Quality of Life and the Carbon Footprint: A Zip-Code Level Study Across the United States

Abstract

Much sustainability scholarship has examined the environmental dimensions of subjective and objective well-being. As an alternative measure of human well-being, we consider the notion of quality of life and draw on a framework from the sustainability literature to study its association with ecological impact, specifically the carbon footprint. We conduct a quantitative analysis, combining zip-code level data on quality of life and the carbon footprint per household for the year 2012 across the continental United States (n=29,953). Findings consistently show a significant, negative association between quality of life and the carbon footprint. Our findings point to the potential advantages of utilizing robust objective measures of quality of life that extends beyond economic well-being and life expectancy alone. Furthermore, our findings question the conventional wisdom that sustainability requires sacrifices, while suggesting opportunities for how increased levels of sustainability may be achieved while retaining high levels of quality of life.

Keywords

quality of life subjective well-being carbon footprint spatial regression zip code united states

Introduction

There is a long history of social science research examining the social drivers of environmental impacts (e.g., Rees and Wackernagel 1996; Wackernagel and Yount 2000; York et al., 2003). Concurrent with this scholarship has also been an interest in the cultural and social psychological implications of environmental change (e.g., Leiserowitz et al., 2006; Markowitz et al., 2012). Relatedly, environmental social scientists and psychologists have been studying the subjective and material well-being of individuals in the context of modern environmental change (e.g., Ambrey & Daniels., 2017; Biedenweg 2017; Cloutier et al., 2014; Maibach et al., 2010; Meijer and Van Beek 2011; Norgaard Kari, 2011). Indeed, there is much interest in the connection between sustainability and subjective well-being (e.g., Ambrey & Daniels., 2017; Cloutier et al., 2014; Helliwell 2014); some research indicates that people who are happier tend to live in a more sustainable way (e.g., Brown et al., 2005) and that there is broad support among the public for living more sustainably (e.g., Bowerman 2014). Meanwhile, in developed countries, like the United States, there is also a commonly made argument that living more sustainably will require sacrifices of life quality, both subjective and material, and that people are unwilling to make these sacrifices, given that they might not think that environmental changes will affect them personally (Forbes 2018; Hall et al., 2018; Oskamp 2000).

The quality of life literature as it relates to sustainability can be largely grouped into two camps: those focusing upon subjective measures of well-being (e.g., happiness) and those focusing on objective measures of well-being (e.g., life expectancy and gross domestic product (GDP)). Much of the sustainability research on well-being focuses on subjective well-being (e.g., Ambrey & Daniels., 2017; Cloutier et al., 2014) and in general they find that sustainability is positively correlated with increases in subjective well-being, while increased carbon intensity is inversely related to subjective well-being. At the same time, there has been a considerable amount of research that focuses upon objective well-being and sustainability or sustainable development (see Constanza et al., 2007 for a review of this literature). Although there is some nuance to this literature, most have argued that economic development, increases in GDP, are generally correlated with increases in quality of life as measured by health, social, and economic outcomes. However, those same increases in objective quality of life outcomes incur a cost in terms of sustainability (Dietz, Rosa, and York 2012; Jorgenson 2014). Although the former literature is more generally regarded as part and parcel of the quality of life/well-being literature, we draw on this second line of research by focusing on a unique means of measuring objective quality of life/well-being and combining it with carbon footprint data to provide new insights into this relationship.

In the following analysis, we aim to make a contribution to this topic by making a distinction between subjective well-being and quality of life and asking whether, at the local level across the United States, there is a statistical association between quality of life and a particular environmental impact: the carbon footprint. To answer this question, with coverage for nearly all zip-codes across the contiguous United States (n = 29,953), we integrate information from two primary data sets. First, from SimplyAnalytics (2021), we collect information for the dependent variable quality of life as well as other demographic and economic factors to be included as control variables. This quality of life variable has been used in a range of academic scholarship, including the sustainability literature (Choudhury, et al., 2021; Clement, et al., 2017; Tsai et al., 2011). It is a weighted composite, containing nine separate sub-indices: the Weather Index, Total Crime Index, Earthquake Index, Culture Index, Amusement Index, Restaurants Index, Medical Index, Religion Index, and Education Index. Second, the information for the carbon footprint per household, the primary independent variable, comes from Ummel (2014; 2016). We discuss these data in greater detail below; here, we highlight two features of this variable. First, the use of average household environmental impacts, rather than per-capita impacts, is consistent with the sustainability literature, in which scholars argue that the ecological dimensions of population size are most appropriately captured through household dynamics (Bradbury, et al., 2014; Liu, et al., 2003). Second, a carbon footprint measure, with direct and indirect sources of emissions, is a valuable source of information previously analyzed by environmental social scientists (e.g., Pattison et al., 2014; Clement et al., 2017). Measured at the level of the zip-code, we compare OLS and spatial regression models to estimate the cross-sectional association between quality of life and average household carbon footprint, controlling for a host of other factors. Before describing the data and analysis, we first review the relevant literature, deriving an exploratory hypothesis to assess the association between quality of life and the carbon footprint.

Dimensions of Environmental Impact and Human Well-Being

The ecological footprint, originally developed by Rees and Wackernagel (1996), is a tool which measures humanity’s environmental impact. Conceptually, the ecological footprint is intended to represent the area of land (and water) required to produce the natural resources consumed by human activities and also the space needed to appropriate all the waste generated as a result of this consumption. Empirically, readily accessible information on natural resource consumption, especially at the national level (see Wackernagel and Yount 2000), has allowed scholars to analyze variation in the ecological footprint over time and across space (e.g., Jorgenson and Clark 2011). Indeed, environmental social scientists have long treated the ecological footprint as a primary variable in evaluating whether nations are developing sustainably (e.g., York et al., 2003; Moran et al., 2008; Figge et al., 2016; Jorgenson and Clark 2011).

While the ecological footprint continues to be used as a common metric of environmental change, some scholars have highlighted its limitations (e.g., Van Den Bergh and Grazi 2010; Wackernagel and Yount 2000; Fiala 2008). Among other concerns, these scholars note that the ecological footprint is disproportionately weighted by the total amount of fossil fuel that humans consume. Indeed, a vast area of forested land is needed to sequester all of the carbon dioxide that humans produce. The Global Footprint Network (2019) estimates that 60% of humanity’s overall ecological footprint is due to the demands placed on ecosystems to cycle anthropogenic CO2 out of the atmosphere. Taking this issue into account, some environmental scholars have begun to construct alternative footprint variables. The ecological footprint is a composite measure and is made up of individual components that can be disaggregated and studied separately. Rather than utilize the ecological footprint as a singular measure of impact, scholars began to focus specifically on the carbon emissions component, developing data sets to track variation in the carbon footprint as a stand-alone metric of environmental change (Ambrey & Daniels., 2017; Jones and Kammen 2014; Ummel 2014, 2016). For our analysis, we follow this approach and incorporate data on the carbon footprint, which we discuss below.

Wackernagel and Yount (2000) argue that the ecological footprint can be refined and strengthened with both new applications and new methods. We address several of these issues, namely, the focus on regional footprints and the application of footprint measures on the human dimensions of satisfaction—quality of life. Meanwhile, in the sustainability literature, the notion of human well-being is also treated as a multifaceted concept with different operational measures (e.g., Biendenweg 2017; Moran et al., 2008). Before reviewing this literature, we first note that there are two common metrics of human well-being: Quality of Life and Subjective Well-Being. The distinction between quality of life and subjective well-being is often blurry as they are sometimes used interchangeably in the literature (Camfield and Skevington 2008; Moro et al., 2008; Veenhoven 2001).¹ However, as variables, the information used to measure subjective well-being is different than the information used to measure quality of life. Subjective well-being, often based on individual-level survey data, is used to refer to an individual’s perception of their own satisfaction with life, while quality of life is a weighted index derived from objective measures of local climate, safety, access to resources and amenities, among other factors (e.g., Kelly et al., 2017). Even though some social scientists have debated the extent to which there is a meaningful difference between the two, at least in terms of whether one or the other better measures the concept of human well-being (e.g., Moro et al., 2008; Veenhoven 2001), the sustainability literature does treat these as distinct concepts, with the majority of the work looking at the environmental dimensions of subjective well-being (e.g., Ambrey & Daniels., 2017; Cloutier et al., 2014; Ericson et al., 2014; Lenzen and Cummins 2013; Verhofstadt et al., 2016; Welsch and Jan, 2011). Here, we briefly review that literature, noting that there has been far less attention on the ecological implications of quality of life.

Environmental Dimensions of Human Well-Being

Traditionally, objective measures of quality of life have been assessed in relatively straightforward ways—whether economic development, via increased GDP, is associated with increased life expectancy (for overview see Riley 2008). In the sustainability literature, a number of works have outlined this research yet found that there is an ecological cost to improvements (Jorgenson 2014; Dietz et al., 2012) and some have identified a path towards a more inclusive means of measuring quality of life (Costanza et al., 2007). However, the critical point of this literature is largely that much of the research on quality of life fails to capture either the dimensions of sustainability or a more complete array of entities that would fully capture these relationships, particularly a broad objective measure of quality of life. We hope to build on the work of others, such as Constanza et al. (2007), by suggesting a more thorough accounting of quality of life measures with a clear measure of sustainability—the carbon footprint. Our approach, however, remains within the underdeveloped camp of research on objective measures of quality of life. However, it is worth noting that the one area of research that has attempted to carve out a more thorough understanding of quality of life is the literature on subjective well-being and sustainability, a literature we turn to next.

Cloutier et al. (2014) compared four different sustainability indices to the Gallup Healthways Well-Being Index for major cities across the U.S. The authors ranked cities in order of best to worst, highest to lowest for The Green City Index, SustainLane US City Rankings, Popular Science US City Rankings, Our Green Cities Index, and the Gallup Healthways Well-being index. The benefit of using all four sustainability indices is that each one measures a different aspect of sustainability. The Green City Index evaluates cities based on environmental performance, such as energy, land use, waste, and CO2; SustainLane is designed to measure how prepared a city is for an unknown future by evaluating such factors as public transport ridership, air quality, land use, local food and agriculture, and housing affordability, among others; The Popular Science index exclusively considers electricity, transportation, green living, and recycling and green perspective; and Our Green Cities looks at the number of different sustainability related programs a city has undertaken (Cloutier et al., 2014). The results of their research indicate that all four sustainability measures were positively correlated with the happiness index, with two having a significant correlation. In a nation-wide, individual-level panel study in Australia, Ambrey and Daniels (2017) find a similar result, showing that carbon footprints and individual well-being are inversely related.

The associations between individual components of both sustainability and happiness indices have also been looked at by some researchers. Lenzen and Cummins (2013) identify common aspects of subjective well-being and carbon footprint, using survey data of Australian lifestyles. In alignment with previous research, they found that higher income levels are associated with increased emissions. However, as Lenzen and Cummins mention, other researchers have found that once gross income reaches $100k carbon emissions continue to increase but well-being does not. They also conclude that while owning a vehicle is associated with increased well-being, living in an area with a high level of vehicle ownership is associated with a decrease in subjective well-being. Interestingly, their research shows that a higher level of academic qualification is associated with an increase in emissions. They attribute this to the increase in socio-economic level that usually accompanies increasing level of education.

Verhofstadt, Ootegem, Defloor, and Bleys (2016: 80) use survey data from a region of Northern Belgium to look at the link between the “environmental sustainability of an individual’s lifestyle” using ecological footprint and certain factors of subjective well-being. Their findings indicate no overall significant correlation between the two but did identify that “the main item that is good for both subjective well-being and ecological footprint is the consumption of seasonal products and fresh products” and that not using electricity for home heating both “reduced [ecological footprint] and increased [subjective well-being]” (Verhofstadt et al., 2016: 83-84). They explain the implication of not finding any significant correlation: No significant correlation means that “having a lower footprint is not associated with reporting a higher level of well-being” but also, and importantly, that “having a lower footprint does not reduce one’s level of subjective well-being” (83). In a related study, using ecological footprints at the nation-state level to account for environmental stressors as part of their model for environmentally efficient well-being, Dietz, York, and Rosa (1999: 119) found “no evidence that adversely stressing the environment improves human well-being, net of affluence and human capital.” Fritz and Koch (2014) also conducted analyses at the nation-state level by considering four dimensions: ecological sustainability, social inclusion, quality of life, and economic development. Their results indicate that even though there is no single clustering of nations that results in a high amount of social inclusion, quality of life, economic development, and ecological sustainability, there are cases (Costa Rica and Uruguay) that provide clues of how to achieve this outcome and can be used in guiding policy decisions.

Two other factors that have been identified as being positive for both subjective well-being and sustainability are empathy and compassion (Ericson et al., 2014). Ericson et al. (2014) suggest that at least part of the relationship may be explained by how both empathy and compassion have the potential to frame environmentalism as an ethical issue which, in turn, may result in behavioral changes. Empathy and compassion might also play into sustainability in another way: both are known to increase pro-social behavior and this may spill over into more pro-environmental choices. Pro-environmental consumption choices have been identified by some research as contributing to a rise in life satisfaction (Welsch and Jan, 2011). One example of this is the proposed idea of ecological citizenship in which pro-environmental behaviors are practiced as a part of everyday shopping behaviors (Seyfang 2005) which at least has the potential to increase life satisfaction.

Summary and Hypothesis about Quality of Life

Based on the above, brief literature review, we observe that sustainability research has found either an inverse association or no statistically significant association between environmental impact and subjective well-being. These findings suggest, at the very least, that efforts to foster sustainability do not compromise the improvement of subjective well-being, and vice versa (i.e., the improvement of subjective well-being does not undermine efforts to foster sustainability). Likewise, studies of objective quality of life largely focus on economic (GDP) and health outcomes (life expectancy) and generally note that sustainability and quality of life are inversely related. We draw on these findings to hypothesize the ecological dimensions of quality of life as an alternative measure of human well-being. Again, while sometimes used interchangeably in the literature, we reiterate that subjective well-being represents a subjective assessment as reported by the individual, while the quality of life metric we use is composed of factors independently observed by various agencies that collect data on weather, crime, education, and local amenities, among other topics. In this light, the data collection methodology for quality of life is comparable to the methodology used to measure environmental impacts, such as the carbon footprint, as we discuss below. Likewise, much of the subjective well-being research has been focused on individual-level behaviors and perceptions about reducing carbon emissions, while objective studies of quality of life are generally fixated on economic well-being. Thus, our contribution is to provide a means to systematically analyze objective measures of quality of life within the context of a comprehensive carbon emissions measure: the carbon footprint. By doing so, the results can inform our understanding of how these two concepts might be related (Wackernagel and Yount 2000) and may provide clues as to how to address policy concerns in an era of climate change. To that end, given the findings from the subjective well-being literature, we propose the following hypothesis on the relationship between the carbon footprint and quality of life:

Hypothesis

There is a negative association or no statistically significant association between quality of life and the carbon footprint.

Data and Analysis

The variables for this study come from two primary data sets: SimplyAnalytics (2021) and Ummel (2014;2016). The information for the carbon footprint, the primary independent variable, comes from Ummel (2014; 2016), and SimplyAnalytics provides the information for eight of the independent variables. We utilize ArcMap to generate the land area, longitude, and latitude variables. All variables are observed for the year 2012, with coverage for more than 70% of all zip-codes across the United States (n = 29,953). From 2012, these zip-codes contained just over 95% of the entire population of the United States. Table 1 displays the variable descriptions.

Table 1.

Description of Variables^a.

Variable	Source	Description
1. Quality of Life	SimplyAnalytics (2021)	The Quality of Life variable is a weighted composite developed by Easy Analytic Software Inc. (SimplyAnalytics 2015), comprised of nine separate sub-indices: the Weather Index, Total Crime Index, Earthquake Index, Culture Index, Amusement Index, Restaurants Index, Medical Index, Religion Index, and Education Index (https://www.easidemographics.com/mdbhelp/html/quality_of_life_1.htm). Each sub-index is described below, with its weight used in the calculation of the Quality of Life composite. As the magnitude of the weight increases, the sub-index has more influence on the overall Quality of Life variable. For instance, the weather index has the biggest weight, and the religion index has the smallest weight. The Weather Index is the weighted average of ten factors concerning temperature, precipitation, daylight and cloud cover: (1) Annual Maximum Average Temperature, (2) Annual Normal Heating Degree Days, (3) Annual Normal Cooling Degree Days, (4) Percent of Possible Sunshine, (5) Mean Sky Cover (Sunrise to Sunset), (6) Mean Number of Clear Days, (7) Annual Maximum Average Temperature minus Annual Minimum Average Temperature, (8) Mean Number of Snow Days, (9) Average Annual Precipitation, and (10) Average Annual Snowfall. (Weight = Positive 20) The Crime Index a model of the combined likelihoods of the following felony crime categories to be present in each area: (1) Aggravated Assault, (2) Burglary, (3), Forcible Rape, (4), Forcible Robbery, (5) Larceny, (6) Motor Vehicle Theft, and (7) Murders. (Weight = Negative 2) The Earthquake Index is based on a seismic measure called “effective peak acceleration,” which is based on the degree of protection provided for in public and occupant buildings against the potential of earthquakes. (Weight = Negative 1) The Culture Index is derived from the number of employees working in the following industries: (1) Museums, Historical Sites, and like Institutions, and (2) Performing Arts Companies. (Weight = Positive 3) The Amusement Index is derived from the number of employees working in the following industries: (1) Amusement Parks and Arcades, (2) Gambling Industries, (3) Other Amusement and Recreation Industries, and (4) Spectator Sports. (Weight = Positive 3) The Restaurants Index is derived from the number of employees working in the following industries: (1) Full Service Restaurants, (2) Limited Service Eating Places, (3) Special Food Services, and (4) Drinking Places. (Weight = Positive 5) The Medical Index is derived from the number of employees working in seventeen different medical industries, such as hospitals, community clinics, nursing care facilities, rehabilitation services, among others. (Weight = Positive 3) The Religion Index is derived from the number of employees working in the Religious Organizations industry. (Weight = Positive 1) The Education Index is derived from the number of employees working in the following industries: (1) Elementary and Secondary Schools, (2) Junior Colleges, (3) Colleges, Universities, & Professional Schools, (4) Business Schools and Computer and Management Training, (5) Technical and Trade Schools, (6) Other Schools and Instruction, and (7) Educational Support Services. (Weight = Positive 2)
2. Carbon Footprint Per Household	Ummel (2014; 2016)	The carbon footprint variable measures the average household fossil fuel consumption (both direct and indirect) for each zip-code. The information for this variable is derived from household-level expenditure data obtained from the Bureau of Labor Statistics’ Consumer Expenditure Survey (CEX). Household-level expenditures were averaged over the years 2008–2012 for 52 spending categories, including specific items such as air travel, beef, electricity, major and small appliances, public transportation, and telecommunications. Specific emissions intensity factors for each of the 52 categories are derived from different academic and governmental sources; for instance, fuel price and life-cycle data from the Energy Information Agency is used to generate intensity factors for energy consumption; the Environmental Protection Administration’s eGRID program provides intensity factors for electricity generation; and the MIT Airline Data Project was used to help estimate an intensity factor for airline travel. With these emission intensity factors, the expenditure data from the CEX can be used to calculate the amount of carbon associated with the amount spent on that particular category. We note that Citizens’ Climate Lobby first converted these into per household carbon burden dollar amount based on a price of $15 per metric ton of CO₂ equivalent. Ummel utilized simulation techniques to aggregate an average household carbon footprint to the zip-code level for the entire continental USA. For our study, we take these data at the zip-code level and divide the per household dollar burden by $15, yielding an estimate for the carbon footprint of households. We then compute the natural logarithm of these values to obtain our primary independent variable, which is a technique employed by other scholars using these data (e.g., Pattison et al., 2014; Clement et al., 2017)
3. Population Density	SimplyAnalytics (2021)	Number of residents per square mile of land area. Log-transformed for bi-variate and regression analyses
4. Median Household Income	SimplyAnalytics (2021)	The median income of all households. Log-transformed for bi-variate and regression analyses
5. Median Household Size	SimplyAnalytics (2021)	The median number of people per household. Log-transformed for bi-variate and regression analyses
6. Unemployment	SimplyAnalytics (2021)	The number of people unemployed divided by the total labor force multiplied by 100. Log-transformed for bi-variate and regression analyses
7. Percent Bachelor's or Higher	SimplyAnalytics (2021)	The number of people with a bachelor’s degree or higher divided by the total population multiplied by 100. Log-transformed for bi-variate and regression analyses
8. Percent 25 or Older	SimplyAnalytics (2021)	The number of people who are 25 years or older divided by the total population multiplied by 100. Log-transformed for bi-variate and regression analyses
9. Percent White	SimplyAnalytics (2021)	The number of people who are non-Hispanic white divided by the total population multiplied by 100. Log-transformed for bi-variate and regression analyses
10. Area	ArcGIS	The area of the zip-code in square miles. Log-transformed for bi-variate and regression analyses
11. Latitude	ArcGIS	The latitude of the centroid of the zip-code
12. Longitude	ArcGIS	The longitude of the centroid of the zip-code

^aNote: Given our focus on the association between quality of life and the carbon footprint, we only include brief descriptions of the control variables.

Dependent Variable: Quality of Life

From the SimplyAnalytics database, we utilize the Quality of Life index, the data for which are generated by Easy Analytic Software, Inc. (SimplyAnalytics 2015). As a partner with SimplyAnalytics, Easy Analytic Software, Inc. (EASI) has developed demographic and economic data used by a variety of individuals and organizations in both the private and public sectors, including for use in academic research (e.g., Crowley 2021; Emerson et al., 2001). The EASI Quality of Life variable has been used in scholarly publications by academics in a range of different disciplines (Choudhury, et al., 2021; Clement, et al., 2017; Tsai et al., 2011). This variable is a weighted composite developed by EASI, comprised of nine separate sub-indices: the EASI Weather Index, EASI Total Crime Index, Earthquake Index, Culture Index, Amusement Index, Restaurants Index, Medical Index, Religion Index, and Education Index. The indices for culture, amusement, restaurants, medical, religion, and education are based on the number of people employed in each industry and are intended to reflect the availability of these resources in a locality; EASI Weather Index uses the closest weather station to determine a proxy score for the impact of weather and is based on numerous factors including annual maximum temperature, mean number of days of snow, and average annual precipitation; EASI Crime index models the likelihood of various types of crime to occur in a given area; the Earthquake index is based on the measure of effective peak acceleration, the same factor used in federal building requirements. See Table 1 for a complete description of the quality of life variable. These measures are in line with those used in previous environmental research using Quality of Life (Clement et al., 2017; Moro et al., 2008). The higher the values for the quality of life variable, the higher the quality of life. For the bivariate and regression analyses, we compute the natural logarithm of these values to ensure that slope estimates are standardized as elasticities.

Primary Independent Variable: Carbon Footprint

The information for the primary independent variable comes from Ummel (2014, 2016), measuring the average household carbon footprint (CF) at the zip-code level. The information for this variable is based on household-level expenditure data obtained from the Bureau of Labor Statistics’ Consumer Expenditure Survey (CEX). Household-level expenditures were averaged over the years 2008–2012 for 52 spending categories, including specific items such as air travel, beef, electricity, major and small appliances, public transportation, and telecommunications. Specific emissions intensity factors for each of the 52 categories are derived from different academic and governmental sources; for instance, fuel price and life-cycle data from the Energy Information Agency is used to generate intensity factors for energy consumption; the Environmental Protection Administration’s eGRID program provides intensity factors for electricity generation; and the MIT Airline Data Project was used to help estimate an intensity factor for airline travel. With these emission intensity factors, the expenditure data from the CEX can be used to calculate the amount of carbon associated with the amount spent on that particular category. We note that Citizens’ Climate Lobby first converted these into per household carbon burden dollar amount based on a price of $15 per metric ton of CO₂ equivalent. Ummel utilized simulation techniques to aggregate an average household carbon footprint to the zip-code level for the entire continental USA. For our study, we take these data at the zip-code level and divide the per household dollar burden by $15, yielding an estimate for the carbon footprint of households. For the bivariate and regression analyses, we then compute the natural logarithm of these values to obtain our primary independent variable, which is a technique employed by other scholars using these data (e.g., Pattison et al., 2014; Clement et al., 2017).

Independent Variables: Controls

Drawing from previous sustainability studies (e.g., Ambrey & Daniels., 2017; Clement et al., 2017; Cloutier et al., 2014; Pattison et al., 2014), we incorporate a suite of variables to control for demographic, economic, and geographic factors. The information for these variables also comes from SimplyAnalytics. The seven control variables include population density, median household income, median household size, unemployment, percent bachelor’s or higher, percent 25 or older, and percent white. Population density equals the number of residents per square mile of land area; median household income equals the middlemost income for all households in the zip-code; median household size represents the typical number of people living in a housing unit; unemployment measures the percent of the labor force who are actively seeking but without paid work; for educational attainment, we include the percent of the population with a Bachelor’s degree or higher; for age structure, we collect information on the percent of the population who is 25 years or older; and, lastly, we include a race variable to measure the percent of the population who is non-Hispanic white. To control for geographic variation in zip-codes, we utilize GIS software to compute the land area and determine the longitude and latitude of the centroid in each zip-code used in our study. These geographic measures control for factors that do not change from year to year, such as size of the zip-code, the length of the day, and when the sun rises and sets. In this way, these geographic factors help to minimize omitted variable bias in a cross-sectional analysis with a single year of data.

For the bivariate and regression analyses, we compute the natural logarithm of all variables (except longitude and latitude); with logged values for both the dependent and independent, the slope estimates are interpreted as elasticities, representing, generally speaking, the percent change in the dependent variable for every one percent change in the independent variable, controlling for the other variables in the model.

Analysis

To test whether there is an association between quality of life and the carbon footprint, we estimate three separate cross-sectional regression models: a standard ordinary least squares (OLS) model, a spatial lag model, and a spatial error model. Comparing results from OLS and spatial regression models highlights any spatially-induced bias in slope estimates. The generic equations for these three models are as follows

{Model 1 OLS: y}_{i} {= a + x}_{ik} β_{k} + ε_{i}

{Model 2 Spatial Lag: y}_{i} = a + ρ {Wy}_{i} {+ x}_{ik} β_{k} + ε_{i}

{Model 3 Spatial Error: y}_{i} {= a + x}_{ik} β_{k} + ν_{i}; ν_{i} = λ W ν_{i} + ε_{i}

wherein, y_i indicates the values of the dependent variable for the i^th zip-code; a represents the model intercept, or constant; and, x_ik indicates the value of the k^th predictor for the i^th zip-code, with β_k representing the association of the k^th predictor with the dependent variable. The spatial lag term ρ represents the average spatial correlation of the dependent variable in the i^th zip-code with neighboring zip-codes, which is based on the spatial weights matrix W. In our study, W is the same for all zip-codes and is constructed using a row-standardized, first-order queen contiguity specification, where the weight equals “1/# of neighbors” for any zip-code that touches the i^th case and “0” otherwise. Thus, for Model 2, the spatial lag for the i^th zip-code is equal to the weighted average of quality of life for all zip-codes that immediately border the i^th case. For the spatial error model, the error term υ_i is decomposed into two parts. The first part estimates the spatial error term λ, which is based on the same contiguity weights matrix W, and the second part ε_i represents all the leftover unobserved variation in the dependent variable.

Results and Discussion

Table 2 displays the univariate and bivariate statistics, and Table 3 displays the results from the regression analyses. In Table 2, we observe that there is a significant, negative bivariate correlation between carbon footprint and quality of life (r = −0.049; p < 0.01). In other words, without controlling for any other variables, zip-codes with a higher carbon footprint per household tend to have a lower quality of life. To test whether this association holds in a multivariate analysis after incorporating additional control variables, we turn to the regression results in Table 3. Before we discuss the coefficients for the specific variables, as a way to compare fit between models, we report the Moran’s I (residuals), the R², and AIC for the three models. Based on these fit statistics, the spatial error model (Model 3) exhibits the best fit, with the highest R² and the lowest AIC and Moran’s I (residuals).² Specifically, we note that the OLS and Spatial Lag models exhibit high magnitudes of spatial autocorrelation, which suggests that the slope estimates from Models 1 and 2 are affected by spatially induced bias. We observe that the signs for three variables switch and the significance levels for two variables change. In Model 3, after minimizing any spatially induced bias, the slope estimates for all variables are highly significant at p < 0.001.

Table 2.

Univariate and Bivariate Statistics^a.

Variable	Mean	SD	1	2	3	4	5	6	7	8	9	10	11	12
1. Quality of Life	94.773	37.617	1.000
2. Carbon Footprint Per Household	43.772	6.241	−0.049	1.000
3. Population Density	1090.260	4549.104	0.157	−0.196	1.000
4. Median Household Income	51948.000	24059.140	0.307	0.498	0.099	1.000
5. Median Household Size	2.534	0.717	0.456	0.205	0.059	0.594	1.000
6. Unemployment	8.007	4.965	0.304	−0.287	0.102	−0.013	0.418	1.000
7. Percent Bachelor’s or Higher	19.937	13.759	0.308	0.284	0.224	0.767	0.327	−0.060	1.000
8. Percent 25 or Older	64.749	17.797	0.342	0.060	0.036	0.532	0.769	0.326	0.367	1.000
9. Percent White	77.991	27.747	0.023	0.271	−0.160	0.481	0.585	0.021	0.261	0.767	1.000
10. Area	131.765	263.161	0.045	−0.042	−0.107	−0.083	0.043	−0.021	−0.085	0.085	0.078	1.000
11. Latitude	38.827	4.750	−0.681	0.017	−0.015	0.082	−0.102	−0.150	0.071	0.015	0.196	0.026	1.000
12. Longitude	−90.462	13.669	−0.287	0.045	0.050	0.048	−0.044	0.017	0.021	0.014	0.062	−0.270	0.006	1.000

^aNote: The means and standard deviations are computed using unlogged values. To be consistent with the regression analyses, all bivariate correlations are estimated using log-transformed variables, with the exception of latitude and longitude. All bivariate correlations are significant at p<0.01, except for the following pairs of variables: carbon footprint and longitude; latitude and longitude; unemployment and median household income.

Table 3.

Results from the Regression of Quality of Life on Carbon Footprint (n=29,953).

	Model 1		Model 2		Model 3
	OLS		Spatial Lag		Spatial Error
	B	SE	b	SE	b	SE
Primary Variable
Carbon Footprint	−0.404***	0.010	−0.291***	0.009	−0.101***	0.008
Control Variables
Population Density	0.053***	0.001	0.043***	0.001	0.049***	0.001
Median Household Income	0.209***	0.006	0.123***	0.005	0.032***	0.004
Median Household Size	−0.085***	0.013	−0.115***	0.011	−0.160***	0.008
Unemployment	0.005*	0.002	−0.005**	0.002	−0.007***	0.001
Percent Bachelor's or Higher	0.031***	0.003	0.030***	0.002	0.035***	0.002
Percent 25 or Older	0.002	0.012	−0.019	0.010	−0.049***	0.006
Percent White	−0.039***	0.003	−0.017***	0.003	0.009***	0.002
Area	0.038***	0.001	0.035***	0.001	0.035***	0.000
Latitude	−0.057***	0.000	−0.036***	0.000	−0.058***	0.002
Longitude	−0.009***	0.000	−0.006***	0.000	−0.012***	0.001
Constant	4.416***	0.066	2.822***	0.060	5.134***	0.106
Rho			0.373***	0.004
Lambda					0.981***	0.002
Moran's I (Residuals)	0.713		0.537		−0.010
R2	0.781		0.841		0.948
AIC	−18492		−28023		−59848

Note: *p<0.05; **p<0.01; ***p<0.001.

With that in mind, we now discuss the slope estimates of the specific variables. Looking at Model 3, the spatial error model, results show positive associations between quality of life and the following variables: population density, median household income, percent bachelor’s, percent white, and land area. Quality of life has negative associations with the following variables: carbon footprint, median household size, percent unemployed, latitude and longitude. From Models 1 to 3, the coefficient signs for unemployment, percent 25 or older, and percent white switch and significance levels for unemployment and percent 25 or older change. To reiterate, given that the spatial error model exhibits the best fit, we focus on the results from Model 3.

Looking at Model 3, controlling for spatial dependency in the error term, there is still a negative association between carbon footprint and quality of life (b = −0.101; p<0.001), which is consistent with the bivariate correlation displayed in Table 2. Thus, controlling for a host of demographic, economic, and geographic variables, as well as spatial autocorrelation in the error term, zip-codes with a higher environmental impact, at least in terms of the carbon footprint, tend to have a lower quality of life. In other words, localities with a higher quality of life tend to consume less fossil fuel, both directly and indirectly. The finding of an inverse association between an objective measure of life quality and the carbon footprint is compatible with findings from previous studies on the environmental implications of subjective well-being (e.g., Ambrey & Daniels., 2017).

Before we elaborate on the theoretical and practical implications of this finding, we briefly discuss the findings for the control variables. While we analyze zip code-level data, we note that these results also generally reaffirm previous quantitative research on individual well-being (e.g., Ambrey & Daniels., 2017). For instance, the slope estimates for the socio-economic status variables (household income, unemployment and educational attainment) suggest that a higher quality of life is positively associated with affluent and highly educated zip-codes and zip-codes with lower unemployment. We also find that several demographic factors are significantly associated with quality of life. While quality of life is positively associated with the population density of a zip-code, it is negatively associated with the size of the typical household. While Ambrey and Daniels. (2017), in an Australian study of individual well-being, did not find a significant estimate for their race variable, we note that for the United States, quality of life is positively associated with the proportion of the population who is white. Similarly, while Ambrey and Daniels observed that age structure has a complex relationship with well-being, the slope estimate for our age variable indicates that zip-codes with more youthful populations tend to have a higher quality of life.

Conclusion

In the above analysis, we examined the association between carbon footprint and human well-being, specifically in terms of quality of life. While our focus was on quality of life, the results of the analysis are consistent with what has been found in previous scholarship on individual well-being (e.g., Ambrey & Daniels., 2017; Cloutier et al., 2014), highlighting the implications of well-being for sustainability. Specifically, the results of our analysis show a consistent, negative association between the dependent variable quality of life and carbon footprint. At the level of the zip-code, this finding suggests that a bigger environmental impact, at least in terms of the carbon emissions embedded in the goods and services consumed in a locality, does not necessarily compromise improvements in quality of life. In fact, the negative slope estimate indicates that a reduction in society’s environmental impact has the potential to result in a higher quality of life. Following previous research on individual well-being (e.g., Ambrey & Daniels., 2017; Cloutier et al., 2014), this finding continues to challenge the claim, as mentioned above, that in order to live more sustainably humans must make sacrifices. It also challenges some of the research on objective measures of well-being that focus on economic well-being; namely, this research suggests that a narrow focus on economic well-being may not capture all elements of quality of life and therefore our findings are consistent with calls for more robust considerations of quality of life measures (Constanza et al., 2007). The conventional wisdom holds that human society must refrain from certain activities and experiences in the pursuit of sustainability; based on the results of our study, we propose that this is not the case for quality of life if a broader understanding of quality of life is employed. To be clear, this does not mean that changes in lifestyle will not be required, but rather it suggests that high levels of life quality can be maintained and potentially improved without increasing the carbon footprint. Although this research is not focused on the policy implications of sustainability and quality of life, the results certainly suggest that policymakers may reconsider the conventional wisdom that a more sustainable society requires sacrifice. Furthermore, it suggests that policymakers and researchers may need to develop and deploy more robust understandings of quality of life that go beyond economic well-being and life expectancy in developing strategies to understand and reduce carbon emissions. Although such policy implications are beyond the scope of this paper, we do encourage additional research to better understand the opportunities and trade-offs for how increased levels of sustainability may be achieved while retaining high levels of quality of life.

While the results of our study clearly point to this implication, here we acknowledge a few limitations of our analysis. First, while we followed the example of previous literature in using a multifaceted measure for human well-being (e.g., Ambrey & Daniels., 2017; Biedenweg 2017; Moran et al., 2008), we acknowledge that the quality of life metric we used for our dependent variable, which comes from EASI Demographics (see SimplyAnalytics 2021), is a composite index. On the one hand, the factors included in EASI Demographics’ quality of life metric are differentially weighted. For instance, the EASI’s weather index (i.e., local climate), although consistent with the literature (e.g., Albouy et al., 2016), is weighted more heavily than the other factors in the index. On the other hand, some scholars may argue that other factors not included in EASI Demographics’ variable (e.g., natural park space or social capital) are relevant to a measure of quality of life (e.g., Biedenweg 2017; Hamdan et al., 2014; Kelly et al., 2017). While our study shows that EASI Demographics’ composite index is significantly related to the carbon footprint, future scholars can begin to evaluate separately the different factors of this index and assess whether to include other items in a quality of life variable.

Second, in terms of our primary independent variable, as described in the data and analysis section, the carbon footprint from Ummel (2014; 2016) is not derived from an exact accounting of all the activities taking place within every single household in all zip-codes; rather, it is derived from survey expenditure data, which makes assumptions about the carbon intensity involved in producing the goods consumed by the households in the locality. While this method allows for wide geographic coverage, it is similar to cross-national data on ecological footprints, which yields not exact accounting but an estimate of the natural resources embedded in the goods and services consumed. Other data collection efforts to build carbon emissions inventories, for example, Vulcan Project (Gurney et al., 2009), have incorporated information from the direct monitoring of emissions from point of source. When combined with indirect sources of CO2, social scientists have used these data to approximate the carbon footprint (e.g., Pattison et al., 2014). All the same, these data are only an approximation of the footprint and do not equal the combined immediate and remote environmental impacts associated with the total amount of resources consumed within a locality.

Lastly, as already mentioned above, our study has only a single wave of information, providing a cross-sectional snapshot of the association between quality of life and the carbon footprint. Thus, as is the case for any analysis using only a single wave of data, we do not employ a theoretical framework in which we propose that changes in the carbon footprint lead to (or cause) changes in quality of life. Rather, our focus is on the association between the two variables. For future research, longitudinal data analysis techniques would be useful in testing the mechanisms through which changes in an independent variable determine changes in a dependent variable (see Allison, 2009). However, the currently available data for the carbon footprint are not tracked over time and thus cannot be incorporated into a longitudinal analysis, which would not only help to minimize omitted variable bias but also allow researchers to estimate change over time in the variables of interest. In our regression model, we control for time-invariant factors such as land area, longitude, and latitude; however, a longitudinal model, in which the researcher can incorporate unit fixed effects, would do this automatically (Allison, 2009). Moreover, when longitudinal data become available, scholars can assess whether the association between the carbon footprint and quality of life holds over time or changes in magnitude and/or direction. Similarly, while we controlled for spatial autocorrelation, which did not influence the results for the quality of life variable, future scholarship can use analytic techniques to test whether the slope estimate for quality of life exhibits other spatial dynamics (e.g., spatial spillover or spatial heterogeneity).

To conclude, this study has observed that the carbon footprint and quality of life are inversely related at the zip-code level across the United States in 2012. This result presents a challenge to the common argument that a more sustainable way of life requires society to make sacrifices; our findings suggest we may need to reconsider this conventional wisdom, at least when it comes to quality of life measured in a broad manner. Recognizing that the two are negatively associated (or even if we had found that there was no significant association) entails important policy considerations, particularly given the anti-science sentiment being expressed in current political discussion. To be clear, we do not believe that these results provide a justification for “business as usual,” but quite the opposite. It suggests that even though changes in lifestyle may be necessary to address society’s most daunting environmental problems, such as climate change, our approach to resolve these problems may point to ways to reduce carbon footprints while maintaining high levels of quality of life. Given that social psychological scholarship on environmental change has emphasized the importance of cognitive barriers (e.g., Norgaard 2009), policy framed around increasing the quality of life and its inverse association with the carbon footprint would likely be well-received (Helliwell 2014). If the policy discourse can be focused on the potential increase in quality of life associated with carbon footprint reduction, this focus may facilitate action in terms of reducing carbon emissions. Nevertheless, given the above limitations of our analysis, we encourage future scholarship to explore further the ecological implications of quality of life.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Matthew Thomas Clement

Notes

References

Albouy

Graf

Kellogg

Wolff

(2016). Climate Amenities, Climate Change, and American Quality of Life. Journal of the Association of Environmental and Resource Economists, 3(1), 205-246. https://doi.org/10.1086/684573.

Allison

P. D.

(2009). Fixed Effects Regression Models: SAGE.

Ambrey

C. L.

Daniels

(2017). Happiness and footprints: assessing the relationship between individual well-being and carbon footprints. Environment, Development and Sustainability, 19, 895–920. https://doi.org/10.1007/s10668-016-9771-1

Biedenweg

(2017). A Comparative Study of Human Well-Being Indicators Across Three Puget Sound Regions. Society & Natural Resources, 30(3), 362–376. https://doi.org/10.1080/08941920.2016.1209606

Bowerman

(2014). How much is too much? A public opinion research perspective. Sustainability: Science, Practice and Policy, 10(1), 14–28. https://doi.org/10.1080/15487733.2014.11908122

Bradbury

Peterson

M. N.

Liu

(2014). Long-term dynamics of household size and their environmental implications. Population and Environment, 36, 73–84. https://doi.org/10.1007/s11111-014-0203-6

Camfield

Skevington

S. M.

(2008). On Subjective Well-being and Quality of Life. Journal of Health Psychology, 13(6), 764–775. https://doi.org/10.1177/1359105308093860

Choudhury

Foroughi

Larson

(2021). Work‐from‐anywhere: The productivity effects of geographic flexibility. Strategic Management Journal, 42(4), 655–683. https://doi.org/10.1002/smj.3251

Clement

M. T.

Pattison

Habans

Pattison

(2017). Scaling down the “Netherlands Fallacy”: A local-level quantitative study of the effect of affluence on the carbon footprint across the United States. Environmental Science & Policy, 78, 1–8. https://doi.org/10.1016/j.envsci.2017.09.001

10.

Cloutier

Larson

Jambeck

(2014). Are sustainable cities “happy” cities? Associations between sustainable development and human well-being in urban areas of the United States. Environment, Development and Sustainability, 16(3), 633–647. https://doi.org/10.1007/s10668-013-9499-0

11.

Costanza

Fisher

Ali

Beer

Bond

Boumans

Danigelis

N. L.

Dickinson

Elliott

Farley

Gayer

D. E.

Glenn

L. M.

Hudspeth

Mahoney

McCahill

McIntosh

Reed

Rizvi

S. A. T.

Rizzo

D. M.

Snapp

(2007). Quality of Life: An Approach Integrating Opportunities, Human Needs, and Subjective Well-Being. Ecological Economics, 61(2-3), 267–276. https://doi.org/10.1016/j.ecolecon.2006.02.023

12.

Crowley

(2021). Social Vulnerability Factors and Reported Post-Disaster Needs in the Aftermath of Hurricane Florence. International Journal of Disaster Risk Science, 12, 13–23. https://doi.org/10.1007/s13753-020-00315-5

13.

Dietz

Rosa

E. A.

York

(2012). Environmentally Efficient Well-Being: Is There a Kuznets Curve? Applied Geography, 32(1), 21–28. https://doi.org/10.1016/j.apgeog.2010.10.011

14.

Emerson

M. O.

Chai

K. J.

Yancey

(2001). Does Race Matter in Residential Segregation? Exploring the Preferences of White Americans. American Sociological Review, 66(6), 922–935. https://doi.org/10.2307/3088879

15.

Ericson

Kjønstad

B. G.

Barstad

(2014). Mindfulness and Sustainability. Ecological Economics, 104, 73–79. https://doi.org/10.1016/j.ecolecon.2014.04.007

16.

Fiala

(2008). Measuring sustainability: Why the ecological footprint is bad economics and bad environmental science. Ecological Economics, 67(4), 519–525. https://doi.org/10.1016/j.ecolecon.2008.07.023

17.

Figge

Kay

Offermans

(2016). The effects of globalization on Ecological Footprints: an empirical analysis. Environment, Development, and Sustainability, 19, 863–876. https://doi.org/10.1007/s10668-016-9769-8

18.

Fritz

Koch

(2014). Potentials for prosperity without growth: Ecological sustainability, social inclusion and the quality of life in 38 countries. Ecological Economics, 108, 191–199. https://doi.org/10.1016/j.ecolecon.2014.10.021

19.

Gurney

K. R.

Mendoza

D. L.

Zhou

Fischer

M. L.Chris C.

Miller

C. C.

Geethakumar

de la Rue du Can

(2009). High Resolution Fossil Fuel Combustion CO2 Emission Fluxes for the United StatesThe Vulcan Project: High Resolution Fossil fuel Combustion CO2 Emissions Fluxes for the United States. Environmental Science & Technology, 43(14), 5535–5541. https://doi.org/10.1021/es900806c

20.

Hall

M. P.

Lewis

N. A.

Jr. Ellsworth

P. C.

(2018). Believing in climate change, but not behaving sustainably: Evidence from a one-year longitudinal study. Journal of Environmental Psychology, 56, 55–62. https://doi.org/10.1016/j.jenvp.2018.03.001

21.

Hamdan

Yusof

Marzukhi

M. A.

(2014). Social Capital and Quality of Life in Urban Neighborhoods High Density Housing. Procedia - Social and Behavioral Sciences, 153, 169–179. https://doi.org/10.1016/j.sbspro.2014.10.051

22.

Helliwell

J. F.

(2014). Social norms, happiness, and the environment: closing the circle. Sustainability: Science, Practice and Policy, 10(1), 78–84. https://doi.org/10.1080/15487733.2014.11908126

23.

Hogan

M. J.

Leyden

K. M.

Conway

Goldberg

Walsh

McKenna-Plumley

P. E.

McKenna-Plumley (2016). Happiness and health across the lifespan in five major cities: The impact of place and government performance. Social Science & Medicine, 162, 168–176. https://doi.org/10.1016/j.socscimed.2016.06.030

24.

Jones

Kammen

D. M.

(2014). Spatial distribution of U.S. household carbon footprints reveals suburbanization undermines greenhouse gas benefits of urban population density. Environmental Science & Technology, 48(2), 895–902. https://doi.org/10.1021/es4034364

25.

Jorgenson

A. K.

(2014). Economic development and the carbon intensity of human well-being. Nature Climate Change, 4(3), 186–189. https://doi.org/10.1038/nclimate2110

26.

Jorgenson

A. K.

Clark

(2011). Societies consuming nature: A panel study of the ecological footprints of nations, 1960-2003. Social Science Research, 40(1), 226–244. https://doi.org/10.1016/j.ssresearch.2010.09.004

27.

Kelly

Ruther

Ehresman

Nickerson

(2017). Placemaking as an economic development strategy for small and midsized cities. Urban Affairs Review, 53(3), 435–462. https://doi.org/10.1177/1078087416657895

28.

Leiserowitz

A. A.

Kates

R. W.

Parris

T. M.

(2006). Sustainability values, attitudes, and behaviors: A review of multinational and global trends. Annual Review of Environment and Resources, 31, 413–444. https://doi.org/10.1146/annurev.energy.31.102505.133552

29.

Lenzen

Cummins

(2013). Happiness versus the environment-A case study of Australian lifestyles. Challenges, 4(1), 56–74. https://doi.org/10.3390/challe4010056

30.

Liu

Daily

G. C.

Ehrlich

P. R.

Luck

G. W.

(2003). Effects of household dynamics on resource consumption and biodiversity. Nature, 421(6922), 530–533. https://doi.org/10.1038/nature01359

31.

Maibach

E. W.

Nisbet

Baldwin

Akerlof

Diao

(2010). Reframing climate change as a public health issue: An exploratory study of public reactions. BMC Public Health, 10, 299. https://doi.org/10.1186/1471-2458-10-299

32.

Markowitz

E. M.

Goldberg

L. R.

Ashton

M. C.

Lee

(2012). Profiling the “pro-environmental individual”: A personality perspective. Journal of Personality, 80(1), 81–111. https://doi.org/10.1111/j.1467-6494.2011.00721.x

33.

Meijer

K. S.

Van Beek

(2011). A framework for the quantification of the importance of environmental flows for human well-being. Society & Natural Resources, 24(12), 1252–1269. https://doi.org/10.1080/08941920.2010.545866

34.

Moran

D. D.

WackernagelGoldfinger

Kitzes

J. A.

Goldfinger

S. H.

Boutaud

(2008). Measuring sustainable development–Nation by nation. Ecological Economics, 64(3), 470–474. https://doi.org/10.1016/j.ecolecon.2007.08.017

35.

Moro

Brereton

Ferreira

Clinch

J. P.

(2008). Ranking quality of life using subjective well-being data. Ecological Economics, 65(3), 448–460. https://doi.org/10.1016/j.ecolecon.2008.01.003

36.

Norgaard

K. M.

(2009). Cognitive and behavioral challenges in responding to climate change. Policy research working paper. World Bank. https://openknowledge.worldbank.org/handle/10986/9066

37.

Norgaard Kari

(2011). Living in denial: Climate change, emotions, and everyday life. MIT Press.

38.

Oskamp

(2000). Psychology of promoting environmentalism: Psychological contributions to achieving an ecologically sustainable future for humanity. Journal of Social Issues, 56(3), 373-390. https://doi.org/10.1111/0022-4537.00173

39.

Pattison

Habans

Clement

M. T.

(2014). Ecological modernization or aristocratic conservation? Exploring the impact of affluence on carbon emissions at the local Level. Society & Natural Resources, 27(8), 850–866. https://doi.org/10.1080/08941920.2014.911996

40.

Rees

Wackernagel

(1996). Urban ecological footprints: Why cities cannot be sustainable-And why they are a key to sustainability. Environmental Impact Assessment Review, 16(4–6), 223–248. https://doi.org/10.1016/S0195-9255(96)00022-4

41.

Riley

J. C

(2008). Low income, social growth, and good health. University of California Press.

42.

Seyfang

(2005). Shopping for sustainability: Can sustainable consumption promote ecological citizenship? Environmental Politics, 14(2), 290–306. https://doi.org/10.1080/09644010500055209

43.

SimplyAnalytics (2015). “Easy Analytic Software, Inc. (EASI) – Methodology Demographics, Related Data, and Life Stage Clusters. ” http://www.simplyanalytics.com/documents/EASI_Methods_2015_Updates.pdf (Accessed August 17, 2021).

44.

SimplyAnalytics (2021). EASI/MRI Consumer Expenditure Data. SimplyAnalytics database. (Accessed August 17, 2021).

45.

Tsai

Mares

A. S.

Rosenheck

R. A.

(2011). A geographic analysis of chronically homeless adults before and after enrollment in a multi-site supported housing initiative: Community characteristics and migration. American Journal of Community Psychology, 48, 341–351. https://doi.org/10.1007/s10464-010-9363-4

46.

Ummel

(2014). Who pollutes? A household-level database of America’s greenhouse gas footprint. CGD working paper 381. Washington, DC: Center for Global Development.

47.

Ummel

(2016). “Impact of CCL’s proposed carbon fee and dividend policy: A high-resolution analysis of the financial effect on U.S. Households”. Paper prepared for Citizens’ climate Lobby (CCL). April 2016. v1.4.

48.

Van Den Bergh

Grazi

(2010). On the policy relevance of ecological footprints. Environmental Science & Technology, 44(13), 4843–4844. https://doi.org/10.1021/es1003582

49.

Veenhoven

(2001). Quality-of-Life and Happiness: Not Quite the Same (pp. 67–95). ” Centro Scientifico Editore. http://hdl.handle.net/1765/8753

50.

Verhofstadt

Van Ootegem

Defloor

Bleys

(2016). Linking individuals’ ecological footprint to their subjective well-being. Ecological Economics, 127(C), 80–89. https://doi.org/10.1016/j.ecolecon.2016.03.021

51.

Wackernagel

Yount

(2000). Footprints for sustainability: The next steps. Environment, Development and Sustainability, 2, 21–42. https://doi.org/10.1023/A:1010050700699

52.

Welsch

Kühling

(2011). Are pro-environmental consumption choices utility-maximizing? Evidence from subjective well-being data. Ecological Economics, 72, 75–87. https://doi.org/10.1016/j.ecolecon.2011.04.015

53.

York

Rosa

E. A.

Dietz

(2003). Footprints on the Earth: The environmental consequences of modernity. American Sociological Review, 68(2), 279–300. https://doi.org/10.2307/1519769