Abigail P. Dumalus December 2018

Macroeconomics of Life Satisfaction Inequality (Working Paper, Unpublished)

Abstract

Not much is known about the variations that macroeconomic movements bring to the significance of how people’s subjective life satisfaction (LS) assessments are aggregately distributed. Now more than ever, leaders, policymakers, academia, the media, and the general public may find it worthwhile to know more about LS inequality. The question “How do macroeconomic shifts influence LS inequality?” is closely examined. Whereas most subjective well-being (SWB) studies in economic literature have been dealing with causes and consequences of overall happiness across societies, this paper investigates the macroeconomics of LS inequality. Using internationally comparable upward-looking LS inequality and measures of macroeconomic shifts within a nested cross-country dataset, robust and conclusive empirical evidence is provided to highlight the fundamental role of macroeconomic movements in shaping perceived differences in LS. Findings in this research adequately bolster the substance of how such changes in the macroeconomic landscape have a bearing on shaping SWB or happiness on the aggregate level. Results also carefully explore how disparities in well-being are adversely or favorably motivated by extreme LS responses due to macroeconomic fluctuations.

This study makes a twofold contribution to the literature on the effects of macroeconomic movements on well-being. First, using the Cowell-Flachaire status-inequality method, this paper constructs the first multi-country estimate of LS inequality, which could be a more suitable measure of overall welfare. The distribution of LS illustrates a more inclusive and substantial investigation of the impacts of any inequalities that may spring from macroeconomic shifts in society. Second: simultaneously considering how these macroeconomic inequalities across different groupings, this study enriches the ‘essence’ of the misery index by providing some insights that ought to refocus the criteria for optimising overall welfare according to a particular country’s LS inequality level, income inequality level, development status, or geographic area. These groupings not only inform the current set of economic, social, political, and cultural factors that influence public policy, but also determine relevant policies that work towards attaining the ‘ideal’ LS inequality for a specific context.

JEL classification: D3, I3 Keywords: life satisfaction, happiness, subjective well-being, inequality, welfare

I. Introduction

Since Di Tella et al. (2003) published their groundbreaking “Macroeconomics of Happiness” paper, it has been broadly established that movements in macroeconomic policy variables influence happiness in nations (Deaton, 2012; Mohseni-Cheraghlou, 2013; Blanchflower et al., 2014; Helliwell & Huang, 2014; Cahill et al., 2015; Frijters et al., 2015; Hariri et al., 2015; Ratcliffe & Taylor, 2015; Mertens & Beblo, 2016). Their empirical contribution has proposed a new approach to examine patterns in how people’s subjective well-being (SWB) evaluations vary as macroeconomic conditions affect their lives in society. For instance, others have recently centered on the consequences of economic and financial crises in the context of individual well-being (e.g. Hoynes et al., 2012; McInerney et al., 2013; Chadi, 2015). Although existing findings are still mixed, interest in the relationship between SWB and income inequality has been significantly growing (e.g. Hagerty, 2000; Clark, 2003; Blanchflower & Oswald, 2004; Alesina et al., 2004; Graham & Felton, 2006; Smyth & Qian, 2008). Macroeconomic shifts are considered to produce substantial income and, arguably, non-income shocks at the individual level, which, influence people’s decision-making and behaviour, and, eventually, subjective well-being (SWB). Remarkable macroeconomic fluctuations could yield both economic and social insecurity, which modify individual assessment of opportunities and intensifies the sense of uncertainty to manage appropriately over time.

Researchers have also tackled the negative outcomes of aggregate inflation, and unemployment on individual happiness. Inflation’s adverse effects on SWB usually involve: worse living standards for people, lower wages due to rising prices of goods and services, political instability, and more risky investment climate, among others (Wolfers 2003; Gandelman and Hernández-Murillo 2009; Ruprah and Luengas 2011; Blanchflower et al., 2014). Previous literature has offered empirical evidence on inflation having consistently negative effect on SWB across individuals (e.g. Graham & Pettinato, 2001; Di Tella et al., 2001, 2003; Wolfers, 2003). Investigating data from the United States and Europe, Alesina et al. (2004) have seen that inflation has a more negative impression on SWB for those with right-leaning political ideology than those who with left-leaning attitudes. On the other hand, unemployment lowers SWB due to either reduced economic security, or general negative externalities (Wolfers 2003; Paul & Moser 2009; Bell & Blanchflower, 2010; Luechinger et al. 2010; Green 2011). Whilst becoming unemployed frequently suggests giving up one’s primary income source, it also has non-pecuniary, psychological stress, which might involve potential loss of self-esteem, social stigma, reduction in social interactions at the workplace, impairment in structure of daytime activities, and so forth (Clark & Oswald, 1994; Helliwell, 2003; Layard et al., 2012). For society at large, unemployment has undesirable external consequences, such as marriage and social problems, suicide risks, alcoholism, and higher prevalence of criminal behaviour (Winkelmann & Winkelmann, 1998; Kassenboehmer et al., 2009). There have been studies that weigh between the costs incurred from inflation and unemployment. Frey and Stutzer (2000) have observed that unemployment has more unfavorable costs on people’s SWB than inflation. Di Tella and MacCulloch (2006) have offered evidence that relate political beliefs with preferences for inflation and unemployment: in that the latter and its outcomes are more distressing for

2 left-wing individuals than right-wing cohorts. Although most scholars acknowledge that unemployment has more deleterious outcomes for individual well-being, the tangible relative unemployment-inflation trade-off remains ambiguous (Frey & Stutzer, 2002; Blanchflower et al., 2014).

Before studies on SWB or individual happiness came into existence, economic literature has been exploring the idea of utility for centuries. According to Bentham (1789/2008), utility is that feature that goods have in providing pleasure or preventing displeasure. Hence, maximising utility has been perceived as either the enrichment of collective pleasure or the reduction of collective displeasure. Pigou (1920) has supported utilitarianism: given an overall (i.e. social) income, society’s welfare could reach its optimum level via reallocation of income from richer to poorer individuals until perfect income equality would be reached. Obviously, this view has seemed paradoxical as social welfare maximisation could be solely attained with the culmination of perfect income equality. Other economists have argued that in a society with varying individual utilities, income equality does not necessarily indicate equality in utilities; thus, the optimal degree of inequality may actually be different from zero (Dalton, 1920; Tinbergen, 1970). This debate not only accentuated how diverging hypotheses about individual utilities leads to quite different solutions for maximising overall welfare, but also initiated the idea of relativity as utility functions represented specific individual tastes for goods and services in the economy.

Much earlier, both Adam Smith (1776/1981) and Karl Marx (1847) have mentioned the idea of interdependent utilities, in which each person’s utility depends on the utility of others. Related to this, Veblen (1898) has observed that individual consumption is also useful for conveying position and status in society. Almost a century later, Frank (1985) has developed Veblen’s observation in a theory about positional goods. From another perspective, people not only differ in their own assessment of goods, but they also differ in the way their utilities depend on the evaluation done by others on these goods. The social welfare optimisation problem has now gained another aspect because utility is simultaneously dependent on absolute status and relative status. With relative income hypothesis, Duesenberry (1949) has noticed that households care about communal consumption standards, and that this interest causes savings rates to go up as a household’s position in the income distribution rises. Moreover, he has also learned that individuals are likely to overestimate upward comparisons whilst downward comparisons are overlooked. Bearing this in mind, addressing the social welfare optimisation puzzle has become even more interesting with asymmetric social comparisons.

In the same year, Stouffer et al. (1949) have released their work on World War II American soldiers’ reported well-being. Their findings have noted that these soldiers’ happiness primarily hinged on the specific context and reference group, which were fellow soldiers with whom they served in the army. Runciman (1966) has further shed light on this concept as relative deprivation in his social justice theory. Relative deprivation has been explained as the situation in which: an individual is deprived of status or any good, then she perceives other people having these resources, and she desires to own these resources. With Runciman’s work, the role of the reference group has been pushed into the limelight. This reference group – i.e. other persons with whom people juxtapose themselves – differs from one

3 individual to another and could be categorised based on different characteristics. This becomes another constraint in the social welfare maximisation problem. In empirical economic research, Yitzhaki (1979) has formalized the notion of relative deprivation by proposing to measure any individual relative deprivation as the aggregation of incomes of people richer than the person being studied. Across a population, adding up all these income distances measures the overall relative deprivation measure for an entire society. At the societal level, this is virtually the same as the absolute Gini index (i.e. the Gini multiplied by the mean). The work done by Yitzhaki has merged the discussion on relative deprivation and relative income with the discussion on summation of individual utilities. This Yitzhaki indicator is effectively a relative income measure whilst being used to estimate the degree of equality of a society. The higher the inequality in society, the more relatively deprived this society would be.

Before Yitzhaki’s research was published, economist Richard Easterlin (1974) has found that although happiness and income are directly related both within and among economies, long-run happiness does not go up as incomes grow. The search for an answer to the Easterlin paradox has also brought up the idea that, relative income becomes more consequential than absolute income in understanding happiness. Average happiness and absolute income are not necessarily associated whereas poorer individuals tend to be less happy than individuals. After overcoming initial resistance to the view that self-evaluated happiness is related to utility, economists have gradually been leaving their own mark. Early investigations have included works, among others, from Kapteyn et al. (1978), van Herwaarden and Kapteyn (1979), Layard (1980), and Frank (1985). Khaneman and Tversky (1979) have seen that when people make choices, they care more about changes from a reference point than the status quo. Much recently, Frey and Stutzer (2002) have claimed that most researchers assume that higher SWB results from greater income growth, as individuals are empowered to have better opportunities to enlarge their consumption possibilities when they earn more. Hagerty and Veenhoven (2003) have shown that national income is directly associated with national happiness level, using a broad time-series data set for 21 developed and developing countries from 1958 to 1996. As the Easterlin paradox underscored this peculiar relationship between sustained economic growth and average happiness levels, this result has become one of the most enduring issues in SWB research. Becchetti and Pelloni (2013) have confirmed this paradox in many advanced economies; however, their findings contradict the same empirical curiosity, as the difference in average happiness levels in both low- and high-income countries is significant and positive. Stevenson and Wolfers (2008a), Sacks et al. (2011) and Ma and Zhang (2014) have challenged the famous Easterlin paradox within the last few years.

Happiness research has become one of the most salient matters in economics in recent decades, as economists continue to pay attention to both theoretical and empirical aspects of SWB, i.e. people’s own accounts of their well-being (Dolan et al., 2008; MacKerron, 2012; Weimann et al., 2015; Clark, 2018). Whilst economics had generally maintained greater emphasis on objective dimensions of well-being, contemporary literature underscores the relevance of studying individual happiness with subjective measures (Stiglitz et al., 2010; Deaton, 2012). Given the observed suitability of subjective responses in evaluating well-being, and because of the availability of large-scale surveys with happiness and life

4 satisfaction questions, many researchers have proposed using SWB measures to supplement objective measurements of progress and policy (Layard, 2005; Senik, 2009; Fitoussi & Stiglitz, 2013; Helliwell et al., 2016). Broad acceptance for SWB measures (e.g. life satisfaction self-assessments) is currently well- established in the literature (see Easterlin, 1974; Ferrer-i-Carbonell & Frijters, 2004; Kahneman & Krueger, 2006), and hinges on fulfillment of public policy relevance requirements: validity, reliability, and interpersonal comparability (Krueger & Schkade, 2008; Helliwell & Barrington-Leigh, 2010; Exton et al., 2015). This burgeoning surge of SWB research has been presenting a number of stimulating issues pertinent to current debates, which include economic development, poverty, and inequality (Decancq et al., 2015; Nikolova, 2016). As many initiatives have been done to construct and incorporate measures based on individual-centered responses, policymakers have been gradually considering subjective life evaluations as a guideline to inform policy formulation and implementation (for an overview, see Barrington-Leigh & Escande, 2017).

Whilst it is crucial for policymakers to better appreciate individual happiness across life domains in order to improve overall societal welfare, most studies on SWB (hereafter life satisfaction, well-being, welfare, and happiness are interchangeably used) have mainly focused on happiness levels, (hence averages), along with its determinants. Relatively fewer SWB studies, however, examine how well- being is distributed across individuals or within one’s life cycle (for an overview, see Clark et al., 2014). In stark contrast, causes and effects of how economic resources are distributed continue to enjoy sustained attention (e.g. Wilkinson & Pickett, 2010; Stiglitz, 2012; Piketty, 2014), as well-being has been usually measured in terms of economic variables, including income, consumption and/or wealth (Atkinson & Bourguignon, 2000). Economists remain strongly interested in income (e.g. Lorenz, 1905; Gini, 1921; Atkinson, 1970; Gottschalk & Smeeding, 1997) for several good reasons: it is relatively easier to estimate; it is known to be one of the most significant resources in modern market economies; and, it can be affected by economic and social policies. On the other hand, it has its own problems. First, income merely provides insights into life outcomes rather than life chances (UNDP, 1990); having equal incomes does not necessarily lead to equal outcomes because individuals have varying abilities to utilise resources (i.e. similar income levels result in different capabilities for different people) (Sen, 2000). Second, it is hardly an inclusive metric for well-being (Ringen, 2006; Gandelman & Porzecanski, 2013). Nonetheless, researchers have long understood the usefulness in analysing both central tendency and dispersion facets of income distributions.

Deliberately focusing on monetary variables has generated constructive information but it also has encountered significant limitations. It stands to reason that people not only worry about consequences arising from economic fluctuations, such as income inequality, but are also concerned about other inequalities, such as social opportunities, education, health, human development, housing, legal rights, longevity, social interactions (Hicks, 1997; Thomas et al., 2001; Grimm et al., 2010; Neal & Rick, 2014; Wang & Parker, 2014; Case & Deaton, 2015). One approach to work around these limitations is to find another measure of well-being, and explore its distribution to estimate the degree of differences in individual happiness of people in society. Considering SWB distribution among people seems to be

5 reasonable enough to yield practical policy implications. The case being made for measuring the distribution of individual happiness does not recommend that conventional economic indicators should be wholly ignored, but rather that happiness distribution could be viewed at least as a similarly convincing and pertinent barometer to explore economic, and social inequalities. Assuming that happiness per se cannot be reallocated across people, recognising the causes of disparity is particularly helpful in the context of SWB inequality. Consciously comprehending SWB inequality drivers would enable policymakers identify domains in which intervention is essential; hence, suitable policies and programmes are established to lessen the dispersion of SWB, thereby achieving social cohesion and advancing overall well-being (Ovaska & Takashima, 2010; Becchetti et al., 2014). Also, individual attitudes toward social comparisons, social mobility, fairness perceptions, and egalitarian tastes may be clarified, to some extent, by people’s assessment of the happiness distribution, and not simply by looking at how economic resources are allocated. Veenhoven (2005a) posits that non-monetary inequalities cannot be aptly estimated using inequality measures using monetary inputs, such as income.

Whereas utilitarian policies are simply in service of the pursuit of enhancing total happiness, egalitarian policies eventually undertakes to coordinate everyone’s welfare, not simply increasing their income levels, where the former is basically represented by the latter. With fairness considerations in mind, risk- averse individuals, who consider happiness inequality as a difficulty, would prefer being in a society where happiness disparities are more likely reduced (for an overview, see Ferrer-i-Carbonell & Ramos, 2014). Policymakers would argue that redistribution of incomes across people appears to be more practical, as it would be implausible to transfer SWB, for instance, life satisfaction, from one person to another (van Praag, 2011). Aside from being a relevant metric for determining general welfare of citizens (Veenhoven, 2005a), SWB inequality offers worthwhile insight into people’s living standards and features of their respective neighbourhoods. Some scholars suggest that the possibility of social conflict is gauged by the happiness gap between the less fortunate and the rich (Tullock, 1971; Gurr, 1994; Brown, 1996; Guimaraes & Sheedy, 2012).

Current literature on comparative SWB inequality have hit its stride in the past few decades. Some studies have observed a decline in happiness inequality in advanced economies in previous decades. According to Veenhoven’s (2005b) study of European countries during the period 1973-2001, falling inequality in happiness is found despite rising income inequality. Both Stevenson and Wolfers (2008b) and Dutta and Foster (2012) have seen decreasing happiness differences among Americans in the period 1972-2006, as non-pecuniary factors (e.g. institutional and technological changes) are shaping the distribution of happiness, resulting in increased opportunities, autonomy, and freedom of choice among individuals. In another study covering the United States, Brooks (2008) has presented empirical evidence that income inequality is much greater and gradually increasing over time, whereas happiness inequality is at a fairly lower status and stable in the long run. Ovaska and Takashima (2010) have determined in their cross-country analysis that inequalities in individual health status and income are positively correlated with happiness inequality whilst poor institutional quality is aggravating differences in SWB. In an earlier study by Ott (2005), selected institutional conditions (e.g. government

6 consumption, social security, subsidies, transfers) are not only increasing happiness levels, but also curbing happiness inequality. In their decomposition of happiness inequality trends in Germany from 1991 to 2007, Becchetti et al. (2014) have noticed that income inequality is not the root cause of discrepancies in happiness among Germans, rather it is the variance with education categories. Likewise, Piketty (2014) has notable empirical findings that exacerbating income inequality is hardly influencing the happiness distribution, and does not fundamentally extend beyond other life domains. Findings by Clark et al., (2014) have uncovered that varying income inequality does not predict happiness inequality in many countries, among them Australia, Germany, United Kingdom, and United States. Yang et al. (2015) have observed that growing income inequality and regional heterogeneity broaden happiness differences, whilst policies that enhance education and economic status help minimise SWB disparities, thereby promoting social harmony in China. Clark et al., (2016) have shown that happiness distributions are improving in countries with consistent income growth but declining during periods of plummeting real gross domestic product (GDP) per capita, indicating that shrinking happiness inequality is related to greater availability of public goods, such as social protection, public infrastructure, health, and education. Goff et al. (2016) have argued that the dispersion of happiness provides a stronger comprehensive estimation of inequality than income inequality. Niimi (2018) has studied that Japan’s happiness inequality is adversely influenced by household incomes, finding that people’s perception of their relative income status has significant impact on the dispersion of happiness.

As an index of well-being, life satisfaction integrates all the different life aspects that are relevant for a person: if its value is higher (lower), then she perceives having a more satisfied (dissatisfied) life. At the societal level, life satisfaction represents overall quality of life: if its value is higher (lower), then life is better (worse) for those citizens. Life satisfaction distribution portrays the inequality of SWB among individuals: if its value is relatively scattered, then some members of a society have much a better life than others; on the other hand, if its value is concentrated, then members of a society are experiencing a similar quality of life. As life satisfaction encompasses all life domains; its dispersion can be seen as a broad indicator of inequality, subsuming not only income differences, but also dissimilarities in other aspects of life for individuals. Although it has been important for individuals to care about overall life satisfaction level, it probably does them more good than harm to take an interest in its distribution, too: for any given average life satisfaction, everyone prefers the variance of life satisfaction to be lower. This implies fewer people have low subjective life satisfaction. Also, society’s overall welfare ought to put more emphasis on people who are miserable than on people with high life satisfaction.

Whilst a considerable chunk of the literature has been discussing well-being averages/levels, this paper refocuses the spotlight on variations/inequalities or, more broadly, distributional aspects of well-being within the macroeconomic landscape. SWB is estimated by people’s life satisfaction evaluations. Defined as heterogeneity in self-evaluated life satisfaction responses in a society, life satisfaction inequality makes sense because individual well-being hinges on the idea of social comparison: people derive satisfaction not only from their absolute status, but also from their relative status compared to a specific reference group. In the literature, this “reference group” either takes an external (i.e. particular

7 demographic entities, such as own family, friends, co-workers, people in the same community, state, region, or people in other countries) or internal (i.e. past individual status) definition. Studies that try to explain the Easterlin paradox have noticed that a person’s relative status matters more for happiness (Frey & Stutzer, 2000; Easterlin, 2001; van Praag & Ferrer-i-Carbonell, 2004; Clark et al., 2008). Each person cares about how her life measures up to a specific benchmark.

After the much-publicised global financial crisis that started in 2007, the “Great Recession” has been a period of general economic decline during the late 2000s and early 2010s. The International Monetary Fund has concluded that it had been the worst global crisis since the Great Depression in the 1930s (IMF, 2009). These predominantly unforeseen and severe events have redirected the interest of macroeconomists on the consequences of recessions on both individual and social well-being (Grusky et al., 2011). Based on recent studies that have explored the effects of this period (e.g. Graham et al., 2010; Greve, 2012; Gudmundsdottir, 2013), the Great Recession has affected the economic and social welfare of billions of individuals across the globe. More specifically, papers on the outcomes of stock market fluctuations, during the 2008 crisis have found that the weakening of financial indicators during the 2008 crisis had an impact on the psychological and evaluative well-being of most people (Bollen et al., 2010; Deaton, 2012; Lin et al., 2015). In a 2009 social survey conducted among Greeks, Panas (2013) has learned that their personal well-being was affected by the economic crisis. Bjørnskov (2014) has observed that decreases in SWB during downturns have been significantly more pronounced in European countries with relatively higher degree of market regulation. On the contrary, Helliwell et al. (2014) have uncovered that American communities with more entrenched social engagement endured less intense reductions in life evaluations in response to rising unemployment throughout the 2008 crisis.

Not much is known about the variations that macroeconomic movements (even in non-crisis periods) bring to the significance of how subjective life satisfaction assessments are aggregately distributed. Hence, the present research aims to extend the discussion by looking into the macroeconomics of life satisfaction inequality, as it appears rational to expect these macroeconomic conditions to affect the distribution of SWB. Now more than ever, leaders, policymakers, academia, the media, and the general public may find it worthwhile to know more about life satisfaction inequality. “How do these macroeconomic shifts influence SWB inequality?” The literature on endogenous macroeconomic policy has long explored the idea of social preference functions in models of a benevolent government that aims to optimise the utility of the representative consumer (Barro & Gordon, 1983).

Measuring life satisfaction inequality presents an undertaking quite different from the standard problem of measuring income or wealth inequality. The struggle principally lies in the measurement of life satisfaction itself: life satisfaction cannot be presumed to be directly and explicitly observable; thus, it may not make sense to treat it as a continuous variable. As a consequence, indirect methods could involve obtaining an individual’s self-reported life satisfaction, or simple modeling using observables that are thought to be related to life satisfaction. These indirect methods could be tricky. Hence, one of the objectives of the present research is to explore a practical approach to inequality measurement in the

8 SWB context and the extent to which various assumptions about life satisfaction influence inequality comparisons. Perhaps, there are two reasons why employing an indirect method to life satisfaction measurement is obviously tricky. First, life satisfaction could not be taken as a monetary-equivalent measure. Second, it ought to be treated as an ordinal variable rather than a continuous one. Conventional methods of inequality analysis and standard properties of inequality indices may not be applicable.

The results from this paper go towards the identification of a more suitably based definition of life satisfaction and of life-satisfaction-inequality measures. The present paper carries out the following: 1. Provide an alternative estimation of life satisfaction inequality using a measure of status that is not artificially imposed through an arbitrary cardinalisation strategy, but based on an upward- looking perspective of the distribution of life satisfaction. In order to compute inequality measurements, this concept of status is derived from the distribution of life satisfaction responses. This is an application of the cardinalisation method employed in the Cowell-Flachaire (2017) status-inequality approach. 2. Provide the first multi-country estimate of life satisfaction inequality using a status-inequality approach from a harmonised panel dataset that spans more than three decades and covers 46 countries with at least four country-wave periods. 3. Examine simultaneously in a single model the macroeconomic determinants of upward-looking status-inequality across countries and over time, taking into account drivers, such as income inequality, economic growth, unemployment, and inflation. 4. Examine whether patterns of life satisfaction inequality vary with particular macroeconomic shifts across different groupings based on life satisfaction inequality level, income inequality level, status of development, and geographic location.

The paper is organised as follows. Section 2 contains some necessary theoretical background. Section 3 introduces the dataset and explains the paper’s empirical strategy. Section 4 presents the results. Section 5 discusses the regression findings on the macroeconomics of life satisfaction inequality.

9 II. Conceptual Framework

Inequality Comparisons Using Ordinal Data

Some Existential Considerations and Methodological Implications

The use of survey-based measures of SWB has gradually become mainstream across the social sciences, even in economics. In spite of this, researchers have different assumptions when it comes to dealing with measures of SWB data. Asking individuals a single question about how satisfied or happy they are with their lives in general or with specific life domains is traditionally how SWB data is collected. These data are commonly regarded as ordinally comparable, and increasingly also cardinally comparable from one person to another. For the latter, there ought to be a unique linear relationship between true utility (i.e. well-being in economics) and measured SWB, though the basis for this information is still limited. The cardinality assumption is generally explained based on statistical requirements, and intermittently requirements for interpretation, rather than on reason. According to Ferrer-i-Carbonell & Frijters (2004), it is often viewed as a necessary assumption, owing to the importance of individual effects in specifying cross-sectional differences found in SWB data and their ability to include information that, otherwise, could bring about bias. Perhaps, instead of justifying its usage due to its practicality, this cardinality assumption should be taken into account on the basis of its reasonability and implications.

Scholars also differ in their interpretations of SWB data. On the one hand, SWB is defined by the way it is measured. Using this approach seems to be convenient but it does not take away issues related to bias and arbitrary measurement due to scale restrictions. On the other hand, SWB data are more commonly understood as measures of true utility in economic analyses. This interpretation suggests certain assumptions about the response function by which true utility is transformed into observed survey answers. The main issue with the latter interpretation is that the response function for SWB cannot be directly observed, since true utility is an abstract (i.e. latent) idea. Layard et al. (2008) have argued that the cardinality assumption is both justified and reasonable, and that SWB data could be used to estimate marginal utilities since information about the utility function’s curvature is necessary for correct policy design. Oswald (2008), in opposition, has argued: rather than treating SWB data as cardinal measures of utility, estimating marginal rates of substitution is sufficient in policy applications.

Hirschauer et al. (2015) have recognised that this matter should be tackled more closely so as to have more robust SWB measurements, including life satisfaction inequality, in economic analysis and policy applications. For most of previous SWB inequality studies, this may be contentious because of their methodologies, resorting to cardinalisation conversions of SWB data. These studies have conventionally undertaken analyses, which are based on standard inequality measures. In turn, such traditional measures usually work with cardinal data, because these are mean-based, i.e. inequality is regarded as a deviation from the mean (Allison & Foster, 2004). Conversely, life satisfaction responses are unnaturally suitable

10 for representation on cardinal scales. Either ordinal data may be measurable but its measurement scale is not exactly known, or it is entirely immeasurable (Blair & Lacy, 2000). Conventionally cardinal variables, like income, have natural zero points and, often open scales. This is in contrast to ordinal variables, like happiness and satisfaction, which are commonly measured on scales that are bounded from above and below (e.g. zero to five, zero to ten). Van Praag (1991) has been one of that documented that the boundedness of scale issue might be causing the downward bias on SWB data. This issue has not yet been cited for its implications for empirical research, so future studies may provide more clarity.

Another critical issue in the literature has been the indeterminacy of interpersonal transfers using SWB data. Standard inequality measures generally have two properties: (1) when there is perfect equality, these take the minimum value of zero, and (2) these typically fulfill the Pigou-Dalton transfer principle, whereby a mean-preserving transfer from a relatively poorer to a richer individual always intensifies inequality. Since the mean is absent for ordinal data, mean-preserving transfers are futile. To illustrate: “what would transferring satisfaction from a satisfied to a less satisfied person exactly imply?” Then, an interpersonal transfer of SWB appears to be a peculiar concept. A number of researchers argue: depending on whether the transfer is ample to re-position individuals to different categories, the impact of the transfer principle remains unspecified (Blair & Lacy, 2000; Kobus, 2015). Kalmijn and Veenhoven (2005) have explained that interpersonal transfers are necessarily possible when the underlying variable qualifies for a “capacity” interpretation. To have an idea of what “capacity” represents, they offer this analogy: volume is a “capacity” variable because it can be summed up; conversely, temperature is an “intensity” variable that seems unusual to be aggregated. Following Kalmijn and Veenhoven (2005), life satisfaction ought to be seen as an intensity variable, whereas income fits the qualification for being a capacity variable. The latter can be collapsed, so to speak, into a total value of a population; this aggregate figure then becomes susceptible to the (re-) distribution. Obviously, redistributing a total amount of life satisfaction for a population seems to be strange. From these existential considerations, SWB studies should ignore standard inequality measures due to the reality that these conventional indicators depend on a capacity interpretation of underlying distributions.

Inequality measurement chiefly has three ingredients: (a) equalisand (the thing that may be unequally distributed), (b) unit of analysis (the economic/social entities among whom the distribution occurs), and (c) aggregation method (the mathematical formula that collects the information in the distribution). Standard inequality measurement theory requires that the first ingredient is a cardinal quantity with a particular cardinalisation. However, SWB studies are basically constrained by the underlying nature of the equalisand itself, so conventional tools of inequality analysis become irrelevant. This measurability problem, cited in SWB literature strands that focus on inequality in satisfaction (e.g. Stevenson & Wolfers, 2008; Yang, 2008; Oswald & Wu, 2011) or health status (e.g. van Doorslaer & Jones, 2003; Costa-i-Font & Cowell, 2013), brings up serious theoretical and practical concerns. Thus, findings from existing literature ought to be treated with some caution. In the literature, there have been two main approaches to making inequality comparisons when the underlying equalisand is ordinal.

11 Measurement of SWB Inequality with Ordinal Data

Results from such ordinal inequality approaches have not only shown rather limited propositions, but also encountered interpretation and application difficulties. One approach is the imputation of some artificial index of individual well-being status as a function of the categories. In particular instances, this imputation has been explicitly or implicitly applied to entities without any natural ranking, such as endowment or attribute vectors; a utility function is necessary to impose order on the data. Specifically, this is akin to the standard measurement of multi-dimensional inequality: the “utility” of these factors are derived, and then the inequality of utility is computed, in which the appropriate aggregator is the utility function (Maasoumi, 1986; Tsui, 1995). This “status” approach has been fraught with significant objections: arbitrariness of cardinalisation, or of the method of grouping components. Albeit the ensuing well-being index seems to be sensible over an extensive subset of categories, outliers (i.e. extreme values) might not be appropriately incorporated in this index, so their inclusion, or lack thereof, could have implications for inequality comparisons. Another approach, derived from the health literature, to ordinal inequality measurement entails a reworking of inequality-ranking strategies, which are contingent upon first-order dominance criteria (Allison & Foster, 2004; Apouey, 2007; Abul Naga & Yalcin, 2008, 2010; Zheng, 2011). It has been established that the median could be used as an equality concept comparable to the use of the mean in standard inequality analysis. Abul Naga and Yalcin (2010) have raised special issues regarding the comparison of distributions with different medians. Analogously to the case of ordinal variables, this approach might face some trouble if quantiles are not well-defined.

Cowell and Flachaire (2017) have recently introduced an alternative version of the status approach to inequality analysis. Their adaptation tackles some aforementioned existential and methodological issues by directly using ordinal data. According to their approach, the notion of status is essential to making inequality comparisons using ordinal data: as long as the underlying data could be ranked, it is sufficient to build an inequality index, which has a natural intuition and easy empirical implementation. Conceptually, “status” is carefully defined as a person’s location within a distribution. In practical terms, this is her position in society. This is closely related with the idea of individuals engaging in social comparisons. Moreover, this particular status concept is comparable to concepts used in poverty and relative deprivation literature (Runciman, 1966; Yitzhaki, 1979) and, more recently, in studies on the inequality of opportunity (de Barros et al., 2008).

Recall that true utility is an abstract idea that cannot be directly observed, whilst individual survey responses to life satisfaction or happiness questions are the SWB data that could be estimated to represent this utility concept. Each person’s status depends on her utility, the distribution of utilities, and the population size. By focusing on status instead of directly measuring utility, the Cowell-Flachaire index (푰푪푭) bypasses both the ordinal data problem and the more famous cardinal data problem. As a result, the status takes care of the issue with the equalisand, which is SWB data. The only complication left to be resolved is that of the aggregation method. As a preliminary step towards a parametric inequality measure for ordinal data, the mergers principle (see also Yitzhaki, 1979) ought to be invoked

12 in order to put some structure on the index. According to Cowell & Flachaire (2017), each individual’s status would be unchanged by combining categories of the distribution to which the individual does not belong. This alternative status-inequality measurement is simply the summation of every individual status ‘gap’ from a specific reference point. What matters in this approach is the precise definition of status. Recall that this appears to be an analogue of the Yitzhaki index: overall relative deprivation is the aggregation of differences in income between each person and all other persons with a higher income in the former’s reference group, taking into account the size of the reference group, i.e. population. Going back to the status-inequality method, the 푰푪푭 becomes a relative status measure that can be used to estimate the degree of equality of a society. Note that self-reported life satisfaction responses have been the basis instead of applying an unnecessary cardinal procedure on the original ordinal data. In case the thing that is unequally distributed happens to have a natural cardinalisation (e.g. income), it makes more sense to define it based on its underlying nature. With the 푰푪푭, status is directly observable (that happens to be ordinal in nature), rather than income. Where ordinal information is clear – as with life satisfaction – then another strategy has to be applied.

Assume that ordinal data simply conveys information on ranks alone, and nothing else: there is a particular number of people in each category 풌 = ퟏ, ퟐ, … , 푲, but these categories could be arranged in increasing order of their desirability. Then a simple argument shows that, if there are 풏풌 persons in category 풌 = ퟏ, ퟐ, … , 푲, then status of individual 풊 who is currently in category 풌 (풊) ought to be a function of either the ‘upward-looking’ concept, 푲

∑ 풏풍 풍=풌(풊) or its ‘downward-looking’ counterpart 풌(풊)

∑ 풏풍 풍=ퟏ

푲 It may be suitable to normalise by the size of the total population 풏 = ∑풌=ퟏ 풏풌, so that the status of individual 풊 is given by either an upward-looking version, 풔풊, in which her status depends on the proportion of people with similar or higher status, that is: 푲 ퟏ 풔 = ∑ 풏 풊 풏 풍 풍=풌(풊)

or, a corresponding downward-looking version, 풔′풊, in which her status depends on the proportion of people with similar or lower status, that is: 풌(풊) ퟏ 풔′ = ∑ 풏 풊 풏 풍 풍=ퟏ

13 Using either the upward- or downward-looking versions, her status ought to lie between zero and one. If everyone is in the same category (i.e. everyone has the same status), then perfect equality is attained. Hence, perfect equality is the suitable, exogenous reference point. In the case of a peer-inclusive definition of status, choosing perfect equality reproduces the ordinal variable with the status concept. For ordinal data, inequality is simply the total ‘distance’, in terms of proportions (rather than absolute differences), from perfect equality.

To reiterate, the present status-inequality measurement problem, then, is fundamentally summarising the information in the vector 풔: = (풔ퟏ, 풔ퟐ, … , 풔풏) in relations to the equality vector (ퟏ, ퟏ, … , ퟏ). Based on their axiomatic approach, Cowell and Flachaire (2017) have observed that inequality ought to take the form of an index in the following family, indexed by 휹

풏 ퟏ ퟏ [ ∑ 풔휹 − ퟏ] , 풊풇휹 ≠ ퟎ 휹(휹 − ퟏ) 풏 풊 휹 풊=ퟏ 푰푪푭(풔; ퟏ) = 풏 ퟏ − ∑ 풍풐품 풔 , 풊풇 휹 = ퟎ 풏 풊 { 풊=ퟏ in which 휹, a parameter that is less than 1 and specifies the desired sensitivity of the index to a particular 휹 portion of the life satisfaction distribution: for high values of 휹, this 푰푪푭 is particularly sensitive to values of 풔풊 close to one. If status is given by 풔풊, then this is an index of ordinal inequality based on an upward- looking status concept; however, if status is represented by 풔′풊, then this index, respectively, is defined as ordinal inequality rooted on downward-looking status. This family is remarkably similar to the well- known generalised entropy class of inequality indices (Cowell, 1980; Shorrocks, 1980). Since it has been established that 휹 < ퟏ, such admissible family of status-inequality could also be analogous to Atkinson indices that are written in their ordinally-equivalent form:

풏 ퟏ⁄ ퟏ 휹 ퟏ − [ ∑ 풔휹] , 풊풇휹 < ퟎ 풐풓 ퟎ < 휹 < ퟏ 풏 풊 풊=ퟏ 푨휹(풔; ퟏ) = ퟏ 풏 ⁄풏

ퟏ − [∏ 풔풊] , 풊풇 휹 = ퟎ { 풊=ퟏ

Derived from a particular distribution of persons across categories, both upward-looking and downward- looking versions could be utilised in the analysis of ordinal data. As the notion of status is essential to the approach to determining ordinal inequality, the choice of status variable must affect how inequality comparisons are made. As the status vector approaches the value of equality, the ordinal inequality 휹 index, 푰푪푭, comes closer to zero. For upward-looking status in which (ퟏ ≥ 풔ퟏ ≥ 풔ퟐ ≥. . . ≥ 풔풏 > ퟎ): people have been observed to move to higher categories. On the other hand, downward-looking status in which

(ퟎ < 풔′ퟏ ≤ 풔′ퟐ ≤. . . ≤ 풔′풏 ≤ ퟏ): people have been perceived to move to lower categories. In other words,

14 inequality ought to increase if a person migrates downwards (upwards) from a higher (lower) status. 휹 Accordingly, 푰푪푭 ought to be theoretically appropriate for considering comparisons of inequality in terms of subjective life satisfaction. The meaning of values on the underlying ordinal scale influences the choice between versions of a person’s status. When higher values convey more (less) desirable situations, then upward- (downward-) looking status is selected. In the next section, this is considered in order to explore how life satisfaction responses of individuals are distributed.

As mentioned in the first section, Easterlin (1974) has underscored this happiness-income paradox: for a specific country, individuals who earn higher incomes probably report higher life satisfaction; whereas for comparisons across countries and higher-income countries, the average level of satisfaction remains flat vis-à-vis income. Later studies have either confirmed the lack of influence of income on life satisfaction for higher-income countries, or found a significant positive impact. Whether income has an effect of life satisfaction among higher-income countries has remained controversy for researchers. Many empirical studies have relied on particular transformations of life satisfaction responses, or established comparisons of the averages of these answers in order to derive composite well-being indices (e.g. Deaton, 2008; Inglehart et al., 2008). For instance, life satisfaction evaluations have been cardinally interpreted on a linear scale: that is, equidistance is assumed for values attached to each successive response category. Especially when these scales have appropriate symmetry of responses around an ostensible middle category, such equidistance has denoted that intervals between points could approximately be regarded as equal so these can be employed in parametric analysis (e.g. Ng, 1997; Ferrer-i-Carbonell & Frijters, 2004; Kristoffersen, 2011). However, cardinal interpretations have been seen to be challenging in the literature on Likert-type scale surveys (see Knapp, 1990; Norman, 2010). Some studies have presumed that ordinal data simply offer information on ranking/ordering, so these ought to be treated as purely ordinal with non-parametric statistics (Kuzon et al., 1996; Jamieson, 2004). For example, marathon results would simply provide the ranking of those who reached the finish line without any information on the differences in time between runners, or their respective arrival records. With an alternative approach that depends on the nature of the underlying data, ordinal inequality 휹 indices, such as the 푰푪푭 , must be solely defined on individual ranks; hence, this specific inequality measure ought to be insensitive to cardinal interpretation of these life satisfaction answers. Beyond existential considerations and methodological implications, the distribution of life satisfaction could offer another way to capture how macroeconomic movements are influencing people’s lives.

15 Life Satisfaction Inequality: Reframing the ‘Misery Index’

Macroeconomic movements are complex phenomena that are able to shape SWB in several different and interrelated ways. There are some essential variables (such as income inequality, economic growth, unemployment, and inflation) that play a fundamental role in determining overall welfare. This appears to be the basis for economist Arthur Okun’s misery index. Even prior to the emergence of literature focused on happiness economics, this specific index for well-being has been found to be constructive for empirical research and policy formulation. It has been established as the first attempt to summarise a range of macroeconomic variables into a single measure so as to monitor the general ‘health’ status of the economy during a business cycle. In its original version, the misery index has considered two crucial targets of macroeconomic policy (inflation and unemployment) in a simple aggregate disutility function. This estimates the level of economic discomfort as the unweighted sum of inflation and unemployment rates (Mankiw, 2010). Although interestingly simple, the intuition behind Okun’s misery index has been adopted in different useful applications (Setterfield, 2009; Blanchflower et al., 2014). When the United States began going through concurrent increases in its inflation and unemployment (i.e. “stagflation”) in the 1970s, the social implications associated with these two macroeconomic indicators have become more apparent and palpable to everyone in society. Okun has suggested the construction of an economic discomfort index as a means of setting a straightforward yet objective measure of economic “dissatisfaction”. A higher level of either these variables has adverse effects on overall social welfare. Hence, the misery index has been viewed as an inverse measure of economic well-being (Nessen, 2008).

At first glance, Okun’s approach seems to be too unsophisticated: it only allows for two facets of national economic performance, and it assigns equal weights for the inflation rate and the unemployment rate. These comments could lead to the propensity to dismiss the index, as a rudimentary and unreliable interpretation; quite the opposite, the misery index prevails as a worthwhile fundamental tool for two reasons. First, it seems to offer a useful estimate of the impact of macroeconomic conditions on overall well-being, as indicated by specific statistics, for instance, the suicide rate (Yang & Lester, 1999), consumer sentiment (Lovell & Tien, 2000), crime rate (Lean & Tang, 2009), and poverty rate (Lechman, 2009). Second, it has become an intuitive notion. Other researchers have extended the “spirit” of the misery index along two, partly overlapping, strands. In one strand, some have tried to improve the original incarnation of the index by including more indicators (e.g. GDP growth rate, real long-term interest rate, house and share prices, and so forth). These “augmented misery indices” have been further enhanced by adding (with weighting factors) new variables to get a full composite indicator of national macroeconomic performance (Setterfield, 2009). In another strand, the misery index has been viewed as a basis in applied research on macroeconomic loss functions (Mayer, 2003). Seminal studies by Di Tella et al. (2001) and Welsch (2007), among others, have explored the relation between SWB and macroeconomic performance in the interest of cultivating reliable social welfare functions that could be utilised to assess effects of policies and shocks on overall well-being (Blanchflower et al., 2014).

16 Instead of simply aggregating each person’s self-reported life satisfaction response to obtain an overall or average life satisfaction in that society, life satisfaction inequality (based on the status approach using ordinal data) could be considered a more appropriate measure of overall welfare. Briefly, the distribution of life satisfaction ought to provide a more inclusive and meaningful estimation of the effects of any inequalities that arise from movements in key macroeconomic variables that have an impact on life satisfaction. In addition to considering the consequences of such macroeconomic fluctuations for how life satisfaction is distributed, there also have been discussions in the literature and empirical evidence suggesting that individuals are or at least ought to be more satisfied in a society wherein they likely face greater equality of opportunities, rather than that of outcomes. Beyond these ostensible links between macroeconomic variables and life satisfaction, macroeconomic inequalities could be argued to generate particular challenges for other key non-pecuniary provisions for life satisfaction: access to education and health, good governance, safety, security, and social trust, among others. From another vantage, these inequalities in the distribution of various non-economic supports to life satisfaction still endure without arguing that macroeconomic movements have driven such disparities.

Happiness economics have directly confronted some of the assumptions in traditional welfare economics that established a restrictive definition of welfare. Rather than being limited to the notion that utility is simply dependent on income, many of these studies have considered a relatively broader concept of well-being, which, among others, encompasses interdependent preferences, hedonic adaptation effects, and experienced and procedural utility functions (Graham, 2008; Kőzegi & Rabin, 2008). Over the past couple of decades, various well-being theories have been closely investigated, including aspirational and distributional fairness models. Whilst the former has focused on the impression of aspirations and social expectations on individual happiness, the latter has centered on the equality and distributional facets of well-being. Apart from taking into account a relatively more comprehensive understanding of welfare or utility, it has been known that all economies have to contend with the constraint of scarce resources. Then, any equilibrium hinges on the preferences of individuals in a society. Consequently, if people’s preferences are skewed towards a distribution of happiness that appears to be imbalanced, then a less- than-desirable equilibrium could emerge. Simply, an uneven life satisfaction distribution may result in unproductive outcomes and behaviour among people. Arguably, life satisfaction comparisons may have prompted people to recognise that others are deriving more pleasure in their lives and are more satisfied.

A person’s well-being has been seen to be shaped by the gap between aspirations and achievements, based on aspirational well-being theories (Inglehart, 1990; Michalos, 1991). Two separate mechanisms – hedonic adaptations (Helson, 1964; Brickman & Campbell, 1971) and social comparisons (Olson et al., 1986) – have initiated and paved the way for the emergence of these personal aspirations. ‘Habit formation’ and ‘interdependent preference’ models have been discussed in relation to such aspirations among people in economic literature. For habit formation models (e.g. Pollak, 1970; Carroll et al., 2000), individuals’ preferences and tastes have been presumed to change over time; but, they have become used to past income and consumption levels. Hence, individual well-being is determined by change and fades with constant consumption (Stutzer, 2004). Regarding interdependent preference

17 models, personal aspirations and social expectations have significantly motivated happiness levels: people’s well-being has been relatively shaped by how they evaluate their experiences compared to relevant others (e.g. Duesenberry, 1949; Clark & Oswald, 1998; Easterlin, 2001). Thus, life satisfaction is particularly driven by relative positional concerns in terms of monetary and material gains (Easterlin, 2003). Clark and Oswald (1998) have found that people often feel a sense of satisfaction/contentment (or dissatisfaction/envy) when they achieve a relatively higher (or lower) status within their specific reference groups. In Easterlin’s (2001) life-cycle model, higher income levels yield greater happiness for people (from diverse socioeconomic backgrounds) at the beginning of their adult life cycle, assuming similar material aspirations, whereby permitting better realisation of their personal aspirations. Given that people’s material aspirations are likely to increase according to the growth in their income levels over the life cycle, this positive relationship between income and happiness weakens as individuals become older. During younger years, social comparisons with different socioeconomic backgrounds act as a key driver in influencing aspirations. In contrast, social comparisons turn out to be limited by specific reference groups within comparable personal material and consumption experiences, and socioeconomic ranks during later years. Easterlin (2003) has further claimed that enrichment (or erosion) in non-monetary life aspects shows a relatively more obvious impression on people’s fulfillment (or frustration) in reaching desired goals, leading to improved (or impaired) life satisfaction.

Other interdependent preference models have been fixated on the equality and distributional angles of life satisfaction. Thurow (1971) and Becker (1974) have put forward the notion that individual well- being is meaningfully shaped by both the average social income and its relative distribution among people in society. Given that individual incomes are unchanged, increasing average social income raises (reduces) the happiness of unselfish (selfish) individuals. Conversely, allowing for a constant average social income and depending on the relative status of people, rising income inequality ought to diminish (develop) the happiness of generous (grudging) persons. Furthermore, these models have been aligned with Rawls (1971) “Theory of Justice”, assuming that individuals are instinctively opposed to inequality because they show an inclination for altruism. Appreciating the distribution of life satisfaction is not only fundamental for exploring the consequences of economic inequality on well-being, but also significant for improvement and monitoring of all-inclusive well-being and social cohesion policy measures, since the manifestation of a wide life satisfaction disparity among people or groups could lead to social conflict. Based on ‘expected utility’ theories of social protest, life satisfaction gaps have had indirect effects on social disruption: people would engage in dissenting behaviour only if expected benefits are relatively higher than costs (Tullock, 1971). On the other hand, ‘discontent’ theories have posited that life satisfaction inequality is inversely related to social unity (Gurr, 1994; Brown, 1996), whereby imposing a clear independent effect on social disruption (Guven et al., 2010; Becchetti et al., 2014). Hence, life satisfaction inequality could be seen as a primary source of resentment and social disorder, whilst income disparity is an unintended one. Whereas a social group may be less fortunate than another group but if the former finds other sources of happiness, income inequality alone would not immediately engender social conflicts. Next, relevant hypotheses on the link between life satisfaction inequality and macroeconomic movements, respectively, are formulated.

18 The Link Between Life Satisfaction Inequality and Macroeconomic Movements

Income Inequality

Hypothesis 1 – Higher income inequality leads to greater life satisfaction inequality.

A person’s status within the distribution of life satisfaction may be determined by how her income level compares to that of other individuals in society. If she perceives belonging to a higher income rank, then she would feel more satisfied than others whose incomes remain at the bottom portion of the income distribution. In this case, there appears to be a wider income gap between her and relevant others. Poorer people could view this disparity as a reason to experience decreased life satisfaction. However, if these people earn significant increases in their incomes that make it more likely for them to become as fortunate as richer individuals, then their own life satisfaction levels come closer to those who are already highly satisfied with life, in general. Thus, countries would have higher life satisfaction inequality when there is higher income inequality.

Economic Growth

Hypothesis 2 – Higher economic growth reduces life satisfaction inequality.

Although there are numerous non-monetary factors driving self-reported life satisfaction, income is still regarded as a primary input for people to say that they are satisfied with their lives. If income growth translates into improved individual happiness levels, then individuals who are able to take advantage of this good fortune are more likely to face an enlarged set of consumption opportunities. As a result, their life satisfaction levels ultimately go up. Moreover, differences in their life satisfaction responses ought to be much smaller. Hence, countries with greater economic growth would likely encounter diminished life satisfaction inequality.

19 Unemployment

Hypothesis 3 – Higher rate of unemployment increases life satisfaction inequality.

When an individual loses her job, it seems less likely for her to have the same life satisfaction level as when she was still employed. Ostensibly, she would not only recognise a drastic change in her overall status in society, but also experience the consequences of this joblessness. Holding the labour status of employed people around her as unchanged, these relevant others would feel more satisfied than her since they still manage to keep a source of income that affords them better opportunity possibilities. In turn, this recently-unemployed person realises the increased gravity of not having access to resources that coincide with job security, whereby significantly lowering her satisfaction with life vis-à-vis that of those still working. Therefore, countries would struggle with more life satisfaction inequality given higher unemployment rates.

Inflation

Hypothesis 4 – Higher rate of inflation intensifies life satisfaction inequality.

If there is a sustained increase in the price level of goods and services and incomes remain constant, any individual would face the reality of consuming less than before the price increase. Moreover, this may imply lower living standards for everyone. Rising prices also entail higher input costs, which not only erode firms’ profits, but also decelerate overall economic activity, imposing adverse ramifications for employment and investment opportunities. For those who might incur more losses due to higher inflation, they turn out to have lower life satisfaction than who are less harmed. In view of that, countries with growing inflation rates would probably tackle increasing life satisfaction inequality.

20 III. Data and Measurements

Data and Empirical Strategy

Subjective Well-Being Inequality Variable

Merging the latest six waves of the World Values Survey (WVS) and four waves of the European Values Study (EVS), the primary data source to be utilised in the present study is the 1981-2014 Integrated Values Surveys (IVS). It covers information on attitudes, beliefs, ideas, opinions, preferences, and values of individuals across the world (EVS, 2015; WVS, 2015). This database has: 1) a significant advantage for comparable, multidimensional micro-data on a range of economic, social, political, and cultural issues; and, 2) a standardised worldwide structure for examining cross-country patterns across different world regions. Life satisfaction, the underlying data source for the dependent variable (SWB inequality), is derived from this data set. Table 1 presents the countries, sorted by the number of waves; it also shows for each country the mean of macroeconomic variables across all waves. Figure 1 presents the distribution of upward-looking life satisfaction inequality by country-wave period.

The first step is to choose an appropriate ordinal variable for estimating SWB, which has both affective and cognitive components (Diener et al., 2002). The affective aspect of SWB encompasses emotions, feelings and moods of individuals, whilst the cognitive counterpart relates to personal evaluations of their current life situation in comparison with expectations (Haller & Hadler, 2006; Veenhoven, 2008). It is standard for national and international surveys to include individual-level information on measures of self-evaluated well-being in ordinal form. In the IVS data set, there are two SWB measures: feeling of happiness (i.e. affective), which portrays short-run SWB (Larsen & Prizmic, 2008); and, life satisfaction (i.e. cognitive), which depicts long-run SWB (Andrews & Withey, 1976). Life satisfaction is, arguably, more reasonable and suitable than happiness, since it is relatively more stable in capturing individuals’ state of being over their life cycles. People weigh the inconsistencies between actual accomplishments across life domains and ambitions vis-à-vis these domains (Campbell et al., 1976; Michalos, 1985).

The ordinal measure of SWB is based on answers to the question: “All things considered, how satisfied are you with your life as a whole these days?” Personal assessments of life satisfaction with potential responses in ten categories ranging from 1 (‘dissatisfied’) to 10 (‘satisfied’) are drawn. Aggregating individual-level answers in order to construct country-level distributions of life satisfaction in each time period, this study uses the Cowell and Flachaire (2017) approach to make international comparisons of life satisfaction inequality. Table 2 illustrates the respondents by country, and inequality indices based on two versions (in percent) for the latest country-wave. In this study, life satisfaction adopts an upward- looking version: higher values on the scale suggest a more advantageous situation. Inequality is lowered when the life satisfaction distribution is skewed towards higher categories; or, equality becomes more likely when people are inclined to report better life satisfaction in society.

21

Figure 1: Distribution of Life Satisfaction Inequality in the IVS, 1981–2014 (countries with at least 4 waves)

22

Table 1: Countries in the Integrated Values Surveys Dataset Coverage Macroeconomic Variables (mean levels across waves) Country Number of Income Real GDP Per Unemployment First Year Latest Year Inflation Rate Waves Inequality Capita Rate Argentina ퟔ ퟏퟗퟖퟒ ퟐퟎퟏퟑ ퟒퟑ. ퟕퟖퟑ ퟖ. ퟑퟒퟕ ퟗ. ퟕퟖퟎ ퟏퟑퟒ. ퟎퟗퟏ Australia ퟒ ퟏퟗퟖퟏ ퟐퟎퟏퟐ ퟒퟓ. ퟗퟓퟎ ퟒퟖ. ퟔퟖퟔ ퟔ. ퟏퟐퟓ ퟒ. ퟔퟗퟎ Belarus ퟓ ퟏퟗퟗퟎ ퟐퟎퟏퟏ ퟑퟏ. ퟐퟒퟎ ퟑ. ퟗퟗퟖ ퟎ. ퟕퟕퟐ ퟔퟒퟕ. ퟖퟖퟎ Belgium ퟒ ퟏퟗퟖퟏ ퟐퟎퟎퟗ ퟒퟓ. ퟖퟎퟎ ퟑퟓ. ퟓퟖퟖ ퟖ. ퟗퟕퟎ ퟑ. ퟎퟑퟔ Bulgaria ퟓ ퟏퟗퟗퟏ ퟐퟎퟎퟖ ퟑퟒ. ퟕퟐퟎ ퟒ. ퟖퟔퟓ ퟏퟐ. ퟗퟒퟖ ퟐퟖퟑ. ퟑퟓퟕ Canada ퟒ ퟏퟗퟖퟐ ퟐퟎퟎퟔ ퟒퟒ. ퟎퟓퟎ ퟑퟗ. ퟖퟕퟑ ퟖ. ퟎퟖퟎ ퟓ. ퟎퟕퟐ Chile ퟓ ퟏퟗퟗퟎ ퟐퟎퟏퟏ ퟓퟐ. ퟎퟐퟎ ퟗ. ퟗퟐퟗ ퟕ. ퟔퟓퟔ ퟖ. ퟕퟗퟒ China ퟓ ퟏퟗퟗퟎ ퟐퟎퟏퟐ ퟒퟔ. ퟗퟔퟎ ퟐ. ퟒퟕퟗ ퟑ. ퟗퟔퟐ ퟓ. ퟓퟗퟒ Colombia ퟒ ퟏퟗퟗퟕ ퟐퟎퟏퟐ ퟓퟑ. ퟑퟓퟎ ퟓ. ퟓퟐퟓ ퟏퟐ. ퟐퟎퟓ ퟏퟏ. ퟑퟒퟐ Czech Republic ퟒ ퟏퟗퟗퟏ ퟐퟎퟎퟖ ퟒퟑ. ퟏퟐퟓ ퟏퟓ. ퟏퟔퟕ ퟓ. ퟐퟔퟑ ퟔ. ퟕퟔퟗ Denmark ퟒ ퟏퟗퟖퟏ ퟐퟎퟎퟖ ퟒퟑ. ퟓퟕퟓ ퟒퟖ. ퟕퟗퟎ ퟔ. ퟖퟑퟓ ퟓ. ퟎퟖퟏ Estonia ퟓ ퟏퟗퟗퟎ ퟐퟎퟏퟏ ퟒퟕ. ퟑퟖퟎ ퟏퟏ. ퟕퟗퟒ ퟖ. ퟎퟏퟒ ퟐퟒ. ퟏퟑퟖ Finland ퟔ ퟏퟗퟖퟏ ퟐퟎퟎퟗ ퟒퟒ. ퟓퟏퟕ ퟑퟕ. ퟏퟒퟗ ퟖ. ퟓퟕퟐ ퟑ. ퟖퟐퟔ France ퟓ ퟏퟗퟖퟏ ퟐퟎퟎퟖ ퟒퟖ. ퟑퟒퟎ ퟑퟓ. ퟖퟑퟐ ퟖ. ퟒퟒퟎ ퟒ. ퟑퟒퟗ Georgia ퟒ ퟏퟗퟗퟔ ퟐퟎퟏퟒ ퟒퟗ. ퟏퟕퟓ ퟐ. ퟒퟔퟎ ퟏퟒ. ퟗퟔퟖ ퟏퟑ. ퟓퟑퟖ Germany ퟕ ퟏퟗퟖퟏ ퟐퟎퟏퟑ ퟒퟕ. ퟑퟖퟔ ퟑퟕ. ퟎퟕퟒ ퟕ. ퟒퟕퟔ ퟐ. ퟒퟓퟕ Hungary ퟔ ퟏퟗퟖퟐ ퟐퟎퟎퟗ ퟓퟐ. ퟕퟑퟑ ퟏퟎ. ퟕퟖퟗ ퟕ. ퟐퟕퟐ ퟏퟐ. ퟔퟐퟐ Iceland ퟒ ퟏퟗퟖퟒ ퟐퟎퟎퟗ ퟑퟎ. ퟔퟐퟓ ퟑퟒ. ퟓퟖퟎ ퟑ. ퟏퟏퟓ ퟏퟒ. ퟗퟖퟎ India ퟓ ퟏퟗퟗퟎ ퟐퟎퟏퟒ ퟒퟔ. ퟖퟔퟎ ퟎ. ퟗퟐퟑ ퟒ. ퟎퟎퟎ ퟕ. ퟏퟑퟓ Ireland ퟒ ퟏퟗퟖퟏ ퟐퟎퟎퟖ ퟓퟎ. ퟎퟐퟓ ퟑퟐ. ퟕퟑퟑ ퟖ. ퟗퟑퟑ ퟕ. ퟑퟐퟖ Italy ퟓ ퟏퟗퟖퟏ ퟐퟎퟎퟗ ퟒퟔ. ퟓퟔퟎ ퟑퟐ. ퟑퟒퟎ ퟗ. ퟑퟐퟖ ퟓ. ퟕퟑퟗ Japan ퟔ ퟏퟗퟖퟏ ퟐퟎퟏퟎ ퟒퟐ. ퟒퟓퟎ ퟑퟗ. ퟎퟒퟔ ퟑ. ퟔퟏퟑ ퟏ. ퟎퟐퟖ Latvia ퟒ ퟏퟗퟗퟎ ퟐퟎퟎퟖ ퟓퟏ. ퟓퟐퟓ ퟖ. ퟔퟖퟏ ퟏퟓ. ퟒퟓퟎ ퟏퟔ. ퟎퟗퟒ

23

Table 1: Countries in the Integrated Values Surveys Dataset Coverage Macroeconomic Variables (mean levels across waves) Country Number of Income Real GDP Per Unemployment First Year Latest Year Inflation Rate Waves Inequality Capita Rate Lithuania ퟒ ퟏퟗퟗퟎ ퟐퟎퟎퟖ ퟒퟖ. ퟏퟐퟓ ퟖ. ퟕퟖퟏ ퟏퟐ. ퟔퟗퟖ ퟕ. ퟑퟗퟎ Malta ퟒ ퟏퟗퟖퟑ ퟐퟎퟎퟖ ퟒퟐ. ퟑퟓퟎ ퟏퟒ. ퟖퟎퟎ ퟔ. ퟗퟗퟎ ퟐ. ퟎퟏퟑ Mexico ퟕ ퟏퟗퟖퟏ ퟐퟎퟏퟐ ퟒퟖ. ퟏퟕퟏ ퟖ. ퟒퟏퟓ ퟑ. ퟖퟑퟏ ퟐퟎ. ퟐퟐퟐ Moldova ퟒ ퟏퟗퟗퟔ ퟐퟎퟎퟖ ퟓퟔ. ퟐퟕퟓ ퟏ. ퟏퟐퟑ ퟕ. ퟐퟓퟓ ퟏퟑ. ퟔퟑퟒ Netherlands ퟔ ퟏퟗퟖퟏ ퟐퟎퟏퟐ ퟒퟓ. ퟖퟖퟑ ퟒퟑ. ퟒퟖퟕ ퟓ. ퟒퟏퟑ ퟐ. ퟗퟏퟔ Nigeria ퟒ ퟏퟗퟗퟎ ퟐퟎퟏퟏ ퟒퟓ. ퟑퟐퟓ ퟏ. ퟓퟖퟗ ퟒ. ퟎퟑퟖ ퟐퟒ. ퟒퟗퟑ Norway ퟓ ퟏퟗퟖퟐ ퟐퟎퟎퟖ ퟒퟎ. ퟖퟎퟎ ퟕퟑ. ퟎퟎퟎ ퟑ. ퟔퟏퟎ ퟒ. ퟐퟒퟖ Peru ퟒ ퟏퟗퟗퟔ ퟐퟎퟏퟐ ퟓퟑ. ퟒퟎퟎ ퟒ. ퟎퟏퟑ ퟒ. ퟔퟐퟖ ퟒ. ퟕퟗퟑ Poland ퟕ ퟏퟗퟖퟗ ퟐퟎퟏퟐ ퟒퟕ. ퟕퟓퟕ ퟗ. ퟎퟑퟏ ퟗ. ퟐퟑퟎ ퟏퟏퟖ. ퟗퟎퟎ Romania ퟔ ퟏퟗퟗퟑ ퟐퟎퟏퟐ ퟑퟗ. ퟖퟑퟑ ퟔ. ퟑퟐퟑ ퟔ. ퟔퟓퟓ ퟔퟑ. ퟑퟕퟑ Russia ퟔ ퟏퟗퟗퟎ ퟐퟎퟏퟏ ퟒퟕ. ퟖퟔퟕ ퟖ. ퟖퟕퟑ ퟕ. ퟒퟑퟓ ퟗퟐ. ퟗퟑퟐ Serbia ퟒ ퟏퟗퟗퟔ ퟐퟎퟎퟖ ퟓퟎ. ퟎퟎퟎ ퟒ. ퟎퟔퟎ ퟏퟓ. ퟏퟐퟎ ퟓퟒ. ퟕퟖퟒ Slovakia ퟓ ퟏퟗퟗퟎ ퟐퟎퟎퟖ ퟒퟏ. ퟏퟖퟎ ퟏퟎ. ퟗퟕퟔ ퟏퟎ. ퟐퟖퟖ ퟖ. ퟐퟖퟎ Slovenia ퟔ ퟏퟗퟗퟐ ퟐퟎퟏퟏ ퟑퟗ. ퟎퟔퟕ ퟏퟗ. ퟓퟐퟔ ퟔ. ퟖퟕퟑ ퟑퟗ. ퟗퟏퟒ South Africa ퟔ ퟏퟗퟖퟐ ퟐퟎퟏퟑ ퟔퟓ. ퟕퟏퟕ ퟔ. ퟑퟗퟕ ퟐퟎ. ퟓퟏퟕ ퟖ. ퟕퟑퟓ South Korea ퟓ ퟏퟗퟖퟐ ퟐퟎퟏퟎ ퟑퟏ. ퟗퟔퟎ ퟏퟑ. ퟕퟐퟗ ퟑ. ퟔퟓퟐ ퟓ. ퟏퟎퟔ Spain ퟖ ퟏퟗퟖퟏ ퟐퟎퟏퟏ ퟒퟔ. ퟔퟏퟑ ퟐퟔ. ퟔퟎퟐ ퟏퟓ. ퟒퟏퟏ ퟓ. ퟐퟏퟗ Sweden ퟕ ퟏퟗퟖퟐ ퟐퟎퟏퟏ ퟒퟔ. ퟔퟕퟏ ퟒퟑ. ퟒퟔퟎ ퟔ. ퟓퟏퟕ ퟑ. ퟒퟎퟎ Switzerland ퟒ ퟏퟗퟖퟗ ퟐퟎퟎퟖ ퟒퟎ. ퟐퟎퟎ ퟔퟖ. ퟗퟑퟖ ퟐ. ퟖퟐퟖ ퟏ. ퟕퟖퟓ Turkey ퟔ ퟏퟗퟗퟎ ퟐퟎퟏퟏ ퟒퟓ. ퟒퟖퟑ ퟗ. ퟎퟓퟗ ퟖ. ퟖퟕퟑ ퟑퟔ. ퟎퟗퟎ Ukraine ퟓ ퟏퟗퟗퟔ ퟐퟎퟏퟏ ퟐퟖ. ퟖퟖퟎ ퟐ. ퟔퟔퟐ ퟖ. ퟎퟓퟒ ퟐퟗ. ퟎퟓퟏ United Kingdom ퟕ ퟏퟗퟖퟏ ퟐퟎퟎퟗ ퟓퟏ. ퟒퟓퟕ ퟑퟑ. ퟖퟏퟗ ퟔ. ퟓퟖퟑ ퟒ. ퟐퟐퟗ United States ퟔ ퟏퟗퟖퟐ ퟐퟎퟏퟏ ퟒퟕ. ퟔퟏퟕ ퟒퟎ. ퟖퟑퟑ ퟔ. ퟒퟒퟕ ퟑ. ퟖퟐퟐ

24

Table 2: Number of People per Code Number in the Life Satisfaction Question and Inequality Index, with Upward- and Downward-Looking Status (latest country-wave period) in percent, 휹 = ퟎ – Life Satisfaction + Inequality Country 풅풐풘풏 풖풑 ퟏ ퟐ ퟑ ퟒ ퟓ ퟔ ퟕ ퟖ ퟗ ퟏퟎ 푰푪푭 푰푪푭 Argentina (2013) ퟕ ퟒ ퟏퟏ ퟑퟎ ퟔퟏ ퟏퟏퟎ ퟐퟑퟖ ퟑퟎퟕ ퟏퟒퟗ ퟏퟎퟑ ퟕퟔ. ퟐퟔퟕ ퟔퟗ. ퟎퟗퟐ Australia (2012) ퟏퟒ ퟐퟔ ퟒퟐ ퟔퟏ ퟏퟐퟎ ퟏퟎퟎ ퟐퟐퟔ ퟒퟒퟔ ퟐퟔퟑ ퟏퟔퟒ ퟕퟗ. ퟏퟔퟕ ퟔퟗ. ퟑퟔퟐ Belarus (2011) ퟑퟗ ퟓퟗ ퟏퟒퟖ ퟏퟒퟗ ퟑퟎퟏ ퟐퟏퟖ ퟐퟔퟐ ퟐퟏퟖ ퟕퟖ ퟓퟗ ퟕퟗ. ퟔퟗퟓ ퟕퟖ. ퟏퟓퟐ Belgium (2009) ퟏퟓ ퟏퟎ ퟐퟗ ퟑퟓ ퟗퟎ ퟏퟎퟐ ퟐퟔퟒ ퟓퟏퟒ ퟐퟕퟐ ퟏퟕퟕ ퟕퟔ. ퟒퟕퟕ ퟔퟕ. ퟐퟑퟔ Bulgaria (2008) ퟏퟒퟑ ퟖퟒ ퟏퟒퟎ ퟗퟔ ퟐퟓퟓ ퟏퟑퟏ ퟏퟕퟖ ퟏퟖퟑ ퟏퟎퟎ ퟏퟔퟕ ퟕퟗ. ퟎퟐퟏ ퟕퟕ. ퟔퟕퟓ Canada (2006) ퟏퟐ ퟏퟕ ퟐퟓ ퟓퟐ ퟏퟐퟑ ퟏퟓퟖ ퟑퟔퟏ ퟕퟐퟔ ퟑퟒퟖ ퟑퟑퟓ ퟕퟔ. ퟒퟓퟖ ퟔퟔ. ퟑퟏퟎ Chile (2011) ퟔ ퟗ ퟐퟏ ퟒퟏ ퟖퟎ ퟏퟐퟔ ퟐퟏퟎ ퟐퟔퟏ ퟏퟐퟑ ퟏퟏퟏ ퟕퟖ. ퟕퟖퟐ ퟕퟎ. ퟗퟕퟗ China (2012) ퟐퟕ ퟒퟎ ퟕퟖ ퟏퟑퟐ ퟐퟔퟏ ퟑퟏퟕ ퟒퟎퟖ ퟓퟔퟒ ퟐퟒퟗ ퟏퟕퟔ ퟖퟎ. ퟏퟕퟏ ퟕퟑ. ퟒퟗퟖ Colombia (2012) ퟖ ퟔ ퟏퟔ ퟐퟓ ퟕퟖ ퟔퟔ ퟏퟓퟔ ퟑퟒퟑ ퟐퟒퟕ ퟓퟔퟕ ퟕퟓ. ퟖퟏퟒ ퟓퟓ. ퟎퟔퟐ Czech Republic (2008) ퟐퟑ ퟐퟗ ퟕퟓ ퟏퟎퟑ ퟏퟕퟎ ퟏퟓퟒ ퟐퟖퟗ ퟒퟒퟒ ퟑퟏퟓ ퟐퟎퟓ ퟖퟎ. ퟗퟏퟑ ퟕퟏ. ퟔퟔퟗ Denmark (2008) ퟐퟕ ퟏퟎ ퟏퟐ ퟏퟓ ퟓퟏ ퟒퟑ ퟏퟐퟕ ퟑퟗퟔ ퟑퟕퟐ ퟒퟓퟎ ퟕퟒ. ퟒퟒퟗ ퟓퟕ. ퟗퟔퟐ Estonia (2011) ퟐퟗ ퟒퟕ ퟏퟎퟒ ퟏퟐퟖ ퟐퟓퟏ ퟐퟐퟐ ퟐퟕퟕ ퟐퟖퟕ ퟏퟐퟏ ퟔퟏ ퟖퟎ. ퟑퟕퟓ ퟕퟔ. ퟔퟖퟖ Finland (2009) ퟏퟏ ퟖ ퟑퟐ ퟒퟏ ퟒퟖ ퟔퟎ ퟏퟕퟗ ퟑퟒퟔ ퟐퟕퟔ ퟏퟐퟓ ퟕퟕ. ퟏퟓퟓ ퟔퟔ. ퟕퟐퟐ France (2008) ퟑퟒ ퟐퟏ ퟓퟐ ퟔퟓ ퟏퟔퟓ ퟏퟑퟕ ퟐퟓퟒ ퟒퟐퟕ ퟏퟖퟕ ퟏퟓퟖ ퟕퟗ. ퟕퟔퟓ ퟕퟏ. ퟓퟏퟒ Georgia (2014) ퟖퟓ ퟔퟏ ퟏퟐퟐ ퟏퟏퟔ ퟑퟎퟏ ퟗퟔ ퟏퟔퟎ ퟏퟐퟓ ퟒퟐ ퟗퟐ ퟕퟕ. ퟗퟔퟗ ퟕퟕ. ퟗퟒퟎ Germany (2013) ퟐퟐ ퟐퟑ ퟓퟗ ퟔퟑ ퟏퟕퟑ ퟏퟔퟎ ퟑퟓퟎ ퟔퟒퟎ ퟑퟏퟖ ퟐퟑퟓ ퟕퟖ. ퟓퟏퟖ ퟔퟗ. ퟐퟔퟑ Hungary (2009) ퟑퟎ ퟐퟕ ퟗퟐ ퟏퟎퟐ ퟏퟖퟕ ퟏퟓퟎ ퟏퟓퟔ ퟏퟓퟓ ퟓퟗ ퟒퟒ ퟕퟗ. ퟖퟐퟎ ퟕퟖ. ퟏퟏퟑ Iceland (2009) ퟔ ퟖ ퟔ ퟏퟔ ퟐퟓ ퟑퟖ ퟗퟔ ퟐퟔퟕ ퟐퟏퟖ ퟏퟐퟑ ퟕퟒ. ퟑퟖퟒ ퟔퟑ. ퟒퟎퟎ India (2014) ퟏퟗퟔ ퟐퟒퟐ ퟏퟐퟒ ퟏퟏퟔ ퟏퟔퟓ ퟏퟕퟖ ퟏퟖퟑ ퟐퟑퟖ ퟔퟖ ퟕퟏ ퟕퟔ. ퟎퟕퟔ ퟖퟎ. ퟐퟎퟑ Ireland (2008) ퟗ ퟒ ퟖ ퟐퟐ ퟓퟑ ퟏퟎퟒ ퟏퟒퟗ ퟐퟔퟑ ퟐퟒퟔ ퟏퟓퟏ ퟕퟕ. ퟓퟖퟓ ퟔퟔ. ퟖퟏퟑ Italy (2009) ퟑퟕ ퟏퟖ ퟓퟐ ퟔퟔ ퟏퟒퟖ ퟏퟓퟒ ퟐퟔퟗ ퟑퟒퟒ ퟏퟕퟓ ퟐퟑퟗ ퟖퟎ. ퟕퟐퟒ ퟕퟎ. ퟖퟖퟖ Japan (2010) ퟑퟏ ퟐퟖ ퟖퟓ ퟏퟒퟏ ퟐퟗퟓ ퟐퟔퟗ ퟒퟒퟗ ퟔퟐퟎ ퟐퟕퟏ ퟏퟗퟐ ퟕퟗ. ퟕퟔퟑ ퟕퟐ. ퟗퟕퟖ Latvia (2008) ퟑퟕ ퟑퟑ ퟗퟑ ퟏퟎퟕ ퟐퟐퟎ ퟐퟑퟎ ퟐퟔퟑ ퟑퟎퟒ ퟏퟑퟓ ퟕퟔ ퟖퟎ. ퟓퟎퟔ ퟕퟔ. ퟏퟑퟓ

25

Table 2: Number of People per Code Number in the Life Satisfaction Question and Inequality Index, with Upward- and Downward-Looking Status (latest country-wave period) in percent, 휹 = ퟎ – Life Satisfaction + Inequality Country 풅풐풘풏 풖풑 ퟏ ퟐ ퟑ ퟒ ퟓ ퟔ ퟕ ퟖ ퟗ ퟏퟎ 푰푪푭 푰푪푭 Lithuania (2008) ퟒퟓ ퟒퟕ ퟗퟏ ퟗퟖ ퟐퟐퟗ ퟏퟕퟏ ퟐퟓퟒ ퟐퟔퟖ ퟏퟔퟓ ퟏퟎퟎ ퟖퟏ. ퟏퟔퟖ ퟕퟔ. ퟓퟏퟎ Malta (2008) ퟐퟐ ퟏퟖ ퟑퟎ ퟒퟎ ퟏퟎퟔ ퟗퟖ ퟏퟕퟏ ퟑퟖퟓ ퟐퟒퟎ ퟑퟖퟕ ퟕퟗ. ퟐퟑퟏ ퟔퟑ. ퟒퟒퟑ Mexico (2012) ퟐퟖ ퟐퟓ ퟏퟑ ퟐퟕ ퟖퟐ ퟔퟗ ퟏퟑퟗ ퟒퟐퟎ ퟑퟐퟕ ퟖퟕퟎ ퟕퟒ. ퟏퟎퟔ ퟓퟎ. ퟓퟔퟑ Moldova (2008) ퟗퟓ ퟒퟔ ퟏퟎퟏ ퟗퟖ ퟏퟔퟔ ퟏퟒퟗ ퟐퟎퟕ ퟐퟓퟔ ퟐퟐퟑ ퟏퟖퟔ ퟖퟏ. ퟒퟐퟎ ퟕퟓ. ퟎퟐퟏ Netherlands (2012) ퟒ ퟔ ퟐퟕ ퟑퟒ ퟔퟐ ퟏퟕퟔ ퟒퟗퟓ ퟕퟑퟑ ퟐퟕퟕ ퟔퟖ ퟕퟎ. ퟗퟓퟗ ퟔퟒ. ퟐퟒퟓ Nigeria (2011) ퟔퟗ ퟔퟗ ퟏퟏퟕ ퟏퟔퟖ ퟏퟕퟐ ퟐퟏퟗ ퟑퟒퟑ ퟑퟐퟖ ퟏퟓퟒ ퟏퟐퟎ ퟖퟏ. ퟐퟖퟐ ퟕퟔ. ퟕퟑퟐ Norway (2008) ퟖ ퟔ ퟏퟖ ퟏퟒ ퟓퟏ ퟒퟎ ퟏퟐퟎ ퟑퟑퟔ ퟐퟖퟗ ퟐퟎퟕ ퟕퟓ. ퟏퟎퟔ ퟔퟐ. ퟖퟐퟓ Peru (2012) ퟐퟎ ퟐퟎ ퟐퟓ ퟒퟖ ퟏퟖퟎ ퟏퟓퟕ ퟐퟏퟑ ퟏퟗퟑ ퟏퟎퟎ ퟐퟓퟎ ퟕퟗ. ퟖퟓퟓ ퟔퟗ. ퟑퟎퟕ Poland (2012) ퟐퟐ ퟕ ퟏퟔ ퟑퟕ ퟏퟑퟕ ퟏퟎퟔ ퟏퟕퟎ ퟐퟓퟕ ퟏퟏퟔ ퟗퟕ ퟕퟖ. ퟑퟖퟑ ퟕퟏ. ퟗퟖퟕ Romania (2012) ퟖퟐ ퟑퟒ ퟓퟐ ퟔퟕ ퟐퟎퟎ ퟏퟕퟐ ퟐퟕퟒ ퟐퟗퟕ ퟏퟐퟐ ퟏퟗퟎ ퟖퟎ. ퟎퟔퟐ ퟕퟑ. ퟕퟓퟗ Russia (2011) ퟖퟒ ퟖퟐ ퟏퟒퟏ ퟏퟗퟒ ퟒퟔퟕ ퟑퟔퟏ ퟒퟎퟒ ퟒퟏퟖ ퟏퟓퟕ ퟏퟓퟕ ퟖퟎ. ퟐퟎퟓ ퟕퟕ. ퟏퟏퟖ Serbia (2008) ퟒퟔ ퟒퟎ ퟖퟓ ퟔퟎ ퟏퟖퟑ ퟏퟑퟗ ퟐퟎퟕ ퟑퟏퟒ ퟏퟔퟕ ퟐퟓퟒ ퟖퟏ. ퟓퟑퟑ ퟕퟏ. ퟓퟓퟔ Slovakia (2008) ퟐퟏ ퟐퟕ ퟕퟖ ퟔퟖ ퟏퟔퟗ ퟏퟑퟓ ퟐퟎퟓ ퟑퟓퟑ ퟐퟖퟗ ퟏퟑퟓ ퟖퟎ. ퟖퟐퟔ ퟕퟐ. ퟎퟖퟑ Slovenia (2011) ퟖ ퟖ ퟑퟐ ퟐퟗ ퟏퟐퟔ ퟏퟏퟕ ퟏퟓퟏ ퟐퟖퟏ ퟏퟔퟑ ퟏퟒퟔ ퟕퟗ. ퟏퟔퟏ ퟕퟎ. ퟑퟏퟎ South Africa (2013) ퟏퟒퟓ ퟕퟐ ퟏퟒퟑ ퟐퟐퟐ ퟒퟏퟒ ퟒퟑퟕ ퟔퟐퟗ ퟕퟏퟒ ퟑퟒퟗ ퟑퟗퟔ ퟖퟏ. ퟎퟏퟐ ퟕퟒ. ퟒퟒퟓ South Korea (2010) ퟏퟒ ퟔ ퟓퟕ ퟔퟕ ퟏퟗퟕ ퟏퟔퟒ ퟐퟕퟎ ퟐퟒퟓ ퟏퟏퟒ ퟓퟓ ퟕퟖ. ퟏퟒퟔ ퟕퟒ. ퟒퟖퟑ Spain (2011) ퟏퟏ ퟏퟏ ퟐퟖ ퟓퟔ ퟏퟔퟑ ퟏퟕퟖ ퟑퟎퟏ ퟐퟔퟗ ퟖퟓ ퟔퟔ ퟕퟕ. ퟒퟏퟖ ퟕퟐ. ퟔퟗퟑ Sweden (2011) ퟓ ퟒ ퟐퟓ ퟐퟖ ퟖퟐ ퟖퟎ ퟐퟒퟗ ퟑퟕퟖ ퟐퟏퟐ ퟏퟒퟏ ퟕퟔ. ퟐퟓퟔ ퟔퟕ. ퟗퟑퟑ Switzerland (2008) ퟏퟕ ퟓ ퟐퟕ ퟑퟏ ퟒퟕ ퟓퟏ ퟏퟒퟔ ퟑퟖퟕ ퟑퟏퟑ ퟐퟒퟓ ퟕퟔ. ퟏퟑퟓ ퟔퟑ. ퟓퟓퟐ Turkey (2011) ퟑퟑ ퟐퟓ ퟑퟗ ퟓퟓ ퟏퟑퟔ ퟏퟔퟕ ퟑퟏퟐ ퟑퟔퟓ ퟐퟓퟑ ퟐퟏퟔ ퟖퟎ. ퟑퟑퟎ ퟕퟏ. ퟎퟔퟒ Ukraine (2011) ퟏퟎퟐ ퟓퟔ ퟏퟑퟏ ퟏퟎퟓ ퟐퟒퟕ ퟏퟖퟔ ퟐퟏퟗ ퟐퟓퟎ ퟏퟎퟑ ퟏퟎퟏ ퟕퟗ. ퟗퟖퟒ ퟕퟖ. ퟏퟓퟎ United Kingdom (2009) ퟐퟑ ퟏퟖ ퟒퟓ ퟓퟓ ퟖퟓ ퟏퟐퟒ ퟐퟓퟒ ퟒퟐퟏ ퟐퟗퟒ ퟐퟑퟗ ퟕퟗ. ퟓퟗퟔ ퟔퟖ. ퟒퟔퟏ United States (2011) ퟐퟎ ퟐퟔ ퟔퟎ ퟖퟓ ퟏퟒퟖ ퟏퟓퟕ ퟑퟗퟎ ퟔퟖퟑ ퟒퟒퟖ ퟏퟗퟗ ퟕퟖ. ퟐퟓퟎ ퟔퟖ. ퟕퟔퟏ

26 Macroeconomic Variables

For income inequality data, the present research uses Gini estimates from the latest Standardized World Income Inequality Database (SWIID) assembled by Solt (2016), incorporating data from the World Bank’s PovcalNet, the UN Economic Commission for Latin America and the Caribbean, OECD Income Distribution Database, Eurostat, the Socio-Economic Database for Latin America and the Caribbean generated by CEDLAS and the World Bank, scholarly articles, and various national statistical offices around the world. Motivated by the World Income Inequality Database (WIID) and the Luxembourg Income Study (LIS) data, the SWIID has been widely used by those undertaking cross-country research on income inequality’s causes and consequences (Solt, 2015b). By standardising incomes, the SWIID enhances continuity and comparability of Gini coefficient data over time and across countries without considerable loss of coverage (Solt 2009, 2016), using the LIS dataset as the benchmark. Income inequality, commonly measured by the Gini coefficient ranging from zero (perfect equality) to 100 (perfect inequality), could either be estimated using market (gross) incomes (i.e. before government taxes and transfers), or disposable (net) incomes (i.e. after government taxes and transfers). This study employs the SWIID (version 6.1) with Gini estimates of both gross and net income inequality, covering 192 countries from 1960 to 2014 for 5,119 country-years (Solt, 2016). Whereas net income inequality seems more fitting since it may represent actual income distribution that influences living standards, gross income inequality could better illustrate the income distribution level that a society aspires to eventually attain. In the short run, redistribution is crucial in raising less fortunate people’s incomes, thereby supporting them to benefit from opportunities, which otherwise are only available to those in the upper part of the income stepladder. It could be more sensible to allow the highest number of citizens possible to be given access to the same prospects for advancement, thereby empowering them to earn adequate income levels before redistribution.

For economic growth data, this study obtains from the United Nations Statistics Division’s (UNSD) National Accounts Main Aggregates Database (UN, 2017). Involving over 200 countries and areas, this database offers a series of analytical national accounts tables from 1970 onwards. Most recent available data are put together by collaboration among the Economic Statistics Branch of the UNSD, international statistical agencies, and national statistical services of countries covered in the database. The basis for this databank is the annual gathering of official national accounts data from the United Nations National Accounts Questionnaires, which are done and reported by respective countries themselves to the UNSD. Accordingly, economic growth in this study is measured in terms of real gross domestic product (GDP) per capita. Estimates for this specific variable are standardised based on 2010 US dollar constant prices.

For unemployment and inflation data, the current study employs internationally comparable figures from the World Development Indicators (World Bank, 2017) database. Known as the primary World Bank collection of development indicators, it includes data for up to 56 years – from 1960 to 2016. Gathered from officially recognised international sources, this database presents the most accurate and up-to-date global development data available for national, regional, and global estimates. In the present research,

27 estimates for unemployment, as measured by the proportion of the labour force (i.e. economically active members of the total population that are jobless but available for and seeking employment), are expressed in percentage. Estimates for inflation, as measured by the consumer price index, capture the annual percentage change in the cost to the average consumer of getting a basket of goods and services.

Control Variables

This study controls for specific individual-level characteristics. Current knowledge on SWB is the basis for selecting these control variables, which have been found to have a bearing on individual well-being. Sernau (2014) has found that an individual’s social status may be classified according to two main types of characteristics: 1) ‘assigned’ characteristics; and, 2) ‘achieved’ characteristics. The former refers to ones that have been present at birth or attributed by others, and over which a person has little or no power. Examples usually include biological/physiological sex, eye shape, gender identity, parentage, parental social status, place of birth, sexual orientation, skin colour, and town size. The other type refers to ones that a specific individual chooses or earns by virtue of her abilities. These may encompass, among others, employment status, family size, financial situation, level of education, marital status, and other indicators of merit. Often, someone’s social status has been a mix of assigned and achieved factors. In others, however, assigned statuses alone have been considered in establishing a person’s social status. Thus, little or no mobility is observed. Opportunities for equality remain to be reached.

Whether the size of places (based on population figures) where people live tend to be larger or smaller vis-à-vis the average size is included to capture an aspect of status assigned to them. Other control variables are people’s tendency to report their educational attainment compared to the average level, and the size of their family compared to everyone else’s; these denote traits obtained through personal efforts. Table 3 presents all pertinent variables, along with their corresponding descriptions.

Table 3: Variable Summary Variable Description Life satisfaction inequality index Inequality in life satisfaction (based on upward-looking status) Whether people tend to have more children (1 = more children, 0 = fewer Children children) Whether people tend to have higher educational attainment (1 = attained Education higher level, 0 = attained lower level) Whether people tend to live in larger towns (1 = larger towns, 0 = smaller Town size towns) Gini index (market) Inequality in household market (pre-tax, pre-transfer) income GDP per capita Real GDP at constant 2010 US dollars (thousands) Inflation rate Percent change in average prices Unemployment rate Percent change in unemployment

28 Empirical Strategy

In using the Cowell and Flachaire (2017) status-inequality method, life satisfaction responses are then summarised into life satisfaction distributions in order to determine SWB inequality without resorting to arbitrary cardinalisation of ordinal data. Country-specific differences (i.e. how each person views herself in the distribution in comparison with everyone else with similar or higher status) in self-evaluations in country-wave periods nested within countries are regressed on macroeconomic variables that may be exerting an impact on the distribution of well-being in society. This study employs a multilevel linear regression model to estimate life satisfaction inequality, subject to periodic variations in macroeconomic variables, such as market income inequality, real GDP per capita, and, separately, annual unemployment and inflation rates, with year-fixed effects, and appropriate control variables.

Before the model is laid out, the life satisfaction (풍풊풇풆풔풂풕풊풕풄) of every individual, 풊, during time 풕 in country 풄 are assumed to be shaped by time-invariant individual-level characteristics, 푿풊풕풄 , also comprising both country-wave-level intercept (휶ퟎ풕풄), and random individual-level effect (풆풊풕풄):

풍풊풇풆풔풂풕풊풕풄 = 흁ퟎ풕풄 + 흁ퟏ풕풄 ∗ 푿풊풕풄 + 풆풊풕풄

The model starts from the country-wave level. SWB inequality for a particular period is viewed from the standpoint of a person’s perception of her status within the life satisfaction distribution during time 풕 for country 풄. The country-wave intercept (흁풕풄 ) hinges on a country intercept (흆풄 ), a number of key macroeconomic variables, which involve: market income inequality ( 푮풊풏풊풕풄 ), real GDP per capita

(푮푫푷풑풄풕풄), annual unemployment rate (푼풕풄), and annual inflation late (푰풏풇풕풄) in each country, aggregated time-invariant characteristics of individuals, 푿풕풄, and a random country-wave effect (휺풕풄):

흁풕풄 = 흆풄 + 휸 ∗ 푮풊풏풊풕풄 + 흋 ∗ 푮푫푷풑풄풕풄 + 휽 ∗ 푼풕풄 + 흅 ∗ 푰풏풇풕풄 + 휷 ∗ 푿풕풄 + 휺풕풄

On the country-level, the overall mean intercept (휶), and a random country effect (풖풄) are included:

흆풄 = 휶 + 풖풄

Hence, the baseline regression model takes the following specification, that is:

푳푺풊풏풆풒풕풄 = 휶 + 휸 ∗ 푮풊풏풊풕풄 + 흋 ∗ 푮푫푷풑풄풕풄 + 휽 ∗ 푼풕풄 + 흅 ∗ 푰풏풇풕풄 + 휷 ∗ 푿풕풄 + 휺풕풄 + 풖풄

29 IV. Results

Estimations of the Link Between Life Satisfaction Inequality and Macroeconomic Movements

To examine the relationship between life satisfaction inequality and macroeconomic movements, these country-wave points are regarded as repeated cross-sectional data. The present research offers strong empirical evidence on how income inequality, economic growth, unemployment, and inflation influence the disparity in life satisfaction evaluations in society. Running the multilevel linear regression model, how upward-looking LS inequality is influenced by periodic variations in macroeconomic variables is estimated accordingly. Through this analysis, life satisfaction inequality has an inverse and highly significant relationship with income inequality and economic growth, respectively, whilst it is positively linked to unemployment and inflation. Additionally, the study determines whether each corresponding impact on life satisfaction inequality results from fewer or more reports of ‘extreme’ (i.e. higher and lower categories) life satisfaction evaluations. After merging the 1981-2014 IVS harmonised panel dataset with available macroeconomic estimates, empirical analyses are confined to 235 observations that cover more than three decades across 46 unique countries, having been surveyed for a minimum of four country-wave periods. Below, Table 4 presents standard descriptive statistics of all variables. In the appendix, Table A1 respective correlations of these variables are shown.

Table 4: Key Descriptive Statistics Variable Mean SD Min Max Life satisfaction inequality index ퟕퟏ. ퟎퟏퟎ ퟔ. ퟑퟓퟔ ퟓퟎ. ퟓퟔퟎ ퟖퟏ. ퟕퟎퟎ Children ퟎ. ퟎퟓퟏ ퟎ. ퟐퟐퟏ ퟎ ퟏ Education ퟎ. ퟔퟎퟎ ퟎ. ퟒퟗퟏ ퟎ ퟏ Town size ퟎ. ퟑퟖퟑ ퟎ. ퟒퟖퟕ ퟎ ퟏ Gini index (market) ퟒퟓ. ퟕퟐퟐ ퟕ. ퟎퟕퟏ ퟐퟓ. ퟒퟎퟎ ퟔퟖ. ퟐퟎퟎ GDP per capita ퟐퟏ. ퟑퟔퟗ ퟏퟖ. ퟖퟕퟑ ퟎ. ퟓퟑퟑ ퟗퟏ. ퟒퟏퟒ Inflation rate ퟐퟑ. ퟏퟎퟓ ퟕퟎ. ퟗퟗퟕ −ퟏ. ퟖퟏퟕ ퟕퟎퟎ. ퟎퟎퟎ Unemployment rate ퟕ. ퟗퟔퟏ ퟒ. ퟖퟔퟕ ퟎ. ퟏퟓퟎ ퟐퟓ. ퟑퟕퟎ

30 Impact of Income Inequality on Life Satisfaction Inequality

First, Figure 2 plots LS inequality (푳푺풊풏풆풒풕풄) and income inequality (푮풊풏풊풕풄) for every country-wave period. Estimating without covariates, the coefficient on 푮풊풏풊풕풄 is (as shown in the first column of Tables 5.1 to 5.4) negative and highly statistically significant, establishing the inverse relationship between 푳푺풊풏풆풒풕풄 and 푮풊풏풊풕풄. The next column indicates that this finding is nevertheless robust, whilst controlling for how people tend to report whether they have more or less children, whether they have higher or lower educational attainment, and whether they live in larger or smaller towns. The magnitude of this coefficient exhibits that a percentage point increase in 푮풊풏풊풕풄 is associated with an expected 0.11 reduction in 푳푺풊풏풆풒풕풄 (or a 0.15 percent reduction from the mean of 푳푺풊풏풆풒풕풄), holding all other factors constant. That is, life satisfaction inequality is reduced in country-wave periods with higher income inequality. Remarkably, the data have shown antithetical evidence for Hypothesis 1: whereas incomes are disparately distributed among people, greater income inequality does not inevitably lead to more differences in their life satisfaction responses, especially for those who already live in countries with high-LS inequality levels. Additionally, the same finding is observed in non-western, and low-income inequality level countries. People who are residing in western countries are the exception to this result, since higher income inequality usually means higher life satisfaction inequality.

Impact of Economic Growth on Life Satisfaction Inequality

Second, Figure 3 depicts 푳푺풊풏풆풒풕풄 and economic growth, in terms of real GDP per capita (푮푫푷풑풄풕풄).

Column 1 across Tables 5.1 to 5.4 indicates that the coefficient on 푮푫푷풑풄풕풄 is similarly negative and highly statistically significant, confirming an opposing relationship between 푳푺풊풏풆풒풕풄 and 푮푫푷풑풄풕풄. This result is strengthened even with the inclusion of the same set of controls. This coefficient states that a unit increase in 푮푫푷풑풄풕풄 brings about a 0.17 decline in 푳푺풊풏풆풒풕풄, which is the same as a 0.24 percent reduction from the mean of the outcome variable, keeping all other variables unchanged. Simply, life satisfaction inequality becomes much less apparent in country-wave periods that have greater economic growth. The data have established strong empirical support for Hypothesis 2: when people realise that improving overall economic progress would suggest better individual living standards, differences in their life satisfaction evaluations are minimised. This observation is unmistakably engaging for people in many countries, irrespective of the level of LS inequality, income inequality, and development. Particularly, the extent of narrowing of these life satisfaction gaps is felt the most in countries who are still not members of more advanced economies. In terms of geographical area, only in western countries does higher economic growth significantly diminishes life satisfaction inequality.

31 Impact of Unemployment on Life Satisfaction Inequality

Third, Figure 4 presents 푳푺풊풏풆풒풕풄 and unemployment rate during each country-wave period (푼풕풄). In

Tables 5.1 to 5.4, the first column states that the coefficient on 푼풕풄 is positively influencing 푳푺풊풏풆풒풕풄 and their relationship is highly significant. Adding covariates does not weaken the robustness of this direct association, although there is a slight decrease in its magnitude. This particular coefficient reveals that raising 푼풕풄 by a percentage point causes a 0.16 upsurge in 푳푺풊풏풆풒풕풄, which is practically a 0.22 percent expansion from the average level of gaps in life satisfaction assessments, provided all other things being equal in society. Hence, life satisfaction inequality is increasing when unemployment is going up. As expected, the data offers persuasive and verifiable proof for Hypothesis 3: as individuals become aware of the enormity of the implications of having fewer employment opportunities, life satisfaction levels among people would considerably vary and become more diverse. With or without controls in the model, this finding is statistically significant for people who belong in countries with markedly apparent high LS inequality, and those who find themselves in advanced economies. For western countries, rising unemployment widens life satisfaction inequality after accounting for compositional differences among individuals in terms of family size, educational attainment, and community size. Exclusive of these differences, people living in non-western countries, and counterparts facing low LS inequality and high- income inequality experience higher disparity in life satisfaction.

Impact of Inflation on Life Satisfaction Inequality

Finally, Figure 5 charts 푳푺풊풏풆풒풕풄 and inflation rate (푰풏풇풕풄) in corresponding country-wave intervals. As seen across Tables 5.1 to 5.4, both column 1 (without covariates) and 2 (including covariates) suggest that the coefficient on 푰풏풇풕풄 is positive and highly statistically significant, substantiating the direct relationship between 푳푺풊풏풆풒풕풄 and 푰풏풇풕풄. The magnitude of this coefficient suggests that boosting 푰풏풇풕풄 by a percentage point triggers a 0.02 increase in 푳푺풊풏풆풒풕풄 (i.e. a 0.02 percent growth from the average level of life satisfaction inequality). Thus, the distribution of life satisfaction among people is more likely broadening when inflation is rising. The data have predictably offered empirical support for Hypothesis 4: for those who care about their consumption and production options compared with their access to resources, growth in prices of goods and services would be another constraint to their living standards, whereby distinctly causing more dispersion in how people evaluate their lives. With the attachment of differences (as estimated by individual-level controls), countries that are experiencing low- and high- income inequality, and high-LS inequality are significantly concerned with escalating inflation. Similarly, life satisfaction responses appear to be more unequally distributed in developing and non-western societies. Even though more developed and western countries seem more accustomed to managing inflation shifts, life satisfaction differences remain to be moderated.

32 Figure 2: Life Satisfaction Inequality & Income Inequality Figure 3: Life Satisfaction Inequality & Real GDP Per Capita

Figure 4: Life Satisfaction Inequality & Unemployment Rate Figure 5: Life Satisfaction Inequality & Inflation Rate

33

Table 5.1: Multilevel Linear Estimates of Macroeconomic Movements by Life Satisfaction Inequality Level Dependent Variable: Total Low-LS Inequality High-LS Inequality Life Satisfaction (1) (2) (3) (4) (5) (6) Inequality Income inequality −ퟎ. ퟏퟕퟎ *** −ퟎ. ퟏퟎퟔ ** −0 .011 0. 020 −ퟎ. ퟏퟎퟖ *** −ퟎ. ퟎퟖퟖ ***

(푮풊풏풊풕풄) (ퟎ. ퟎퟓퟗ) (ퟎ. ퟎퟓퟑ) (0.101) (0.093) (ퟎ. ퟎퟑퟎ) (ퟎ. ퟎퟐퟗ) Real GDP per capita −ퟎ. ퟏퟒퟑ*** −ퟎ. ퟏퟔퟕ*** −0.012 −ퟎ. ퟎퟓퟔ** −ퟎ. ퟏퟎퟑ*** −ퟎ. ퟎퟗퟖ***

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟐퟎ) (ퟎ. ퟎퟏퟖ) (0.026) (ퟎ. ퟎퟐퟔ) (ퟎ. ퟎퟐퟏ) (ퟎ. ퟎퟐퟏ) Unemployment rate ퟎ. ퟐퟑퟔ** ퟎ. ퟏퟓퟓ** ퟎ. ퟐퟕퟕ** 0.147 ퟎ. ퟏퟏퟔ** ퟎ. ퟏퟎퟑ**

(푼풕풄) (ퟎ. ퟎퟖퟓ) (ퟎ. ퟎퟕퟔ) (ퟎ. ퟏퟐퟐ) (0.117) (ퟎ. ퟎퟒퟕ) (ퟎ. ퟎퟒퟒ) ퟎ. ퟎퟏퟗ*** ퟎ. ퟎퟏퟔ*** 0.031 0.070 ퟎ. ퟎퟎퟕ** ퟎ. ퟎퟎퟖ** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟓ) (0.055) (0.052) (ퟎ. ퟎퟎퟑ) (ퟎ. ퟎퟎퟑ)

ퟕퟕ. ퟓퟎퟒ*** ퟕퟒ. ퟕퟔퟐ*** ퟔퟒ. ퟑퟖퟏ*** ퟔퟒ. ퟑퟎퟏ*** ퟖퟎ. ퟒퟔퟒ*** ퟕퟖ. ퟓퟎퟗ*** Constant (휶) (ퟐ. ퟖퟐퟖ) (ퟐ. ퟓퟔퟖ) (ퟒ. ퟔퟑퟕ) (ퟒ. ퟐퟒퟎ) (ퟏ. ퟓퟗퟗ) (ퟏ. ퟔퟕퟒ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟏퟐퟒ ퟏퟐퟒ ퟏퟏퟏ ퟏퟏퟏ R-squared ퟎ. ퟒퟎퟑ ퟎ. ퟓퟒퟑ ퟎ. ퟐퟗퟐ ퟎ. ퟒퟐퟕ ퟎ. ퟓퟖퟏ ퟎ. ퟔퟒퟐ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

Table 5.2: Multilevel Linear Estimates of Macroeconomic Movements by Income Inequality Level Dependent Variable: Total Low-Income Inequality High-Income Inequality Life Satisfaction (1) (2) (3) (4) (5) (6) Inequality Income inequality −ퟎ. ퟏퟕퟎ *** −ퟎ. ퟏퟎퟔ ** −ퟎ. ퟐퟔퟔ ** −ퟎ. ퟏퟓퟕ ** −0 .081 −0 .116

(푮풊풏풊풕풄) (ퟎ. ퟎퟓퟗ) (ퟎ. ퟎퟓퟑ) (ퟎ. ퟎퟗퟏ) (ퟎ. ퟎퟕퟕ) (0.161) (0.141) Real GDP per capita −ퟎ. ퟏퟒퟑ*** −ퟎ. ퟏퟔퟕ*** −ퟎ. ퟏퟕퟔ*** −ퟎ. ퟏퟗퟔ*** −ퟎ. ퟎퟔퟒ* −ퟎ. ퟏퟏퟐ***

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟐퟎ) (ퟎ. ퟎퟏퟖ) (ퟎ. ퟎퟐퟒ) (ퟎ. ퟎퟐퟏ) (ퟎ. ퟎퟑퟖ) (ퟎ. ퟎퟑퟒ) Unemployment rate ퟎ. ퟐퟑퟔ** ퟎ. ퟏퟓퟓ** 0.123 0.098 ퟎ. ퟑퟓퟗ** 0.168

(푼풕풄) (ퟎ. ퟎퟖퟓ) (ퟎ. ퟎퟕퟔ) (0.110) (0.091) (ퟎ. ퟏퟐퟒ) (0.114) ퟎ. ퟎퟏퟗ*** ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟏퟓ*** ퟎ. ퟎퟏퟒ*** ퟎ. ퟎퟓퟓ** ퟎ. ퟎퟒퟎ* Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟒ) (ퟎ. ퟎퟐퟓ) (ퟎ. ퟎퟐퟐ)

ퟕퟕ. ퟓퟎퟒ*** ퟕퟒ. ퟕퟔퟐ*** ퟖퟑ. ퟓퟐퟖ*** ퟕퟖ. ퟒퟒퟐ*** ퟔퟗ. ퟓퟐퟕ*** ퟕퟑ. ퟒퟐퟕ*** Constant (휶) (ퟐ. ퟖퟐퟖ) (ퟐ. ퟓퟔퟖ) (ퟑ. ퟖퟑퟏ) (ퟑ. ퟑퟓퟕ) (ퟖ. ퟐퟓퟕ) (ퟕ. ퟐퟗퟖ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟏퟏퟎ ퟏퟏퟎ ퟏퟐퟓ ퟏퟐퟓ R-squared ퟎ. ퟒퟎퟑ ퟎ. ퟓퟒퟑ ퟎ. ퟔퟔퟑ ퟎ. ퟕퟕퟗ ퟎ. ퟑퟖퟓ ퟎ. ퟓퟒퟕ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

34

Table 5.3: Multilevel Linear Estimates of Macroeconomic Movements by Development Status Dependent Variable: Total Non-OECD OECD Life Satisfaction (1) (2) (3) (4) (5) (6) Inequality Income inequality −ퟎ. ퟏퟕퟎ *** −ퟎ. ퟏퟎퟔ ** −0 .148 −0 .099 −0 .047 0. 035

(푮풊풏풊풕풄) (ퟎ. ퟎퟓퟗ) (ퟎ. ퟎퟓퟑ) (0.103) (0.107) (0.083) (0.071) Real GDP per capita −ퟎ. ퟏퟒퟑ*** −ퟎ. ퟏퟔퟕ*** −ퟎ. ퟓퟖퟓ** −ퟎ. ퟒퟎퟔ* −ퟎ. ퟏퟑퟐ*** −ퟎ. ퟏퟔퟗ***

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟐퟎ) (ퟎ. ퟎퟏퟖ) (ퟎ. ퟐퟏퟑ) (ퟎ. ퟐퟑퟗ) (ퟎ. ퟎퟐퟑ) (ퟎ. ퟎퟐퟎ) Unemployment rate ퟎ. ퟐퟑퟔ** ퟎ. ퟏퟓퟓ** 0.178 0.076 ퟎ. ퟑퟏퟖ*** ퟎ. ퟐퟒퟓ**

(푼풕풄) (ퟎ. ퟎퟖퟓ) (ퟎ. ퟎퟕퟔ) (0.155) (0.160) (ퟎ. ퟏퟎퟐ) (ퟎ. ퟎퟖퟖ) ퟎ. ퟎퟏퟗ*** ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟓퟒ*** ퟎ. ퟎퟒퟏ*** 0.010 ퟎ. ퟎퟏퟒ** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟏퟓ) (ퟎ. ퟎퟏퟔ) (0.008) (ퟎ. ퟎퟎퟕ)

ퟕퟕ. ퟓퟎퟒ*** ퟕퟒ. ퟕퟔퟐ*** ퟖퟒ. ퟓퟎퟒ*** ퟕퟗ. ퟑퟒퟏ*** ퟕퟏ. ퟑퟕퟕ*** ퟔퟕ. ퟔퟕퟐ*** Constant (휶) (ퟐ. ퟖퟐퟖ) (ퟐ. ퟓퟔퟖ) (ퟖ. ퟎퟔퟓ) (ퟖ. ퟔퟖퟐ) (ퟑ. ퟖퟓퟒ) (ퟑ. ퟐퟔퟓ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟕퟓ ퟕퟓ ퟏퟔퟎ ퟏퟔퟎ R-squared ퟎ. ퟒퟎퟑ ퟎ. ퟓퟒퟑ ퟎ. ퟓퟖퟗ ퟎ. ퟔퟐퟗ ퟎ. ퟒퟒퟎ ퟎ. ퟔퟐퟏ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

Table 5.4: Multilevel Linear Estimates of Macroeconomic Movements by Geographical Location Dependent Variable: Total Non-West West Life Satisfaction (1) (2) (3) (4) (5) (6) Inequality Income inequality −ퟎ. ퟏퟕퟎ *** −ퟎ. ퟏퟎퟔ ** −ퟎ. ퟐퟔퟓ *** −ퟎ. ퟏퟔퟕ ** ퟎ. ퟏퟗퟎ ** ퟎ. ퟐퟎퟖ **

(푮풊풏풊풕풄) (ퟎ. ퟎퟓퟗ) (ퟎ. ퟎퟓퟑ) (ퟎ. ퟎퟖퟖ) (ퟎ. ퟎퟖퟑ) (ퟎ. ퟎퟖퟔ) (ퟎ. ퟎퟕퟔ) Real GDP per capita −ퟎ. ퟏퟒퟑ*** −ퟎ. ퟏퟔퟕ*** 0.030 −0.082 −ퟎ. ퟐퟎퟎ*** −ퟎ. ퟐퟎퟎ***

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟐퟎ) (ퟎ. ퟎퟏퟖ) (0.080) (0.076) (ퟎ. ퟎퟐퟓ) (ퟎ. ퟎퟐퟐ) Unemployment rate ퟎ. ퟐퟑퟔ** ퟎ. ퟏퟓퟓ** ퟎ. ퟐퟔퟑ* 0.069 0.117 ퟎ. ퟏퟓퟑ*

(푼풕풄) (ퟎ. ퟎퟖퟓ) (ퟎ. ퟎퟕퟔ) (ퟎ. ퟏퟒퟕ) (0.139) (0.103) (ퟎ. ퟎퟗퟐ) ퟎ. ퟎퟏퟗ*** ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟒퟏ** ퟎ. ퟎퟐퟓ* 0.008 ퟎ. ퟎퟏퟏ* Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟏퟓ) (ퟎ. ퟎퟏퟒ) (0.007) (ퟎ. ퟎퟎퟔ) ퟕퟕ. ퟓퟎퟒ *** ퟕퟒ. ퟕퟔퟐ *** ퟕퟖ. ퟗퟓퟖ *** ퟕퟔ. ퟔퟏퟑ *** ퟔퟒ. ퟐퟔퟑ *** ퟔퟏ. ퟖퟑퟐ *** Constant (휶) (ퟐ. ퟖퟐퟖ) (ퟐ. ퟓퟔퟖ) (ퟓ. ퟔퟗퟏ) (ퟓ. ퟐퟐퟓ) (ퟑ. ퟗퟖퟐ) (ퟑ. ퟓퟐퟕ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟏퟎퟎ ퟏퟎퟎ ퟏퟑퟓ ퟏퟑퟓ R-squared ퟎ. ퟒퟎퟑ ퟎ. ퟓퟒퟑ ퟎ. ퟒퟔퟓ ퟎ. ퟓퟗퟎ ퟎ. ퟓퟗퟖ ퟎ. ퟔퟗퟖ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

35 Mechanisms in the Link Between Life Satisfaction Inequality and Macroeconomic Movements

To see whether reductions (increases) in 푳푺풊풏풆풒풕풄 associated with higher 푮풊풏풊풕풄 and 푮푫푷풑풄풕풄 (growing

푼풕풄 and 푰풏풇풕풄) arise from fewer or more extreme life satisfaction reports (i.e. high LS or low LS), the same model is estimated with proportions of those who answered 9 or 10 (풉풊품풉_풔풂풕), and those who stated 1 or 2 (풍풐풘_풔풂풕) on the 10-point scale. First, rising income inequality is associated with a boost in high LS among non–western countries and those that already face relatively starker income gaps; in low–income inequality societies, the likelihood of reporting low LS is much lower. Curiously, those who are in western countries tend to convey more low–LS responses in the face of swelling income disparity. Second: higher economic growth leads to a reduction in low LS in many countries, regardless of LS inequality level, income inequality level, and geographical location. For advanced economies and western societies, greater economic growth is also observed to result in more high–LS evaluations. No matter the income inequality level, LS inequality is decreasing on account of an increase in high LS. Third, soaring unemployment is found to diminish high–LS answers among people facing either high– LS inequality or high–income inequality. Apart from declining high–LS reports due to increased rates of joblessness, people in developed economies also are more likely to convey low LS. Finally, escalating inflation is associated with fewer high–LS reports from people living with high–LS inequality and those belonging in low-income inequality countries. In non–western and developing economies, heightening price growth triggers fewer high LS and more low LS among people. Thus, greater income inequality and economic growth are linked to fewer low–LS views and more high–LS reports, whilst soaring unemployment and inflation are directly related to increasing low–LS views and less high–LS opinions.

Table 6: Multilevel Linear Estimates of Macroeconomic Movements by Life Satisfaction Inequality Level Dependent Variable: Low LS Low LS High LS High LS Extreme LS Responses (1) (2) (3) (4) −ퟎ. ퟏퟔퟓ *** −ퟎ. ퟏퟏퟐ *** ퟎ. ퟒퟎퟐ *** ퟎ. ퟐퟔퟒ ** Income inequality (푮풊풏풊 ) 풕풄 (ퟎ. ퟎퟒퟗ) (ퟎ. ퟎퟒퟔ) (ퟎ. ퟏퟐퟏ) (ퟎ. ퟏퟏퟏ) −ퟎ. ퟏퟒퟒ*** −ퟎ. ퟏퟓퟒ*** ퟎ. ퟐퟑퟖ*** ퟎ. ퟐퟕퟔ*** Real GDP per capita (푮푫푷풑풄 ) 풕풄 (ퟎ. ퟎퟏퟕ) (ퟎ. ퟎퟏퟔ) (ퟎ. ퟎퟒퟐ) (ퟎ. ퟎퟑퟗ) ퟎ. ퟏퟕퟏ** ퟎ. ퟏퟑퟒ** −ퟎ. ퟒퟎퟕ** −ퟎ. ퟐퟕퟕ* Unemployment rate (푼 ) 풕풄 (ퟎ. ퟎퟕퟏ) (ퟎ. ퟎퟔퟕ) (ퟎ. ퟏퟕퟔ) (ퟎ. ퟏퟔퟏ) ퟎ. ퟎퟏퟒ** ퟎ. ퟎퟏퟐ** −ퟎ. ퟎퟑퟔ** −ퟎ. ퟎퟑퟏ** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟏퟑ) (ퟎ. ퟎퟏퟏ)

ퟏퟏ. ퟒퟗퟓ*** ퟖ. ퟗퟗퟎ*** ퟗ. ퟕퟕퟑ*** ퟏퟔ. ퟎퟔퟗ*** Constant (휶) (ퟐ. ퟑퟕퟎ) (ퟐ. ퟐퟓퟐ) (ퟓ. ퟖퟒퟑ) (ퟓ. ퟒퟐퟑ) Includes: Year fixed effects Yes Yes Yes Yes Other covariates++ No Yes No Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ R-squared ퟎ. ퟓퟐퟏ ퟎ. ퟓퟗퟖ ퟎ. ퟑퟒퟏ ퟎ. ퟒퟕퟒ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

36 Table 6.1: Multilevel Linear Estimates of Macroeconomic Movements Where the Outcome Variable is Either Low or High LS Levels (by Life Satisfaction Inequality Level) Low-LS Inequality High-LS Inequality Dependent Variable: Low LS Low LS High LS High LS Low LS Low LS High LS High LS Extreme LS Responses (1) (2) (3) (4) (5) (6) (7) (8) 0. 021 0. 022 0. 146 0. 109 −ퟎ. ퟏퟔퟒ ** −0 .119 ퟎ. ퟑퟐퟓ *** ퟎ. ퟐퟗퟕ *** Income inequality (푮풊풏풊 ) 풕풄 (0.030) (0.029) (0.210) (0.203) (ퟎ. ퟎퟕퟔ) (0.077) (ퟎ. ퟎퟕퟗ) (ퟎ. ퟎퟕퟗ) Real GDP per capita −ퟎ. ퟎퟑퟖ*** −ퟎ. ퟎퟒퟐ*** 0.009 0.076 −ퟎ. ퟐퟒퟎ*** −ퟎ. ퟐퟑퟓ*** −0.034 −0.040

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟎퟖ) (ퟎ. ퟎퟎퟖ) (0.054) (0.056) (ퟎ. ퟎퟓퟒ) (ퟎ. ퟎퟓퟓ) (0.056) (0.056) 0.038 0.044 −ퟎ. ퟒퟒퟑ* −0.300 ퟎ. ퟐퟏퟖ* 0.193 −ퟎ. ퟐퟐퟗ* −ퟎ. ퟐퟎퟒ* Unemployment rate (푼 ) 풕풄 (0.035) (0.036) (ퟎ. ퟐퟓퟐ) (0.255) (ퟎ. ퟏퟏퟗ) (0.117) (ퟎ. ퟏퟐퟒ) (ퟎ. ퟏퟐퟎ) −0.013 −0.004 0.013 −0.058 0.011 0.012 −ퟎ. ퟎퟏퟕ** −ퟎ. ퟎퟏퟗ** Inflation rate (푰풏풇 ) 풕풄 (0.016) (0.016) (0.114) (0.113) (0.008) (0.008) (ퟎ. ퟎퟎퟗ) (ퟎ. ퟎퟎퟖ)

1.214 1.156 ퟑퟏ. ퟒퟑퟔ*** ퟑퟏ. ퟕퟏퟓ*** ퟏퟓ. ퟐퟐퟎ*** ퟏퟐ. ퟎퟓퟒ** 6.399 ퟗ. ퟖퟖퟐ** Constant (휶) (1.351) (1.314) (ퟗ. ퟔퟎퟐ) (ퟗ. ퟐퟓퟒ) (ퟒ. ퟎퟔퟒ) (ퟒ. ퟒퟐퟏ) (4.240) (ퟒ. ퟓퟏퟕ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes No Yes Observations ퟏퟐퟒ ퟏퟐퟒ ퟏퟐퟒ ퟏퟐퟒ ퟏퟏퟏ ퟏퟏퟏ ퟏퟏퟏ ퟏퟏퟏ R-squared ퟎ. ퟒퟑퟗ ퟎ. ퟒퟖퟓ ퟎ. ퟐퟕퟖ ퟎ. ퟑퟓퟎ ퟎ. ퟓퟒퟐ ퟎ. ퟓퟕퟖ ퟎ. ퟒퟒퟏ ퟎ. ퟓퟎퟔ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

Table 6.2: Multilevel Linear Estimates of Macroeconomic Movements Where the Outcome Variable is Either Low or High LS Levels (by Income Inequality Level) Low-Income Inequality High-Income Inequality Dependent Variable: Low LS Low LS High LS High LS Low LS Low LS High LS High LS Extreme LS Responses (1) (2) (3) (4) (5) (6) (7) (8) −ퟎ. ퟑퟑퟏ *** −ퟎ. ퟐퟔퟑ *** ퟎ. ퟔퟐퟏ *** ퟎ. ퟒퟎퟖ ** 0. 060 0. 066 ퟎ. ퟕퟏퟒ ** ퟎ. ퟕퟎퟏ ** Income inequality (푮풊풏풊 ) 풕풄 (ퟎ. ퟎퟗퟏ) (ퟎ. ퟎퟖퟔ) (ퟎ. ퟐퟎퟓ) (ퟎ. ퟏퟖퟓ) (0.123) (0.116) (ퟎ. ퟑퟐퟎ) (ퟎ. ퟐퟗퟓ) Real GDP per capita −ퟎ. ퟏퟒퟑ*** −ퟎ. ퟏퟓퟒ*** ퟎ. ퟑퟎퟑ*** ퟎ. ퟑퟒퟗ*** −ퟎ. ퟏퟎퟑ*** −ퟎ. ퟏퟐퟎ*** ퟎ. ퟏퟑퟗ* ퟎ. ퟏퟗퟓ**

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟐퟒ) (ퟎ. ퟎퟐퟑ) (ퟎ. ퟎퟓퟒ) (ퟎ. ퟎퟓퟎ) (ퟎ. ퟎퟐퟗ) (ퟎ. ퟎퟐퟖ) (ퟎ. ퟎퟕퟓ) (ퟎ. ퟎퟕퟏ) 0.119 0.087 −0.132 −0.099 0.158 0.074 −ퟎ. ퟕퟓퟔ*** −ퟎ. ퟓퟎퟏ** Unemployment rate (푼 ) 풕풄 (0.110) (0.102) (0.247) (0.219) (0.096) (0.093) (ퟎ. ퟐퟒퟖ) (ퟎ. ퟐퟑퟕ) ퟎ. ퟎퟎퟗ* 0.008 −ퟎ. ퟎퟑퟎ** −ퟎ. ퟎퟐퟕ** ퟎ. ퟎퟗퟏ*** ퟎ. ퟎퟖퟐ*** −ퟎ. ퟎퟖퟒ* −0.055 Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟓ) (0.005) (ퟎ. ퟎퟏퟐ) (ퟎ. ퟎퟏퟎ) (ퟎ. ퟎퟏퟗ) (ퟎ. ퟎퟏퟖ) (ퟎ. ퟎퟓퟎ) (0.047)

ퟏퟖ. ퟖퟏퟐ*** ퟏퟔ. ퟎퟐퟖ*** −3.976 6.256 −1.330 −1.077 1.847 −0.203 Constant (휶) (ퟑ. ퟖퟑퟏ) (ퟑ. ퟕퟒퟎ) (8.599) (8.040) (6.352) (5.994) (16.468) (15.266) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes No Yes Observations ퟏퟏퟎ ퟏퟏퟎ ퟏퟏퟎ ퟏퟏퟎ ퟏퟐퟓ ퟏퟐퟓ ퟏퟐퟓ ퟏퟐퟓ R-squared ퟎ. ퟔퟎퟓ ퟎ. ퟔퟕퟖ ퟎ. ퟓퟖퟕ ퟎ. ퟔퟗퟏ ퟎ. ퟓퟗퟑ ퟎ. ퟔퟓퟖ ퟎ. ퟑퟑퟒ ퟎ. ퟒퟔퟎ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

37 Table 6.3: Multilevel Linear Estimates of Macroeconomic Movements Where the Outcome Variable is Either Low or High LS Levels (by Development Status) Non-OECD OECD Dependent Variable: Low LS Low LS High LS High LS Low LS Low LS High LS High LS Extreme LS Responses (1) (2) (3) (4) (5) (6) (7) (8) −0 .114 −0 .021 ퟎ. ퟑퟖퟖ ** 0. 226 0. 015 0. 036 0. 160 −0 .007 Income inequality (푮풊풏풊 ) 풕풄 (0.104) (0.103) (ퟎ. ퟏퟕퟏ) (0.172) (0.050) (0.048) (0.194) (0.172) Real GDP per capita −ퟎ. ퟓퟏퟔ** −0.201 ퟏ. ퟏퟕퟐ*** 0.636 −ퟎ. ퟎퟖퟗ*** −ퟎ. ퟏퟎퟎ*** ퟎ. ퟐퟐퟗ*** ퟎ. ퟑퟎퟔ***

(푮푫푷풑풄풕풄) (ퟎ. ퟐퟏퟓ) (0.229) (ퟎ. ퟑퟓퟑ) (0.382) (ퟎ. ퟎퟏퟒ) (ퟎ. ퟎퟏퟒ) (ퟎ. ퟎퟓퟒ) (ퟎ. ퟎퟒퟗ) 0.132 −0.014 −0.110 0.095 ퟎ. ퟏퟒퟑ** ퟎ. ퟏퟓퟔ** −ퟎ. ퟔퟒퟓ** −ퟎ. ퟓퟎퟗ** Unemployment rate (푼 ) 풕풄 (0.156) (0.153) (0.257) (0.256) (ퟎ. ퟎퟔퟏ) (ퟎ. ퟎퟓퟗ) (ퟎ. ퟐퟑퟕ) (ퟎ. ퟐퟏퟒ) ퟎ. ퟎퟒퟖ*** ퟎ. ퟎퟑퟐ** −ퟎ. ퟏퟎퟑ*** −ퟎ. ퟎퟖퟑ*** 0.002 0.003 −0.017 −0.024 Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟏퟓ) (ퟎ. ퟎퟏퟓ) (ퟎ. ퟎퟐퟒ) (ퟎ. ퟎퟐퟓ) (0.005) (0.005) (0.017) (0.017)

13.068 3.751 −4.807 10.992 2.701 1.287 ퟐퟐ. ퟎퟎퟎ** ퟐퟗ. ퟔퟑퟏ*** Constant (휶) (8.122) (8.328) (13.362) (13.889) (2.291) (2.205) (ퟖ. ퟗퟑퟗ) (ퟕ. ퟗퟔퟕ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes No Yes Observations ퟕퟓ ퟕퟓ ퟕퟓ ퟕퟓ ퟏퟔퟎ ퟏퟔퟎ ퟏퟔퟎ ퟏퟔퟎ R-squared ퟎ. ퟔퟐퟑ ퟎ. ퟔퟗퟏ ퟎ. ퟔퟔퟕ ퟎ. ퟕퟐퟎ ퟎ. ퟓퟎퟑ ퟎ. ퟓퟔퟒ ퟎ. ퟑퟏퟓ ퟎ. ퟒퟖퟔ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

Table 6.4: Multilevel Linear Estimates of Macroeconomic Movements Where the Outcome Variable is Either Low or High LS Levels (by Geographical Location) Non-West West Dependent Variable: Low LS Low LS High LS High LS Low LS Low LS High LS High LS Extreme LS Responses (1) (2) (3) (4) (5) (6) (7) (8) −ퟎ. ퟏퟖퟓ ** −0 .087 ퟎ. ퟔퟏퟐ *** ퟎ. ퟑퟓퟗ ** ퟎ. ퟏퟐퟎ ** ퟎ. ퟏퟐퟓ ** −ퟎ. ퟑퟕퟖ ** −ퟎ. ퟒퟏퟓ ** Income inequality (푮풊풏풊 ) 풕풄 (ퟎ. ퟎퟖퟎ) (0.077) (ퟎ. ퟏퟔퟓ) (ퟎ. ퟏퟒퟖ) (ퟎ. ퟎퟓퟗ) (ퟎ. ퟎퟓퟒ) (ퟎ. ퟏퟖퟖ) (ퟎ. ퟏퟕퟒ) Real GDP per capita −ퟎ. ퟏퟓퟎ** −ퟎ. ퟐퟑퟏ*** −0.212 0.001 −ퟎ. ퟎퟗퟑ*** −ퟎ. ퟎퟗퟒ*** ퟎ. ퟑퟖퟓ*** ퟎ. ퟑퟖퟓ***

(푮푫푷풑풄풕풄) (ퟎ. ퟎퟕퟑ) (ퟎ. ퟎퟕퟏ) (0.150) (0.135) (ퟎ. ퟎퟏퟕ) (ퟎ. ퟎퟏퟔ) (ퟎ. ퟎퟓퟓ) (ퟎ. ퟎퟓퟏ) 0.215 0.053 −0.328 0.081 0.099 ퟎ. ퟏퟐퟖ* −0.257 −0.307 Unemployment rate (푼 ) 풕풄 (0.134) (0.130) (0.277) (0.248) (0.071) (ퟎ. ퟎퟔퟔ) (0.224) (0.210) ퟎ. ퟎퟓퟎ*** ퟎ. ퟎퟑퟗ** −ퟎ. ퟎퟕퟖ** −ퟎ. ퟎퟒퟔ* 0.001 0.003 −0.013 −0.018 Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟏퟒ) (ퟎ. ퟎퟏퟑ) (ퟎ. ퟎퟐퟗ) (ퟎ. ퟎퟐퟓ) (0.005) (0.005) (0.016) (0.014)

ퟏퟏ. ퟏퟗퟒ** ퟗ. ퟔퟑퟗ* 7.469 14.463 −1.396 −2.738 ퟑퟖ. ퟔퟖퟐ*** ퟒퟑ. ퟏퟗퟏ*** Constant (휶) (ퟓ. ퟏퟖퟑ) (ퟒ. ퟖퟔퟕ) (10.695) (9.321) (2.749) (2.530) (ퟖ. ퟔퟔퟕ) (ퟖ. ퟎퟖퟕ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Other covariates++ No Yes No Yes No Yes No Yes Observations ퟏퟎퟎ ퟏퟎퟎ ퟏퟎퟎ ퟏퟎퟎ ퟏퟑퟓ ퟏퟑퟓ ퟏퟑퟓ ퟏퟑퟓ R-squared ퟎ. ퟓퟖퟕ ퟎ. ퟔퟔퟗ ퟎ. ퟒퟕퟔ ퟎ. ퟔퟑퟖ ퟎ. ퟓퟎퟗ ퟎ. ퟔퟎퟐ ퟎ. ퟓퟓퟒ ퟎ. ퟔퟐퟗ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

38 Robustness Checks

In a sequence of robustness checks (available in Tables A2 to A5) in the appendix, the present study confirms those findings pertaining to the negative effects of income inequality and economic growths, as well as positive effects of unemployment and inflation, are robust to alternate empirical methodologies. Utilising different measures of income inequality, such as the Gini coefficient based on disposable (post- tax, post-transfer) income, the share of (post-tax, post-transfer) income held by either the top 10%, or top 20% of income earners, and the Gini coefficient (based on disposable income) from another income inequality database, the World Income Inequality Database, version 3.4 (WIID 3.4) released in January 2017 by the United Nations University’s World Institute for Development Economics Research (UNU- WIDER, 2017). Similar results are replicated when using the natural logarithm of LS inequality as the outcome variable. Thus, the strength of the empirical model employed in this paper seems persuasive to thrive across a gamut of specifications.

Linearity in the impact of income inequality and economic growth on LS inequality has been verified. Nonlinearities in the income inequality and economic growth variables have been modeled as alternative specifications by getting their log-linear transformations, respectively. Also, income inequality is captured by the Gini coefficient as dummy variables with a specific range (18-35, 35-50, 50-70). To see whether economic growth has a curvilinear relationship with LS inequality, real GDP per capita squared is included along with its main effect. Whereas it turns out that the quadratic economic growth term is not significant, decreasing LS inequality is still associated with better economic prospects. Remarkably, the natural log of economic growth also contributes to minimised differences in life satisfaction. Therefore, the linearity of economic growth having an inverse link with LS inequality holds up quite efficiently, as in the original model. For the Gini coefficient, those range dummies confirm that increasing income differences lower LS inequality. Although the log-linear version (before- and after- taxes and transfers) of the Gini coefficient is significant, this result does not essentially deviate from the original interpretation of the highly significant and negative effect of income inequality on LS inequality. Hence, the present research maintains the linearity of the Gini coefficient.

39 V. Discussion of Findings

Whereas most SWB studies in economic research have been exploring the causes and consequences of overall happiness across societies, this paper recalibrates the discussion by highlighting the significance of variations/inequalities in life satisfaction. The present research investigates the macroeconomics of life satisfaction inequality. Using internationally comparable upward-looking life satisfaction inequality and measures of macroeconomic shifts within a nested cross-country dataset, robust and conclusive empirical evidence (over more than thirty years across 46 countries) is provided to highlight the fundamental role of macroeconomic movements in shaping perceived differences in life satisfaction: 1) income inequality and economic growth assert negative and highly significant effects; whilst 2) unemployment and inflation impart positive and highly significant impressions on the distribution of life satisfaction in society. Findings in this research adequately bolster the substance of how changes in the macroeconomic landscape have a bearing on shaping happiness on the aggregate level. These results also closely explore how these disparities in well-being are adversely or favorably motivated by extreme life satisfaction responses from either side of the spectrum due to macroeconomic fluctuations.

This study makes a twofold contribution to the literature on the effects of macroeconomic movements on well-being. First, using the Cowell-Flachaire status-inequality method, this paper constructs the first multi-country estimate of LS inequality, which could be a more suitable measure of overall welfare. In other words, the distribution of life satisfaction illustrates a more inclusive and substantial investigation of the impacts of any inequalities that may spring from macroeconomic shifts in society. Second: simultaneously exploring how these macroeconomic inequalities across different groupings, this study enriches the ‘essence’ of the misery index by providing some insights that ought to refocus the criteria for optimising overall welfare according to a country’s life satisfaction inequality level, income inequality level, development status, or geographic area. These groupings, plausibly, inform the current set of economic, social, political, and cultural factors that influence public policy. Moreover, these determine relevant policies that work towards attaining the ‘ideal’ LS inequality for a specific context.

Moderating Effects of Macroeconomic Movements on Life Satisfaction Inequality

Contrary to Hypothesis 1, this paper finds that individuals are more likely to have less dispersed life satisfaction assessments when income differences become more palpable. For people in non–western countries and those currently tackling high–income inequality, they have a tendency to report higher life satisfaction. Their counterparts living in western countries and those already managing quite well with lower income inequality are less likely to report low life satisfaction. An intuitive reason for decreasing LS inequality associated with higher income inequality is that people are more likely to consider these growing income gaps as indications of improvement in social standing and better life prospects. Indeed, the increase in high–LS responses and the respective decline in low–LS reports enable the distribution of well-being to become less unequal in society.

40

That higher economic growth has a significantly positive impact on well-being has been extensively covered in existing economic literature. The present study offers empirical evidence for Hypothesis 2 in that individuals are more inclined to have less dissimilar life satisfaction evaluations in the face of enhanced economic expansion. No matter the geographic location, and degrees of income inequality or LS inequality: people are more likely to have higher life satisfaction, whereby reducing the probability of reporting low life satisfaction levels. In western countries and those experiencing greater intensities of economic development, high LS responses are becoming more ubiquitous. Undeniably, people are more likely to perceive rising and sustained economic progress as a credible signal for leading happier lives. Due to more prevalent high–LS and fewer low–LS opinions, well-being disparity is greatly curbed.

Based on current economic studies, soaring unemployment has undesirable effects on well-being. This paper delivers convincing empirical reinforcement for Hypothesis 3; when individuals recognise the repercussions of fewer job opportunities, they tend to have more diverse life satisfaction judgments. This finding not only affects those people who are personally forced out of their careers, but also those who are familiar with these displaced workers and everyone else who still have their sources of income. For the former group, being unemployed erodes possibilities for security and stability. For the latter, increasing unemployment endangers their potential to get the most out of life. In countries with either high–income inequality or high–LS inequality, people are less likely to communicate high–LS responses as they are, respectively, coping with more ostensible differences in life domains directly affecting their overall life satisfaction. People in advanced economies are more likely to declare low–LS appraisals, along with decreasing high–LS answers, as they are better equipped to contrast life with work and one that has joblessness in it. Because of declining high–LS answers and more noticeable low–LS evaluations, life satisfaction differences are further broadened.

From established economic research, rising inflation implies lower wages, worse living standards, and increased economic and political instability, and more investment risks. In turn, these dire consequences obviously dampen well-being. The present research uncovers empirical proof for Hypothesis 4 in that differences in life satisfaction are heightened with persistent growth in prices of goods and services. For those experiencing low–income inequality and high–LS inequality in society, the likelihood of reporting high–LS opinions is lower. In developing and non–western countries, people either react with increasing low–LS reports or fewer high–LS assessments, since the aforesaid costs of inflation are more likely exaggerated. Predictably, individuals view higher inflation as another challenge to fulfill their needs and desires. On account of the increase in low–LS assessments and continued waning of high–LS statements from people, differences in well-being turn out to be more pronounced.

41 Policy Implications

Based on strong empirical evidence in this paper, LS inequality in any country is apparently a more comprehensive proxy for aggregate well-being, in that it is able to encapsulate the reality that there are winners and losers in every macroeconomic movement. The reason why policymakers ought to devote more serious consideration and time looking at the relationship between life satisfaction inequality and these shifts is that this stipulates a compelling panorama on which to chronicle and appreciate both economic and non-economic changes that transpire in these macroeconomic fluctuations. Beyond the salient linkages, macroeconomic inequalities induce specific obstacles for other vital non-monetary life domains (e.g. education, health, personal relationships, community, politics, etc.) that shape well-being. When it comes to policy implications, estimating the distribution of people’s life satisfaction suggests that governments should focus not on raising absolute happiness levels but on alleviating inequality in happiness in society. Intuitively, policy targeting LS inequality levels needs to prudently consider the efficient (i.e. making thorough use of scarce resources) and pragmatic (i.e. making decisions and actions that are suitable in practice) coordination of various economic levers. Whether these policy interventions stimulate high LS or suppress low LS responses, achieving the ‘ideal’ levels of LS inequality could potentially optimise the number of winners whilst having a relatively smaller group of losers.

42 Table A1: Correlations Life Gini satisfaction GDP per Unemployment Inflation Variable index Children Education Town size inequality capita rate rate (market) index Life satisfaction inequality index ퟏ. ퟎퟎퟎ Gini index (market) −ퟎ. ퟎퟓퟒ ퟏ. ퟎퟎퟎ GDP per capita −ퟎ. ퟒퟗퟖ −ퟎ. ퟎퟔퟑ ퟏ. ퟎퟎퟎ Unemployment rate ퟎ. ퟐퟐퟓ ퟎ. ퟒퟐퟔ −ퟎ. ퟐퟑퟑ ퟏ. ퟎퟎퟎ Inflation rate ퟎ. ퟐퟐퟕ −ퟎ. ퟏퟕퟒ −ퟎ. ퟐퟐퟎ −ퟎ. ퟎퟒퟖ ퟏ. ퟎퟎퟎ Children −ퟎ. ퟎퟗퟗ −ퟎ. ퟎퟒퟓ −ퟎ. ퟏퟔퟗ −ퟎ. ퟎퟗퟒ −ퟎ. ퟎퟐퟕ ퟏ. ퟎퟎퟎ Education ퟎ. ퟐퟗퟓ −ퟎ. ퟎퟔퟖ ퟎ. ퟎퟐퟒ ퟎ. ퟎퟎퟓ −ퟎ. ퟎퟕퟎ −ퟎ. ퟎퟒퟕ ퟏ. ퟎퟎퟎ Town size −ퟎ. ퟏퟒퟒ ퟎ. ퟎퟗퟐ −ퟎ. ퟎퟕퟔ ퟎ. ퟎퟑퟑ ퟎ. ퟎퟎퟏ −ퟎ. ퟎퟔퟒ −ퟎ. ퟎퟕퟐ ퟏ. ퟎퟎퟎ

43 Table A2: Robustness Checks: Different Income Inequality Measures Dependent Variable: LS Inequality (1) (2) (3) (4) (5) Income inequality Disposable income inequality (SWIID) −ퟎ. ퟏퟖퟕ***

(푮풊풏풊_풅풊풔풑풕풄) (ퟎ. ퟎퟒퟔ) Disposable income inequality (WIID) −ퟎ. ퟏퟗퟒ***

(푮풊풏풊_풅풊풔풑_풘풊풊풅풕풄) (ퟎ. ퟎퟒퟏ) Share of top 10% of income earners in −ퟎ. ퟐퟖퟑ***

the country (푺풉풂풓풆풕풐풑ퟏퟎ풕풄) (ퟎ. ퟎퟒퟗ) Share of top 20% of income earners in −ퟎ. ퟐퟕퟐ***

the country (푺풉풂풓풆풕풐풑ퟐퟎ풕풄) (ퟎ. ퟎퟒퟔ) Market Income Inequality −ퟐ. ퟏퟓퟔ*

(ퟏ 풊풇 ퟑퟓ < 푮풊풏풊 ≤ ퟓퟎ, ퟎ 풐풕풉풆풓풘풊풔풆) (ퟏ. ퟐퟑퟕ) Market Income Inequality −ퟐ. ퟕퟓퟕ*

(ퟏ 풊풇 ퟓퟎ < 푮풊풏풊 ≤ ퟕퟎ, ퟎ 풐풕풉풆풓풘풊풔풆) (ퟏ. ퟒퟐퟎ)

−ퟎ. ퟐퟎퟐ*** −ퟎ. ퟐퟎퟓ*** −ퟎ. ퟐퟎퟕ*** −ퟎ. ퟐퟏퟏ*** −ퟎ. ퟏퟔퟑ*** Real GDP per capita (푮푫푷풑풄 ) 풕풄 (ퟎ. ퟎퟐퟎ) (ퟎ. ퟎퟏퟗ) (ퟎ. ퟎퟏퟗ) (ퟎ. ퟎퟏퟗ) (ퟎ. ퟎퟏퟗ) ퟎ. ퟏퟓퟓ** ퟎ. ퟏퟓퟑ** ퟎ. ퟏퟔퟒ** ퟎ. ퟏퟔퟒ** ퟎ. ퟏퟑퟐ* Unemployment rate (푼 ) 풕풄 (ퟎ. ퟎퟔퟗ) (ퟎ. ퟎퟔퟕ) (ퟎ. ퟎퟔퟔ) (ퟎ. ퟎퟔퟓ) (ퟎ. ퟎퟕퟑ) ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟏퟓ** ퟎ. ퟎퟏퟒ** ퟎ. ퟎퟏퟓ*** ퟎ. ퟎퟏퟔ*** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) ퟕퟔ. ퟑퟔퟎ *** ퟕퟔ. ퟖퟎퟖ *** ퟕퟕ. ퟔퟎퟕ *** ퟖퟏ. ퟒퟑퟑ *** ퟕퟐ. ퟒퟖퟔ *** Constant (휶) (ퟐ. ퟎퟓퟐ) (ퟏ. ퟗퟔퟒ) (ퟏ. ퟖퟖퟒ) (ퟐ. ퟑퟑퟎ) (ퟏ. ퟖퟖퟐ) Includes: Year fixed effects Yes Yes Yes Yes Yes Other covariates++ Yes Yes Yes Yes Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ R-squared ퟎ. ퟓퟔퟗ ퟎ. ퟓퟖퟏ ퟎ. ퟔퟎퟎ ퟎ. ퟔퟎퟐ ퟎ. ퟓퟒퟐ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

44 Table A3: Robustness Checks: Non-Linear Income Inequality and Economic Growth Measures Dependent Variable: LS Inequality (1) (2) (3) (4) (5) (6) Income inequality Natural log of market income inequality −ퟒ. ퟗퟏퟏ** −ퟒ. ퟓퟒퟎ**

(SWIID) (풍풏푮풊풏풊_풎풌풕풕풄) (ퟐ. ퟑퟒퟒ) (ퟐ. ퟐퟓퟒ) Natural log of disposable income −ퟓ. ퟐퟑퟑ*** −ퟓ. ퟓퟗퟒ***

inequality (SWIID) (풍풏푮풊풏풊_풅풊풔풑풕풄) (ퟏ. ퟕퟒퟎ) (ퟏ. ퟔퟖퟐ) Natural log of disposable income −ퟎ. ퟏퟕퟑ*** −ퟎ. ퟐퟎퟗ*** inequality (WIID) (ퟎ. ퟎퟒퟑ) (ퟎ. ퟎퟒퟐ) (풍풏푮풊풏풊_풅풊풔풑_풘풊풊풅풕풄)

Economic Growth Natural Log of Real GDP per capita −ퟐ. ퟒퟖퟏ*** −ퟑ. ퟎퟐퟑ*** −ퟑ. ퟎퟕퟕ***

(풍풏푮푫푷풑풄풕풄) (ퟎ. ퟑퟐퟗ) (ퟎ. ퟑퟔퟒ) (ퟎ. ퟑퟒퟖ) −ퟎ. ퟏퟔퟖ*** −ퟎ. ퟐퟒퟕ*** −ퟎ. ퟐퟕퟕ*** Real GDP per capita (푮푫푷풑풄 ) 풕풄 (ퟎ. ퟎퟓퟒ) (ퟎ. ퟎퟓퟕ) (ퟎ. ퟎퟓퟓ) Real GDP per capita squared −0.000 −0.001 −0.001 ퟐ (푮푫푷풑풄풕풄) (0.001) (0.001) (0.0001)

ퟎ. ퟐퟑퟓ*** ퟎ. ퟐퟑퟒ*** ퟎ. ퟐퟒퟎ*** ퟎ. ퟏퟓퟐ** ퟎ. ퟏퟓퟑ** ퟎ. ퟏퟔퟓ** Unemployment rate (푼 ) 풕풄 (ퟎ. ퟎퟕퟕ) (ퟎ. ퟎퟕퟑ) (ퟎ. ퟎퟕퟏ) (ퟎ. ퟎퟕퟓ) (ퟎ. ퟎퟕퟏ) (ퟎ. ퟎퟔퟖ) ퟎ. ퟎퟏퟖ*** ퟎ. ퟎퟏퟖ*** ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟏퟔ*** ퟎ. ퟎퟏퟑ** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟔ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) ퟗퟏ. ퟗퟑퟑ *** ퟗퟐ. ퟓퟔퟗ *** ퟖퟎ. ퟑퟓퟑ *** ퟖퟕ. ퟑퟏퟖ *** ퟗퟎ. ퟐퟗퟔ *** ퟕퟖ. ퟑퟏퟔ *** Constant (휶) (ퟖ. ퟕퟓퟗ) (ퟔ. ퟒퟒퟒ) (ퟐ. ퟒퟏퟖ) (ퟖ. ퟒퟎퟖ) (ퟔ. ퟏퟏퟗ) (ퟐ. ퟐퟑퟕ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ Yes Yes Yes Yes Yes Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ R-squared ퟎ. ퟓퟎퟐ ퟎ. ퟓퟏퟑ ퟎ. ퟓퟐퟗ ퟎ. ퟓퟒퟑ ퟎ. ퟓퟓퟖ ퟎ. ퟓퟖퟓ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

45 Table A4: Robustness Checks: Non-Linear Life Satisfaction Inequality as Outcome Variable Dependent Variable: Log of LS Inequality (1) (2) (3) (4) (5) (6) Income inequality Market income inequality (SWIID) −ퟎ. ퟎퟎퟏ**

(푮풊풏풊_풎풌풕풕풄) (ퟎ. ퟎퟎퟏ) Disposable income inequality (SWIID) −ퟎ. ퟎퟎퟐ***

(푮풊풏풊_풅풊풔풑풕풄) (ퟎ. ퟎퟎퟏ) Disposable income inequality (WIID) −ퟎ. ퟎퟎퟑ***

(푮풊풏풊_풅풊풔풑_풘풊풊풅풕풄) (ퟎ. ퟎퟎퟎퟏ) Share of top 10% of income earners in −ퟎ. ퟎퟎퟒ***

the country (푺풉풂풓풆풕풐풑ퟏퟎ풕풄) (ퟎ. ퟎퟎퟏ) Share of top 20% of income earners in −ퟎ. ퟎퟎퟒ***

the country (푺풉풂풓풆풕풐풑ퟐퟎ풕풄) (ퟎ. ퟎퟎퟏ) Market Income Inequality −0.028

(ퟏ 풊풇 ퟑퟓ < 푮풊풏풊 ≤ ퟓퟎ, ퟎ 풐풕풉풆풓풘풊풔풆) (0.019) Market Income Inequality −ퟎ. ퟎퟑퟕ*

(ퟏ 풊풇 ퟓퟎ < 푮풊풏풊 ≤ ퟕퟎ, ퟎ 풐풕풉풆풓풘풊풔풆) (ퟎ. ퟎퟐퟏ) −ퟎ. ퟎퟎퟐ*** −ퟎ. ퟎퟎퟑ*** −ퟎ. ퟎퟎퟑ*** −ퟎ. ퟎퟎퟑ*** −ퟎ. ퟎퟎퟑ*** −ퟎ. ퟎퟎퟐ*** Real GDP per capita (푮푫푷풑풄 ) 풕풄 (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) ퟎ. ퟎퟎퟐ* ퟎ. ퟎퟎퟐ** ퟎ. ퟎퟎퟐ** ퟎ. ퟎퟎퟐ** ퟎ. ퟎퟎퟐ** ퟎ. ퟎퟎퟐ* Unemployment rate (푼 ) 풕풄 (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) ퟎ. ퟎퟎퟎퟐ** ퟎ. ퟎퟎퟎퟐ*** ퟎ. ퟎퟎퟎퟐ*** ퟎ. ퟎퟎퟎퟐ** ퟎ. ퟎퟎퟎퟐ** ퟎ. ퟎퟎퟎퟐ** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) ퟒ. ퟑퟏퟎ *** ퟒ. ퟑퟑퟕ *** ퟒ. ퟑퟒퟔ *** ퟒ. ퟑퟓퟕ *** ퟒ. ퟒퟏퟒ *** ퟒ. ퟐퟕퟖ *** Constant (휶) (ퟎ. ퟎퟑퟗ) (ퟎ. ퟎퟑퟏ) (ퟎ. ퟎퟐퟗ) (ퟎ. ퟎퟐퟖ) (ퟎ. ퟎퟑퟓ) (ퟎ. ퟎퟐퟖ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ Yes Yes Yes Yes Yes Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ R-squared ퟎ. ퟓퟏퟖ ퟎ. ퟓퟒퟔ ퟎ. ퟓퟔퟎ ퟎ. ퟓퟕퟖ ퟎ. ퟓퟖퟎ ퟎ. ퟓퟏퟖ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

46 Table A5: Robustness Checks: Non-Linear Life Satisfaction Inequality as Outcome Variable & Non-Linear Income Inequality and Economic Growth Measures Dependent Variable: Log of LS Inequality (1) (2) (3) (4) (5) (6) Income inequality Natural log of market income inequality −ퟎ. ퟎퟔퟕ* −ퟎ. ퟎퟔퟐ*

(SWIID) (풍풏푮풊풏풊_풎풌풕풕풄) (ퟎ. ퟎퟑퟓ) (ퟎ. ퟎퟑퟒ) Natural log of disposable income −ퟎ. ퟗퟕퟔ*** −ퟎ. ퟎퟖퟎ***

inequality (SWIID) (풍풏푮풊풏풊_풅풊풔풑풕풄) (ퟎ. ퟎퟐퟔ) (ퟎ. ퟎퟐퟓ) Natural log of disposable income −ퟎ. ퟎퟎퟑ*** −ퟎ. ퟎퟎퟑ*** inequality (WIID) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (풍풏푮풊풏풊_풅풊풔풑_풘풊풊풅풕풄)

Economic Growth Natural Log of Real GDP per capita −ퟎ. ퟎퟑퟓ*** −ퟎ. ퟎퟒퟐ*** −ퟎ. ퟎퟒퟑ***

(풍풏푮푫푷풑풄풕풄) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) (ퟎ. ퟎퟎퟓ) −ퟎ. ퟎퟎퟐ** −ퟎ. ퟎퟎퟑ*** −ퟎ. ퟎퟎퟒ*** Real GDP per capita (푮푫푷풑풄 ) 풕풄 (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) Real GDP per capita squared −2.500 −8.280 −0.00001 ퟐ (푮푫푷풑풄풕풄) (0.000) (0.000) (0.00)

ퟎ. ퟎퟎퟑ*** ퟎ. ퟎퟎퟑ*** ퟎ. ퟎퟎퟑ*** ퟎ. ퟎퟎퟐ* ퟎ. ퟎퟎퟐ** ퟎ. ퟎퟎퟐ** Unemployment rate (푼 ) 풕풄 (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) (ퟎ. ퟎퟎퟏ) ퟎ. ퟎퟎퟎퟑ*** ퟎ. ퟎퟎퟎퟑ*** ퟎ. ퟎퟎퟎퟐ*** ퟎ. ퟎퟎퟎퟐ** ퟎ. ퟎퟎퟎퟐ*** ퟎ. ퟎퟎퟎퟐ** Inflation rate (푰풏풇 ) 풕풄 (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) (ퟎ. ퟎퟎퟎ) ퟒ. ퟓퟒퟒ *** ퟒ. ퟓퟔퟖ *** ퟒ. ퟑퟗퟓ *** ퟒ. ퟒퟕퟖ *** ퟒ. ퟓퟑퟏ *** ퟒ. ퟑퟔퟒ *** Constant (휶) (ퟎ. ퟏퟑퟏ) (ퟎ. ퟎퟗퟕ) (ퟎ. ퟎퟑퟔ) (ퟎ. ퟏퟐퟔ) (ퟎ. ퟎퟗퟐ) (ퟎ. ퟎퟑퟒ) Includes: Year fixed effects Yes Yes Yes Yes Yes Yes Other covariates++ Yes Yes Yes Yes Yes Yes Observations ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ ퟐퟑퟓ R-squared ퟎ. ퟒퟕퟖ ퟎ. ퟒퟗퟎ ퟎ. ퟓퟎퟖ ퟎ. ퟓퟏퟗ ퟎ. ퟓퟑퟒ ퟎ. ퟓퟔퟑ Notes: * p<0.1, ** p<0.05, *** p<0.01; standard errors are in parentheses. ++Include children, education, and town size.

47 References

Abul Naga, R. and Yalcin, T. (2010). Median Independent Inequality Orderings. SIRE Discussion Papers, No. 118. Scotland, UK: Scottish Institute for Research in Economics.

Abul Naga, R. and Yalcin, T. (2008). Inequality Measurement for Ordered Response Health Data. Journal of Health Economics: 27(6), 1614 – 1625.

Alesina, A., Di Tella, R., and MacCulloch, R. (2004). Inequality and Happiness: Are Europeans and Americans Different? Journal of Public Economics: 88(9-10), 2009–2042.

Allison, R. and Foster, J. (2004). Measuring Health Inequality Using Qualitative Data. Journal of Health Economics: 23(3), 505 – 524.

Andrews, F. and Withey, S. (1976). Social Indicators of Well-Being. New York, New York: Plenum Press.

Apouey, B. (2007). Measuring Health Polarization with Self-Assessed Health Data. Health Economics: 16(9), 875–894.

Atkinson, A. (1970). On the Measurement of Inequality. Journal of Economic Theory: 2(3), 244–263.

Atkinson, A. and Bourguignon, F. (eds.) (2000). Handbook of Income Distribution, Volume 2. New York, New York: Elsevier.

Barrington-Leigh, C. and Escande, A. (2017). Measuring Progress and Well-being: A Comparative Review of Indicators. Social Indicators Research: 135(3), 893–925.

Barro, R. and Gordon, D. (1983). A Positive Theory of Monetary Policy in A Natural Rate Model. Journal of Political Economy: 91(4), 589–610.

Becchetti, L., Massari, R., and Naticchioni, P. (2014). The Drivers of Happiness Inequality: Suggestions for Promoting Social Cohesion. Oxford Economic Papers: 66(2), 419–442.

Becchetti, L. and Pelloni, A. (2013). What are We Learning from the Life Satisfaction Literature? International Review of Economics: 60(2), 113–155.

Becker, G. (1974). A Theory of Social Interactions. Cambridge, Massachusetts: National Bureau of Economic Research.

Bell, N. and Blanchflower, D. (2010). UK Unemployment in the Great Recession. National Institute Economic Review: 214(1), 3–25.

48 Bentham, J. (2008). An Introduction to the Principles of Morals and Legislation. New York, New York: Barnes & Noble. (Original work published in 1789).

Bjørnskov, C. (2014). Do Economic Reforms Alleviate Subjective Well-Being Losses of Economic Crises? Journal of Happiness Studies: 15(1), 163–182.

Blair, J. and Lacy, M. (2000). Statistics of Ordinal Variation. Sociological Methods and Research: 28(3), 251–280.

Blanchflower, D., Bell, D., Montagnoli, A., and Moro, M. (2014). The Happiness Trade-Off Between Unemployment and Inflation. Journal of Money, Credit and Banking: 46(2), 117–141.

Blanchflower, D. and Oswald, A. (2004). Well-Being Over Time in Britain and the USA. Journal of Public Economics: 88(7), 1359–1386.

Bollen, J., Mao, H., and Zeng, X. (2010). Twitter Mood Predicts the Stock Market. Journal of Conceptual Science: 2(1), 1–8.

Brickman, P. and Campbell, D. (1971). Hedonic Relativism and Planning the Good Society. In M. Appley (ed.), Adaptation-Level Theory. New York, New York: Academic Press.

Brooks, A. (2008). Gross National Happiness. Why Happiness Matters in America – and How We Can Get More of It. New York, New York: Basic Books.

Brown, M. (1996). The Causes and Regional Dimensions of Internal Conflict. In M. Brown (ed.) The International Dimensions of Internal Conflict. Cambridge, Massachusetts: Massachusetts Institute of Technology Press.

Cahill, K., McNamara, T., Pitt-Catsouphes, M., and Valcour, M. (2015). Linking Shifts in the National Economy with Changes in Job Satisfaction, Employee Engagement and Work–Life Balance. Journal of Behavioral and Experimental Economics: 56(1), 40–54.

Campbell, A., Converse, P., and Rodgers, W. (1976). The Quality of American Life: Perceptions, Evaluations, and Satisfactions. New York, New York: Russell Sage Foundation.

Carroll, C., Overland, J., and Weil, D. (2000). Saving and Growth with Habit Formation. American Economic Review: 90(3), 341–355.

Case, A. and Deaton, A. (2015). Rising Morbidity and Mortality in Midlife Among White Non-Hispanic Americans in the 21st Century. Proceedings of the National Academy of Sciences: 112(49), 15078– 15083.

Chadi, A. (2015). Concerns About the Euro and Happiness in Germany During Times of Crisis. European Journal of Political Economy: 40(1), 126–146.

49

Clark, A. (2018). Four Decades of the Economics of Happiness: Where Next? Review of Income and Wealth: 64(2), 245–269.

Clark, A. (2003). Inequality Aversion and Income Mobility: A Direct Test. DELTA Working Paper, No. 11. Paris, France: DELTA.

Clark, A., Fleche, S., and Senik, C. (2016). Economic Growth Evens Out Happiness: Evidence from Six Surveys. Review of Income and Wealth: 62(3), 405–419.

Clark, A., Flèche, S., and Senik, C. (2014). The Great Happiness Moderation. In A. Clark and C. Senik (eds.) Happiness and Economic Growth: Lessons from Developing Countries. Oxford, UK: Oxford University Press.

Clark, A., Frijters, P., and Shields, M. (2008). Relative Income, Happiness, and Utility: An Explanation for the Easterlin Paradox and Other Puzzles. Journal of Economic Literature: 46(1), 95–144.

Clark, A. and Oswald, A. (1998). Comparison-Concave Utility and Following Behaviour in Social and Economic Settings. Journal of Public Economics: 70(1), 133–155.

Clark, A. and Oswald, A. (1994). Unhappiness and Unemployment. Economic Journal: 104(424), 648– 659.

Costa-i-Font, J. and Cowell, F. (2013). Measuring Health Inequality with Categorical Data: Some Regional Patterns. Research on Economic Inequality: 21(1), 53–76.

Cowell, F. (1980). On the Structure of Additive Inequality Measures. Review of Economic Studies: 47(3), 521–531.

Cowell, F. and Flachaire, E. (2017). Inequality with Ordinal Data. Economica: 84(334), 290–321.

Dalton, H. (1920). The Measurement of the Inequality of Incomes. Economic Journal: 30(119), 348– 361. de Barros, R., Ferreira, F., Chanduvi, J., and Vega, J. (2008). Measuring Inequality of Opportunities in Latin America and the Caribbean. New York, New York: Palgrave Macmillan.

Deaton, A. (2012). The Financial Crisis and the Well-Being of Americans. Oxford Economic Papers: 64(1), 1–26.

Deaton, A. (2008). Income, Health and Well-Being Around the World: Evidence from the Gallup World Poll. Journal of Economic Perspectives: 22(2), 53–72.

50 Decancq, K., Fleurbaey, M., and Schokkaert, E. (2015). Inequality, Income and Well-Being. In: A. Atkinson and F. Bourguignon (eds.) Handbook of Income Distribution, Volume 2A. New York, New York: Elsevier.

Diener, E. and Biswas-Diener, R. (2002). Will Money Increase Subjective Well-Being? Social Indicators Research: 57(2), 119–169.

Di Tella, R. and MacCulloch, R. (2006). Some Uses of Happiness Data in Economics. Journal of Economic Perspectives: 20(1), 25–46.

Di Tella, R., MacCulloch, R., and Oswald, A. (2003). The Macroeconomics of Happiness. Review of Economics and Statistics: 85(4), 809–827.

Di Tella, R., MacCulloch, R., and Oswald, A. (2001). Preferences Over Inflation and Unemployment: Evidence from Surveys of Happiness. American Economic Review: 91(1), 335–341.

Dolan, P., Peasgood, T., and White, M. (2008). Do We Really Know What Makes Us Happy? A Review of the Economic Literature on the Factors Associated with Subjective Well-Being. Journal of Economic Psychology: 29(1), 94–122.

Duesenberry, J. (1949). Income, Saving and the Theory of Consumer Behavior. Cambridge, Massachusetts: Harvard University Press.

Dutta, I. and Foster, J. (2012). Inequality of Happiness in US: 1972-2010. Review of Income and Wealth: 59(3), 393–415.

Easterlin, R. (2003). Explaining Happiness. Proceedings of the National Academy of Sciences: 100(19), 11176–11183.

Easterlin, R. (2001). Income and Happiness: Towards a Unified Theory. Economic Journal: 111(473), 465–484.

Easterlin, R. (1974). Does Economic Growth Improve the Human Lot? Some Empirical Evidence. In: P. David and W. Melvin (eds.) Nations and Households in Economic Growth. Stanford, California: Stanford University Press.

European Values Study (EVS). (2015). EVS Longitudinal Data File 1981-2008. Cologne, Germany: University of Tilburg with GESIS Data Archive. ZA4804 Data file Version 3.0.0. Online: http://www.gesis.org/?id=10379

Exton, C., Smith, C., and Vandendriessche, D. (2015). Comparing Happiness Across the World. OECD Statistics Directorate Working Papers, No. 62. Paris, France: OECD Publishing.

51 Ferrer-i-Carbonell, A. and Frijters, P. (2004), The Effect of Methodology on the Determinants of Happiness. Economic Journal: 114(497), 641–659.

Ferrer-i-Carbonell, A. and Ramos, X. (2014) Inequality and Happiness. Journal of Economic Surveys: 28(5), 1016–1027.

Fitoussi, J. and Stiglitz, J. (2013). On the Measurement of Social Progress and Well-Being: Some Further Thoughts. Global Policy: 4(3), 290–293.

Frank R. (1985). The Demand for Unobservable and Other Non-Positional Goods. American Economic Review: 75(1), 101– 116.

Frey, B. and Stutzer, A. (2002). What Can Economists Learn from Happiness Research? Journal of Economic Literature: 40(2), 402–35.

Frey, B. and Stutzer, A. (2000). Happiness, Economy and Institutions. Economic Journal: 110(466), 918–938.

Frijters, P., Johnston, D., Shields, M., and Sinha, K. (2015). A Lifecycle Perspective of Stock Market Performance and Well-Being. Journal of Economic Behavior and Organization: 112(1), 237–250.

Gandelman, N. and Hernández-Murillo, R. (2009). The Impact of Inflation and Unemployment on Subjective Personal and Country Evaluations. Federal Reserve Bank of St. Louis Review: 91(3), 107– 126.

Gandelman, N. and Porzecanski, R. (2013). Happiness Inequality: How Much is Reasonable? Social Indicators Research: 110(1), 257–269.

Gini, C. (1921). Measurement of Inequality of Incomes. Economic Journal: 31(121), 124–126.

Goff, L., Helliwell, J., and Mayraz, G. (2016). The Welfare Costs of Well-being Inequality. NBER Working Paper No. 21900, Cambridge, Massachusetts: National Bureau of Economic Research.

Gottschalk, P. and Smeeding, T. (1997). Cross-National Comparisons of Earnings and Income Inequality. Journal of Economic Literature: 35(2), 633–687.

Graham, C. (2008). The Economics of Happiness. In: S. Durlauf and L. Blume (eds.) The New Palgrave Dictionary of Economics Online, Second Edition. Hampshire, UK: Palgrave Macmillan.

Graham, C., Chattopadhyay, S., and Picon, M. (2010). Adapting to Adversity: Happiness and the 2009 Economic Crisis and the United States. Social Research: 77(2), 715–748.

Graham, C. and Felton, A. (2006). Inequality and Happiness: Insights from Latin America. Journal of Economic Inequality: 4(1), 107–122.

52

Graham, C. and Pettinato, S. (2001). Happiness, Markets and Democracy: Latin America in Comparative Perspective. Journal of Happiness Studies: 2(3), 237–268.

Green, F. (2011). Unpacking the Misery Multiplier: How Employability Modifies the Impacts of Unemployment and Job Insecurity on Life Satisfaction and Mental Health. Journal of Health Economics: 30(2), 265–276.

Greve, B. (2012). The Impact of the Financial Crisis on Happiness and Affluent European Countries. Journal of Comparative Social Welfare: 28(3), 183–193.

Grimm, M., Harttgen, K., Klasen, S., Misselhorn, M., Munzi, T., and Smeeding, T. (2010). Inequality in Human Development: An Empirical Assessment of 32 Countries. Social Indicators Research: 97(2), 191–211.

Grusky, D., Western, B., and Wimer, C. (eds.). (2011). The Great Recession. New York, New York: Russell Sage Foundation.

Guimaraes, B. and Sheedy, K. (2012). A Model of Equilibrium Institutions. CEPR Discussion Papers, No. 1123. London, UK: Centre for Economic Performance.

Gudmundsdottir, G. (2013). The Impact of the Economic Crisis on Happiness. Social Indicators Research: 110(3), 1083–1101.

Gurr, T. (1994). Peoples Against States: Ethnopolitical Conflict and the Changing World System. International Studies Quarterly: 38(3), 347–377.

Guven, C., Senik, C., and Stichnoth, H. (2010). You Can’t Be Happier Than Your Wife. Happiness Gaps and Divorce. Journal of Economic Behavior and Organization: 82(1), 110–130.

Hagerty, M. (2000). Social Comparisons of Income in One's Community: Evidence from National Surveys of Income and Happiness. Journal of Personality and Social Psychology: 78(4), 764–771.

Hagerty, M. and Veenhoven, R. (2003). Wealth and Happiness Revisited: Growing National Income Does Go with Greater Happiness. Social Indicators Research: 64(1), 1–27.

Haller, M. and Hadler, M. (2006). How Social Relations and Structures Can Produce Happiness and Unhappiness: An International Comparative Analysis. Social Indicators Research: 75(2), 169–216.

Hariri, J., Bjørnskov, C., and Justesen, M. (2015). Economic Shocks and Subjective Well-Being: Evidence from a Quasi-Experiment. World Bank Policy Research Working Paper, No. 7209. Washington, DC: World Bank.

53 Helliwell, J. (2003). How's Life? Combining Individual and National Variables to Explain Subjective Well-Being. Economic Modelling: 20(2), 331–360.

Helliwell, J. and Barrington-Leigh, C. (2010). Viewpoint: Measuring and Understanding Subjective Well-Being. Canadian Journal of Economics: 43(3), 729–753.

Helliwell, J. and Huang, H. (2014). New Measures of the Costs of Unemployment: Evidence from the Subjective Well-Being of 3.3 million Americans. Economic Inquiry: 52(4), 1485–1502.

Helliwell, J., Huang, H., and Wang, S. (2014). Social Capital and Well-Being in Times of Crisis. Journal of Happiness Studies: 15(1), 145–162.

Helliwell, J., Layard, R., and Sachs, J. (2016). World Happiness Report 2016, Update (Vol. I). New York, New York: UN Sustainable Development Solutions Network.

Helson, H. (1964). Adaptation-Level Theory. Oxford, UK: Harper and Row.

Hicks, D. (1997). The Inequality-Adjusted Human Development Index: A Constructive Proposal. World Development: 25(8), 1283–1298.

Hirschauer, N., Lehberger, M., and Musshoff, O. (2015). Happiness and Utility in Economic Thought— or: What Can We Learn from Happiness Research for Public Policy Analysis and Public Policy Making? Social Indicators Research: 121(3), 647–674

Hoynes, H., Miller, D., and Schaller, J. (2012). Who Suffers During Recessions? Journal of Economic Perspectives: 26(3), 27–47.

Inglehart, R. (1990). Culture Shift in Advanced Industrial Society. Princeton, New Jersey: Princeton University Press.

Inglehart, R., Foa, R., Peterson, C., and Welzel, C. (2008). Development, Freedom, and Rising Happiness: A Global Perspective (1981–2007). Perspectives on Psychological Science: 3(4), 264–285.

International Monetary Fund (IMF). (2009). World Economic Outlook: Crisis and Recovery. Washington, DC: IMF.

Jamieson, S. (2004). Likert Scales: How to (Ab)use Them. Medical Education: 38(12), 1217–1218.

Kahneman, D. and Krueger, A. (2006). Developments in the Measurement of Subjective Well-Being. Journal of Economic Perspectives: 20(1), 3–24.

Kahneman, D. and Tversky, A. (1979). Prospect Theory: An Analysis of Decision Under Risk. Econometrica: 47(2), 263–291.

54 Kapteyn, A., van Praag, B., and van Herwaarden, F. (1978). Individual Welfare Functions and Social Reference Spaces. Economics Letters: 1(2), 173–177.

Kassenboehmer, S. and Haisken-DeNew, J. (2009). You’re Fired! The Causal Negative Effect of Entry Unemployment on Life Satisfaction. Economic Journal: 119(536), 448–462.

Knapp, T. (1990). Treating Ordinal Scales as Interval Scales: An Attempt to Resolve the Controversy. Nursing Research: 39(2), 121–123.

Kobus, M. (2015): Polarization Measurement for Ordinal Data. Journal of Economic Inequality: 13(2), 275–297.

Kőzegi, B. and Rabin, M. (2008). Choices, Situations, and Happiness. Journal of Public Economics: 92(8), 1821–1832.

Kristoffersen, I. (2011). The Subjective Well-Being Scale: How Reasonable is the Cardinality Assumption? UWA Discussion Paper No. 15. Melbourne, Australia: University of Western Australia Department of Economics.

Krueger, A. and Schkade, D. (2008). The Reliability of Subjective Well-Being Measures. Journal of Public Economics: 92(8/9), 1833–1845.

Kuzon, W., Urbanchek, M., and McCabe, S. (1996). The Seven Deadly Sins of Statistical Analysis. Annals of Plastic Surgery: 37(3), 265–272.

Larsen, R. and Prizmic, Z. (2008). Regulation of Emotional Well-Being: Overcoming the Hedonic Treadmill. In M. Eid and R. Larsen (eds.) The Science of Subjective Well-Being. New York, New York: Guilford Press.

Layard, R. (2005). Happiness: Lessons From A New Science. London, UK: Penguin Books.

Layard, R. (1980). Human Satisfaction and Public Policy. Economic Journal: 90(363), 737–750.

Layard, R., Clark, A., and Senik, C. (2012). The Causes of Happiness and Misery. In J. Helliwell, R. Layard and J. Sachs (eds.) World Happiness Report. New York, New York: The Earth Institute, Columbia University.

Layard, R., Mayraz, G., and Nickell, S. (2008). The Marginal Utility of Income. Journal of Public Economics: 92(8-9), 1846–1857.

Lean, H. and Tang, C. (2009). New Evidence from the Misery Index in the Crime Function. Economics Letters: 102(2), 112–115.

55 Lechman, E. (2009). Okun’s and Barro’s Misery Index as An Alternative Poverty Assessment Tool: Recent Estimations for European Countries. MPRA Paper No. 37493. Munich, Germany: University Library of Munich.

Lin, C., Chen, C., and Liu, T. (2015). Do Stock Prices Drive People Crazy? Health Policy & Planning: 30(2), 206–214.

Lorenz, M. (1905). Methods of Measuring the Concentration of Wealth. Publications of the American Statistical Association: 9(70), 209–219.

Lovell, M. and Tien, P. (2000). Economic Discomfort and Consumer Sentiment. Eastern Economic Journal: 26(1), 1–8.

Luechinger, S., Meier, S., and Stutzer, A. (2010). Why Does Unemployment Hurt the Employed? Evidence from the Life Satisfaction Gap Between the Public and the Private Sector. Journal of Human Resources: 45(4), 998–1045.

Ma, Y. and Zhang, Y. (2014). Resolution of the Happiness-Income Paradox. Social Indicators Research: 119(2), 705–721.

Maasoumi, E. (1986). The Measurement and Decomposition of Multidimensional Inequality. Econometrica: 54(4), 991–997.

MacKerron, G. (2012). Happiness Economics from 35,000 Feet. Journal of Economic Surveys: 26(4), 705–735.

Mankiw, N. (2010). Macroeconomics. New York, New York: Worth Publishers.

Marx, K. (1847). Relation of Wage-Labour to Capital. Wage Labour and Capital (translated by Friedrich Engels in 1891). Online: http://www.marxists.org/archive/marx/works/1847/wage-labour/ch06.htm

Mayer, T. (2003). The Macroeconomic Loss Function: A Critical Note. Applied Economics Letters: 10(6), 347–349.

McInerney, M., Mellor, J., and Nicholas, L. (2013). Recession Depression: Mental Health Effects of the 2008 Stock Market Crash. Journal of Health Economics: 32(6), 1090–1104.

Mertens, A. and Beblo, M. (2016). Self-Reported Satisfaction and the Economic Crisis of 2007–2010: Or How People in the UK and Germany Perceive A Severe Cyclical Downturn. Social Indicators Research: 125(2), 537–565.

Michalos, A. (1991). Global Report on Student Well-Being (Vol. 1). New York, New York: Springer- Verlag.

56 Michalos, A. (1985). Multiple Discrepancies Theory (MDT). Social Indicators Research: 16(4), 347– 413.

Mohseni-Cheraghlou, A. (2013). Labor Markets and Mental Well-Being: Labor Market Conditions and Suicides in the United States (1979–2004). Journal of Socio-Economics: 45(1), 175–186.

Neal, D. and Rick, A. (2014). The Prison Boom and the Lack of Black Progress after Smith and Welch. NBER Working Paper Series No. 20283. Chicago, Illinois: National Bureau of Economic Research.

Nessen, R. (2008). The Brookings Institution’s Arthur Okun: Father of the Misery Index. The Brookings Institution. Washington, DC. Online: http://www.brookings.edu/opinions/the-brookings-institutions- arthur-okun-father-of-the-misery-index/

Ng, Y. (1997). A Case for Happiness, Cardinalism, and Interpersonal Comparability. Economic Journal: 107(445), 1848–1858.

Niimi, Y. (2018). What Affects Happiness Inequality? Evidence from Japan? Journal of Happiness Studies: 19(2), 521–543.

Nikolova, M. (2016). Happiness and Development. IZA Discussion Paper No. 10088. Bonn, Germany: Institute for the Study of Labor.

Norman, G. (2010). Likert Scales, Levels of Measurement and the ‘Laws’ of Statistics. Advances in Health Science Education: 15(5), 625–632.

Olson, J., Herman, C., and Zanna, M. (1986). Relative Deprivation and Social Comparison: The Ontario symposium (Vol. 4). Hillsdale, New Jersey: Erlbaum.

Oswald, A. (2008). On the Curvature of the Reporting Function from Objective Reality to Subjective Feelings. Economics Letters: 100(3), 369–372.

Oswald, A. and Wu, S. (2011). Well-Being Across America. Review of Economics and Statistics: 93(4), 1118–1134.

Ott, J. (2005). Level and Inequality of Happiness in Nations: Does Greater Happiness of a Greater Number Imply Greater Inequality in Happiness? Journal of Happiness Studies: 6(4), 397–420.

Ovaska, T. and Takashima, R. (2010). Does A Rising Tide Lift All the Boats? Explaining the National Inequality of Happiness. Journal of Economic Issues: 44(1), 205–224.

Panas, E. (2013). Homeorhesis and Indication of Association between Different Types of Capital on Life Satisfaction: The Case of Greek under Crisis. Social Indicators Research: 110(1), 171–186.

57 Paul, I. K. and Moser, K. (2009). Unemployment Impairs Mental Health: Meta-Analyses. Journal of Vocational Behavior: 74(3), 264–282.

Pigou, A. (1920). The Economics of Welfare. London, UK: MacMillan and Co.

Piketty, T. (2014). Capital in the Twenty-First Century. Boston, Massachusetts: Harvard University Press.

Pollak, R. (1976). Interdependent Preferences. American Economic Review: 66(3), 309–320.

Ratcliffe, A. and Taylor, K. (2015). Who Cares About Stock Market Booms and Busts? Evidence from Data on Mental Health. Oxford Economic Papers: 67(3), 826–845.

Rawls, J. (1971). A Theory of Justice. Cambridge, Massachusetts: Harvard University Press.

Ringen, S. (2006). Reflections on Inequality and Equality. WZB Discussion Papers. Berlin, Germany: Social Science Research (WZB).

Runciman, W. (1966). Relative Deprivation and Social Justice. London, UK: Routledge and Kegan Paul.

Ruprah, J. and Luengas, P. (2011). Monetary Policy and Happiness: Preferences Over Inflation and Unemployment in Latin America. Journal of Socio-Economics: 40(1), 59–66.

Sacks, D., Stevenson, B., and Wolfers, J. (2011). Subjective Well-Being, Income, Economic Development, and Growth. In C. Sepúlveda, A. Harrison, and J. Lin (eds.) Development Challenges in a Postcrisis World. Washington, DC: World Bank.

Sen, A. (2000). Development as Freedom. New York, New York: Anchor Books.

Senik, C. (2009). Income Distribution and Subjective happiness: A Survey. OECD Social Employment and Migration Working Papers, No. 96. Paris, France: OECD Publishing.

Sernau, S. (2014). Social Inequality in A Global Age (4th edition). Thousand Oaks, California: Sage Publications, Inc.

Setterfield, M. (2009). An Index of Macroeconomic Performance. International Review of Applied Economics: 23(5), 625–649.

Shorrocks, A. (1980). The Class of Additively Decomposable Inequality Measures. Econometrica: 48(3), 613–625.

58 Smith, A. (1981). An Inquiry into the Nature and Causes of the Wealth of Nations. 2 volumes. In R. Campbell, A. Skinner, and W. Todd (eds.) Voume II of the Glasgow Edition of the Works and Correspondence of Adam Smith. Indianapolis, Indiana: Liberty Fund. (Original work published in 1776).

Smyth, R. and Qian, X. (2008). Inequality and Happiness in Urban China. Economics Bulletin: 4(23), 1– 10.

Solt, F. (2016). The Standardized World Income Inequality Database. Social Science Quarterly: 97(5), 1267–1281.

Solt, F. (2015a). Economic Inequality and Nonviolent Protest. Social Science Quarterly: 96(5), 1314– 1327.

Solt, F. (2015b). On the Assessment and Use of Cross-National Income Inequality Datasets. Journal of Economic Inequality: 13(4), 683–691.

Solt, F. (2010). Does Economic Inequality Depress Electoral Participation? Testing the Schattschneider Hypothesis. Political Behavior: 32(2), 285–301.

Solt, F. (2009). Standardizing the World Income Inequality Database. Social Science Quarterly: 90(2), 231–242.

Stevenson, B. and Wolfers, J. (2008a). Economic Growth and Subjective Well-Being: Reassessing the Easterlin Paradox. Brookings Papers on Economic Activity: 39(1), 1–102.

Stevenson, B. and Wolfers, J. (2008b). Happiness Inequality in the United States. Journal of Legal Studies: 37(2), 33–79.

Stiglitz, J., Sen A., and Fitoussi, J. (2010) Mis-measuring Our Lives: Why GDP Doesn’t Add Up. Report by the Commission on the Measurement of Economic Performance and Social Progress. New York, New York: The New Press.

Stiglitz, J. (2012). The Price of Inequality: How Today’s Divided Society Endangers Our Future. New York, New York: W.W. Norton.

Stouffer, S. DeVinney, L., and Suchmen, E. (1949). The American Soldier: Adjustment During Army Life. Vol. 1. Princeton, New Jersey: Princeton University Press.

Stutzer, A. (2004). The Role of Income Aspirations in Individual Happiness. Journal of Economic Behavior and Organization: 54(1), 89–109.

Thomas, V., Wang, Y., and Fan, X. (2001). Measuring Education Inequality: Gini Coefficients of Education. World Bank Policy Research Working Paper, No. 2525. Washington, DC: World Bank.

59 Thurow, L. (1971). The Income Distribution as a Pure Public Good. Quarterly Journal of Economics: 85(2), 327–336.

Tinbergen, J. (1970). A Positive and Normative Theory of Income Distribution. Review of Income and Wealth: 16(3), 221–234.

Tsui, K. (1995). Multidimensional Generalizations of the Relative and Absolute Inequality Indices: The Atkinson-Kolm-Sen Approach. Journal of Economic Theory: 67(1), 251–265.

Tullock, G. (1971). The Paradox of Revolution. Public Choice: 11(1), 89–100.

United Nations. (2000). National Accounts Main Aggregates Database. New York, New York: Economics and Statistics Branch, United Nations Statistics Division. Online: https://unstats.un.org/unsd/snaama/introduction.asp

United Nations Development Programme (UNDP). (1990). Human Development Report 1990: Concept and Measurement of Human Development. New York, New York: UNDP

United Nations University’s World Institute for Development Economics Research (UNU-WIDER). (2017). World Income Inequality Database (WIID3.4). Online: http://www.wider.unu.edu/project/wiid- world-income-inequality-database van Doorslaer, E. and Jones, A. (2003). Inequalities in Self-Reported Health: Validation of A New Approach to Measurement. Journal of Health Economics: 22(1), 61–87. van Herwaarden, F. and Kapteyn, A. (1979). Empirical Comparison of the Shape of Welfare Functions. Economics Letters: 3(1), 71–76. van Praag, B. (2011). Well-Being Inequality and Reference Groups: An Agenda for New Research. Journal of Economic Inequality: 9(1), 111–127. van Praag, B. and Ferrer-i-Carbonell, A. (2004). Happiness Quantified: A Satisfaction Calculus Approach. Oxford, UK: Oxford University Press.

Veblen, T. (1898). The Theory of the Leisure Class. New York, New York: MacMillan.

Veenhoven, R. (2008). Sociological Theories of Subjective Well-Being. In M. Eid and R. Larsen (eds.) The Science of Subjective Well-Being: A Tribute to Ed Diener. New York, New York: Guilford Publications.

Veenhoven, R. (2005a). Inequality of Happiness in Nations. Journal of Happiness Studies: 6(4), 351– 355.

60 Veenhoven, R. (2005b). Return of Inequality in Modern Society? Test by Dispersion of Life-Satisfaction Across Time and Nations. Journal of Happiness Studies: 6(4), 457–487.

Wang, W. and Parker, K. (2014). Record Share of Americans Have Never Married: As Values, Economics, and Gender Patterns Change. Washington, DC: Pew Research Center.

Weimann, J., Knabe, A., and Schöb, R. (2015). Measuring Happiness: The Economics of Well-Being. Cambridge, Massachusetts: Massachusetts Institute of Technology Press.

Welsch, H. (2007). Macroeconomics and Life satisfaction. Journal of Applied Economics: 10(2), 237– 251.

Wilkinson, R. and Pickett, K. (2010). The Spirit Level: Why Equality Better for Everyone. London, UK: Penguin.

Winkelmann, L. and Winkelmann, R. (1998). Why are the Unemployed So Unhappy? Evidence from Panel Data. Economica: 65(257), 1–15.

Wolfers, J. (2003). Is Business Cycle Volatility Costly? Evidence from Surveys of Subjective Well- Being. International Finance: 6(1), 1–26.

World Bank. (2017). World Development Indicators 2017. Washington, DC: World Bank. Online: https://openknowledge.worldbank.org/handle/10986/26447

World Value Survey (WVS). (2015). WVS 1981-2014 official aggregate v.20150418. Madrid, Spain: World Values Survey Association (WVSA), with ASEP/JDS. Online: http://www.gesis.org/?id=10379

Yang, B. and Lester, D. (1999). The Misery Index and An Index of Misery. Psychological Reports: 84(3), 1086.

Yang, J. Liu, K., and Zhang, Y. (2015). Happiness Inequality in China. MPRA Paper No. 66623. Munich, Germany: University Library of Munich.

Yang, Y. (2008). Social Inequalities in Happiness in the United States, 1972 to 2004: An Age-Period- Cohort Analysis. American Sociological Review: 73(2), 204–226.

Yitzhaki, S. (1979). Relative Deprivation and the Gini Coefficient. Quarterly Journal of Economics: 93(2), 321-324.

Zheng, B. (2011). A New Approach to Measure Socioeconomic Inequality in Health. Journal of Economic Inequality: 9(4), 555–577.

61