*3.1. Data*

As mentioned before, this section will present the data used in this empirical investigation. Thirty-one countries from the European region were used, namely Austria (AT), Belgium (BE), Bulgaria (BG), Croatia (HR), Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (GR), Hungary (HU), Iceland (IS), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), Luxembourg (LU), Malta (MT), the Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Republic of Cyprus (CY), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden (SE), Turkey (TR), and the United Kingdom (UK). Moreover, as mentioned earlier, this group of countries was selected because they have experienced a rapid increase in the obesity epidemic and social, economic, and environmental transformations that have facilitated this problem in the last three decades.

Data for the period between 1991 and 2016 was utilised in this investigation. Please note that data for the variable Y only begins in 1991. Moreover, the time series of this investigation goes until 2016 due to data availability for the variable OBESE (see, Our World in Data 2022). The variables used in this empirical investigation are shown in Table 1 below. The variables EC, Y, and CO2, were first transformed into per capita values. Per capita values allow disparities to be controlled for population growth over time and within countries (e.g., Fuinhas et al. 2022; Koengkan et al. 2020). After this, all variables were transformed into natural logarithms ("Log").


**Table 1.** Description of variables and summary statistics.

Notes: Obs., Std.-Dev., Min. and Max denote the number of observations, the standard deviation, the minimum and the maximum, respectively.

To capture the effect of obesity on environmental degradation, the econometric model has to include other variables that also explain pieces of the explained variable, the socalled control variables. Thus, the model uses variables that are in line with economic theory. Furthermore, the variables have support from the literature. For instance, variables Y\_PC, EC, and UP have been used to justify the increase in CO2 emissions, and GLOBA has been used as a proxy for environmental degradation (e.g., Koengkan and Fuinhas 2021a; Hdom and Fuinhas 2020; Wang et al. 2018). On the other hand, the variable OBESE, taken from the Our World in Data (2022), has not been used in literature to capture the rise in environmental degradation. Therefore, our study, by including this variable, is pioneering. The panel of countries and the variables used in this investigation are presented in this subsection. The methodology pursued in this investigation will follow the strategy presented in Figure 1 below.

52

**Figure 1.** Methodology strategy. The authors created this figure.

Therefore, after presenting the variables and the methodology strategy that this investigation will follow, it is also necessary to present the methodological approach for our empirical analysis.

#### *3.2. Methodology*

As mentioned before, our empirical analysis will use the QvM model approach. This method was developed by Machado and Silva (2019) as an alternative for quantile regression. Consistent with Kazemzadeh et al. (2022) and Koengkan et al. (2020), this method can differentiate individual effects in panel data models. Moreover, Machado and Silva (2019) also add that this method can explain how the regressor affects the entire conditional distribution. Therefore, this method can be adapted to provide estimates in cross-sectional models with endogenous variables. Koengkan et al. (2020) reveal that his method is based on moment conditions, not on conditional means, to identify the conditional means under exogeneity and that it allows the identification of the exact structural quantile function. Thus, given these advantages indicated by Koengkan et al. (2020) and Machado and Silva (2019), this empirical investigation opted to use this methodological approach.

After briefly presenting the methodological approach and its advantages, it is time to show the equation where the QvM is constructed. For this, Equation (1) is presented:

$$Y\_{it} = a\_i + X\_{it}' \beta + \left(\delta\_i + Z\_{it}' \gamma\right) \mathcal{U}\_{it} \tag{1}$$

where *Yit*, *X it* comes from a panel of n individuals *i* = 1, ... , *n* over *T* periods, with *P δ<sup>i</sup>* + *Z itγ* > 0 = 1. The parameters (*α*1, *<sup>δ</sup>i*), *<sup>i</sup>* = 1, ... , *<sup>n</sup>*, catch the individual *i* fixedeffects, and *Z* is a *k*-vector of known differentiable (with probability 1) transformations of the components of *X*, with the element *l* given by *Zl* = *Zl*(*X*), *l* = 1, ... , *k*. The sequence { *Xit*} is *i.i.d.*, for any fixed *I*, and independent across *t*. *Uit* are *i.i.d.*, across *i* and *t*, statistically independent of *Xit*, and normalised to satisfy the moment condition *E*(*U*) = 0 ∧ *E*(|*U*|) = 1 (e.g., Koengkan et al. 2020).

Before estimating the model regression, it is advised to assess the statistical proprieties of variables. Therefore, a battery of preliminary tests is applied (see Table 2 below).

**Table 2.** Preliminary tests.


Indeed, after the regression of the QvM model, it is necessary to apply the postestimation tests to identify if the models are adequate. Table 3 below shows the postestimation tests that will be used in this empirical investigation.

**Table 3.** Post-estimation tests for the QvM model.


#### Stata Commands

After presenting the preliminary tests, the QvM model and the post-estimation test, we must show the Stata commands we used in this empirical investigation. Table 4 below shows the Stata commands used.

#### **Table 4.** Stata commands.


Indeed, the preliminary tests, model regression, and post-estimation tests will be accomplished using Stata 17.0. The following section will show the results and discussions.

#### **4. Results and Discussions**

As previously explained, this section will present the results and the possible explanations for the macroeconomic impact of the obesity epidemic on environmental degradation. The preliminary tests indicated that the variables used have characteristics such as (i) lowmulticollinearity among independent variables (as shown in Table A1 in Appendix A); (ii) cross-sectional dependence in the logs of variables (as shown in Table A2 in Appendix A); (iii) variables with orders of integration borderline I(0) and I(1) (as shown

Table A3 in Appendix A); and (iv) the presence of panel fixed-effects (as shown Table A4 in Appendix A). This last result is significant because fixed effects are required in the QvM model. Therefore, the fixed-effects estimator is the most suitable for accomplishing this empirical analysis. Therefore, the empirical results of these tests are vital to identifying the characteristics of the group of countries under study and the possible methodologies to be applied.

The next step after the preliminary tests is to carry out the model regression. The 25th, 50th, and 75th quantiles were calculated to assess the non-linearities of the effect of the obesity epidemic on environmental degradation. We utilised these quantiles to simplify the exhibition of empirical results. Furthermore, we used several quantiles (e.g., 5th, 10th, 15th, and others). It can be seen that there is no information loss, as all independent variables pointed to the same effect of the dependent variable.

Moreover, a dummy variable was added to the model because, during the analysis, the European countries suffered some shocks, such as economic and political. Indeed, if these shocks are not considered, it could produce inaccurate results and misinterpretations during model regression. Therefore, this empirical analysis added dummy variables that represent these shocks. In the literature, the inclusion of dummy variables needs to follow the following triple criterion of choice developed by Fuinhas et al. (2017). For example, (i) the potential relevance of recorded economic and political events at the country level; (ii) the occurrence of international events known to have disturbed the European countries; and (iii) a significant disturbance in the estimated residuals. Therefore, the dummy variables added to the model regression are IDEUROPE\_2015 (Europe, the year 2015). The dummy variables called "IDEUROPE\_2015" represent a decrease in all countries' GDP in the model. This event was caused by the persistent effects of the European debt crisis (often also referred to as the eurozone crisis or the European sovereign debt crisis (Koengkan and Fuinhas 2021a). Table 5 below shows the QvM model regression results with the dummy variable's inclusion. The QvM model results without including the dummy variable can be seen in Table A5 in Appendix A.


**Table 5.** QvM estimation (controlling for shocks).

Notes: \*\*\*, \*\* and \* denote statistically significant at the 1%, 5%, and 10% levels, respectively.

The QvM model regression results indicate that in the 25th, 50th and 75th quantiles, the obesity epidemic, electricity consumption, and urbanisation process increase CO2 emissions. That is, they encourage environmental degradation by increasing CO2 emissions. Nevertheless, economic growth decreases emissions of CO2 in the European region. Moreover, the results from the QvM model also show non-linear behaviour. The estimated coefficients values vary as we go up (or down) in the quantiles' regression. Thus, the empirical results answer the central question of our empirical investigation. Moreover, the post-estimation tests (e.g., the Modified Wald test) indicate that the estimator of this study is adequate (as shown in Table 6 below).

**Table 6.** Post-estimation test.


Notes: \*\*\* denotes statistically significant at the 1% level; H0 of Wald test: the coefficients for all variables are jointly equal to zero.

After finding the positive effect of the obesity epidemic on environmental degradation, it is necessary to ascertain whether the results found by the QvM model regression are reliable and robust when we perform a change in the econometric method approach. Indeed, this approach to finding if the model is robust or not was previously used by Koengkan et al. (2020) and Fuinhas et al. (2017).

After identifying that the obesity epidemic encourages environmental degradation by increasing CO2 emissions, the next step is to answer the following question: What is the possible explanation for this phenomenon? One possible way of explaining this effect is that the obesity epidemic is caused by the increased consumption of processed foods from multinational food corporations, fast-food chains and multinational supermarket chains, as well as the food production on farms, as indicated by some authors (e.g., Koengkan and Alberto Fuinhas 2021b; Fox et al. 2019; Gerbens-Leenes et al. 2010; Popkin 1998). The increased consumption of processed foods from multinational food corporations and farms will positively affect energy consumption from non-renewable energy sources.

Another explanation for this phenomenon is related to the reduction of outdoor activities, which exacerbates the problem of obesity. Consequently, this reduction will encourage intensive motorised transportation, screen-viewing leisure activities, and the use of home appliances, as indicated by some authors (e.g., Koengkan and Alberto Fuinhas 2021b; Bell et al. 2002; Sobal 2001). Indeed, Koengkan and Alberto Fuinhas (2021b) also add that the increase in the use of home appliances and motorised transportation has implications for the energy demand from fossil fuel energy sources, where the consumption of these kinds of sources increases considerably.

Additionally, the positive impact of the obesity epidemic on CO2 emissions is also indirectly related to economic growth, globalisation, and urbanisation. Therefore, the levels of obesity have been related to increasing economic activity, where economic development causes effects on dietary changes (e.g., Springmann et al. 2016). Indeed, the transition from low to high income caused by this process tends to induce some individuals to consume fatty and energy-dense foods of animal origin. Therefore, the increase in income contributes to the rise in obesity levels, except in countries where home-produced food is predominant (e.g., Roskam et al. 2010; Gerbens-Leenes et al. 2010).

The globalisation process also causes an increase in obesity by the dietary changes. According to Popkin (1998), Fox et al. (2019), Koengkan and Alberto Fuinhas (2021b) and Koengkan et al. (2021), the process of globalisation will contribute to the food chain extension. As mentioned above, this extension will enable economies of scale in food production processes. Consequently, the economies of scale in food production processes will allow a diet rich in energy-caloric foods. Food consumption with high sugar and salt contents is less expensive and accessible to lower-income classes. Moreover, the unhealthy supply of processed foods is related to the increase in multinational food companies, supermarkets, and fast-food chains caused by globalisation.

The urbanisation process also plays a role in the increase in the obesity problem. The process of urbanisation allows better accessibility to food due to supermarkets, multinational supermarkets, and fast-food chains offering a ready supply of processed foods, which consequently causes the decline of farm stands and open markets with healthier foods (Reardon et al. 2003). This same process also exposes people to the mass media marketing of food and beverages that influence traditional diets (Hawkes 2006). Moreover, urbanisation increases car use and reduces walking or biking for transportation or leisure, contributing to obesity, and obesity increases car use. Moreover, all these explanations align with the findings of Koengkan and Alberto Fuinhas (2021b). They found that economic growth, globalisation, and urbanisation positively affect the overweight problem in the European region. They consequently encourage energy consumption from non-renewable energy sources and subsequently increase CO2 emissions/environmental degradation. Furthermore, the explanations above align with the results from the complementary analysis that this investigation carried out (as shown in Table A6, in Appendix A).

Indeed, the positive effect of electric power consumption on CO2 emissions could be related to some factors highlighted in this empirical investigation. On the one hand, it can be linked to electricity consumption in groups of not environmentally responsible countries. Consequently, electricity generation from fossil fuels is linked with increases in CO2 emissions. These findings may also signal that the panel of countries under study could depend on fossil fuel sources for economic growth. On the other hand, it may be linked to the inefficiency of renewable energy policies that stimulate the development and consumption of renewable energy sources. Several authors have already found this impact (e.g., Fuinhas et al. 2021; Ozcan et al. 2020; Muhammad et al. 2020; Adedoyin et al. 2020; Koengkan and Alberto Fuinhas 2021b; Yazdi and Dariani 2019; Salahuddin et al. 2019; Fuinhas et al. 2017).

However, the negative effect of economic growth on CO2 emissions could be related to some factors highlighted in this empirical investigation. A justification for the negative impact can be linked to a strong depression/recession affecting people's consumption behaviour. Accordingly, it affected energy-intensive sectors, electricity consumption, and, finally, the emissions of CO2. A U-shaped relationship between economic growth and CO2 emissions could be another possible explanation for this negative impact. An increase in economic growth initially leads to a decline in CO2 emissions levels, consequently reaching a threshold. Indeed, economic activity intensifies environmental degradation. Indeed, the country's industrialisation increases pollution. The policies limiting the levels of industrial pollution can be another explanation. Those policies promote the embracing of environmentally friendly techniques and processes of production.

Consequently, environmentally friendly technologies were promoted, and this promotion contributes to producing and consuming renewable energy by industries and families. However, some authors found a negative impact on economic growth and CO2 emissions (e.g., Koengkan and Alberto Fuinhas 2021b; Muhammad et al. 2020; Aye and Edoja 2017).

Moreover, the positive effect of urbanisation on CO2 emissions could be related to increased urban populations, positively affecting the demand for energy from fossil fuel energy sources, households, the transport sector and industries. Additionally, this positive effect could be related to the low energy efficiency improvement caused by the slow introduction of new energy technologies, low diversification of energy sources and low environmental regulation efficiency. It encourages industries' and families' acquisition of environmentally friendly technologies (Koengkan and Alberto Fuinhas 2021b). Figure 2 below summarises the effect of independent variables on the dependent variable.

This section showed the results from the primary model and the robustness check, the possible explanations for the impact of the obesity epidemic on environmental degradation, and a brief explanation of the impact of other variables. The next section will show the conclusions of this experimental investigation.

**Figure 2.** Summarization of the effect of the independent variables on the dependent variable.

#### **5. Conclusions**

This empirical investigation approached the macroeconomic effect of the obesity epidemic on environmental degradation using CO2 emissions as a proxy in a panel of thirty-one European countries from 1991 to 2016. As stated above, the QvM model was used to carry out an empirical study. This study's preliminary tests indicated that the variables used have low-multicollinearity characteristics, cross-sectional dependence in logarithms, and variables have I(0) or borderline I(1) order of integration, non-presence cointegration between the variables, and also fixed-effects.

The QvM and fixed-effect models regression results show that the obesity epidemic, electricity consumption, and urbanisation encourage environmental degradation by increasing CO2 emissions. At the same time, economic growth decreases the emissions of CO2 in the European region. Indeed, the post-estimation test results for the QvM model indicate that this study's estimator is adequate.

The obesity epidemic increases the environmental degradation problem in three ways. First, the obesity epidemic is caused by increased consumption of processed foods from fast-food and multinational supermarket chains and multinational food corporations. The increase in food production will positively impact fossil fuel energy sources' energy consumption. Second, obesity reduces physical and outdoor activities, increasing the intensive use of motorised transportation, home appliances, and screen-viewing leisure activities, consequently increasing energy consumption from non-renewable energy sources. Finally, a third possible way can be related indirectly to economic growth, globalisation, and urbanisation. This empirical evidence leads to a supplementary research question: What can be done to reverse the influence of the obesity epidemic on environmental degradation in the European region?

Several initiatives need to be created to reduce the effect of the obesity epidemic on environmental degradation. The first initiative is related to policies that reduce the sale of foods with high calorie-energy close to schools; the second initiative is to create policies that restrict the sale and consumption of unhealthy foods through taxation. Unhealthy food will cost more with the introduction of taxes and, consequently, encourage healthier foods. The third initiative is related to developing policies that encourage the generalised practice of physical activity and its importance. The fourth initiative is related to reducing lobbying by multinational food corporations through policies encouraging local producers. Finally, the fifth and last initiative is associated with creating policies that encourage the food sector to produce foods that are more healthy and have the least possible impact on the environment.

However, this problem is not limited to reducing the obesity problem. It is necessary to change production and consumption in the European region. Indeed, several initiatives have already been implemented to reduce the consumption of non-renewable energy in the region. However, it is essential to do more to reverse this situation. For example, policymakers need to create more measures to reduce the barriers to products and technologies that improve energy efficiency and produce green energy. This reduction could benefit households and industries by acquiring renewable energy technologies and reducing these products' prices.

Regarding food production, it is necessary to introduce policies that encourage, (i) better productivity, where the efficiency improvements can lead to a 33% reduction in land use, a 12% reduction in water use, and a 16% reduction in production emissions; (ii) reduce livestock emissions, where the increase in productivity and efficiency gains can reduce land use, feed requirements, and GHG emissions per gallon of milk or pound of meat; (iii) reduce the consumption of fertiliser, as the use of these substances emits nitrous oxide, a potent greenhouse gas (the introduction of techniques including nitrification inhibitors can replace applications of fertilisers); (iv) introduce renewable energy and energy efficiency technologies supported by fiscal and financial instruments that help farmers gain access to renewable energy and energy-efficient technologies; and (v) reduce waste and food loss, with support of fiscal and financial instruments to help farmers to improve their equipment and energy efficiency in farm buildings. In addition, these policies can reduce the consumption of fossil fuels and emissions.

This study is a kick-off regarding the effect of the obesity epidemic on environmental degradation and other aspects such as energy consumption, economic growth, and urbanisation. Therefore, this investigation is in the initial stages of maturation, which will supply a solid foundation for second-generation research regarding this topic.

#### *Limitations of the Study*

As we already know, this empirical investigation is not free of limitations. Indeed, the preliminary limitations of this empirical investigation stem from (i) the presence of a short period. In this investigation we used the period between 1991 and 2016. Indeed, this period was used due to data availability for the variable OBESE. Moreover, more time is necessary to capture the dynamic effects of the variables OBESE, EC, Y\_PC, and UP; (ii) the inexistence of literature that approaches the macroeconomic effect of the obesity epidemic on environmental degradation. The lack of this kind of literature makes difficult the elaboration of deeper discussions regarding the results found; and (iii) the European countries are firmly integrated and are mainly developed ones. This former characteristic limits the generalisation of our results to diverse contexts.

The limitations mentioned above are usually found in investigations in their early stages of maturation. Developing second-generation studies regarding this topic is essential to overcoming these limitations. Despite the limitations in this investigation, this study could draw meaningful conclusions.

**Author Contributions:** J.A.F.: Writing—review and editing, Supervision, Funding acquisition, and Project administration. M.K.: Conceptualisation, Writing—Original draft, Supervision, Validation, Data curation, Investigation, Formal analysis, and Visualisation. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was financially supported by the research unit on Governance, Competitiveness and Public Policy, UIDB/04058/2020 and UIDP/04058/2020, funded by national funds through FCT—Fundação para a Ciência e a Tecnologia; and by the CeBER R&D unit, funded by national funds through FCT—Fundação para a Ciência e a Tecnologia, I.P., project UIDB/05037/2020.

**Data Availability Statement:** Data available on request from the corresponding author.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **Appendix A**

**Table A1.** VIF-test.


#### **Table A2.** Pesaran CD-test.


Notes: \*\*\* denotes statistical significance at the 1% level; Ho for CD-test: cross-section independence.

**Table A3.** Panel Unit Root test (CIPS-test).


Notes: \*\*\* and \*\* denote statistically significant at the 1% and 5% levels, respectively.

**Table A4.** Hausman test.


Notes: \*\*\* denotes statistically significant at the 1% level.


#### **Table A5.** QvM estimation without dummy variable.

Notes: \*\*\* and \* denote statistically significant at the 1% and 10% levels, respectively.

**Table A6.** QvM estimations (complementary analysis).


Notes: \*\*\*, \*\* and \* denote statistically significant at the 1%, 5%, and 10% levels, respectively.

#### **References**


countries-1971-2017 (accessed on 11 February 2022).

