emigration of skilled health professionals, as well as practices of taking healthy diets, and in contrast, the negative influence could be the consumption of unhealthy diets and restriction of new medical technologies. Government spending on health also does not seem to exert a significant impact on life expectancy in Bangladesh [39] and sub-Saharan Africa [40]. One key answer could be due to the corresponding governance that facilitates access to equitable and quality health services [41] and efficiency in the allocation and management of health funds to target the desired goal of accessibility to all [42]. In other words, out-of-pocket is still the main factor in the improvement in life expectancy even in developing countries, implying their income is strongly determining the decision to receive medical treatment. Within private health spending, two strands of empirical findings can also be observed positive significance [40,43] and non-significance [44].

Bilas et al. [17] argue that countries with improper education and healthcare development tend to suffer from achieving sustainable development. Most countries in the world are considering healthcare as a basic right for everyone that could improve the individual’s welfare. As a result, health expenditure and literature on the health economy have increased during the last decade. Cremieux et al. [45] examine the markedly increased drug expenditures, which also reflect higher healthcare costs on health outcomes in Canada. The growing utilization of pharmaceutical products may also be a result of the cost-containing strategy of a more outpatient-focused. Moreover, quality pharmaceutical products’ availability may also help to promote life expectancy for those suffering from lymphoma, leukemia, and AIDS, as well as upgrade quality of life by comforting issues such as anemia, pain, and depression. With the call for the Millennium Development Goals (MDGs) in 2000, a strong commitment to developing water services, urban planning, and so on has pushed the agendas to successfully generate positive outcomes on health, including improving life expectancy. Halicioglu [36] suggests that proper urban development equipped with health facilities and information installed completely may help to improve life expectancy. Nonetheless, it is also possible that as in the case of most cities in developing countries, congestion, pollution, and expensive access to medical care may hamper the aim of expanding the life expectancy of urban dwellers. Similar conclusions were obtained by Fuchs [46], Bokhari et al. [47], and Kulkarni [48]), that although the contribution of healthcare spending could be big, the contribution to health outcomes is minimal. Quality of delivery and the establishment of a financial system are among the conditions that need to be improved first.

Hauck et al. [49] also share the conclusion that poor sanitation and water quality could be the primary breeding grounds for contagious diseases. Although efforts have been made to prepare a community-level water infrastructure, poor countries generally suffer a shortage of funds to maintain them. Private healthcare could have been confirmed by Moreno-Serra and Smith [50] as capable of bringing down child and adult mortality but could be too expensive for everyone to have access to it. Owumi and Eboh [42] echo the idea that overdependence on out-of-pocket health expenditure could push the poor to the precipice of catastrophic health spending, especially if their spending on health exceeds their income, individually or collectively, as a household.

3. Methodology

3.1. Data and Model Specification

As indicated in the introduction, this study uses panel data from the emerging seven countries: China, India, Brazil, Turkey, Russia, Mexico, and Indonesia, for the period from 1990 to 2017. This period was carefully chosen based on data availability for the variables and sample of the emerging seven countries in question. The data comprise health outcomes, measured by life expectancy at birth, total (years) and mortality rate, and infant (per 1000 live births). The independent variable of interest is ecological footprint—EFConsPerCap (constant per capita), including alternative degrading indicators such as CO₂ emissions (metric tons per capita), CH₄–nitrous oxide emissions (thousand metric tons of CO₂ equivalent), and N₂O–methane emissions (kt of CO₂ equivalent). We also control for GDP per capita (constant 2015 USD) and life-preservative measures such as current health expenditure per capita (current USD), people using at least basic drinking water services (% of the population), and Urban population growth (annual %). Most of the variables are obtained from World Bank Development indicators except for EFConsPerCap (constant per capita), which was sourced from the Global Footprint Network. The baseline model (on which this study is premised) expresses health outcomes as a function of LEF, LEFSQ, LGDPPC, WA, Hexp, and URB as follows:

$$lnHO=f\left(LEF, LEFSQ, LGDPPC, WA, Hexp, URB\right)$$

(1)

The above model can be converted into regression models, as shown below. Equation (2) regresses health outcome (life expectancy) on ecological footprint and its squared term, with two control variables (GDP per capita and one life preservative measure variable proxied by water access). Equation (3) studies the effect of ecological footprint on health outcomes by adding alternative measures of life preservation (proxied by current health expenditure), while Equation (4) incorporates an urbanization variable (as an additional measure of life preservation). Supplementary analyses were carried out to test whether the effects of ecological footprint on health outcomes are sensitive to alternative proxy measures of health outcome (such as mortality rate); life expectancy and mortality rate were used as dependent variables in order to test the robustness of the findings. Last but not least, we used dynamic ordinary least squares to check the robustness of FMOLS estimators.

Model 1

$$\mathit{l}\mathit{n}\mathit{H}\mathit{O}={\mathit{\alpha }}_0+\mathit{\alpha }\mathit{L}\mathit{E}{\mathit{F}}_{\mathit{i}\mathit{t}}+ \mathit{\alpha }\mathit{L}\mathit{E}\mathit{F}\mathit{S}{\mathit{Q}}_{\mathit{i}\mathit{t}} + \mathit{\alpha }\mathit{G}\mathit{D}\mathit{P}\mathit{P}{\mathit{C}}_{\mathit{i}\mathit{t}} +\mathit{\alpha }\mathit{W}{\mathit{A}}_{\mathit{i}\mathit{t}}+{\mathit{\epsilon }}_{\mathit{t}}$$

(2)

Model 2

$$lnHO={\alpha }_0+\alpha LE{F}_{it}+ \alpha LEFS{Q}_{it} + \alpha GDPP{C}_{it} + \alpha Hex{p}_{it}+{\epsilon }_t$$

(3)

Model 3

$$lnHO={\alpha }_0+\alpha LE{F}_{it}+ \alpha LEFS{Q}_{it} + \alpha GDPP{C}_{it} +\alpha UR{B}_{it}+{\epsilon }_t$$

(4)

In all the models, subscripts i and t represent the cross-sectional units (i.e., the emerging seven countries in this case), and t is the year of study (1990 to 2017), respectively. HO is the health outcome variables (measured by life expectancy and mortality rate); LEF is the ecological footprint; $LEFSQ$ is the squared term of ecological footprint; $LGDPPC$ is the GDP per capita; $WA$ measures water access; $Hexp$ represents current health expenditure; ${\alpha }_0$ represents the intercept; and the rest of the alphas represent the slope of LEF, LEFSQ, LGDPPC, WA, Hexp, and URB, respectively. εt is the error term of the regression.

In panel data, setting cross-sectional dependency issues is not uncommon. Accordingly, the cross-sectional dependence (CD) test advanced by Pesaran [51] is used to detect any correlation among the cross-sections. Other standard specification tests, such as panel unit root tests, were carried out to ascertain the absence or presence of long-run features of the variables used in this paper. We specifically used the second-generation panel unit root test [52] to identify the presence of stationarity in the data series. We also undertook Pedroni cointegration tests to establish whether or not there exists a long-run association for those variables with long-run appearances [53,54].

3.2. Estimation Strategy

To estimate the precise impact of the independent variables (LEF, LEFSQ, LGDPPC, WA, Hexp, URB) on the health outcomes, this study employed panel fully modified ordinary least square (FMOLS) and dynamic ordinary least square (DOLS). A notable advantage of using FMOLS and DOLS is that they account for serial correlation and endogeneity issues, thus providing reliable long-run estimations. Following Zhang et al. [55], FMOLS and DOLS are expressed through Equations (3), (5) and (6), respectively.

$${\widehat{\phi }}_{\mathit{F}\mathit{M}}=\left({\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{N}} {\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{T}}{\left({\mathit{x}}_{\mathit{i}\mathit{t}}-{\overline{x}}_{\mathit{i}}\right)}^{\mathbf{1}} \right) \left({\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{N}} {\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{T}}\left({\mathit{x}}_{\mathit{i}\mathit{t}}-{\overline{x}}_{\mathit{i}}\right){\widehat{\mathit{y}}}_{\mathit{i}\mathit{t}}^{*}+\mathit{T}.{\widehat{\Delta }}_{\mathit{\epsilon }\mathbf{\mu }}\right)$$

(5)

where ${\widehat{\Delta }}_{\epsilon \mu }$ signifies the serial correlation of the correction term, while ${\widehat{y}}_{it}^{*}$ denotes endogeneity correction. DOLS estimator has also been used to correct for serial correlation as well as endogeneity. Panel DOLS, on the other hand, can be expressed as follows:

$${\mathit{Y}}_{\mathit{i}\mathit{t} }={\mathit{\alpha }}_{\mathit{i} }+{\mathit{x}}_{\mathit{i}\mathit{t}}\mathit{\delta }{\displaystyle \sum }_{\mathit{j}=\mathit{q}\mathbf{1}}^{\mathit{j}=\mathit{q}\mathbf{2}}{\mathit{x}}_{\mathit{i}\mathit{t}}\mathbf{\Delta }{\mathit{x}}_{\mathit{i}\mathit{t}+\mathit{j}}+{\mathit{\nu }}_{\mathit{i}\mathit{t} }$$

(6)

The estimated coefficient of DOLS is given by

$${\widehat{\phi }}_{\mathit{D}\mathit{O}\mathit{L}\mathit{S}} ={\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{N}} \left({\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{T}} {\mathit{Z}}_{\mathit{i}\mathit{t}}{\mathit{Z}}_{\mathit{i}\mathit{t}}^{\mathbf{1}}\right) \left({\displaystyle \sum }_{\mathit{t}=\mathbf{1}}^{\mathit{T}} {\mathit{Z}}_{\mathit{i}\mathit{t}}{\widehat{\mathit{y}}}_{\mathit{i}\mathit{t}}^{*}\right)$$

(7)

where $Z_{it}$ = $(x_{it}-{\overline{x}}_i$, $\mathsf{\Delta }{x}_{it+q}$) is 2(q + 1), and X1 represent the independent variables.

4. Empirical Results and Discussion

4.1. Descriptive Statistics

Before discussing the empirical results obtained from the estimation of PFMOLS and PDOLS models, we first report the descriptive analysis. Precisely,

Table 1. Summary statistics.

	LEF	LCO2	LCH4	LN2O	LLE	LMORT	LGDPPC	LWATER	LHEXP	URB
Mean	0.829	1.045	12.800	11.543	4.241	3.220	70.070	4.508	5.278	2.292
Median	0.984	0.986	12.937	11.408	4.247	3.248	76.124	4.540	5.696	2.418
Maximum	1.922	2.683	14.032	13.212	4.346	4.484	87.963	4.594	6.939	5.081
Minimum	−0.273	−0.440	10.517	10.153	4.058	1.705	39.290	4.322	2.755	−0.467
Std. Dev.	0.540	0.784	0.886	0.869	0.064	0.626	13.637	0.068	1.183	1.279
Skewness	−0.347	0.270	−0.971	0.284	−0.521	−0.144	−0.743	−0.818	−0.535	−0.243
Kurtosis	2.340	2.285	3.247	1.960	2.684	2.334	2.198	2.527	1.973	2.939
Jarque–Bera	7.423	6.550	29.391	10.754	9.668	4.305	23.304	15.210	11.549	1.957
Probability	0.024	0.038	0.000	0.005	0.008	0.116	0.000	0.000	0.003	0.376
Observations	194	196	184	184	196	196	196	126	126	196

presents the mean statistics for all the examined variables. From this table, it is evident that the mean of LGDPPC is the largest (70.070) and differs significantly across countries (maximum = 87.963 and minimum = 39.290). The average CH4 is 12.800 in the countries, and the standard deviation is 0.886. Additionally, logN₂O, LHEXP, and percentage of LWATER show a mean of 11.543, 5.278, and 4.508, respectively. Over the period, the average of logLE and MORT of selected countries is 4.241 and 3.220. In terms of standard deviation, the highest value is for the LGDPPC variable with 13.637, followed by URB and logHEXP with 1.279 and 1.183, respectively.

4.2. Results of Cross-Sectional Dependence and Panel Unit Root Tests

As a standard procedure in this field, before implementing some econometric methods which deal with panel data assessments, it is a common practice to begin by checking the presence of cross-sectional dependence or independence among the variables. The literature is full of evidence suggesting that the findings from conventional unit root tests might be spurious and misleading if the variables are found to be cross-sectional dependent because it is based on the assumption of cross-sectional independence [56]. For this purpose, this study applied the cross-sectional dependence test promulgated by Pesaran [51] to examine the presence of cross-sectional dependence and heterogeneity issues. The null hypothesis of cross-sectional independence is tested against the alternative hypothesis of cross-sectional dependence consistent with the literature [56,57]. If we reject the null hypotheses, it suggests that there is a presence of cross-sectional dependence among all of the variables [51]. Interestingly,

Table 2. Pesaran cross-sectional dependence test.

	Breusch–Pagan LM	Pesaran Scaled LM	Bias-Corrected Scaled LM	Pesaran CD
LEF	157.060 ***	20.995 ***	20.865 ***	6.285 ***
LEFSQ	129.254 ***	16.704 ***	16.574 ***	0.164
GDPpc	470.733 ***	69.395 ***	69.266 ***	21.576 ***
Water	376.340 ***	54.830 ***	54.624 ***	19.399 ***
Hexp	314.232 ***	45.247 ***	45.041 ***	17.687 ***
URB	251.967 ***	35.639 ***	35.509 ***	11.084 ***
Lmort	570.852 ***	84.844 ***	84.714 ***	23.891 ***
LCO2	345.167 ***	50.020 ***	49.890 ***	12.866 ***
LCH4	222.122 ***	31.034 ***	30.904 ***	−1.728 *
LN20	338.852 ***	49.046 ***	48.916 ***	−2.388 **

Probabilities * p < 0.1, ** p < 0.05, *** p < 0.01. Source: computed by the authors.

shows that there is a presence of cross-sectional dependence across all models; hence, we reject the null hypotheses, thus indicating that the observations for different countries are not independent.

Given the fact that the conventional unit root tests are not appropriate in the presence of cross-sectional dependence across all models, this study then employed Pesaran’s [52] CADF and CIPS cross-sectional augmented panel unit root tests which account for cross-sectional dependence consistent with the work of Ummalla et al. [56].

Table 3. Panel unit root test results for the E7 Countries.

	Level	1st Difference
LEF	−1.22899	−3.94728 ***
LEFSQ	−2.18931	−4.58504 ***
GDPpc	−0.66552	−3.27755 ***
Water	−4.48847 ***	−3.31874 ***
Hexp	−1.65101	−3.067 ***
URB	−0.05321	−3.31874 ***
Lmort	−0.69333	−2.10836 ***
LCO2	−1.56472	−3.69986 ***
LCH4	−1.10243	−2.26065 **
LN20	−0.56507	−3.69986 ***

Probabilities ** p < 0.05, *** p < 0.01.

presents the results of the second-generational unit root test, which provides more accurate and reliable results than a first-generation test.

The results reveal that the data series are all stationary at first difference. This finding is important because spurious regression results can occur if nonstationarity is not accounted for in time series analysis.

4.3. Panel Cointegration Test

The cointegration test predicts the existence of a long-run relationship provided that the series is integrated in unique order [57]. Westerlund panel cointegration is often preferred since it accounts for cross-sectional dependency. A panel cointegration test was proposed by Westerlund [58] and Persyn and Westerlund [59], in which the hypothesis is investigated using two different tests. Westerlund [37] formulated the test to establish the long-run association in the presence of cross-sectional dependency.

Table 4. Panel cointegration tests.

Alternative Hypothesis: Common AR Coefs. (Within Dimension)
				Weighted
		Statistic	Prob.	Statistic	Prob.
Panel v-Statistic		−0.365304	0.6426	0.967374	0.1667
Panel rho-Statistic		1.621181	0.9475	0.930278	0.8239
Panel PP-Statistic		0.001367	0.5005	−1.883249 ***	0.0298
Panel ADF-Statistic		−3.272337 ***	0.0005	−2.567167 ***	0.0051
Alternative Hypothesis: Individual AR Coefs. (Between Dimension)
		Statistic	Prob.
Group rho-Statistic		1.449549	0.9264
Group PP-Statistic		−2.735723	0.0031 ***
Group ADF-Statistic		−3.622141	0.0001 ***

Probabilities *** p < 0.01. Source: computed by the authors.

presents the results of the Westerlund panel cointegration test. The results suggest that the null hypothesis of no cointegration can be rejected under the cross-sectional dependency.

4.4. Empirical Results

In this section, we begin our discussion by presenting the estimates (reported in

Table 5. The relationship between ecological footprint and life expectancy.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Variable	FMOLS	FMOLS	FMOLS	DOLS	DOLS	DOLS
LEF	0.00303	0.03382	1.90056	3.42317	0.058499	1.97755
	(0.19734)	(1.82064)	(7.83225)	(10.5954)	(3.14748)	(6.70028)
LEFSQ	−0.04198	−0.04546	−1.27278	−1.71464	−0.03821	−1.27872
	(−6.69183)	(−5.91286)	(−4.84067)	(−10.40755)	(−4.27151)	(−3.66109)
LGDPPC	0.00993	0.00398	0.03975	0.02500	0.00479	0.03955
	(22.90750)	(6.29279)	(18.95265)	(13.23841)	(6.16565)	(13.75878)
LIFE_PRES (LWATER)	11.80400			182.8188
	(12.03149)			(182.8188)
LIFE_PRES (health exp)		0.02048			0.01299
		(6.07476)			(3.93785)
LIFE_PRES (URB)			0.14696			0.15548
			(4.36019)			(3.28815)
R-squared	0.95775	0.94859	0.94314	0.93821	0.94824	0.88042

Note: numbers in ( ) denote t-statistics. Source: computed by the authors.

,

Table 6. The relationship between ecological footprint and mortality rate.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Variable	FMOLS	FMOLS	FMOLS	DOLS	DOLS	DOLS
LEF	−3.82103	−3.48008	−2.87751	−1.07239	−3.69298	−4.17047
	(−468.0943)	(−11.18255)	(−64.47403)	(−1.11295)	(−5.87442)	(−17.55965)
LEFSQ	0.97205	0.81126	0.54205	0.04119	0.81877	1.10578
	(1676.274)	(4.98669)	(11.41772)	(0.11586)	(3.55272)	(6.18006)
LGDPPC	0.07174	0.11440	0.50592	0.04206	0.09011	0.07801
	(189.3385)	(14.43167)	(32.50776)	(5.05298)	(12.57060)	(63.71295)
LIFE_PRES (LWATER)	66.16504			138.0408
	(177859.5)			(4.11969)
LIFE_PRES (health exp)		−0.60519			−0.26339
		(−5.57974)			(−2.74183)
LIFE_PRES (URB)			2.02539			−0.71237
			(32.8008)			(−1.82861)
R-squared	0.28047	0.62886	0.32179	0.86241	0.99116	0.78018

,

Table 7. The impact of CO₂ emission on life expectancy.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Variable	FMOLS	FMOLS	FMOLS	DOLS	DOLS	DOLS
LCO2	0.02924	2.76336	0.12232	0.03538	2.30147	0.27649
	(3.49589)	(4.07713)	(11.6799)	(4.67802)	(2.72782)	(2.05999)
LC2OSQ	−0.01361	−1.78439	−0.04082	−0.01157	−1.32227	−0.12302
	(−4.14000)	(−3.80776)	(−15.77026)	(−4.12169)	(−2.3995)	(−1.99453)
LGDPPC	−0.00018	0.03514	0.00375	−0.00026	0.01910	0.00649
	(−0.37606)	(3.84817)	(6.84098)	(−0.57793)	(0.78444)	(3.19803)
LIFE_PRES (LWATER)	0.27584			0.10654
	(2.63388)			(0.84511)
LIFE_PRES (health exp)		0.18107			0.40155
		(1.73133)			(1.26144)
LIFE_PRES (URB)			0.00159			0.00118
			(0.73452)			(0.11743)
	0.98583	0.98119	0.91737	0.98583	0.98119	0.91737

Note: numbers in ( ) denote t-statistics. Source: computed by the authors.

,

Table 8. The impact of N₂O emission on life expectancy.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Variable	FMOLS	FMOLS	FMOLS	DOLS	DOLS	DOLS
LN2O	0.67075	0.62897	0.68970	0.71274	0.64887	0.69074
	(47.94549)	(37.82371)	(52.50428)	(43.17605)	(29.20665)	(42.5600)
LN2OSQ	−0.02816	−0.02508	−0.02964	−0.03069	−0.02669	−0.02999
	(−25.0070)	(−16.72581)	(−37.34581)	(−21.7448)	(−13.1655)	(−24.4102)
LGDPPC	0.00386	0.00438	0.00416	0.00216	0.00406	0.00458
	(6.48872)	(4.66839)	(6.34085)	(3.69175)	(3.38515)	(8.56019)
LIFE_PRES (LWATER)	31.56068			62.29815
	(1.08184)			(1.68951)
LIFE_PRES (health exp)		0.00648			0.00601
		(1.37455)			(1.00807)
LIFE_PRES (URB)			0.00083			0.00631
			(0.13776)			(0.84286)
	0.95350	0.31109	0.53575	0.84597	0.763682	0.94391

Note: numbers in ( ) denote t-statistics. Source: computed by the authors.

and

Table 9. The impact of CH₄ emission on life expectancy.

	Mode 1	Model 2	Model 3	Model 4	Model 5	Model 6
Variable	FMOLS	FMOLS	FMOLS	DOLS	DOLS	DOLS
LCH4	0.51539	0.6710	0.60926	0.54143	0.55491	0.70012
	(4.47491)	(479.5764)	(296.0696)	(4.59443)	(27.53364)	(30.31017)
LCH4SQ	−0.02137	−0.02771	−0.02415	−0.02249	−0.01944	−0.02976
	(−4.54473)	(−319.8015)	(−195.2354)	(−4.68272)	(−9.89613)	(−17.2211)
LGDPPC	0.00277	0.00321	0.005201	0.00236	0.00396	0.00359
	(2.78778)	(17.34819)	(59.78653)	(2.19645)	(2.20125)	(5.73843)
LIFE_PRES (LWATER)	0.21689			0.19143
	(1.28215)			(1.0978)
LIFE_PRES (health exp)		0.000961			0.00507
		(0.484928)			(0.90973)
LIFE_PRES (URB)			0.023918		73	0.002328
			(26.29132)			(0.33385)
	0.44354	0.327945	0.41618	0.95004	0.67600	0.91075

Note: numbers in ( ) denote t-statistics. Source: computed by authors.

) carried out using both the PFMOLS and PDOLS estimators described in the methodology section. All control variables are converted into a logarithmic form for the empirical estimation. Additionally, these variables are added in a stepwise manner for robustness analysis. Models (1) to (6) in Table 5 regress life expectancy as a dependent variable against economic factors and health outcomes—including ecological footprint, squared ecological footprint, GDP per capita, access to clean drinking water, healthcare expenditure, and urbanization.

We begin our analysis by discussing the empirical results of PFMOLS presented in Models (1) to (3) below. Except for the squared ecological footprint, economic factors such as ecological footprint, GDP per capita, access to drinking water, healthcare expenditure, and urbanization displayed a positive influence on life expectancy. With respect to ecological footprint, the results in Table 5 show that ecological footprint presents a significant positive effect on life expectancy from Models (1) to (3), which is supported by past studies [19,20,22,24]. The results suggest that the current levels of ecological footprints support life expectancy. The literature is full of evidence suggesting that the positive coefficient of ecological footprint on life expectancy might be attributed to natural green environments such as trees, which can lead to faster recovery among surgical patients [19,20]. An alternative reason might be that physical activities in the forests and outdoor parks could improve health conditions [22]. However, Dietz et al. [60] found opposite results and argued that when people utilize an area of land to produce its waste material, the land turns into soil deterioration and land degradation that have a negative impact on the climate and thus reduce human life longevity. Squared ecological footprint has been associated with a decrease in life expectancy, implying that in the long run, ecological footprint might have a significant negative influence on life expectancy.

Meanwhile, the effect of GDP per capita on life expectancy is observed to be statistically significant and positive in all PFMOLS models, although the magnitude is small, consistent with [61,62,63], to mention only a few. From Models (1) to (3), the results demonstrate that a 1% rise in GDP per capita increases life expectancy by 0.0099%, 0.0039%, and 0.0397%, respectively. Implicitly, increased levels of income permit increased access to the consumption of improved quality goods and services, better housing, and medical care services that affect health status [64]. The results are in line with the findings of Halicioglu [36]), who concluded that if high GDP is an indication of strong security enforcement and well-designed laws, leading to a minimum or zero social unrest and violence, then life expectancy can be predicted to be positively improved.

Next is access to drinking water. Access to drinking water enters with a positive and significant impact on life expectancy, and the impact of access to water is substantial in Model (1), indicating that a 1% rise in this variable improves life expectancy by 11.804%. These results reinforce the UN General Assembly declaration that clearly states that every person has the right to enough, continuous, safe, acceptable, physically accessible, and affordable water for personal and domestic use.

In Models (2) and (3), we repeated the investigation by adding healthcare expenditure and urbanization as control variables. Interestingly, the results demonstrate that healthcare expenditures (viewed as a measure of the provision of health facilities to society) are positive and statistically correlated with life expectancy, confirming the findings of Cervantes et al. [65] and Bein et al. [66]. The finding shows that a 1% increase in health expenditure increases life expectancy by 0.0204%. The results suggest that increased healthcare expenditure is associated with greater availability of healthcare services in E7 member countries.

The results further demonstrate that urbanization has a positive and significant effect on life expectancy in the PFMOLS model, which confirms the findings of Kalediene and Petrauskiene [67] for Lithuania. The results suggest that the urban population often enjoys better-quality medical care and means of life, an improved education system, and other enhanced socio-economic amenities, which positively influence health outcomes [68].

For robustness checks, we further estimate an alternative model using DOLS. In Table 5, Models (4) to (6) report the factors influencing life expectancy. Remarkably, the results of the PDOLS model appeared to mimic the same pattern in terms of the direction of the impact and the level of significance as those presented by the FMOLS estimator. For instance, ecological footprint appears to possess a significantly positive impact on life expectancy in all models supporting the results of the PFMOLS estimator. The square ecological footprint is shown to have a negative impact on life expectancy. In line with the results of the PFMOLS model, GDP per capita, healthcare expenditure, and access to water appear to have significant and positive effects on life expectancy in the PDOLS. Overall, the findings from PDOLS demonstrate consistent results with PFMOLS estimates. Therefore, the conclusions advanced earlier for significant variables from PFMOLS also apply to the findings displayed in this part.

Table 6 presents the empirical results of PFMOLS and PDOLS estimators from Models (1) to (6). In this section, we regress mortality rate on ecological footprint, GDP per capita, access to drinking water, healthcare expenditure, and urbanization. There are some noticeable differences between the results presented earlier and the estimates presented in this part. Apart from the levels of significance, the major differences are in terms of the direction of the impact of the coefficients. For instance, ecological footprint has a negative and strongly significant impact on mortality rate across the three first models. The reason might be that there are strong environmental outcomes that are not polluted and expose individuals to an unhealthy setting, which might make them sick, leading to increased death [69]. These results align with the findings of Mays and Smith [69], who showed that increases in ecological footprint correlate with a decrease in mortality rate. A closer look at the squared ecological footprint–mortality nexus indicates a positive and strongly significant relationship. We conclude that in the short run, ecological footprint might have a negative impact on the mortality rate; however, the relationship can be positive in the long run.

GDP per capita appears to have the opposite impact in this model—which enters with positive and statistically significant coefficients in all models; thus, Models (1) to (3). The findings imply that a 1% increase in GDP per capita in Models (1) to (3) results in an increase in mortality rate by 0.0717%, 0.1144%, and 0.5059%, respectively. These results are unexpected given the fact that GDP per capita is often assumed to improve life expectancy at birth through improved economic growth and development and hence results in the prolongation of longevity [70]. Access to drinking water is an important factor influencing mortality rates—which enters positively and significantly, suggesting that the provision of poor water quality to the general public could be the primary breeding ground for contagious waterborne diseases such as malaria and lower respiratory infections, which have important implications for burden of diseases and contributes to higher mortality [49].

Total healthcare expenditure is perceived to have a significant influence on life expectancy because it directly helps reduce mortality and morbidity [30]. Consistent with this thinking, healthcare expenditure appears to enter with the expected negative sign at a 1% level of significance. The results suggest that an increase in healthcare expenditures by 1% is correlated with a reduction in mortality rate of −0.6051%, an indication that higher healthcare expenditure has a long-lasting effect on low-resource communities [30]. These findings are similar to those of Maruthappu et al. [71], who reported that higher public healthcare expenditure is negatively associated with HIV mortality. Urbanization is one of the critical determinants of mortality rate in the current study, positively correlated for Model 3 and negatively correlated for Model 6.

Conceivably, what is more interesting is a comparison of the results of the FMOLS and the DOLS model. In this study, the DOLS estimator was estimated as a robustness check and produced results that are qualitatively similar to those of PFMOLS. For instance, coefficients for ecological footprint reported in Models (4) to (6) once again matter in explaining mortality rate and enter with negative coefficients. Squared ecological footprint, GDP per capita, and access to water still matter in explaining mortality rate—which are positively and significantly rated to mortality rate. Generally, the findings from DOLS demonstrate consistent results with FMOLS estimates. Therefore, the conclusions arrived at earlier for the FMOLS estimates for significant variables also apply to the estimates displayed in this part.

Table 7 reports the PFMOLS and PDOLS estimation results where we replace ecological footprint with ${\mathrm{CO}}_2$ emission (proxy environmental degradation). What is evident from Models (1) to (3) is that ${\mathrm{CO}}_2$ emission has a significant positive effect on life expectancy, suggesting that higher ${\mathrm{CO}}_2$ emission increases life expectancy. Precisely, a 1% increase in ${\mathrm{CO}}_2$ emission, keeping all other variables unchanged, increases life expectancy by 0.0292%, 2.7633%, and 0.1223%, respectively. Therefore, this study finds that ${\mathrm{CO}}_2$ emissions are detrimental to life expectancy since the release of ${\mathrm{CO}}_2$ into the air can result in numerous environmental problems with devastating impacts on human health. On the other hand, squared ${\mathrm{CO}}_2$ emission was found to have a strong negative effect on life expectance. More specifically, a one percent increase in squared ${\mathrm{CO}}_2$ emission, holding all other factors constant, decreases life expectancy by −0.0113613 percent. In fact, the results imply that after ${\mathrm{CO}}_2$ emission reaches a certain threshold level, life expectancy starts to decline.

Remarkably, GDP per capita presents negative but insignificant coefficients on life expectancy in Model (1) but the direction of the impact changes to positive in Models (2) and (3). The difference between the models is due to the life-preserving factor that was included in the model. Access to basic drinking water has a significantly positive effect on life expectancy, and the effect is extensive, suggesting that a 1% rise in this variable increases life expectancy by 0.2758%. These results are to be expected since maintaining a healthy water intake might also improve longevity [70]. Likewise, we conduct some robustness checks to make certain that the estimates discussed thus far are consistent. Models (4) to (6) in Table 7 show the results of DOLS. Based on the DOLS analysis, it is found that ${\mathrm{CO}}_2$ emissions have a significant and positive effect on life expectancy in Models (4) to (6). This indicates that a 1% increase in ${\mathrm{CO}}_2$ emission increases life expectancy by 0.0353%, 2.3014%, and 0.2764%, respectively. These positive estimates are supported by the results of FMOLS. In contrast, squared ${\mathrm{CO}}_2$ emission has a negative impact on expectancy in all models. This suggests that a 1% increase in squared ${\mathrm{CO}}_2$ emission in Models (4) to (6) will reduce life expectancy by 0.0115%, 1.3222%, and 0.1230%, respectively. GDP per capita is negative and insignificant in Model (4), consistent with the estimates of Model (1). However, Models (5) and (6) present positive but insignificant coefficients. Access to water, health expenditure, and urbanization present a positive and significant impact on life expectancy.

Table 8 presents the results of life expectancy, focusing more on nitrous oxide (${\mathrm{N}}_{2}0$). We applied PFMOLS as a preferred estimation technique and PDOLS for robustness checks. Interestingly, nitrous oxide (${\mathrm{N}}_{2}0$) enters with a positive and statistically significant coefficient when the PFMOLS estimator is used, suggesting that a 1% increase in ${\mathrm{N}}_{2}0$ will increase life expectancy by 0.6707%, 0.6289%, and 0.6897% for Models (1) to (3), respectively. On the other hand, squared nitrous oxide presents negative and statistically significant coefficients across Models (1) to (3). The results seem to suggest that nitrous oxide might be influencing life expectancy positively up to a certain point but negatively affecting life expectancy in the long run. GDP per capita remains an important determinant of life expectancy—which enters positively and significantly in Models (1) to (3). Thus, a 1% increase in GDP per capita increases life expectancy by 0.0038%, 0.0043%, and 0.0041%, respectively. Access to drinking water, healthcare expenditure, and urbanization all enter with a positive impact on life expectancy. These results are also in line with those presented earlier.

For robustness analysis, we further estimate a model using DOLS. Models (4) to (6) report the factors influencing life expectancy. In these models, the results follow the same direction and pattern as those presented in Models (1) to (3). For instance, nitrous oxide, GDP per capita, access to water, healthcare expenditure, and urbanization still present positive effects on life expectancy. Correspondingly, squared nitrous oxide once again matters in explaining life expectancy—entering with negative and statistically significant coefficients in all models. These results are consistent with the results of Models (1) to (3). Overall, the estimates from DOLS demonstrate consistent results with FMOLS estimates. The conclusions advanced earlier in Table 1 and Table 3 for significant variables also apply to the results presented in this part.

Table 9 presents the results of life expectancy, focusing more precisely on methane $({\mathrm{CH}}_4$), which is positively and significantly related to life expectancy in Models (1) to (3), a sign that capturing and using ${\mathrm{CH}}_4$ presents prospects to produce new sources of clean energy and alleviate global climate change and hence improve quality of life. However, squared methane enters with negative and statistically significant coefficients. In keeping with the results presented earlier, GDP per capita, access to water, healthcare expenditure, and urbanization remain important determinants of life expectancy—which enter positively in all models.

A closer look at the results of the PDOLS model, the results are similar in the direction of the impact to those of the PFMOLS model. As can be observed, ${\mathrm{CH}}_4$ still matters in explaining life expectancy—which enters positively and significantly in the PDOLS model. Alternatively, squared ${\mathrm{CH}}_4$ has continuously been negative and significantly related to life expectancy in all models. Other control variables such as GDP per capita, healthcare expenditure, and urbanization are still present and enter with positive coefficients.

5. Conclusions

The interlinkages between economic development, increased ecological footprints, and health outcomes are major talking points in E7 nations. Not only are these developing countries in a heightened era of industrialization, urbanization, and economic growth, but the source of this development relies mainly on fossil fuel consumption. According to a study by Tong et al. [72], the E7 countries, which consist of Brazil, China, India, Indonesia, Mexico, Russia, and Turkey, accounted for 26% of the global gross domestic product (GDP) in 2018. However, the same group of countries also contributed more than 40% to the total energy consumption worldwide. This level of energy consumption has shown increasing patterns of the ecological footprint in these countries, patterns that might influence health outcomes. However, the current level of acceptably low ecological footprints in E7 nations has led to conflicting results when observing the casual relationship between ecological footprint and health outcomes, especially over the long run. The aim of this study was to build on the existing literature and further assess the impact of the ecological footprint on health outcomes for E7 nations in the long run.

First, the results of the PFOMLS and DOLS models show that the ecological footprint has a positive impact on life expectancy in E7 countries. These results show support for a strong positive relationship between environmental demands and health outcomes in developing countries, consistent with some development studies [43,73]. The positive relationship could be linked with the social benefits that are associated with an increased ecological footprint; with still an acceptably low level of ecological footprint, these benefits still outweigh the negative health effects of environmental degradation. The social benefits, through a larger ecological footprint, include urbanization, improved public health, and an overall improvement in living standards that all result in higher life expectancy. However, upon observing the relationship over time, we discover a decoupling between environmental demands and well-being. It is found that the squared term of ecological footprint has a negative and significant impact on life expectancy. This indicates that as time progresses, the previously positive relationship between environmental degradation and life expectancy breaks down and becomes negative.

To ensure the validity of our results, we also performed additional robustness checks by replacing the ecological footprint measure with other indicators such as CO₂ emissions, nitrous oxide, and methane. The results show that all of these alternative measures for ecological footprint have a positive impact on life expectancy in the short run but a negative impact in the long run. We can therefore conclude that after reaching a certain threshold, a higher level of ecological footprints would lead to lower life expectancy in E7 countries, mimicking the current decoupling environmental degradation–health outcome nexus in developed nations.

Although the relationship between ecological footprint and well-being is still positive in these E7 nations, our results suggest a decoupling relationship over time and provide evidence that over time the relationship between environmental degradation and health outcomes could become negative, and further environmental degradation would affect health outcomes in these countries. This provides an incentive for policymakers to fast-track the reduction in a country’s ecological footprint through policies that include a shift to more renewable energy plans, promoting sustainable transportation, encouraging energy-efficient buildings, promoting sustainable urban development, strengthening environmental regulations, and promoting sustainable agriculture and forestry. Future research could delve deeper into the relationship between the ecological footprint and population health outcomes of key micro-regional cities, considering the impact of regional development policies. Overall, more serious efforts and strategies are needed to keep the size of ecological footprints to be favorable to human life before these countries’ own ecological footprint becomes a real burden on the health outcomes of their citizens.