Examining the income-SWB satiation point, and the potential moderating effect of income inequality

(1)

Master Thesis – Final Version

MSc in Business Administration: Leadership and Management track

Examining the income-SWB satiation point, and the potential moderating

effect of income inequality

Date of submission: 15.06.2018

Thesis Supervisor: Dr. Richard Ronay

(2)

Statement of originality

This document is written by Student Christian August Stang who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents

(3)

Tables

Table 1. Means, SD and Correlations………..30 Table 2. Regression analysis results ………...32

Figures

Figure 1. Effect of income on experiential SWB…...………..…..31 Figure 2. Effect of income on evaluative SWB………...33 Figure 3. Effect of income inequality on income-evaluative SWB relationship…………...34 Figure 4. Effect of income inequality on income-experiential SWB relationship…....……....35

(5)

Acknowledgement

I want to thank my supervisor Richard Ronay for his feedback and support throughout this process. His excitement for the project was a motivational factor pushing both me and this thesis forward.

(6)

Abstract

Although most research claims there to be a relationship between both income and subjective well-being (SWB) and income inequality and SWB, the interplay between these three variables is not yet fully understood. Due to the global increase in income inequality (The world inequality lab, 2018) and the negative effects income inequality has on societies (Wilkinson & Pickett, 2009), such research is much needed. Therefore, the present study set out to investigate if there is a satiation point where income no longer relates to SWB.

Moreover, it examines the potential moderating effect of income inequality (Gini Coefficient) on this potential satiation point. By separating SWB measures according to its two aspects (evaluative and experiential), the present study further aimed to examine if there are differences in how income inequality might moderate. Through examining a cross-national level dataset from 2011 (European Quality of Life Survey) it was found that there is a satiation point where increased income no longer increases evaluative and experiential SWB – although at very high levels of income. Furthermore, it was found that income inequality has a moderating effect on the relationship between income and SWB – after a certain income level is reached, increased income only increases both SWB aspects when income inequality is high. These findings are discussed and suggestions for future research are made.

(7)

Introduction

“It's like the more money we come across, the more problems we see”

- Notorious B.I.G. “So I must leave, I'll have to go. To Las Vegas or Monaco. And win a fortune in a game. My life will never be the same”

- Abba In the famous song “no money no problems”, Notorious B.I.G sings about how more money seems to entail more problems. On the other hand ABBA sings enthusiastically about how money is the key to happiness in their song “money, money, money”. These might not be scientific contributions to the ongoing debate regarding the link between income and happiness/well-being, but they do touch upon a well-established assumption in most societies: that money affects people’s quality of life and well-being.

The relationship between income and people’s happiness and well-being has been extensively studied (Jebb, Tay, Diener & Oishi, 2018). Besides capturing the interest of famous singers, understanding the nuances of this relationship is important for governments, institutions, companies and people in general, and thus of large interest to scholars as well. Increased understanding of the link and the aspects surrounding it is important because it lays ground for policy-making both on a macro-level within nations, and on a micro-level within families and for individuals.

On an existential level it is important to understand if, how and why income is a key driver of people’s happiness. The operational importance lies in understanding if and how societies and the people who constitute them should emphasize “making more money” as a core value. As people constantly seem to seek higher levels of material wealth in their lives, such an ongoing struggle might tamper with other aspects of people’s life such as sacrifices in personal relationships and time (Diener & Oishi, 2000), and perhaps leave people less

(8)

happy than they might have been if their focus and efforts were directed elsewhere.

Understanding the determining aspects of the interplay between income and happiness thus becomes an important factor when societies move forward, especially when implementing policies intended to make citizens happier.

Accompanying the importance of understanding the link between income and happiness, several questions can be raised. Should societies encourage people to earn more money, even though research has found that people believe money determines their life satisfaction more than it actually does (Aknin, Norton & Dunn, 2009)? Might such societal norms function as a motivational factor and thus benefit the society through increased productivity? If yes, should it be upheld despite the fear of people stepping onto a hedonic treadmill, on which happiness will not increase, despite increased material wealth, due to everyone around them experiencing the same increase in wealth (Hagerty, 2000)? Or might it be that other aspects of life, such as more leisure time, is more important than increased income when trying to maximize happiness and well-being?

The previous questions open up for an endless number of research opportunities, and the present study aims to embrace a few of them. More precisely, it investigates if there is a satiation point where income no longer increases subjective well-being (SWB) in Europe, and whether income inequality (Gini coefficient) moderates the relationship between income and SWB, meaning if this relationship differs due to the degree of income inequality (high or low). Empirical investigations of the moderating effect of income inequality on a purely European sample, seems absent in the existing literature.

Studies investigating SWB tend to differentiate between evaluative and experiential SWB. This means that researchers either ask respondents about how they evaluate their lives, (e.g. how satisfied they are with their lives) or how they experience their lives, (e.g. the amount and degree of happiness experienced during the last days). Accordingly, evaluative

(9)

measures tries to capture people’s overall life evaluation, while experiential measures tries to capture day-to-day happiness (Helliwell, Huang & Wang, 2017; Hudson, Anusic, Lucas & Donnellan, 2017). This differentiation is essential in the present research because it is found that some aspects of SWB experience a satiation point where income no longer increases SWB, while others do not (Kahneman & Deaton, 2010; Stevenson & Wolfers, 2013; Jebb et al., 2018).

The first main objective of the present study is to investigate if there is a satiation point where income no longer leads to increased SWB, and at what income level that might occur. These questions have been investigated before, however mostly on American samples, e.g. Kahneman and Deaton (2010) and Stevenson and Wolfers (2013). The latter study however investigated the relationship for several countries outside the US, but other world regions were not clustered.

A recent study by Jebb et al. (2018) also tested satiation points for several countries across the world, however neither here was Europe tested as one entity. Furthermore, these three studies uses data from the Gallup World Poll (GWP). Cheung and Lucas (2016) notes that studies using GWP tends to find similar results, while they on the other hand found different results when using a different data source. They thus encourage future studies aiming to understand the link between income and SWB, to use other data sources than GWP. The present study answers this call (and is the first to test for satiation points on a purely European sample, N=43.636) by analyzing data from the European Quality of Life Survey. This will hopefully bring new knowledge to the research domain regarding how income relates to SWB in Europe.

Besides contributing to the research field, analysing the income-SWB link in Europe is of practical interest as well. This because understanding when and how income will give rise to well-being is much needed, and will strengthen policy implications according to Tay,

(10)

Zyphur and Batz (2017). Moreover, research shows that people’s life satisfaction is the best predictor for whether governments are re-elected, and thus more important than

unemployment rates, inflation or economic growth (Clark, Flèche, Layard, Powdthavee & Ward, 2017). It is thus not surprising that Inglehart and Klingemann (2000) found that the level of SWB within nations is closely linked to the sustainability of democracies, which is mirrored by Brehm and Rahm (1997) finding that life satisfaction relates to confidence with governments. These notions highlight the importance of understanding aspects related to life satisfaction and well-being when formulating policies. Thus, the present study aims to guide policy-makers in Europe by examining to what extent increased income increases SWB. Perhaps policies intended to encourage material aspiration is much needed, or perhaps policies intended to improve life quality through other methods are more relevant.

Wilkinson and Pickett (2009) confirms the relevancy of the present study by stating that governments and policy-makers might start to measure, value and promote other aspects of life (e.g. community involvement or local environment quality) when they recognize that constantly earning more does not imply greater happiness. This is an anticipation of the results of the present study, but it nonetheless confirms that the findings of the present study might represent valuable knowledge for policy-makers.

The findings might represent valuable knowledge for organizations as well because a longitudinal study by Bowling, Eschleman and Wang (2010) revealed a causal relationship between SWB and job satisfaction. They suggest this is because SWB entails experiencing emotions, in turn influencing satisfaction towards certain domains, e.g. work. Job satisfaction is in turn found to correlate (.30) with job performance (Judge, Thoresen, Bono & Patton, 2001). The present study might thus guide organizations when designing pay structures and employee benefits.

(11)

The second main objective of this study also represents potential novelty to the research domain. Examining income inequality as a moderator on the satiation point (on a purely European sample and based on the differentiation of the two SWB aspects) surpasses previous research. The findings might thus lay grounds for future research in several ways. First, if the moderating effect is apparent, additional research will be needed to understand why income inequality tampers with the income-SWB link. Second, if the moderating effect differs depending on what aspects of SWB that are investigated, this tells us that income inequality plays a more complex role than previous research has revealed. This complexity is also an interesting subject for further investigations. Third, if no moderation is found,

attention can be directed elsewhere, e.g. to perceived income inequality and perceived social mobility – related aspects of income inequality which cannot be inferred from the Gini coefficient which is tested in the present study.

Understanding the moderating effect of income inequality on the relationship between income and SWB might also provide practical implications, as it increases our understanding of how income inequality affects societies and their citizens. Thus, the findings might help governments understand if the degree of inequality in their country relates to why certain aspects of SWB are not achieved through a general increase in income. Income

inequality is increasing across the world (The world inequality lab, 2018), and greater income equality is said to be the key to achieve success in several societal domains such as creating trust, stress reduction and increased general health (Wilkinson & Pickett, 2009): “If we want to do better, we need to be more equal” (p. 149) they state. The present study intends to answer if this is true in terms of doing better = increased SWB. Based on this some suggestions for policy makers can be made.

To summarize, the present study aims to contribute with novel empirical research and practical suggestions by answering the following research questions:

(12)

 Is there in fact a SWB satiation point for income in Europe, and if so: do both evaluative and experiential SWB experience satiation?

 Does income inequality (Gini-coefficient) moderate the satiation point in the relationship between income and evaluative/experiential SWB in Europe?

The thesis is structured as follow: First relevant literature is reviewed in order to formulate hypotheses according to the research questions. Thereafter, method, sample and measures are described. Following this, it will be described how the data is prepared in order to conduct the analyses. Furthermore, the findings of all hypotheses will be described in the result section. The results will be thoroughly discussed thereafter, followed by sections pointing to limitations in the present study and suggestions for future research. Lastly, conclusions are made.

Literature Review

In the following section literature regarding the main subjects of the thesis will be discussed, namely the different aspects of SWB, the relationship between income and SWB, and income inequality. First, due to the diverse views on SWB in the literature, it becomes important to establish a relevant and precise conceptualization that can be applied to the present study. Second, the relationship between income and SWB, and the nature of this relationship will be investigated. This through reviewing the main debate within this research field, namely if it is absolute income or relative income compared to others that determines the income–SWB relationship. Third, when reviewing literature on income inequality it will be important to understand how it relates to SWB, and accordingly how it relates to the income-SWB link. Hypotheses will be formulated throughout this review.

(13)

Conceptualizing Well-Being

Questions regarding people’s life quality have interested philosophers and thinkers across cultures and time, and there are numerous approaches to measuring this “ultimate life goal”. One approach has been to assess objective measures such as health and pollution, while another way is to measure people’s subjective experience of their own lives in terms of subjective well-being (SWB) (Diener & Suh, 1997). It is appropriate to label the latter

approach’s construct “subjective”, as it is something that is experienced by a certain

individual (Campbell, 1976). Such subjective measures help researchers understand people’s cognitive and affective reactions to their own lives. SWB is thus a broad phenomenon that includes people’s 1) global judgements of life satisfaction and 2) their responses to life events and daily life (Diener, Suh, Lucas & Smith, 1999). Accordingly, even though objective aspects might influence quality of life, such aspects are not an inherent part of SWB per se (Diener, 1984; McBride, 2001).

When gathering data, the World Happiness Report emphasizes the distinction between objective aspects and subjective perceptions of life quality. They state that it is imperative to understand how people evaluate their life quality and not interpret measures assumed to influence their life quality (Helliwell et al., 2017). The present study

acknowledges this distinction, and will thus investigate SWB and not objective aspects potentially affecting people’s life quality. This further seems appropriate because measuring SWB allows for uncovering people’s own judgement of their lives based on their own values and standards (Diener, Sapyta & Suh, 1998).

Overall, SWB is defined as “people’s overall evaluation of their lives and their emotional experiences” (Diener et al., 2017, p. 87), and can therefore be categorized into two different aspects: global life evaluation and experiential SWB (Diener et al., 1999; Tay, Chan & Diener, 2014; Helliwell et al., 2017). Global life evaluation is one’s overall evaluation of

(14)

his/her life quality, and captures life circumstances and quality of life in a more stable and complete way than experiential measures (Helliwell et al., 2017). Experiential SWB on the other hand describes a judgement of one’s day to day affective states (Hudson et al., 2017). Therefore, experiential measures are more prone to be affected by temporary mood

differences such as positivity prior to a weekend (Stone, Schneider & Harter, 2012). Nevertheless, both aspects are found to be important and different parts of overall SWB (Lucas, Diener & Suh, 1996), and distinguishing the two aspects is thus important when examining the income-SWB link (Deaton & Stone, 2013).

While there seems to be a broad agreement about the evaluative aspect of SWB, scholars use different terms when examining the experiential aspect. It is referred to as hedonic SWB – daily happiness (Deaton & Stone, 2013), affective SWB – amount and degree of positive and negative emotions experienced during a life period (Diener, 1984; Ng & Diener, 2018) and emotional well-being – the emotional quality of a person’s everyday experience (Kahneman & Deaton, 2010). It is also important to note that the experiential aspect includes both positive and negative feelings (Kahneman & Deaton, 2010; Ng & Diener, 2018).

For the purpose of this study, SWB will be conceptualized according to the reviewed literature. Thus, evaluative SWB will be clustered as one aspect, reflecting people’s overall evaluation of their lives. Experiential SWB on the other hand reflects emotional day-to-day affections (both positive and negative) and will accordingly be clustered as a separate aspect. This separation will allow for measuring the different constructs of SWB (Lucas et al., 1996), and it should be noted that the reliability and validity of both aspects are found not to differ substantially (Hudson et al., 2017). Exactly how both aspects are measured in the present study will be discussed in the method section.

(15)

The Link Between Income and SWB

Even though divergent conclusions regarding the income-SWB link have been drawn, there is robust evidence supporting a positive relationship between the two (Tay et al., 2017; Ng & Diener, 2018), both across nations and over time (Stevenson & Wolfers, 2013). The divided conceptualization of SWB and its two aspects helps us further investigate the logic of this relationship because it has been shown that income is more strongly associated to

evaluative than experiential SWB (Diener, Ng, Harter & Arora, 2010; Kahneman & Deaton, 2010; Helliwell et al., 2017). Although this notion seems widely accepted, the “why’

explaining it seems rather unclear.

The effect of income on SWB. The most buzz-making and tangible contribution to

the understanding of how evaluative and experiential SWB is differently related to income, came with Kahneman and Deaton (2010). They clustered SWB into its two aspects and investigated if income increased them linearly or if there is a satiation point where income no longer contributes to increased SWB. It should be noted that a satiation point does not mean that an individual’s SWB would not increase if that individual suddenly earned more. Rather, it means that people in different income groups above the potential satiation point, would not differ in SWB.

Kahneman and Deaton (2010) found no satiation point for evaluative SWB, even though the curve became less steep at higher levels of income. The higher income, the better life evaluation. However, when people were asked about certain aspects regarding their previous day (positive affects, stress and worrying, i.e. experiential SWB), they found that the curve flattened after approximately $75.000 income. Through their findings, they therefor drew the conclusion – that experiential SWB does not increase after a certain level of income is achieved. On the contrary, evaluative SWB do not satiate and is thus more strongly related to income, at least at higher levels of income.

(16)

Stevenson and Wolfers (2013) did also not find a satiation point for evaluative SWB (they did not test for experiential SWB) when reviewing several datasets containing income and SWB data. Thus, the findings of both studies seem to represent a consensus on this research topic, namely that increased income = increased evaluative SWB no matter how much you earn. However, a very recent study by Jebb et al. (2018) identifies a satiation point at $95.000 globally also for evaluative SWB. They even found preliminary evidence for a turning point for this SWB aspect in certain world regions (Western Europe/Scandinavia, Eastern Europe/the Balkans, Latin America/the Caribbean, Northern America and East Asia), suggesting that evaluative SWB decreases in the highest income groups. They suggest that it is not the high income itself that reduces life evaluation, but the accompanied demands, such as workload and responsibility, that high income might demand. Additionally they suggest that social comparison mechanisms might increasingly come into play, meaning that higher income entails increased unfulfilled material aspiration, leaving evaluative SWB to stagnate.

Despite drawing different conclusions than Kahneman and Deaton (2010), Jebb et al. (2018) state that the two studies do not conflict, due to differences in data and method. More specifically, they claim that Kahneman and Deaton (2010) clustered income groups too broadly rather than treating income as a continuous variable (which Jebb et al. did). This made it plausible that the mean income in the second highest income group was in the lower range, thus also lowering the mean SWB level and creating significant difference towards the highest income group. They thus suggest that Kahneman and Deaton (2010) might have come to the same conclusion as themselves if their method was improved. That Kahneman and Deaton (2010) did find the curve for evaluative SWB to become less steep when moving into the higher income groups, might support such a suggestion. It should also be noted that Jebb et al. (2018) found a satiation point for experiential SWB at $60.00-$75.000, which is in line with the findings of Kahneman and Deaton (2010).

(17)

Jebb et al. (2018) acknowledges that Stevenson and Wolfers (2013) treated income as a continuous variable but they nonetheless criticize their findings on two specific points. First, they claim Stevenson and Wolfers (2013) used raw household income which thus not accounts for household size. This in turn led them to assume that a certain amount of income operates identically for a lone individual and a family of four – thus inflating the satiation estimates. This leaves it understandable that Stevenson and Wolfers did not find satiation. Second, they claim Stevenson and Wolfers (2013) had sparse data in the upper income groups because they apparently excluded the upper 90 % of income distributions. Jebb et al. (2018) allegedly tackled this problem by pooling countries into world regions, while

Stevenson and Wolfers (2013) operated with sparse data above $ 64.000 within each country. Nonetheless, Jebb et al.’s (2018) method also leaves an unanswered question. Even though they examine the satiation point worldwide, they do not test for satiation points for the two aspects of SWB on Europe as a whole, which leaves this specific question still open.

With the previous review in mind, one thing is certain: more research is needed (as different studies finds different results) – especially for determining potential satiation points in Europe. Relying on the findings of Kahneman and Deaton (2010) and Jebb et al. (2018) however, H1 predicts a satiation point where income no longer increases experiential SWB. Moreover, H2 acknowledges that much research found income to have more impact on evaluative measures at higher levels of income. Lack of support for H2 will however provide support for the findings of Jebb et al. (2018) – that also evaluative SWB satiates.

H1: Experiential measures of SWB experience a satiation point where income no longer increases experiential SWB

H2: Evaluative measures of SWB do not experience a satiation point, meaning that income increases evaluative SWB also in the higher income groups. However, the curve is expected to become less steep at higher levels of income

(18)

The studies of Kahneman and Deaton (2010), Stevenson and Wolfers (2013) and Jebb et al. (2018) shed new light on, and partly contradict, the seminal findings of Richard

Easterlin (1974). He concluded that money does not buy happiness beyond a certain point, although he did not specify an exact point. Easterlin argued this because richer countries did not report higher happiness than poorer. Moreover, GDP growth in USA did not increase happiness among the citizens. However, Easterlin did find that within nations, richer people were happier than poorer people. He thus concluded that there is no link between the

economic development of societies, measured by GDP, and its average level of happiness, but that richer people are happier than poorer within countries due to relative wealth. These rather contradictory findings later became known as the Easterlin paradox.

Easterlin explain his within-country findings with the Duesenberry-type model, which states that relative status considerations is an important determinant of happiness (Easterlin, 1974). This aligns with the basic tenets of social comparison theory – that people compare themselves to proximal or similar others and make self-evaluative judgements thereafter (Festinger, 1954). Such social comparison differs from adaptive comparison, where an individual makes comparisons to his/her own experiences and standards (McBride, 2001). Social comparison theory however, suggests that relative income compared to others affects aspects of SWB, which is what Easterlin argued when finding that richer countries were not happier than poorer, while richer people were happier than poor people within countries.

Despite their theoretical contribution, Easterlin’s findings have been questioned because the distinction between evaluative and experiential aspects of SWB is unclear in his studies. Even though he supposedly included both aspects (his data measured evaluative SWB through Cantril’s “Self-anchoring scale” and experiential SWB through a happiness-scale: very happy, fairly happy and not very happy), he do not differentiate findings based on the measures. Many studies such as Kahneman and Deaton (2010) uses Cantril’s

(19)

“Self-anchoring scale”, and the measure itself is tested and found to yield valid results. However, the experiential SWB measure is weak, as such a method assumes that e.g. the distance between very happy and fairly happy is the same as between fairly happy and not very happy (Stevenson & Wolfers, 2013). Thus, Graham (2013) notes that it is possible to come to different conclusions regarding the Easterlin paradox simply by using different SWB questions, which also the findings of Kahneman and Deaton (2010), Stevenson and Wolfers (2013) and Jebb et al. (2018) shows. Deaton and Stone (2013) confirm this by stating that the distinction of SWB measures is essential if one wants to understand the income-SWB link thoroughly. The previous conceptualization of SWB suggested the same.

The absolute and relative income perspective. Despite some measurement issues,

the findings of Easterlin laid the grounds for a prominent debate concerning whether it is an increase in absolute income that relates to SWB, or if it is instead that relative income (i.e., comparisons with proximal others) is determinant. In one camp, research by Ruut Veenhoven (1988, 1991) supports the absolute income perspective and represents a seminal contribution to this debate. He explains that income helps people to meet innate and universal needs that are unconscious prerequisites for functioning in life (Veenhoven, 1988, 1991). He claims that people judge their life positively when needs are met – regardless of their situation compared to others and even when others enjoy substantially better conditions (Veenhoven & Ehrhardt, 1995). He calls this the “livability theory”, wherein income creates livable conditions.

Veenhoven claims that the needs related to obtaining livable conditions are products of evolution and thus crucial for people’s ability to function and survive.

When promoting the absolute perspective, Veenhoven simultaneously expresses doubt about the relative income perspective. He argues that the relative income perspective suggests that happiness is independent of what is objectively “good”, and thus dependent on subjective comparisons to others (Veenhoven, 1991). Accordingly, he suggests that the

(20)

relative income perspective is somewhat randomly constructed due to it involving 1)

individual differences between people and 2) comparison standards which adjust over time. It thus becomes a calculus of life – “as is” against “ought”, or how life should be at a given time (Veenhoven & Ehrhardt, 1995). As such, Veenhoven claims the relative income perspective is “odd” and represents a paradox because it suggests that it makes little sense to promote happiness in societies as it is volatile, which thus makes it strange that both individuals and states constantly strives to improve situations as a means to becoming happier. He suggests this “mismatch” derives due to the theory overlooking the needs perspective and its

unconscious functioning within humans.

Veenhoven further elaborates with evolutionary arguments when promoting the absolute versus the relative perspective. More precisely, he states that the relative income perspective suggests that people’s happiness depends on their conscious comparisons to others, and that it therefor overlooks that need-gratification relates to pleasant affects unconsciously. Because cognitive abilities, i.e. conscious relative reasoning, occurred rather late in evolution, he argues for the unconscious process, i.e. the absolute income perspective. Simply put, he claims income increases SWB unconsciously because it allows people to meet their needs, and not because it allows them to consciously compare their standing to others. Easterlin (1974) on the other hand does not really address the evolutionary aspect when arguing for the relative income perspective. He rather states that his results in favor of the relative income perspective are “a testimony to the adaptability of mankind” (p. 119).

Diener and Oishi (2000) reviewed Veenhoven’s need theory, and state that it needs a Maslovian extension, meaning that it also has to include self-actualization needs. Maslow (1943) defines self-actualization needs as “the desire to become more and more of what one is, to become everything that one is capable of becoming” (p. 382). He further argues that self-actualization is also defined as a human need – meaning that even the very wealthy

(21)

might pursue rewarding activities in pursuit of gratifying this higher order need. Within this perspective, income spending becomes an important factor within the absolute income perspective as well, because fulfilling self-actualization needs might, although not

necessarily, involve spending a lot of money. Accordingly, one might argue that a Maslovian extension to Veenhoven’s Livability theory brings the absolute and relative perspectives conceptually closer. This because in today’s societies, the boundaries to self-actualization rest more on standards and opportunities than on the unconscious need-gratification – pleasant affect link of our distal ancestors.

When discussing the importance of fulfilling self-actualizing needs, Maslow states that discontent and restlessness will develop if they are not met. An interpretation of this is that not fulfilling such needs affects evaluative SWB to a greater extent than experiential SWB. This because evaluative measures captures global evaluation of life quality, while experiential measures captures emotional quality of a person’s everyday experience. Simply put: whether you are/are not able to self-actualize affects your life satisfaction, but not (to the same degree) your daily happiness. Thus, a Maslovian extension of Veenhoven’s livability theory, might suggest that the two perspectives, the absolute and relative, are somewhat overlapping. This because suddenly needs possibly stemming from comparisons to others are brought into the equation also within the absolute income perspective.

Some researchers have expressed doubt about both the relative and absolute

perspective, e.g. Diener, Sandvik, Seidlitz and Diener (1993) who explored the perspectives using both US and international samples. Through their findings they agree with Veenhoven that relative standards and comparison to others do not influence the relationship between income and SWB. Their results do not however give unconditional support for the absolute income perspective either. Several researchers (Hagerty, 2000; Stevenson & Wolfers, 2013; Sacks, Stevenson & Wolfers, 2011) do however argue for the absolute income perspective,

(22)

disputing Easterlin’s paradox and claiming that people in richer countries are in fact happier than those in poorer countries. A recent review of the topic by Tay et al. (2017) also proposes a stronger relationship between the absolute income perspective and SWB. However, they do argue that relative income plays a role for happiness over time.

It is not the goal of this thesis to solve this tension in the literature but to use these existing perspectives in the formulation of the hypotheses. More specifically, two

distinguished perspectives emerge. First, because there is less support for a satiation point for evaluative measures, one might argue that such judgements depends on social comparisons to others and therefore are more strongly related to the relative income perspective or self-actualization needs fulfillment. Said more simply: there is no point where income stops to affect evaluative SWB because this evaluation depends on the relativity towards others. This suggestion is supported by Cheung and Lucas (2016) arguing that social comparison of income plays a part in the relationship between relative income and evaluative SWB.

Second, the literature suggests that experiential SWB is more strongly influenced by the absolute income perspective, because this SWB aspect does experience a satiation point. This might mirror the absolute income perspective, and Veenhoven’s livability theory, because this perspective suggests that income contributes to increased SWB only as far as it helps people meet their needs. Simply put, at a certain income level all your livability needs are met, hence day to day well-being does not increase after that point.

Integrating these perspectives, one might suggest that experiential SWB depend on income solely, but only to a certain level where livability needs are met, while evaluative SWB depends on both income levels and the degree of income inequality. The latter because income inequality reflects the relativity between people in societies. Diener et al. (2010) suggests support for such a proposition – stating that income-related societal circumstances

(23)

might influence people’s life satisfaction substantially. Contrary, they suggest that income increases might have little effect on experiential measures, such as feelings.

The present study attempts to investigate these suggestions by introducing income inequality as a potential moderator on the relationship between income and the two aspects of SWB at the point where those aspects might start to decline or even experience satiation. Accordingly, the present study predicts income inequality to affect the strength of the

relationship between income and SWB. More precisely, it is expected that income inequality moderates the relationship between income and evaluative SWB but not the relationship between income and experiential SWB. The following section will explain the rational for this proposition in detail.

Income Inequality

The effect of income inequality on SWB. Several great thinkers and leaders have

emphasized the importance of income equality in societies, e.g. Mahatma Gandhi who stated: “a nation’s greatness is measured by how it treats its weakest members” (Wilkinson & Picket, 2009, p. 149). The question then becomes if this also applies for greatness in terms of

happiness. According to Yu and Wang (2017) the answer is yes, as they suggest income inequality rather than absolute income predicts happiness. However, their suggestion does not answer the question: are people in low income inequality or high income inequality countries the happiest? This question has been studied by several researchers, and diverging answers have emerged. More precisely, research finds income inequality to 1) affect SWB negatively, 2) not having an overall effect on SWB, and 3) affect SWB positively. These different

findings will be explained below.

First, some researchers have reported a negative relationship between income inequality and life satisfaction such as Cheung and Lucas (2016) and Hagerty (2000). Similarly, Oishi, Kesebir and Diener (2011) found in historical data that Americans were

(24)

happier in times with low income inequality than in times with high income inequality. Alesina, Di Tella and MacCulloch (2004) and Yu and Wang (2017) also finds a negative income inequality-SWB link in their studies.

Yu and Wang (2017) tries to explain why income inequality might be related to happiness1. Among both US and European respondents, they find an inverted U-shaped relationship between income inequality and happiness due to two competing dynamic processes. When inequality is low, they claim it signals aspirational opportunities to those with lower income and thus increases happiness. However, when inequality is high the effect goes from signaling aspirational opportunities to jealousy effects, leaving people with lower income less happy. Furthermore, they found differences between Americans and Europeans, which suggests that there is a difference between actual income inequality, perceived income inequality and perception of whether such inequality is good or bad. They suggest Americans tolerate higher levels of income inequality than Europeans due the common perception of “The American dream”.

Second, some researchers such as Ngamaba, Panagioti and Armitage (2018) report that income inequality is unrelated to SWB. Through a meta-analysis they conclude that there are no statistically significant associations between income inequality and SWB, and that the link is complex. Nevertheless, when performing subgroup analyses they found that income inequality and SWB are positively related in poor countries and negatively related in developed countries. They thus suggest that the income inequality-SWB relationship is moderated by a country’s economic development. On that note, it should be mentioned that analyzing the moderating effect also from a development perspective, is beyond the scope of this thesis.

(25)

Third, research by Rözer and Kraaykamp (2013) found that people in unequal

countries report higher SWB than those living in equal countries. Similarly, Katic and Ingram (2018) report that SWB is higher when income inequality is higher. These findings are

confirmed in a “box fresh” study by Ng and Diener (2018) who found that people in unequal countries reported higher SWB scores than those in more equal countries.

Even though different results emerge in different studies, the present study

acknowledges that there are more evidence in favor of income inequality affecting SWB – thus making it relevant to investigate it as a moderator on the income-SWB link.

The moderating effect of income inequality. Ng and Diener (2018) also investigated

the moderating effect of income inequality (Gini coefficient) on the relationship between income and SWB – the same as the intentions of the present study. They found that income inequality moderated the effects of income on SWB because income had a stronger impact on SWB for people in equal countries than in unequal countries. They thus find that even though increased income inequality is associated with higher SWB, income affects SWB more in equal countries. However, when discussing their conclusions they note that their study was conducted worldwide and not based on a separation of the different aspects of SWB. They thus call for future studies to tackle these limitations – a call the present study answers.

Ng and Diener’s (2018) suggestion: that income affects SWB more in equal countries, aligns with Cheung and Lucas (2016) finding that higher income inequality is associated with stronger relative income effects. This would mean that a person’s income would increase his/her SWB less if his/her neighbor earns more than him/her as opposed to if the neighbor earned equally as much. Accordingly, they suggest that income inequality leads to higher levels of social comparison effects. However, they found this effect to be strongest for low-income individuals, which aligns with the findings of Yu and Wang (2017): that increased income inequality leads to jealousy effects in the lower income groups. As the present study

(26)

aims to understand how potential satiation points (where income no longer increases SWB) are moderated by income inequality, it is the effect on the higher income groups that are of interest, thus, clues should be found elsewhere.

One clue that might explain how income inequality moderates the relationship between higher levels of income and SWB is found in the study of Jebb et al. (2018). They suggest they find a turning point where increased income decreases evaluative SWB due to increased social comparison effects in the higher income groups. When taking on such a perspective, one might expect social comparison, and thus the relative income perspective, to be stronger in equal countries, at least when higher levels of income are achieved. This because countries with higher income are more equal than low income countries (Ng & Diener, 2018). That other studies (Rözer & Kraaykamp, 2013; Katic & Ingram, 2018) found SWB to be lower in equal countries than unequal countries supports this proposition.

The different findings as explained above point to tension in the literature. First, there does not seem to be an agreement on whether income inequality is positive or negative for SWB. The present study partly tests this, through investigating if income inequality moderates the relationship between income and the two aspects of SWB (experiential and evaluative). Second, conflicting clues and suggestions leaves the following question still unanswered: how do social comparison effects come into play; is it the rich, the poor in unequal countries or both that are affected by such effects. Even though the answer is not given, the present study builds on the suggestion of Jebb et al. (2018), namely that social comparison effects affect rich people in equal societies, leaving certain aspects of SWB to possibly stagnate at higher levels of income. It accordingly acknowledges the findings of Ferrer-i-Carbonell (2005), namely that social comparisons mostly go upwards and not downwards.

(27)

In line with the suggestions made throughout this review: that 1) evaluative SWB is more related to the relative income perspective and costly self-actualization needs, and 2) experiential SWB is more related to the absolute income perspective, suggestions for the moderating effect of income inequality can be made. First it is expected that higher income inequality would affect the positive relationship between income and evaluative SWB so that the point of inflection (lessening point) would occur later. This would mean that evaluative SWB rises more at higher income levels in unequal than equal countries. Reason for this being that upwards social comparison effects come into play for richer people in equal countries, leaving the effect of income on evaluative SWB to diminish.

Second, the moderating effect of income inequality on the income-experiential SWB link might be weaker or even absent, because such variables are less related to the relativity towards others, and thus mostly explained by the livability theory, i.e. the absolute income perspective. This would leave income inequality potentially less important, or not important at all, because well-being through fulfilling needs (except perhaps self-actualization needs) do not relate to the relativity towards others. Thus, the following hypotheses are proposed:

H3: Income inequality (Gini Coefficient) moderates the relationship between income and evaluative SWB, meaning that the point of inflection (lessening point) occurs later when income inequality is higher

H4: Income inequality (Gini Coefficient) does not moderate the relationship between income and experiential SWB, meaning that income inequality does not affect when the satiation point occurs

Combined all four hypotheses propose the following: income increases evaluative SWB at higher levels of income, but not experiential SWB. However, income inequality moderates the relationship between income and evaluative SWB so that evaluative SWB only rises at higher income levels when income inequality is high. This because of social

(28)

comparison effects. Income inequality is not predicted to have the same effect on the relationship between income and experiential SWB.

Summary Literature Review

The present study examines if there is a satiation point where income no longer relates to the two aspects of SWB. Furthermore, it investigates the potential moderating effect of country level income inequality on these potential satiation points. The study thus takes on a different approach than most previous studies which have investigated main effects between income and income inequality on the one hand, and SWB on the other. By separating

measures of SWB, the present study further aims to reveal if there are differences in how income inequality might moderate. Ng and Diener (2018) encouraged future research to investigate this exact question. Moreover, the present study answers Cheung and Lucas’ (2016) call: future studies examining the income-SWB link should use other data sources than the Gallup World Poll. The present study does so on a purely European sample which has not been done before.

Despite conflicting evidence, much research points to both a relationship between income and SWB and income inequality and SWB. Furthermore, accurate measurements of SWB are important, especially when attempting to uncover any satiation point, where income no longer is related to SWB. Moreover, research suggests that the link between income and SWB might be affected by the levels of income inequality. This is however only suggested to be the case for evaluative SWB because income’s relationship to this SWB aspect is related to the relativity towards others, i.e. the relative income perspective. The income-experiential SWB link on the other hand is not expected to be affected by income inequality, as income’s relationship to this SWB aspect is explained by the absolute income perspective.

(29)

Method

Sample

The data used in the present study is obtained by European Foundation for the Improvement of Living and Working Conditions. Through their European Quality of Life Survey from 2011, respondents from 33 European countries are interviewed regarding occupation, work-life balance, life satisfaction, positive and negative SWB, societal trust etc. The survey also includes questions regarding household income and surrounding aspects. Additionally, the dataset contains information regarding country level income inequality as expressed by the Gini coefficient. Combined, this database thus includes all variables necessary to investigate the present research questions and their hypotheses. The number of respondents from each country range from 1000-1500, and the total number of respondents are N=43.636.

Measures

The present study uses measures for income, SWB and income inequality (Gini coefficient).

Income: independent variable. Both Kahneman and Deaton (2010) and Jebb et al.

(2018) use gross household income as measure of income. The same is the case in the present study, which thus entails one measure for this variable. However, in the current data,

respondents were asked about their net household income through the question: Q63 Please can you tell me how much your households NET income per month is? That respondents are asked about net (after tax) income, differs from the previous studies reviewed earlier. It will thus be important to keep this in mind when comparing the results to those studies that measured gross income. Furthermore, the dataset includes an income item which adjusts for purchasing power parity and household size based the OECD equivalence scale.

(30)

As stated earlier, Jebb et al. (2018) claim that Kahneman and Deaton’s (2010) findings might be weak due to operating with too wide income intervals (second highest income group spanned between $90.000 and $120.000). On the same note, they criticize Kahneman and Deaton (2010) for treating income categorically, and claim that it is important to treat income as a continuous variable when investigating any potential relationship with SWB. Thus, an important data preparation step (i.e. before starting the analysis), is to determine how income should be treated.

SWB: dependent variable. Because of the richness of the dataset, the present study

combines different evaluative measures into one scale. The items allow the respondents to rate their satisfaction in certain life domains: education, present job, accommodation, health, family life, present standard of living, social life and economic situation. Respondents answer on a 1-10 scale, where 1 is very dissatisfied and 10 is very satisfied.

As the literature review revealed, researchers use different labels when

conceptualizing experiential SWB. The important factor however, is that these variables are able to capture day-to-day affects to a greater extent than the evaluative measure. This

includes both positive and negative emotions. For positive experiential SWB, the items in the present survey allow respondents to express their emotional state over the last two weeks by asking them: “Please indicate for each of the five statements which is closest to how you have been feeling over the last two weeks.” The statements are: I have felt cheerful and in good spirits, I have felt calm and relaxed, I have felt active and vigorous, I woke up feeling fresh and relaxed and My daily life has been filled with things that interest me. The respondents answer on a six-point scale (1. All of the time – 6. At no time). The same method is used to measure negative experiential SWB. The statements are: I have felt particularly tense, I have felt lonely and I have felt downhearted and depressed.

(31)

The present study does not aim to distinguish between positive and negative

emotions, but rather acknowledges that both aspects are part of the overall experiential SWB. As such, the items measuring negative experiential SWB are reverse coded and merged with the items measuring positive experiential SWB.

Income inequality: moderator variable. The Gini coefficient measures the

statistical dispersion of income (or wealth) distribution among a nation’s residents (Yu & Wang, 2017). The scale goes from 0 to 1, where 0 expresses total equality. The study uses a Gini coefficient which was present in the downloaded data file.

Control variables. In accordance with previous studies on these subjects (e.g. Yu &

Wang, 2017; Ng & Diener, 2018) the present study uses the following control variables: GDP (provided on country level), and gender, education, marital status, children in the household and age (provided on the individual level).

Model specification. Given that this study is a cross-country analysis investigating

the effect of income and income inequality on individuals’ SWB, there are two levels/units of analysis – individual and country level. As such, the present study performs multilevel

modeling that allows nesting individuals within countries (Snijders, 2011; Cheung & Lucas, 2016). The model accounts for country-specific effects by including country-specific random intercepts; the remaining variables (i.e. the main variables of interest- income and income inequality as well as all control variables) are included in the model as fixed effects. As previous research points to differences between countries in terms of how income inequality affects the society, it becomes important to account for those differences, as random-effects procedures are often used when there is evidence of heterogeneity among the population effects, the model includes random intercepts (Hedges & Vevea, 1998).

(32)

Data Analysis/Results

Data Preparation

Measuring income: independent variable. When measuring income, it is important

to adjust for household size (Jebb et al., 2018) and purchasing power parity (PPP) within the country of analysis. As mentioned earlier, the dataset already includes an equivalised OECD measure, Y11_Income_percapita, which adjusts for both factors, with the OECD definition of household. To control this measure, a separate measure was created where PPP-income was adjusted for household size using the square root scale. Both measures gave similar results, and it was thus decided to proceed with the already equivalised OECD- measure,

Y11_Income_percapita, because adjusting for OECD’s definition of household is in line with income measurements in previous research.

A frequency check of Y11_Income_percapita revealed 10.815 missing values (24,8 % of the entire sample), those values are assumed to be missing (completely) at random.

Furthermore, due to the relationship between SWB and income being log-linear (Jebb et al., 2018) this measure is transformed using the natural logarithm of income. The transformation also allows obtaining an income distribution which resembles a normal distribution.

Apart from a linear main effect of income, the model also include higher order terms, that is a quadratic and cubic term for income. Those terms allow accounting for the expected non-linear relationship between income and SWB (Fermi, Pasta, Ulam & Tsingou, 1955; Williams, 2015). As such, introducing a quadratic term allows for one bend in the slope, while a cubic income term further enables two bends in the slopes if that occurs within the range of the data (Williams, 2015). These terms also allow for dealing with extreme income values. Furthermore, interactions between all three income terms and Gini were generated; following conventional practices all interactions were created using mean centered variables (which facilitates their interpretation).

(33)

Measuring SWB: dependent variable. Positive and negative experiential SWB are

measured on a six-point scale (1. All of the time – 6. At no time). The five items measuring positive SWB were thus reverse coded such that higher scores indicated higher positive SWB. These were combined with items measuring negative experiential SWB to form one scale (α = .872) (Bland & Altman, 1997).2 The reliability analysis also showed that

Cronbach’s α would not have been higher if any experiential SWB items were excluded, thus making it appropriate to keep all items in the experiential SWB scale. 134 respondents did not answer half or more of the experiential SWB questions, but due to the large number of respondents, they were kept in the analyses as results are not expected to differ.

The eight evaluative SWB items also demonstrated good reliability (α = .811). Cronbach’s α would have increased to .826 if Q40h Economic situation in [country]/How satisfied are you? was removed. However, because the reliability already is satisfactory, the difference is low and the notion that this item is rather relevant for the hypotheses in the present research, it was decided to keep this item in the scale.

One of the items Q40b Your present job / How satisfied are you? had 55,8% missing values. A potential explanation could be the age composition of the sample, because over a third of the respondents (36%) were either between 18-25 and > 64 and, thus likely to not have worked yet or anymore. This explanation appears even more plausible when looking at the variable Y11_Q1 Have you ever had a paid job? which has 44,5% missing values. Moreover, when asked about current situation 44,4 % of the sample answers employed or self-employed, while e.g. 29,5% are retired. Nevertheless, if this item was removed, Cronbach’s α would decrease to .781, leaving it relevant to keep this item in the scale. However, all subsequent analyses on evaluative SWB were conducted on two separate scales

2_{Above recommended level of 0.7/0.8 and below the recommended maximum level > .90. Results above this}

level might have suggested that some items are redundant as they test the same question, but in different forms (Tavakol & Dennick, 2011).

(34)

(one with and one without this item), but results did not differ. Thus, a scale with all the items is considered in the subsequent analysis, and the results obtained when using that scale are reported. It should also be noted that there were 110 respondents who did not answer half or more of the evaluative SWB items, but again: due to the large number of respondents they were kept in the analyses.

To conduct analyses with the scales, they were first standardized. However, this approach was dismissed and the scales was mean centered instead. This because with standardization it is challenging to know if the variables are standardized through the

standard deviation of Y (SWB) at the individual or group level (Bloom, Hill, Black & Lipsey, 2008). Also, standardization across the entire sample would create relative scores for non-relative samples due to there being large differences between European countries. With the mean centering approach, this challenge is avoided because such an approach, as opposed to standardization, allows for preserving the correlation with the raw data in multilevel analyses such as in the present study.

Measuring control variables. Two of the control variables (gender and marital

status) were transformed into dummies as they were categorical. Gender is measured by allowing respondents to answer 1 = Male and 2 = Female, and there were no missing values for this question. Marital status is measured through Q31 Could I ask you about your current marital status? Respondents were able to choose between four categories: 1 = Married or living with a partner, 2 = Separated or divorced and not living with a partner, 3 = Widowed and not living with a partner and 4 = Never married and not living with a partner. A

frequencies check revealed that there were 244 missing values who either refused to answer or did not know their marital status.

Q32 How many children of your own do you have? allowed respondents to state the number of children they had. There were 226 respondents who did not know or did not want

(35)

to answer. Moreover, education is measured by allowing respondents to state their completed level of education ranging from non to tertiary education – advanced level. The present study thus uses the already clustered measure ISCEDsimple which categorizes education levels based on the amount of education. There were 91 missing values for this question. Age is measured through the item HH2b Starting with yourself, what was your age last birthday? and there were no missing values for this item. Furthermore, the dataset already included an item measuring GDP per capita, namely gdppercapita.

While some of the variables have missing values, the number of those is rather negligible given the sample size and are thus not removed. Also, all missing values are assumed to be missing (completely) at random.

Results

After having prepared the data used to analyze the hypotheses, the results are

explained in the following section. First, table 1 provides some descriptive statistics; in more detail, it shows the means, standard deviations and correlations between the variables and interactions in the present study.

(36)

Table 1: Means, Standard Deviations and Correlations

Variables M SD EV SWB EX SWB

Log

income Income Income2 Income3 Gini

Gini x income Gini x income2 Gini x income3 Marital status Gender Nr. of children Age Edu-cation GDP EV SWB 6,87 1,56 (.81) EX SWB 4,36 0,96 .536** _(.87) Log income 6,66 0,90 .358** _.211** _- Income 1156,71 1956,83 .176** _.101** _.563** _- Income2 45,18 11,34 .369** _.220** _.975** _.658** _- Income3 310,92 116,45 .360** _.214** _.942** _.732** _.991** _- Gini coefficient 29,79 3,40 -.200** _-.131** _-.309** _-.111** _-.326** _-.313** _- Gini x income -0,90 2,85 -0,004 0,007 .012* _.057** _.063** _.063** _.065** _- Gini x income2 -0,06 9,75 -.075** _-.073** _.107** _.018** _-.022** _-.044** _.270** _-.356** _- Gini x income3 0,89 63,24 -.016** _0,011 _-.265** _.019** _-.110** _-.070** _-.046** _.531** _-.838** _- Marital status 1,88 1,20 -.065** _-.044** _-.094** _-.063** _-.104** _-.105** _.034** _0,011 _.031** _-0,011 _- Gender 1,57 0,50 -.044** _-.103** _-.064** _-.046** _-.073** _-.075** _.041** _0,000 _.020** _-0,001 _.028** _- Nr. of children 1,59 1,35 -.046** _-.054** _-.066** _-.025** _-.066** _-.063** _-.030** _-.035** _-0,003 _0,003 _-.408** _.070** _- Age 49,52 18,10 -.088** _-.105** _.013* _0,003 _0,008 _0,005 _.045** _-.029** _.016** _-0,002 _-.187** _.035** _.412** _- Education 3,46 6,26 .070** _.036** _.080** _.042** _.084** _.083** _-.044** _-.014* _-.014* _-0,004 _-.023** _-.019** _-.024** _-.063** _- GDP 23416,17 10148,73 .242** _.115** _.508** _.250** _.532** _.522** _-.396** _-.039** _-.121** _-.019** _-.035** _-.019** _.028** _.040** _.169** _-

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

(37)

As aforementioned, to test the hypotheses concerning a non-linear relationship

between income and SWB, a quadratic and a cubic term of income were introduced. Thus, for answering H1 and H2, the following regression equation was tested:

𝐸𝑋/𝐸𝑉 𝑆𝑊𝐵 =

𝛽0 + 𝛽1 𝑖𝑛𝑐𝑜𝑚𝑒 + 𝛽2 𝑖𝑛𝑐𝑜𝑚𝑒2 + 𝛽3 𝑖𝑛𝑐𝑜𝑚𝑒3 + 𝛽4 𝐺𝐷𝑃 + 𝛽5 𝑔𝑒𝑛𝑑𝑒𝑟 + 𝛽6 𝑒𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛 + 𝛽7 𝑚𝑎𝑟𝑖𝑡𝑎𝑙 𝑠𝑡𝑎𝑡𝑢𝑠 + 𝛽8 𝑐ℎ𝑖𝑙𝑑𝑟𝑒𝑛 𝑖𝑛 𝑡ℎ𝑒 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 + 𝛽9 𝑎𝑔𝑒

When testing H1, income is found to have a significant (joint) effect on experiential SWB, p< .05. The main (linear) effect of income is positive (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒} = .26) while both income squared (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}2 = −.17) and income cube (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}3 = −.01) are negative. The results for this model as well as all remaining models fitted as part of this analysis can be found in table 2.

Figure 1 plots the effect of income on experiential SWB, revealing that the effect of income is positive albeit to a point, after which it disappears and even becomes negative.

Figure 13_{. Effect of income on experiential SWB}

3_{The results for all hypotheses were plotted according to the mean range (-9.78 – 4.96) of the income variable.}

As income is measured through the natural logarithm of income, income as absolute value was regenerated using the exponential function on the log value and adding the mean value (μ = 6.6608).

-0,1 0 0,1 0,2 0,3 0,4 0,5 0,6 500 950 1700 3200 5700 10500 19000 42500 Experiential SWB Monthly income

(38)

Table 2: Regression analysis results

Model 1: EX SWB Model 2: EV SWB Model 3: Gini EV SWB Model 4: Gini EX SWB

Variable β coefficient t β coefficient t β coefficient t β coefficient t

Log income 0.261** 29.198 0.621** 45.487 0.628** 41.021 0.261** 25,871 Income2 -0.017** -4.806 -0.018** -3.294 -0.007 -1.204 -0.014** -3,664 Income3 -0.007** -12.000 -0.012** -13.860 -0.013** -11.255 -0.007** -8,787 Gini -0.016 -0.826 -0.008 -0,825 Gini x income 0.009 1.884 0.004 1,285 Gini x income2 -0.012** -6.749 -0.006** -4,78 Gini x income3 -0.002** -6.475 -0.001** -3,374 Married 0.231 1.347 0.171** 6.498 0.18** 6.487 0.036 1,957 Separated -0.231** -10.303 -0.438** -12.768 -0.431** -12.292 -0.214** -9,237 Widowed -0.265** -10.826 -0.170** -4.55 -0.131** -3.377 -0.241** -9,444 Unmarried 0b ₀b ₀b ₀b Male 0.136** 12.528 0.015 0.901 0.008 0.448 0.137** 12,045 Female 0b ₀b ₀b ₀b Nr. of children 0.007 1.497 0.002 0.301 0.001 0.195 0.006 1,101 Age -0.002** -6.78 -0.005** -8.998 -0.005** -8.825 -0.002** -5,455 Education 0.001 1.106 0.005** 3.919 0.004* 3.137 0.0005 0,567 GDP per capita 0.000 0.04 0.00001* 2.646 0.00001 1.760 -0.000002 -0,589

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Table 2. Regression analysis results

This confirms the presence of a satiation point after which income no longer increases experiential SWB. Thus, H1 is supported. In more detail, a satiation point is found for net monthly income at approximately €12.800 where EX SWB = 0.53 (mean centered). SD = €1.957. While experiential SWB actually goes down beyond this satiation point, it never goes below average within the range of these data.

When testing H2, income is found to have a significant (joint) effect on evaluative SWB, p< .05. Also here, the main (linear) effect of income is positive (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}= .62) while both income squared (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}2 = −.02) and income cube (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}3 = −.01) are negative. Figure 2 plots the effect of income on evaluative SWB, revealing that income exerts a positive effect, albeit only up to a point, after which it disappears and becomes slightly negative. Thus, H2 is not supported because a satiation point is also found for evaluative SWB. However, satiation occurs much later than for experiential SWB, around net monthly

(39)

income levels of €28.000 where EV SWB = 1.27 (mean centered). Furthermore, even though a turning effect has also been found, it is smaller than for experiential SWB, especially within the range of the data.

Figure 2. Effect of income on evaluative SWB

When investigating H3 and H4, the following regression equation was tested:

𝐸𝑋/𝐸𝑉 𝑆𝑊𝐵 =

𝛽0 + 𝛽1 𝑖𝑛𝑐𝑜𝑚𝑒 + 𝛽2 𝑖𝑛𝑐𝑜𝑚𝑒2 + 𝛽3 𝑖𝑛𝑐𝑜𝑚𝑒3 + 𝛽4 𝐺𝑖𝑛𝑖 + 𝛽5 𝑖𝑛𝑐𝑜𝑚𝑒 𝑥 𝐺𝑖𝑛𝑖 + 𝛽6 𝑖𝑛𝑐𝑜𝑚𝑒2 𝑥 𝐺𝑖𝑛𝑖 + 𝛽7 𝑖𝑛𝑐𝑜𝑚𝑒3 𝑥 𝐺𝑖𝑛𝑖 + 𝛽8 𝐺𝐷𝑃 + 𝛽9 𝑔𝑒𝑛𝑑𝑒𝑟 + 𝛽10 𝑒𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛

+ 𝛽11 𝑚𝑎𝑟𝑖𝑡𝑎𝑙 𝑠𝑡𝑎𝑡𝑢𝑠 + 𝛽12 𝑐ℎ𝑖𝑙𝑑𝑟𝑒𝑛 𝑖𝑛 𝑡ℎ𝑒 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 + 𝛽13 𝑎𝑔𝑒

When testing H3, and thus the moderating effect of Gini on the relationship between income and evaluative SWB, the overall effect of income is significantly moderated by the level of income inequality in the country, p<.05 (𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒 × 𝐺𝑖𝑛𝑖} = .01, 𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}2_{× 𝐺𝑖𝑛𝑖} = −.01, 𝑏_{𝑖𝑛𝑐𝑜𝑚𝑒}3_{× 𝐺𝑖𝑛𝑖} = −.002). This means that the effects of income on evaluative SWB differs depending on the level of income inequality. Figure 3 visualizes the simple slopes underlying these effects and thus the findings of H3. The graph plots the effect of income on evaluative

-0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 1,2 1,4 500 1000 2500 7000 19000 52000 Evaluative SWB Monthly income

Examining the income-SWB satiation point, and the potential moderating effect of income inequality