Wealth inequality, income inequality, and subjective well-being: A cross-country study

Although it may seem intuitive that the relationship between inequality at the societal level and individual-level well-being is negative, the empirical literature is far from unanimous on this question. In fact, there are good reasons for either a positive relationship, a negative one, or even a lack of any relationship between the two. People may dislike inequality when they expect to be on the losing side, while others may like it if they expect to benefit from it. Growing up during a recession, or during an economic transition, may explain the former prevailing attitude, while perceiving high income mobility perspectives may explain the latter. The outcome would also depend on risk aversion, with more risk-averse individuals willing to accept less inequality. The issue is further complicated by the multidimensional nature of inequality. Empirical literature dealing with the inequality–happiness puzzle has so far focused on income inequality [for summaries, see e.g., Di Tella and MacCulloch, 2006; Verme, 2011; Ferrer-i-Carbonell and Ramos, 2014; Clark and D’Ambrosio, 2015; Schneider, 2016]. To the best of our knowledge, wealth inequality has so far been overlooked in this strand of the literature, even though it occurs at much higher levels than income inequality in most, if not all, countries of the world, and also globally [Chancel et al., 2022]. Moreover, income and wealth, although interrelated, play different roles in people’s lives. Income is a flow, while wealth is a stock; hence, what is true of one is not necessarily true of the other. Income allows for current consumption, whereas wealth offers a safety net in the case of negative income shocks (such as unexpected job losses, uninsured emergencies, etc.) and may serve as collateral. Saving income helps to accumulate wealth and, at the same time, wealth may generate income, such as capital gains or imputed rents. Income and wealth do not always go hand in hand: one may be income-rich and wealth-poor (e.g., earning a lot, but living in a rented property) or income-poor and wealth-rich (e.g., living on a pension, but still owning a property). D’Ambrosio et al. [2020] show that both permanent income and permanent wealth are good predictors of life satisfaction, but their effects differ. Considering this, investigating the effects of wealth inequality on life satisfaction and comparing them to income effects seem justified.

There are several ways to measure the level of inequality. The happiness literature usually focuses on the most-popular Gini index which, although easy to calculate and understand, does not provide information on different parts of income or wealth distribution. It may be that people’s preferences toward inequality differ with respect to the top and the bottom of the distribution, for example. In this study, using cross-country panel regressions, we look at the top 1%, top 10%, middle 40%, and bottom 50% shares of income or wealth distribution as our inequality measures, which is quite rare to come across in this strand of literature.

The present study offers three main contributions. First, to our knowledge, it is the first paper that systematically investigates the wealth inequality–happiness relationship. Second, it reaches beyond the popular Gini index and studies inequality concentrated in different parts of the income or wealth distribution. Third, it precisely defines the income concept and differentiates between pre-tax and post-tax income, rather than using the vague term “income inequality.”

We herein aim to investigate two main research questions: Is there a relationship between decreasing country-level income inequality and increasing citizens’ subjective well-being? Is there a relationship between decreasing country-level wealth inequality and increasing citizens’ subjective well-being? In other words: do people necessarily dislike inequality or does the answer depend on whether we consider income or wealth, how we measure it, or even whether a given country is rich or poor?

We find a quite surprising pattern: when significant, the link between subjective well-being (by which we mean either a feeling of happiness or life satisfaction) and inequality is negative for the middle 40% share and positive for the top 10% and the top 1% shares. The results are stronger and more significant for wealth than for either pre-tax or post-tax income, but all of them yield a similar pattern. In other words, individuals seem to feel happier when the richest get richer and feel less satisfied with a higher share of income or wealth going to the middle class. Interestingly, we also find that the effects for wealth are largely driven by low- and middle-income countries’ citizens. Regarding the bottom 50%, we find some evidence of a positive attitude present among high-income countries’ citizens toward the larger share of income or wealth going to the poorest half.

The paper is structured as follows. Section 2 presents theoretical arguments both in favor of and against a negative inequality–happiness relationship and reviews the relevant literature, with a special focus on the wealth inequality–happiness link. Section 3 discusses the relevant data aspects and Section 4 lays out the methods used in the study. Section 5 presents the results. Finally, Section 6 concludes and discusses the possible limitations of the paper.

The data used in this study come from three main sources. The data on both income inequality and wealth inequality at the national levels come from the World Inequality Database (WID), the data on subjective well-being and other individual-level characteristics come from the Integrated Values Surveys (IVS), and the per capita GDP estimates come from the World Development Indicators, which is the major World Bank set of development indicators.

The usual approach in happiness studies, when modeling the relationship between individual-level well-being and a country-level indicator (such as inequality level), is to combine cross-country and longitudinal data [e.g. Alesina et al., 2004; Verme, 2011]. This involves using the so-called repeated cross-section data, wherein there is a panel of countries (although unbalanced in our case), and the number of individuals nested in the countries forms a different group each year. Their individual characteristics are then employed as covariates in the regression. We thus estimate the following equation: 1SWTijt=βInequality jt+yXijt+δYjy+Cj+Tt+εijt,$$SW{T_{ijt}} = \beta {\rm{Inequality}}{{\rm{\;}}_{jt}} + y{X_{ijt}} + \delta {{\rm{Y}}_{jy}} + {C_j} + {{\rm{T}}_t} + {\varepsilon _{ijt}},$$ where i denotes individuals, j denotes countries, and t denotes time. SWB stands for subjective well-being and is either Happiness or Life Satisfaction. Inequality is one of the twelve possible inequality indices: three possibilities regarding the income or wealth concept (pre-tax income, post-tax income, wealth) multiplied by four measures (top 1%, top 10%, middle 40%, bottom 50% shares). X is a vector of individual-level variables: gender, age, age squared, number of children, marital status, education, employment, self-assessed income position, self-assessed health, religiosity, and trust in other people. Y denotes the log of per capita GDP. In our primary specification, we include country (C_j) and year (T_t) fixed effects. We estimate Eq. (1) using an ordered probit model with standard errors clustered at the country level. We use individual sampling weights.

We start this section by presenting the descriptive statistics of inequality measures in our sample of countries (Table 1), in which it is clear that pre-tax income inequality is higher than post-tax inequality on average, demonstrating that the redistribution policies in most (if not all) countries do work. Moreover, wealth inequality is far higher than income inequality, with the top 10% in terms of wealth reaching the maximum of 86% and a mean of 62%, compared to 61% and 32% for post-tax income, respectively. Regarding our two dependent variables, happiness and life satisfaction, people are “rather happy” on average and choose number 7 on a 1–10 life satisfaction scale.

We now turn to the results of the ordered probit regressions for pre-tax income, post-tax income, and wealth, respectively. Tables 2–4 present average marginal effects for the probability of being very happy or being very satisfied with life, that is for the highest categories of the two variables. For pre-tax income, we observe that only the relationship between happiness and the middle 40% share is significant (Table 2). Interestingly, the sign of the relationship is negative, meaning that people are less happy with the middle class earning more.

Regarding post-tax income, the only significant relationship turns out to be between happiness and the bottom 50% share (Table 3). The positive sign of this relationship may indicate some preferences for redistribution, especially considering the insignificant corresponding relationship with pre-tax income (Table 2). An increase of the bottom 50% post-tax income share by 1 pp. translates to a 0.6 pp. increase in the probability of being very happy. This may seem not large, but the size of this effect is comparable to others found in the literature; for example, a similar study by Brzezinski [2019] finds the effects of the top 10% and top 1% income shares on the probability of being very happy or very satisfied with life in the range between 0.2 pp. and 0.6 pp.

Turning our attention to wealth, we see an interesting pattern. The relationship between the middle 40% share and both happiness and life satisfaction is negative (Table 4), while the relationship between the top 1% share and both happiness and life satisfaction are positive and of a slightly smaller magnitude. The link between the top 10% wealth share and happiness is also positive. We can interpret these results in the following way: increasing the wealth share of the middle class makes people less happy and less satisfied with their lives. At the same time, increasing the wealth share of the richest makes people feel happier. This may seem counterintuitive at first glance, but there are several possible explanations for these findings. First, the resentment toward the enrichment of the middle class may stem from the increasing polarization in many countries. Second, the positive attitude toward the richest getting richer may be a sign of the “tunnel effect,” as described in Section 2.1: respondents may believe that one day they will become rich too, and thus the higher the wealth of this elite group, the better. Moreover, they may attribute the wealth of the richest to their efforts, rather than luck, and simply believe that the richest deserve it. Korom [2023] summarizes the existing literature and concludes that the rich are perceived as “deserving” when their fortunes result from hard work and competencies, rather than from family gifts and bequests. He also finds that, in fact, modern European multimillionaires are the “hybrid rich,” namely entrepreneurs who benefit both from earned and unearned financial resources: on the one hand, the richest European households are the most likely to have received large family transfers, while on the other, they are also likely to have tertiary education and they derive most of their wealth from self-employed businesses [Korom, 2023].

The relationship between wealth inequality and well-being may be further investigated by dividing the sample of countries into “rich” and “poor” ones, based on the World Bank classification [World Bank, 2024]. Table 5 suggests that the results presented in Table 4 are largely driven by low- and middle-income countries. The relationship between wealth inequality and well-being is significant in all but one case, and its direction is the same as discussed before: negative for the bottom 50% and middle 40% and positive for the top 10% and top 1%. Contrary to this, the same relationship for high-income countries is almost always insignificant, except for the link between the bottom 50% and life satisfaction (Table 6). It appears that the “tunnel effect” discussed earlier, coupled with some reluctance to give to the poor, may play a role in catching-up economies, whereas citizens of wealthier countries may be more inclined to support redistribution policies.

5.1

Treatment of country and year fixed effects

The usual approach in the literature is to include country and year fixed effects (i.e., two-way fixed effects) in the regression of this (e.g., Eq. 1) or similar kind. There is, however, a discussion in the literature regarding the use of two-way fixed effects. Kropko and Kubinec [2020] claim that the two-way fixed effects specification is statistically undefined and produces results that are impossible to interpret. Fortunately, the reasoning presented by Kropko and Kubinec [2020] does not necessarily concern our case, because while using country fixed effects, we use data on individuals nested in countries, rather than solely countries. Regardless, the effects estimated in Tables 2–4 should be treated as “within-country, within-year” effects. In this section, we first remove country fixed effects from Eq. (1) and keep the year fixed effects. Second, we do the opposite, i.e., we remove the year fixed effects and keep the country ones. In the former case, we introduce between-country variation to the model; in the latter, we ignore changes over time.

Figure 1 presents the average marginal effects from ordered probit models without country fixed effects for the probability of being very happy and being very satisfied with life, respectively. We can see that the pattern generally holds: the effects, when significant, are negative for the bottom 50% and middle 40% shares and positive for the top 10% and top 1% shares. The effects are usually larger and more significant than in the main specification presented in Tables 2–4, which is especially visible for post-tax income, both for happiness and life satisfaction. The resentment toward the large share of (post-tax) income held by the middle class contrasts with positive attitudes toward the enrichment of the richest. Only the positive attitude toward the bottom 50% post-tax income share, previously interpreted as some preference for redistribution, seems to disappear in the between-country context. Regarding pre-tax income and wealth, the same negative–positive patterns hold, although a little less significant for the life satisfaction question than for the happiness one. We conclude from this that the between-country component in the study of subjective well-being—inequality relationship is not negligible.

Figure 1.

Turning now to Figure 2, in which we exclude year fixed effects, but keep country fixed-effects, we see that the results are less significant than in our primary specification. This is probably because there is relatively little country-level over time variation in both the SWB and inequality data, so the models tend to produce insignificant results. Another possible explanation is that such models mix up the past and the present. The attitudes toward inequality may evolve over time and such models simply ignore this fact, as well as all external shocks like wars or systemic changes. For instance, it may matter for countries that underwent an economic transition from centrally planned to market economies. Grosfeld and Senik [2010] find that, in the case of Poland, there was a structural break in the relationship between income inequality and satisfaction: first, an increase in income inequality was welcomed by the population as a sign of forthcoming opportunities but, after a few years, attitudes changed due to widespread dissatisfaction with the country’s economic outcomes.

Figure 2.

We study the link between subjective well-being and various measures of inequality using cross-country panel regressions. Our most important contribution is the focus on household wealth inequality. We find that individuals are happier with increasing the top 10% and top 1% shares of wealth and less happy with increasing the middle 40% share of wealth.

These effects are even stronger and apply also to post-tax income when we allow for between-country variation. Increasing the bottom 50% share of post-tax income also makes individuals happier, as if they favored income redistribution. When the sample of countries is divided into rich and poor ones, it turns out that these effects for wealth are largely driven by low- and middle-income countries, such as Brazil, China, Ecuador, Egypt, Iraq, Serbia, or Turkey, to name a few. On the other hand, the citizens of rich countries, such as Australia, Canada, Finland, France, Germany, or the United Kingdom, seem to be more supportive of redistribution policies and are more indifferent regarding top wealth inequality. In general, the effects for wealth are more often significant than for income, both pre-tax and post-tax, although all of them yield a similar pattern.

The present study suffers, of course, from certain limitations. First, inequality at the national level, and indeed at any level, is a concept difficult to grasp [Schneider, 2016]. It is unlikely that it is grasped equally well by all respondents, which may introduce bias, even if educational attainment is controlled for. Gimpelson and Treisman [2018] show, using a number of large, cross-national surveys, that people are generally wrong when asked about the level of inequality, and that the misperceptions are sizable. Second, a country population may not be an appropriate reference group. In other words, what matters for people’s subjective well-being may not be inequality at the country level, but rather inequalities in their closest neighborhood, age cohort, or among their co-workers. This may explain the relatively small effects found in the study. This issue requires further investigation, which is largely limited by the data availability. Hvidberg et al. [2023] link survey data to administrative data for Denmark and show that people assess income inequalities within their co-workers and education groups as significantly more unfair than overall inequality. To the best of our knowledge, there is no such study in a cross-country setting. Third, people’s perceptions about the extent of inequality are likely more relevant than objectively measured inequality. Gimpelson and Treisman [2018] show that there is a strong correlation between perceived inequality and the demand for redistribution or reported conflict between the rich and poor, while the same for the actual level of inequality does not hold. Also Hvidberg et al. [2023] show that people tend to underestimate inequality the most within their reference groups.