This chapter focuses on how actual and perceived inequality shape preferences for redistribution. It shows that demand for redistribution is closely related to concern over income disparities and what underlies them, i.e. perceptions of and preferences for economic inequality. Changes in actual inequality, as measured by conventional indicators, are associated with changes in demand for redistribution, but only as long as changes in concern evolve in the same direction. The effect of changes in inequality on demand for redistribution reflects both changes in relative income – by making some people poorer and thus more favourable to government intervention – and people’s own preferences for the aggregate level of inequality. Despite being related to inequality, demand for redistribution has increased only mildly over time and reacted only to a limited extent to rises in concerns and inequality. The chapter assesses possible explanations, based on observational evidence and a review of survey experiments.
Does Inequality Matter?
3. How does inequality shape the demand for redistribution?
Abstract
3.1. Perceptions of inequality and the demand for redistribution
Perceptions of and concern over inequality are key drivers of cross-country differences in preferences for redistribution
How do concern over and perceptions of income inequality and intergenerational persistence influence opinions of redistributive policies? In OECD countries, an average share of 70% of respondents in ISSP and Eurobarometer surveys agree that redistribution is the responsibility of the government (Figure 3.1). Confirming a long tradition of studies with a strong transatlantic perspective (see (Kambayashi and Lechevalier[1]) for a recent survey), the share who believe it is the government’s duty is lowest in the United States. It is low, too, in Australia, Great Britain and New Zealand. It is highest in European countries, particularly Latvia and Portugal. Within Europe, the share tends to be lower in Nordic countries, where redistribution was extensive until the early 2000s. Countries outside the Europe-Anglosphere ambit are scattered across the distribution. In Japan, a lower-than-average share of people deem that the government should reduce differences in income, while Israel, Mexico and Turkey are located in the middle of the distribution.
Figure 3.1. On average, a large share of people believe it is the responsibility of the government to reduce income differences
Note: Respondents are asked their opinion about the statement “It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes.” In Eurobarometer the statement is slightly different – “The government in (OUR COUNTRY) should take measures to reduce differences in income levels” – while the scale of answers is identical. OECD averages are unweighted averages across the OECD countries included in the figure.
Source: OECD calculations from ISSP 2017, and Eurobarometer 471/2017 for Belgium, Estonia, Greece, Ireland, Italy, Luxembourg, Latvia, Malta, Netherlands, Poland, Portugal and Slovenia.
Demand for government intervention in tackling inequalities and concern over income disparities are closely related (Figure 3.2). In countries where concern is greatest, respondents are more likely to believe that reducing income differences is the government’s duty. The main determinants of international differences in concern about income disparities (see Chapter 2) account for as much as 60% of the cross-country variation in demand for redistribution (Table 3.1). A wider gap between perceived and preferred top-bottom earnings disparities is positively associated with demand for redistribution, which suggests that perceptions and preferences are crucial in shaping support for redistributive policies. Also associated with strong demand for redistribution are perceptions of strong generational persistence. By contrast, where people firmly believe in meritocracy, i.e. the importance of hard work, there is less support for redistribution.
Figure 3.2. Concern over income disparities is closely related with the demand for redistribution
Note: see Figure 3.1. The dotted line is the linear fit.
Source: OECD calculations based on ISSP 2017, and Eurobarometer 471/2017 for Belgium, Estonia, Greece, Ireland, Italy, Luxembourg, Latvia, Malta, Netherlands, Poland, Portugal, Romania and Slovak Republic. For Slovenia, redistributive preferences are from Eurobarometer 471/2017, while concern over income disparities are from ISSP 2017.
Table 3.1. The main determinants of concern over income disparities also drive differences in demand for redistribution in all countries
Percentage point increase in the share of respondents who strongly agree with the statements “Income differences are too large” and “It is the responsibility of the government to reduce income differences”, associated with one percent (or one percentage point) increase for various factors
Note: *** denotes statistically significant at the 1% level; ** at the 5%; * at the 10%. Robust standard errors in parentheses. The OLS regressions uses as explanatory variable the average logarithm of the perceived/preferred top-bottom earnings ratio; the regression coefficients can therefore be interpreted as the percentage change in the share of respondents who strongly agree that income disparities are too wide, associated with a 1% change in the perceived top-bottom earnings ratio. The average logarithm of top-bottom earnings is calculated using the reweighted sample to account for missing values. The sample includes all OECD countries available in ISSP 2009 for which it was possible to calculate all variables.
Source: OECD calculations based on ISSP 2009.
Recent results from the 2020 Risks that Matter survey confirm the key role of perceptions of income inequality and intergenerational persistence in shaping demand for redistribution. The survey asks respondents about their preferences for more redistribution than current levels. Countries where people perceive the richest 10%’s income share to be high express strong demand for greater government intervention to reduce income disparities (Figure 3.3).1 The same is true of perceptions of wider intergenerational disparities.
Strong perceptions of income inequality and intergenerational persistence are also associated with demand for more progressive taxation – even more closely, in fact, than with support for general government intervention.
Figure 3.3. Demand for redistribution and progressive taxation is higher in countries where people perceive higher inequality and less social mobility
Note: Blue lines are linear fit lines. The top panel shows answers to the question: “Governments can reduce income differences between the rich and the poor by collecting taxes and providing social benefits. In your country, do you think the government should do more or less to reduce income differences?” The bottom panel shows answers to the question: “Should the government tax the rich more than they currently do in order to support the poor?”. Individuals answering “Can’t choose” are not considered in calculating the shares.
Source: OECD calculations from the 2020 Risks that Matter Survey.
Individuals’ perceptions of country-wide inequality matter as much as their own income in explaining demand for redistribution
At the individual level, demand for more redistribution and progressive taxation varies from one socio-demographic group to another. These differences, reported in Panel A of Figure 3.4 and estimated by keeping other characteristics constant, confirm previous evidence (Alesina and Giuliano, 2011[2]).
Individuals from high-income households are less favourable to redistribution. The finding is consistent with a standard economic model in which individuals balance the personal gains of redistribution with its costs (Meltzer and Richard, 1981[3]; Rueda and Stegmueller, 2019[4]). Gains for high-income households may be greater equality or other indirect benefits, such as increased national productivity thanks to wider access to education. As for costs, they may be aggregate – if redistribution reduces incentive, for example – or specific to the individual, e.g. higher taxes for the rich (see Section 3.3).
The importance of socio-economic status is not confined to an individual’s current situation, but extends to the whole household’s well-being and income. Respondents who are concerned about their household’s well-being in the next few years, for example, are more likely to call for redistribution and progressive taxation (Figure 3.4, Panel B). This confirms that future prospects of upward or downward mobility are an important determinant of people’s preferences for redistribution (Benabou and Ok, 2001[5]). This also because people are, at least to some degree, risk averse, and higher risk aversion is associated with more demand for redistribution as insurance against future shocks (Gärtner, Mollerstrom and Seim, 2017[6]).
For their part, university graduates are less inclined to demand more redistribution, possibly because of their upbringing or their expectations of high earnings later in life. The unemployed and tenants demand more redistribution, presumably because they are more uncertain about the future. Women and older people, too, are more likely to demand greater redistribution (Box 3.1), although the age effect dwindles in older cohorts (its relationship with demand for redistribution is concave).
Analysis of demand for more progressive taxation yields similar results, albeit with some differences. The negative association with disposable income is even more marked, while there are no significant differences between men and women, or between couples and single respondents.
Perceptions of income inequality and intergenerational persistence are important factors in shaping demand for redistribution, even after controlling for different socio-demographic variables. An increase of 40 percentage points in either perception (equivalent to a climb from the 25th to the 75th percentile of the distribution) is associated with a 5‑percentage point rise in the share of respondents wanting more government intervention (Figure 3.4, Panel B). This finding is similar to the difference in demand for redistribution between people at the top and bottom of the income distribution. Perceptions of increases in inequality during the previous 10 years are even more strongly associated with demand for redistribution. All these findings are consistent with Fong (2001[7]), who shows that people’s beliefs about income distribution and reasons for inequality matter as much as personal income in explaining individual preferences for redistribution. Perceived macro-economic trends are also relevant. Pessimistic views of the country’s changing economic situation in the previous 12 months strengthen demand for redistribution, much as the worsening of a household’s financial situation does.
Figure 3.4. Individual perceptions of inequality matter for demand for redistribution
Note: Demand for redistribution refers to respondents who answer “more” or “much more” in response to the question “In your country, do you think the government should do more or less to reduce income differences?”, while for tax progressivity it refers to those who answer “yes” or “definitely yes” to the question “Should the government tax the rich more than they currently do in order to support the poor?” The differences associated with different groups, characteristics and perceptions are estimated with other characteristics kept constant, using a multivariate probit model that includes all the characteristics reported in the table (age squared is also included), plus country fixed effects. The reported differences are average marginal effects. Respondents who answer “Can’t choose” are not considered.
Source: OECD calculations from the 2020 Risks that Matter Survey.
Perceptions of income inequality and intergenerational persistence also shape redistributive preferences, whether or not both or either are perceived to be widespread (Table 3.2). Even when individuals believe that intergenerational persistence is low, their perception that income inequality is high is associated with a stronger demand for redistribution, and vice versa. There is thus no full trade-off between intergenerational persistence and income equality in people’s opinions. If equal opportunities at birth made income inequality fully acceptable, then perceived income inequality would not matter when individuals perceive little incidence of intergenerational persistence. The survey experiment conducted by Amiel et al. (2014[8]) supports the conclusion that people have preferences on both counts. Amiel et al. presented different patterns of intergenerational income mobility and inequality to a number of university students from Israel, Italy and the United Kingdom.2 The conclusion was that respondents value both components and are willing to trade them off against each other only in special circumstances.
Table 3.2. Both perceived income inequality and intergenerational persistence shape demand for redistribution
Percentage share of respondents who demand more redistribution, by type of perceptions
Note: Shares have been calculated keeping all other dimensions (socio-demographic characteristics and other perceptions, apart from the beliefs about the evolution of income inequality in the past year) constant, using a probit model as in Figure 3.4, where perceptions have been added as dummies for each category and fully interacted.
Source: OECD calculations from the 2020 Risks that Matter Survey.
People may value income inequality independently from intergenerational mobility because they hold other beliefs about reasons for inequality. As discussed in Chapter 2, the literature examines opinions that circumstances beyond the control of individuals matter more (or less) than hard work for personal success (Piketty, 1995[9]; Alesina and Angeletos, 2005[10]; Fong, 2001[7]). This chapter, too, has already stressed the importance of such beliefs (Table 3.1). Further confirmation comes from the fact that demand for more progressive taxation is lower in countries where a larger share of people believe that the reason why some individuals live in poverty is not social injustice or bad luck, but laziness or lack of willpower (Figure 3.5).
A stream of literature in experimental economics has tried to address the relevance of beliefs about reasons for inequality by putting individuals in laboratory settings where researchers randomly manipulate the sources of income. Durante, Putterman and van der Weele (2014[11]) find that preferences for greater equality are lower when the initial distribution is assigned according to the participants’ performance in some task (a quiz or skill game), and higher when they are allotted an income arbitrarily (either randomly or relative to the average income in their place of residence). In a redistributive experiment with representative samples of participants from 60 countries, Almås et al. (2020[12]) asked them to choose whether to change the pay gap between two workers in a real-life situation. The results reveal that, when the pay gap depended on the workers’ performance, respondents were more reluctant to narrow the gap.3
Figure 3.5. Demand for more progressive taxation is lower where people believe that poverty is due mostly to lack of personal effort
Source: OECD calculations from the 2018 Risks that Matter Survey.
Box 3.1. Gender differences in redistributive preferences
On average, women demand more redistribution than men do in OECD countries (Figure 3.6). The trend is not attributable to socio-demographic differences (e.g. income, employment status), and is consistent with a large body of evidence from other surveys (Alesina and Giuliano, 2011[2]; Goerres and Jæger, 2015[13]; Luttmer and Singhal, 2011[14]), and is found in almost all countries, although not always statistically significant. The female-male ratio of support for redistribution (controlling for differences in other socio-demographic characteristics) is slightly negative, though not statistically significant, only in Mexico, Estonia and the Netherlands, while it is the highest in Israel and Norway.
Women’s stronger preference for redistribution is confirmed by laboratory experiments in which researchers manipulate the initial income distribution and sources of disparities among participants, then let them choose whether to alter the levels of inequality. Female participants tend to choose lower levels of inequality (Durante, Putterman and van der Weele, 2014[11]). The main difference appears when the reason for inequality is performance rather than luck or socio-economic background (Buser et al., 2020[15]). Although all participants scale down their redistribution preferences when initial disparities are due to differences in performance on some task, women do so by much less.
Figure 3.6. Women demand more redistribution than men
Note: The differences net of other characteristics are estimated using a multivariate probit regression that includes the socio-demographic characteristics as in Figure 3.4.
Source: OECD calculations from the 2020 Risks that Matter Survey.
The gender difference for redistributive preferences is no greater in countries with wide gender wage gaps, which suggests that the difference is not driven simply by the condition of women in the country (Figure 3.7). In Korea and Estonia, which both have wide gender wage gaps, there is no statistically significant difference between men’s and women’s attitudes to redistribution. In Norway and Slovenia, where the gender wage gap is relatively narrow, women are much more favourably disposed to redistribution than men.
Figure 3.7. The gender difference in demand for greater redistribution is not driven by gender wage gaps
Note: The demand for more redistribution is the share of respondents who answer “more” or “much “more” in response to the question whether the government should do more than it currently does to reduce income differences between the rich and the poor. The gender wage gap is evaluated at the median and refers to the latest available year.
Source: OECD calculations from the 2020 Risks that Matter Survey and OECD Earnings Database.
A recent in-survey informational experiment carried out in the United States by Settele (2021[16]) finds that respondents’ concern over gender disparities are influenced by information. The author provides randomly selected respondents with two different estimates of the gender wage gap. Those who are shown a wide gender wage gap express greater concern and say the government should do much more to narrow the disparities. However, support for individual policies varies little, apart from a moderate rise in support for stricter equal pay legislation and more robust affirmative action.
One explanation for the limited support for the different policies is that a sizeable share of respondents has little faith in their effectiveness for reducing gender disparities. Another explanation is that, while overall support for some interventions increases, different respondents express different preferences for different policies, so that no individual policy has much support. Both explanations are consistent with results from experiments in which participants are given information on actual inequality (see Section 3.4).
Experiencing hardship during the COVID-19 crisis is associated with greater demand for redistribution
People who experience health problems, economic hardship, or a worsening of household finances during the COVID‑19 pandemic tend to call for more redistribution and progressive taxation (Table 3.3, Columns 1 and 4). OECD (2021[17]) further shows that household insecurity during the COVID-19 crisis is associated with higher demand for social protection. In a survey in the United States in October 2020, Klemm and Mauro (2021[18]) also find that people who have lost their job or been seriously ill, or whose loved ones have, are more favourable to progressive taxation. These results are in line with evidence from Alesina and Giuliano (2011[2]) that negative shocks boost preferences for redistribution.
Increases in perceived inequalities and perceived household risk may explain the positive association between exposure to hardship during the COVID‑19 crisis and demand for redistribution. The evidence from Risks that Matter shows that both perceived inequalities and risk are factors in explaining the association. Hardship experienced during the COVID-19 crisis heightens perception of inequality. After controlling for the heightened perception, the association between hardship and demand for more redistribution weakens (Columns 2 and 5). Hardship also heightens the perception of household risk. Again, taking the heightened perception of risk into account (Columns 3 and 6) further lessens the relationship between hardship and demand for more redistribution.
Table 3.3. Experiencing hardship during the COVID‑19 pandemic is associated with demand for greater redistribution
Percentage point increase in the shares of respondents who demand more redistribution and progressive taxation associated with changes in different characteristics
Note: * denotes statistically significant at the 10% level, ** at 5%, *** at 1%. Robust standard errors in brackets. The results are average marginal effects from probit regressions, including country fixed effects and weighting by sample weights (rescaled so that weights sum up to 1 within each country). Household and individual characteristics are the same as those included in Table 2.2 and Figure 3.4. Experiencing any health or economic hardship includes having experienced physical or mental health problems because of the pandemic, having experienced job-related disruption during the pandemic, or having had difficulties in making ends meet during the pandemic.
Source: OECD calculations from the 2020 Risks that Matter Survey.
Unfortunately, it is not possible to test whether respondents who experience hardship also wanted greater redistribution before the crisis, because Risks that Matter does not interview the same individuals over time. However, at the country level, average demand for more progressive taxation in the previous Risks that Matter wave (2018) does not correlate with the share of respondents who experience hardship due to COVID‑19 (Figure 3.8). By contrast, there is a positive association between demand for progressive taxation in 2020 and the share of people who experienced hardship.4 In other words, the change in demand for redistribution is positively associated with the reported impact of the COVID-19 crisis on the respondents and their household members, suggesting that the perceived impact of the crisis might have increased preferences for more progressive taxation. This result is consistent with the findings of Giuliano and Spilimbergo (2013[19]), who show that experiencing a recession – particularly when growing up – leads to higher demand for redistribution. Gualtieri et al. (2019[20]), too, provide evidence that experiencing trauma – e.g. an earthquake – increases demand for redistribution.
Figure 3.8. Demand for more progressive taxation increased the most in countries where more respondents reported hardship during the COVID-19 crisis
Note: Demand for more progressive taxation is measured as the share that answer “yes” or “definitely yes” to the question “Should the government tax the rich more than they currently do to support the poor?” (excluding respondents who choose “Can’t choose”). Experiencing health or economic hardship during the pandemic refers to respondents who report that they themselves or a member of the family experienced physical or health problems because of the pandemic, experienced loss of employment during the pandemic, or had difficulties in making ends meet during the pandemic. For consistency with the 2020 wave, the 2018 values have been calculated only on individuals aged 18-64. Exact differences between the two Risks that Matter waves should be interpreted with some caution due to adjustments in sampling methods and coverage.
Source: OECD calculations from the 2018 and 2020 Risks that Matter Surveys.
3.2. Actual inequality and demand for redistribution
Rising income inequality is associated with greater demand for redistribution
When compared at a single point in time, countries which conventional indicators show to have higher levels of inequality (e.g. as measured by the Gini index) do not have higher shares of respondents who agree that it is the government’s duty to reduce income differences (Table 3.4, Column 1). Previous studies report zero (Bussolo et al., 2019[21]) or even negative correlations (Kerr, 2014[22]).
Table 3.4. Changes in actual income inequality explain changes in preferences for redistribution, but the association is driven by changes in concern
Note: *** denotes statistically significant at the 1% level; ** denotes statistically significant at the 5% level; * denotes statistically significant at the 10% level. Standard errors clustered by country in parentheses. Eurobarometer is used for countries not surveyed in ISSP 2017, as in Table 3.1. Data for the Slovak Republic in 2017 are based on Eurobarometer because ISSP 2017 for Slovakia does not include the question on preferences for redistribution. The results are robust to introducing a dummy for the Eurobarometer observations and macro-variables (employment rate, unemployment rate and the logarithm of GDP per head in USD 2015 PPP).
Source: OECD calculations based on ISSP 1987, 1992, 1999, 2009, 2017 and Eurobarometer 2017 for preferences for redistribution; OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD) for inequality and other variables.
Nevertheless, changes in inequality indicators correlate positively with changes in demand for redistribution. Where inequality grows the most, so do preferences for redistribution (Columns 2‑3). These findings are consistent with several papers, though not all, which look at within-country changes in inequality and preferences for redistribution (Kerr, 2014[22]; Olivera, 2015[23]; Kuhn, 2019[24]).5 A positive association between inequality and preferences for redistribution also emerges from studies which look at regional variation within countries, such as Rueda and Stegmueller (2019[4]) and Colagrossi, Karagiannis and Raab (2019[25]).
What seems to matter the most in shaping preferences for redistribution is market inequality within the working-age population. Importantly, though, disposable income differences between generations matter as well. In countries where the elderly are relatively better-off, demand for redistribution is lower. For the working age population, the results may be explained by the prospect of upward mobility (POUM) hypothesis (Benabou and Ok, 2001[5]). POUM conjectures that if people expect to climb the income ladder in the future, they will be more reluctant to support redistribution policies. If all inequalities within the working-age population are held equal, differences between the young and the elderly are an indicator of future income prospects.
The association between income inequality and preferences for redistribution is driven by rising concern over income disparities
Changes in concern over income disparities explain the relationship between changes in inequality indicators and preferences for redistribution. Once levels of concern are taken into account, the correlation between the Gini index and demand for redistribution becomes slight (Table 3.4, Columns 6 and 7). The subjective factors embedded in people’s concern over income disparities – perceptions of and preferences for inequality – thus appear more relevant than actual inequality for explaining the demand for redistribution, as Gimpelson and Treisman (2018[26]) also argue.
It should not be inferred, however, that preferences for redistribution and actual inequality are disconnected. On the contrary, Chapters 2 and 3 show that when actual inequality (as captured by the Gini index) rises, so does concern about it, and demand for redistribution grows. However, when concern does not change with inequality, neither do redistributive preferences – if perceptions do not fully incorporate the new level of inequality, for example, or a change in other relevant subjective factors (e.g. preferred income disparities or belief in hard work) lessen the impact of inequality. Moreover, increased concern is associated with greater demand for redistribution even when actual inequality remains stable.6
There are indications that support for redistribution has increased during the COVID-19 crisis
Recent evidence from the Risks that Matter survey suggests that the cross-country correlation between indicators of inequality and redistributive preferences might have strengthened during the COVID-19 crisis.7 Indeed, demand for both more redistribution and more progressive taxation is stronger in countries with higher levels of inequality in 2020 (Figure 3.9), unlike Risks that Matter 2018, which observed no association (OECD, 2019[27]).8
Figure 3.9. Demand for redistribution is positively associated with income inequality
Note: The Gini coefficient for disposable income refers to 2018, apart from 2017 for Chile, Denmark, Germany, Ireland, Italy, Switzerland, the United States, 2016 for Mexico and the Netherlands, and 2015 for Turkey 2015.
Source: OECD calculations from the 2020 Risks that Matter Survey and OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD).
Demand for more progressive taxation increased the most between 2018 and 2020 in countries where inequality was already widespread before the pandemic (Figure 3.10). This relation is explained by differences in the share of respondents that reported experiencing hardship during COVID-19. Two mechanisms may explain this trend. The first, discussed above, is that the crisis exposed pre-existing inequalities, therefore raising awareness of inequality in countries where levels were already high. The second is that the experience of hardship has been more widespread in countries that were unequal even before the crisis. The experience of hardship thus likely drives demand for progressive taxation either directly or by exposing pre-existing disparities.
Figure 3.10. Recent changes in the demand for progressive taxation have been greater in countries that are more unequal
Note: The share of respondents who demand more progressive taxation refers to those who answer “yes/definitely yes” to the question whether the government should increase progressivity to support the poor (excluding those answering “Can’t choose”). The linear fit net of differences in hardship during the COVID-19 crisis is obtained by first netting out for the share of respondents who report having experienced health or economic hardship during the COVID-19 crisis. For consistency with the 2020 wave, the 2018 values have been calculated only for individuals aged 18‑64. Changes between the two Risks that Matter waves should be interpreted with some caution due to adjustments in sampling methods and coverage.
Source: OECD calculations from the 2018 and 2020 Risks that Matter Surveys, and OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD).
Higher redistribution lowers people’s demand for further intervention
People’s preferences for redistribution depend on its current level. If taxes and transfers already control income disparities effectively, then people are likely to be less concerned about inequality and do not demand more redistribution. Assessing the relationship between the current level of and demand for redistribution is complicated for one important reason – voters’ preferences. They may well determine levels of redistribution, as exemplified by the wide differences in redistribution levels between the United States and Europe.
Risks that Matter partially helps address the relationship between the current level of and demand for redistribution, as it explicitly asks respondents whether they want more (or less) redistribution. It may therefore be expected that, for a given level of market inequality, redistribution that is currently of a high level might reduce demand for more of it.
To help unbundle the association between the current level of and demand for redistribution, it makes sense to measure income inequality and redistribution only within the working-age population for two reasons. First, it appears more salient (Table 3.4) and, second, because assessing redistribution among the elderly is complicated by international differences in pension systems.
The measure of redistribution used is the gap between inequality in market and disposable income, which the literature refers to as the Reynolds-Smolensky (RS) index. The hypothesis is that demand for more redistribution is high in countries where market inequality is also high, and that extensive redistribution reduces demand for further increases. The hypothesis is confirmed by respondents who answer that the government should do much more than it currently does (Figure 3.11, left-hand panel). In the countries where market inequality is greater, so are preferences for redistribution, while more redistribution curbs demand for it. The same is true of demand for more progressive taxation (Figure 3.11, right-hand panel).
Figure 3.11. Demand for more redistribution is lower in countries where the current level of redistribution is higher
Note: Each bubble refers to a country. The RS index is the difference between the Gini index for market income and the Gini index for disposable income for the working-age population. A higher value indicates that taxes and transfers reduce inequality to a larger extent. Mexico and Turkey are not included because data for these countries exclude taxes paid.
Source: OECD calculations from the 2020 Risks that Matter Survey and OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD) .
Even when preferences for redistribution are expressed with respect to the current level, other factors that drive actual levels of and demand for redistribution may still influence cross-country comparisons. As Chapter 2 shows, differences between countries in income equality preferences are both wide and persistent.
To account for persistent country differences, Olivera (2015[23]) focuses on changes in redistribution preferences, using a pseudo-panel approach applied to ESS survey data, combined with market income inequality data from the Standardized World Income Inequality Database and public social protection expenditure (in % of GDP) from Eurostat. He finds that when market income inequality rises, so do redistribution preferences, while they fall when social protection is strong. Table 3.5 shows similar within-country regressions from the ISSP panel combined with IDD data. The signs of the coefficients in Column 1 confirm that redistributive preferences increase when market income inequality grows, and fall when levels of redistribution and social protection are higher. However, the coefficient on the RS index is not statistically different from zero.9 There is a more sizeable negative association with the amount of social expenditure per head (Column 3), which is also used in the literature as a proxy for redistribution (Karabarbounis, 2011[28]).
Table 3.5. Changes in redistribution preferences are negatively associated with changes in redistribution, but the link is weak
Percentage point increase in the shares of respondents who agree it is the government’s duty to reduce income differences associated with a 1% (percentage point for fractions) increase in...
Note: Standard errors clustered by country in parentheses. The RS index is the distance between the Gini coefficients for market income and for disposable income (the higher the index, the stronger is redistribution). The analysis includes the logarithm of GDP per capita because the amount of redistribution is likely to depend on fiscal constraints. The lag refers to the previous wave year (statutory if the country was not observed or actual if the country was observed, and to 5 years before for wave 1987).
Source: OECD calculations based on ISSP 1987, 1992, 1999, 2009, 2017 and Eurobarometer 2017; OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD) and OECD Social Expenditure Database (https://www.oecd.org/social/expenditure.htm).
3.3. Inequality, relative income and preferences for redistribution
High-income individuals demand less redistribution, but not because they perceive smaller disparities
Individuals’ relative income is a key driver of preferences for redistribution. People in the upper part of the income distribution have lower redistribution preferences, although even among them a sizeable share of them agree that the government should do more to reduce income disparities.
One interpretation of differences in preferences between high- and low-income individuals – rooted in the Meltzer and Richard (1981[3]) median voter model – is based on personal gains and losses from redistribution. If individuals cared only about their own consumption, those with above-average income would oppose redistribution, and those on below-average income support it.
An alternative interpretation is that poorer individuals overestimate inequality, while those in the upper income tertile underestimate it. However, evidence from the 2020 wave of Risks that Matter survey does not lend support to this hypothesis (Figure 3.12, Panel A). On average, differences in perceptions of income inequality between the lowest, middle and top household-equivalised income tertiles are quite narrow. In some countries, such as Austria, France, Poland, Korea and Turkey, richer respondents even perceive the richest 10%’s share of income to be greater than other groups do. Indeed, if their social network is narrow and their information limited, the better-off may have better knowledge of the income levels of those at the top of the distribution, but know less about those at the bottom (Cruces, Perez-Truglia and Tetaz, 2013[29]).
Despite their similar perceptions of income inequality, people on high incomes favour less redistribution in most countries (Figure 3.12, Panel B). The few exceptions are countries where demand for redistribution is high in all income groups, as in Chile, Spain, Greece, Lithuania, Mexico and Turkey. In the United States, where there is little support for government intervention in narrowing income disparities, demand for more redistribution is low in all income classes. Generally, the middle income tertile in most countries shows redistribution preferences that are closer to people on low incomes than on high ones.
Figure 3.12. Perceptions of income disparities are similar in all income groups, but preferences for redistribution are not
Source: OECD calculations from the 2020 Risks that Matter Survey.
As for earnings inequality, high-income individuals again do not always perceive lower disparities. In the OECD countries observed in ISSP 2009, individuals in the top income tertile actually report wider perceived top-bottom earnings disparities than other income groups (Figure 3.13). Their preferred disparities help understand how that finding squares with the lower demand for redistribution observed in the top income tertile. In almost all countries, high-income individuals prefer higher earnings disparities. As a result, in the vast majority of countries, the distance between perceived and preferred disparities is lower in the top-income tertile than in other groups.
Figure 3.13. Both perceived and preferred earnings disparities are wider among high-income individuals
Note: Average across 28 OECD countries (see Figure 2.8, Panel B). Individuals in each country and period are divided into tertiles based on the household income variable (equivalised) available in ISSP. Missing data were imputed following the procedure illustrated in Annex A in Ciani et al. (forthcoming[30]).
Source: OECD calculations from ISSP 2009.
A different interpretation of the negative relation between own-income and redistributive preferences is that the better- and worse-off have different beliefs about the reasons for inequality. Piketty (1995[9]) proposes a model in which people draw on their own experience to learn about the actual rate of social mobility in their country. Those who meet with success, and eventually become rich, end up believing that hard work pays and are therefore less favourable to redistribution. Fong (2001[7]) shows that the effect of the belief in hard work on redistributive preferences accounts for a sizeable share of the association between redistributive preferences with income.
It might therefore follow that the rich should take a more optimistic view of the aggregate level of intergenerational upward mobility. However, that is not true of most countries (Figure 3.14). In several, high-income individuals actually believe in weaker intergenerational mobility with respect to what low-income individual believe. Alesina, Stantcheva and Teso (2018[31]), drawing on data for France, Italy, Sweden, the United Kingdom and the United States, suggest that this apparent conundrum is attributable to contradictory views of intergenerational mobility and reasons for personal success. Even if high-income individuals think that most people remain stuck in the same income bracket as their parents, they tend to justify their own position with the belief that, eventually, individual effort pays off. Alesina, Stantcheva and Teso (2018[31]) also find that individuals at the top of the income distribution are more likely to agree that hard work is the main reason for being rich.
Figure 3.14. Richer people are often less optimistic about intergenerational mobility
Note: Intergenerational persistence among the poorest 10% refers to the likelihood that a child from a household in the poorest 10% of the income distribution will remain in the same decile upon becoming adult. Income tertiles are calculated on equivalised household disposable income.
Source: OECD calculations from the 2020 Risks that Matter Survey.
The fact that better-off people are less in favour of redistribution does not mean that only the poor drive demand for redistribution up or down over time. Indeed, in some countries where preferences for more redistribution have grown over time, that growth was stronger among those in the top income tertile of the distribution (Figure 3.15). Cases in point are the United States, Italy, Slovenia and Switzerland, while in Hungary and Australia the increase was greater in the bottom tertile. Even where average redistribution preferences declined, it was not always attributable to the rich. In the United Kingdom and Austria, the drop was more pronounced among the poor. Furthermore, changes in the preferences of the middle income tertile do not always lie in between the top and bottom tertiles. In Germany and Austria, middle-income demand for redistribution declined less than that of high and low income individuals, while in Australia it rose by a smaller extent than for the other two groups.
Figure 3.15. In some countries, the redistribution preferences of the better-off grew more steeply over time
Note: Households are divided into tertiles corresponding to the equivalised income distribution of each of the country in each period (1987‑1992 and 2009) according to the household income variable available in ISSP. Missing data were imputed, see Annex A in Ciani et al. (forthcoming[30]) for details and further information on the collection of income data in ISSP.
Source: OECD calculations based on ISSP 1992, 1987, 2009.
Greater income inequality shapes individual demand for redistribution through both relative income and social preferences
Growth in income inequality may increase demand for redistribution through two channels (Alesina and Giuliano, 2011[2]; Rueda and Stegmueller, 2019[4]):
The first is through changes in people’s relative income. This is the logic behind the Meltzer and Richard (1981[3]) median voter model. When inequality rises, the median voter becomes poorer than the average. Based on models in which voters care only about their individual gains and losses, median voters’ relative impoverishment makes them favourable to redistribution. As the income distribution is skewed, the greater the inequality, the higher the share of individuals who demand more redistribution or progressive taxation.
The second is through people’s preferences for equality. Even assuming that only their own consumption matters to people, the macro-level of inequality matters if it affects their consumption – when, for example, greater inequality reduces GDP per capita growth and limits educational investment (Cingano, 2014[32]), or when inequality and poverty increase the fear of crime (Rueda and Stegmueller, 2015[33]). More broadly, people have preferences for macro-levels of inequality (Alesina and Giuliano, 2011[2]; Clark and D’Ambrosio, 2015[34]).
The importance of both channels has been confirmed experimentally, by re-creating in a laboratory setting redistributive situations where the distribution of income is manipulated by the experimenters (Durante, Putterman and van der Weele, 2014[11]). Rueda and Stegmueller (2019[4]) suggest a way to disentangle the effects of the two channels on observational data – by looking at the effects of respondents’ income and aggregate inequality and how they interact. Figure 3.16 replicates their results on ISSP microdata by looking at variations of country-wide inequality over time (by including country fixed effects). It shows that the effect of relative income is as predicted by the simplest model in which individuals care only about their personal gains or losses from redistribution: individual with higher income are less concerned over income disparities and demand less government intervention. Therefore, when inequality rises, the income of households at the bottom of the distribution falls further away from the average, and they become more favourable to redistribution. On top of the relative income effect, an increase in the macro-level of inequality shifts the entire curve of preferences for redistribution upward, confirming the relevance of the second channel, i.e. people’s preferences for equality. However, macro-level inequality has a stronger effect among the better-off. This finding confirms estimates carried out by Rueda and Stegmueller (2019[4]) on other datasets, ranging from the ESS for European countries to the General Social Survey for the United States.
Figure 3.16. The macro-level effect of inequality on people’s concern over inequality and preferences for redistribution is stronger among the better-off
Note: The lines show predicted share of respondents agreeing with the two statements by relative income of respondents and level of income inequality in the country. Low inequality refers to a value of the Gini coefficient for disposable income of 0.26 (the 25th percentile in the sample), while high inequality refers to a Gini coefficient of 0.34 (the 75th percentile). The shaded areas are 95% confidence intervals using standard errors clustered at the level of country (22 countries). Relative income refers to relative family income (equivalised using the square root of household size) compared to the country-wave average (after winsorising to avoid outliers). The estimates are carried out following Rueda and Stegmueller (2019[4]) approach (See Annex 3.A for more details).
Source: OECD calculations from ISSP 1987, 1992, 1999, 2009, 2017, and OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD).
There are multiple reasons why the preferences of the better-off should respond strongly to macroeconomic levels of inequality. Rueda and Stegmueller (2015[33]; 2020[35]) argue that the negative impact of redistribution on the consumption of the rich is relatively less relevant than its positive impact on the poor. The rich are likely to accept redistribution’s direct impact on higher taxes because they have a social preference for more equality or they care about other indirect effects of inequality on their consumption. Rueda and Stegmueller (2015[33]; 2020[35]) argue that fear of crime also explains why an increase in inequality translates into increased redistribution preferences among the rich.
Another interpretation is that, for the poor, their self-interest is aligned with the thrust of redistribution. And while it might seem puzzling that the rich react more to macro-levels of inequality, it should not be forgotten that the mechanism behind the relative-income channel is still at work – i.e. higher inequality increases the redistribution preferences of low-income individuals by making them even relatively poorer.
Most people believe they belong to the middle class
The impact of the relative-income channel inequality on preferences for redistribution is stronger if people are aware of their position in the income distribution. Often, though, they are not. Evidence from the Compare Your Income webtool shows that most people believe their income is close to the median (Figure 3.17 and Balestra and Cohen (forthcoming[36])). Two opposite forces drive this pattern, which is usually referred to as “middle-income bias” (OECD, 2019[37]; Hoy and Mager, 2021[38]; Cansunar, 2021[39]):10
1. People from the bottom of the income distribution tend to overestimate their relative position. In fact, in almost all countries more than 90% of individuals believe that they are located above the 25th percentile of the distribution.
2. High-income individuals tend to underestimate their position, although such misrepresentation varies from country to country. Greater shares of the better-off think they are lower down the distribution in Belgium, Portugal, Greece, Italy and Spain, while in Canada and Australia such underestimation is lower.
The two forces offset each other. Nevertheless, the average respondent in most countries overestimates his/her position.
Figure 3.17. Most people believe their income is close to the median
Note: Blue dots are for the 10th and 90th percentiles, the box for the 25th and 75th percentiles, while the grey dot is the average.
Source: Balestra and Cohen (forthcoming[36]) on Compare Your Income 2015-2020.
Evidence from several studies confirms that most people tend to believe their income is close to the median. Cruces et al. (2013[29]) show that in Buenos Aires more than 50% overestimate their relative income position and 30% underestimate it. Karadja et al. (2017[40]) state that most Swedes believe they are poorer than they actually are. Bublitz (2020[41]) provides evidence of differences between perceived and actual relative income position in Brazil, France, Germany, Russia, Spain and the United States. More detailed findings on Denmark from Hvidberg et al. (2020[42]) suggest that the divergence between perceived and actual relative income positions is partly explained by the fact that people are better able to assess their relative position with respect to others in specific reference groups – such as cohorts or co-workers – rather than with reference to the entire population. Nevertheless, they still find that poorer individuals largely overestimate their income position relative to the reference groups that matter the most to them, so limiting their demand for redistribution.
Informing people as to their true income position changes their preferences for redistribution – those who overestimate their income become more favourable to redistributive intervention and those who underestimate it less favourable (Cruces, Perez-Truglia and Tetaz, 2013[29]; Karadja, Mollerstrom and Seim, 2017[40]). However, evidence from available experiments suggests that being informed of the facts generally has – on average – a limited effect on their support for redistribution (Box 3.2).
Box 3.2. Informing people of their true income rank changes their attitudes towards redistribution, though only to a small extent
What would the average level of redistributive preferences be if people were to know their actual position in the income distribution? Answering that hypothesis entails associating true relative income with preferences for redistribution. But that is complicated, mainly because people with a perception of relative income are not selected at random and might differ from others in characteristics that cannot be controlled.
For these reasons, the literature uses in-survey experiments to understand what the consequences would be of informing people of their true position in the income distribution. In such experiments, a randomly selected subset of respondents is given the information before they answer standard questions about preferences for redistribution. Given the random selection, these respondents are no different from those who do not receive the information, so comparing the answers of the two groups measures the effect of the information. In most studies, however, the estimated effect is imprecise, with a large confidence interval. Ciani, Fréget and Manfredi (forthcoming[43]) offer a meta-analysis of the available experiments, which yields a more precise average answer. The average effect across studies is slight, with a narrow confidence interval of around zero.
The small overall effect might just be the result of heterogeneous responses from individuals who overestimate or underestimate their relative income position. If both groups were to revise their expectations, those who overestimate would become more favourable towards redistribution, while those who underestimate would become less favourable. Therefore, two margins would compensate each other. Some studies report heterogeneous results depending on prior perception of income position. Focusing on these studies, the meta-analysis finds results that are consistent with the hypothesis that people who overestimate their position increase their demand for redistribution once they are informed about their actual relative standing, while those who underestimate decrease their support. Cruces et al. (2013[29]), for instance, find that in Buenos Aires the impact of information is statistically different from zero only for those who initially overestimate their position, while Karadja et al. (2017[40]) find the opposite pattern in Sweden. The average effects in each group are, however, slight.
There are two possible explanations for the fact that knowing one’s income position has such little effect. The first is that on average the difference between perceived and actual relative position is limited, or alternatively, that the relative income position estimated through national surveys is not necessarily very relevant to gauging people’s opinions about the role of government in redistribution. The second explanation is that other beliefs, which may vary widely from country to country, matter more than relative income.1
Note
1. For the poor who overestimate their position, Hoy and Mager (2021[38]) suggest that the small effect on redistributive preferences might be driven by respondents who use their income as a “benchmark” to evaluate the condition of other individuals. Before receiveing information, they believe their income is a benchmark for the middle class. After discovering that they are actually poor, they also realise that there are fewer poor people than they expected. In fact, Hoy and Mager (2021[38]) find that in most of the countries of their study, poor people who overestimate their position in the income distribution reduce their concerns with inequality when they are provided with information.
3.4. What shapes the association between income inequality and preferences for redistribution?
Concern over income inequality does not fully translate into higher demand for redistribution
Although concern over income disparities influences preferences for redistribution, it does not translate in the same proportions into higher demand for government intervention. According to the latest available data from the ISSP and Eurobarometer, an average of around 80% of the population in OECD countries agrees that income disparities are too wide. However, the share that thinks the government should act to reduce inequality is less than 80% in most countries (Figure 3.18). The widest gaps are observed in English-speaking countries (apart from Ireland; see Benson et al. (2021[44]) for related evidence for the United Kingdom), Switzerland, the Czech Republic and Japan (Kambayashi and Lechevalier, 2021[1]). The gap is especially wide in the United States, which suggests that the key transatlantic divide with respect to European countries in attitudes towards redistribution stems from the different views of the role of government, rather from concern over income disparities (Osberg and Bechert, 2016[45]). The discrepancy between high levels of concern over income disparities and little call for redistributive action is also significant in some European countries (Austria, France, Germany), despite their strong welfare states, as well as in some post-transition countries (the Czech Republic, Hungary, Estonia and the Slovak Republic).
Figure 3.18. People’s demand for redistribution is lower than their concern over income disparities
Note: Respondents are asked their opinion about the statements “Differences in income in [country] are too large” and “It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low income”. In Eurobarometer the statements are slightly different: “Nowadays in [our country] differences in people's incomes are too great” and “The government in [our country] should take measures to reduce differences in income levels”, but the response scale is identical. Data from ISSP are used where available.
Source: OECD calculations from ISSP 2017; from Eurobarometer 471/2017 for Belgium, Estonia, Greece, Ireland, Italy, Luxembourg, Latvia, Netherlands, Poland, Portugal whose data are; for Slovenia, concern are from ISSP and preferences for redistribution from Eurobarometer.
The mismatch between concern about income disparities and preferences for redistribution is manifest in aggregate trends. Over the last three decades, redistribution preferences have increased, but by less than concern over income disparities (Figure 3.19). The average increase on both counts was steepest between the late-1980s/early-1990s and the onset of the Great Recession, though the increase in the demand for redistribution was only half that of concern about inequalities.11
Focusing on countries observed between (approximately) these two points in time and on the share that strongly agree with the statements, only two demonstrate both decreased concern over income disparities and reduced preference for redistribution – New Zealand and Norway (Figure 3.20, upper panel). In all other countries, concern over inequality has grown over time, while changes in preferences for redistribution have been smaller or even negative. Only in Italy and Australia have they increased at a similar pace.
Figure 3.19. Preferences for redistribution have increased by less than concern about income disparities
Note: Unweighted average across countries of the share of people who strongly agree that income disparities (in their country) are too large. Despite the availability of data, the figure does not include Germany in 1987 (only West Germany was surveyed). Nor does it include the Czech Republic and the Slovak Republic in 1992, which still made up Czechoslovakia; although separate samples are available. In 1992, the question referred to the whole of Czechoslovakia. As the aim is tracking the evolution over time, countries that have gaps (Italy and Switzerland) or do not appear in ISSP 2017 are not included. Unlike Figure 2.2, Slovenia is not included because the question on government intervention was not asked in ISSP.
OECD 5: Australia, Austria, Great Britain, Hungary, United States; OECD 8: + Germany, New Zealand, Sweden; OECD 15: + Czech Republic, Denmark, Spain, France, Israel, Japan, Slovak Republic; OECD 20: + Switzerland, Finland, Iceland, Lithuania, Turkey; OECD 13: Australia, Switzerland, Chile, Czech Republic, Germany, Denmark, Finland, United Kingdom, Italy, Japan, New Zealand, Norway, Slovenia.
Source: OECD calculations based on ISSP 1987, 1992, 1999, 2009, 2017, 2019.
In the decade from 2009 to 2019, concern fell slightly, while average demand for redistribution actually grew a little. In New Zealand and the United Kingdom, concern over income disparities grew more than demand for redistribution, while concern contracted sizeably in Italy and Slovenia, as did redistribution preferences, albeit by less. Norway, Germany, Australia and Switzerland actually showed a greater rise in demand for redistribution than in concern, so narrowing the gap between the two.
The evolution from the late 1980s to 2019 is similar if we look at the shares of people who agreed or agreed strongly (Annex Figure 3.A.1). The increase in the preference for greater redistribution during the last decade has, however, been stronger if measured with this share, and brings it more into overall line with the rise of concern, so highlighting the fact that differences lie chiefly in the strength of people’s agreement.
Figure 3.20. Changes in concern about income disparities and preferences for redistribution in selected countries
Note: The initial year depends on when countries are observed. For countries denoted by * it is 1992-93, for the others 1987-88.
Source: OECD calculations based on ISSP 1987, 1992, 2009, 2019.
The association between changes in income inequality and changes in redistribution preferences is weaker than between changes in income inequalities and people’s concern about them (Table 3.6). The difference is driven mostly by the weaker association between the shares of respondents who strongly believe that income inequality is too great and the share that strongly believes it is state’s duty to reduce it. A 1 percentage point increase in the disposable income Gini coefficient is associated with an increase of 1.7 percentage points in concern and only 0.9 in the demand for redistribution.
Table 3.6. Rising inequality increases people’s concern over inequality but has a weaker effect on their preferences for redistribution
Percentage point increase (or score increase for average answer) associated with a 1 percentage point increase in...
Note: * denotes statistically significant at the 10% level, ** at 5%, *** at 1%. All coefficients can be read as percentage point changes.
Source: OECD calculations from ISSP 1987, 1992, 1999, 2009, 2017 and Eurobarometer 2017 for concern over income disparities; OECD Income Distribution Database (https://stats.oecd.org/Index.aspx?DataSetCode=IDD) for the Gini coefficient.
The weaker effect of inequality on redistribution preferences is confirmed by the meta-analysis of in-survey experiments proposed by Ciani, Fréget and Manfredi (forthcoming[43]). Learning the true extent of inequality increases redistribution preferences, but only a little. The effects are generally positive but slight, and weaker than the impact on perceptions of and concern over inequality. In 36 experiments that examined the effect of learning the true magnitude of inequality on people’s perceptions and concern and their redistribution preferences, one standard deviation increase in perceptions/concern is associated with an increase in redistribution preferences of less than 1/5. These results confirm that the increased perceptions and concern over inequality translate only partially into higher preferences for redistribution.
One reason why demand for redistribution responds only partially to growing inequality is that people’s tolerance of it also grows. Trump (2018[46]) finds that showing people that income inequality is higher than they thought increases the level of disparities deemed fair. In a similar experiment in Mexico, Campos-Vazquez et al., (2020[47]) find, however, that informing respondents as to the true extent of income inequality, or the true level of intergenerational mobility, does not affect what they think levels of inequality and intergenerational mobility should be – i.e. their preferred levels. The effect of information about inequality on preferred level of inequality is still a largely unexplored issue as few experimental studies examine and collect preferred levels of inequality. Nevertheless, it is important to mention that rising tolerance of income disparities may explain the limited rise in demand for redistribution when actual inequality grows, but not why concern over income disparities does not.
The perceived effectiveness of redistribution policies influences public support
Different factors may weaken the impact of perceived and actual inequality on the demand for redistribution. To begin with, different individuals may hold very different views about the “feasible and legitimate role of government” (Bechert and Osberg, 2016, p. 1[48]). Some may believe that state redistribution is ineffective in addressing inequality, or that the efficiency costs of redistributive policies (such as a lower labour supply) outweigh their benefits.
Findings from Risks that Matter 2020 show that people’s views of the effectiveness of policies and the potential waste of public resources are associated with demand for redistribution. Accounting for other characteristics, those who claim that many people receive public benefits without deserving them are less likely to support redistribution or progressive taxation (Figure 3.21). This perception may offset increases in preferences for redistribution prompted by perceptions of greater inequality or intergenerational persistence. The belief that benefits are granted undeservedly may be seen both as an indictment of how government policy targets benefits and as an expression of the belief that social benefits claimants cheat. Drawing on data from the European Social Survey, Algan, Cahuc and Sangnier (2016[49]) find that people who think many welfare claimants are not entitled to the benefits which they receive are less supportive of the welfare state.
Figure 3.21. People who believe that public benefits are mistargeted demand less redistribution
Note: Demand for redistribution refers to respondents who answer “more” or “much more” to the question whether the government should do more than it currently does to reduce income differences between the poor and the rich. Support for progressive taxation refers to respondents who answer “yes” or “definitely yes” to the question whether the government should tax the rich more than they currently do to support the poor. The differences associated with different groups/characteristics/perceptions are estimated using a multivariate probit model including all the characteristics in this figure and in Figure 3.4.
Source: OECD calculations from the 2020 Risks that Matter Survey.
The importance of people’s perceptions of policy effectiveness is confirmed by observational and experimental evidence. When asked about specific policies, respondents often fail to recognize their redistributive effect. Bartels (2005[50]), studying the wide support for the tax cuts enacted in the early 2000s in the United States, argues that individuals struggle to evaluate the redistributive effects of tax reforms. Accordingly, Kuziemko et al. (2015[51]) show that informing people how poor families benefit from different government programmes boosts the effect of information about inequality on redistribution preferences. Stantcheva (2020[52]) collects people’s views of tax policies’ consequences for redistribution and finds that they shape support for redistribution. Indeed, in her experiment, raising respondents’ awareness of the implications of progressive taxation for redistribution increases support for such policy.
On the other hand, people may doubt the ability of policies to reduce inequality, even if they are aware of their implications. Lergetporer, Werner and Woessmann (2020[53]) show that providing scientifically based information about the effectiveness of equality-enhancing educational policies increases respondents’ support for them. Similarly, Pellicer, Piraino and Wegner (2019[54]) find that supplying South African respondents with evidence that income inequality is lower in similar countries challenges their belief that inequality is inevitable and increases their preferences for redistribution. Settele (2021[16]) shows that giving people the facts about the wide gender wage gap has only a limited impact on demand for policies to reduce that gap, because a sizeable share of respondents believe that such policies are ineffective (see Box 3.1).
The perceived efficiency costs of redistributive policies – that they reduce the labour supply, for example – may also shape redistribution preferences. Hayes and Guay (2020[55]) supply respondents with information about the possible efficiency costs of inequality-reducing policies. They find that doing so reduces support for them, while telling the truth about benefits has no effect. Mishagina and Montmarquette (2018[56]) also find that informing respondents about the employment and price costs of a minimum wage policy reduces support for it. However, Stantcheva (2020[52]) shows that getting respondents to consider the efficiency of progressive taxation – e.g. labour supply responses and reduced aggregate revenue – affects their redistribution preferences only slightly.
Evidence from laboratory experiments supports the conclusion that efficiency considerations do affect people’s redistributive preferences, but that the effect is slight. Durante, Putterman and van der Weele (2014[11]) – using large group laboratory experiments where researchers manipulate initial income distribution and the efficiency cost of redistribution – find that imposing a large efficiency cost of redistribution prompts participants to lower their demand for redistribution.
Moreover, efficiency costs are less important than social preferences for greater equality and the self-motivated preferences of lower-income groups for more redistribution (the two channels considered in Section 3.3). Almås et al. (2020[12]) run a redistributive experiment in a representative survey of 60 countries, in which people have to make real-world choices about reducing, or not, pay gaps between two workers in a real-life situation. Their findings show that people end up reducing them by less when the researchers enforce an aggregate ‘efficiency’ cost of redistribution, by imposing that only a fraction of the sum taken from the high-wage worker goes to the low-wage worker.12 Nevertheless, they also find that such “efficiency” considerations are less important than the reasons – i.e. performance or luck – why the wage of the two workers was different prior to any redistribution.
The drivers of trust in public institutions shape demand for redistribution
Even when people are concerned about rising inequality, they may not support redistributive policies because they have limited trust in their government. A stream of observational evidence from the United States – started by Heterington (2006[57]) and Rudolph and Evans (2005[58]) – suggests that low level of trust in government reduces support for redistributive policies. This might explain why, in some countries, demand for redistribution has grown only to a limited extent despite rising inequality (Macdonald, 2019[59]). However, low trust in government does not necessarily spell little support for redistributive policies in all countries at all times. Svallfors (1999[60]) and Edlund (2006[61]) show that Swedish and Norwegian people who are wary of the government do not demand lower redistribution – perhaps because there is strong nationwide support for the welfare state in both countries.
The experimental evidence also yields mixed findings about the association between trust in government and preferences for redistribution. Kuziemko et al. (2015[51]) show that, when individuals receive information about actual levels of income inequality in the United States, their trust in government falls, which might be attributable to the limited – albeit positive – effect of information about inequality on preferences for redistribution. Using an experiment in which they prime respondents to be less confident in the integrity of government, they also show that eroding trust has an independent negative effect on demand for redistribution. Lergetporer, Werner and Woessmann (2020[53]), by contrast, argue that less trust does not explain the limited effect of learning the facts about inequality on redistribution preferences, because the effect is not driven by groups who are usually more trustful of government. In a recent experiment carried out in the United States, Peyton (2020[62]) found that boosting respondents’ trust in government by getting them to read an op-ed praising public officials’ integrity, did not lead to any sizeable change in demand for redistribution.
Focusing on the public governance drivers of trust in public institutions – responsiveness, reliability, openness, fairness and integrity (OECD, 2017[63]; Murtin et al., 2018[64]) – helps to bring clarity and reconcile the different findings. In fact, the overall “trust in government” is an outcome, shaped by these drivers. The intensity and importance of the different drivers vary extensively across countries (OECD/Korea Development Institute, 2018[65]; OECD, 2021[66]). Furthermore, there are reasons to expect the different drivers to relate differently to demand for redistribution.
Higher levels of satisfaction with government responsiveness, openness and fairness, for example, could be associated with less demand for more redistribution, because people are already satisfied with the current tax and benefit system (see Section 3.3). Edlund (2006[61]) finds that, for a large share of Swedish respondents, “distrust in the capability of the welfare state is an issue of insufficient resources”, and therefore people who are not satisfied with the welfare state back increased social spending. Furthermore, people might demand more progressive taxation to make up for being treated unfairly by other policies (Scheve and Stasavage, 2016[67]). Conversely, they might demand less if they feel that the government is very open and fair.
People’s beliefs in the integrity of public officials with whom they deal directly also play an important role. Civic-minded citizens are more willing to support a larger welfare state and greater redistribution if they believe that petty corruption is low and are confident that benefits go to those who need them most (Algan, Cahuc and Sangnier, 2016[49]). This dimension of public integrity is therefore likely associated with stronger demand for redistribution.
Integrity at different levels of public institutions may produce different effects. Corruption in the upper echelons of the state – revolving door practices or big business bribes, for example – creates inequalities perceived as unfair and anti-meritocratic. By the same token, when people perceive less top-tier corruption, they may believe that income distribution is fair or more meritocratic to begin with. Consequently, they may demand less redistribution (Alesina and Angeletos, 2005[10]).
Findings from the Trustlab survey (Figure 3.22) confirm the role of some these different drivers of trust in shaping demand for redistribution. People who believe that the government is responsive are less likely to demand more redistribution through progressive taxation. Respondents who see low levels of top-tier government corruption are also less favourable to redistribution. On the opposite, when they perceive low levels of petty corruption they demand more progressive taxation.
Figure 3.22. People demand less redistribution if they believe that the current government is responsive, but demand more if they think petty corruption is widespread
Note: The preferred level of redistribution is calculated as the difference between the Gini index for market income of the country of the respondents minus the respondent’s preferred Gini index of post-tax income. The latter is calculated applying the tax schedule indicated by the respondent to the country’s market income Gini index. This preferred level of redistribution is higher when the respondent chooses a more progressive taxation schedule. The effect of the single drivers are estimated through a regression that controls for a wide set of respondents’ socio-economic characteristics plus country fixed effects. The countries included in the estimates are Germany, Italy, Japan, Slovenia, United Kingdom, United States.
Source: Bonnet et al. (forthcoming[68]) elaboration on the Trustlab survey.
The preferred mix of redistributive policies varies across people and countries
There is a different interpretation of the findings from informational experiment literature on how learning the facts about inequality has little effect on support for redistribution. It is that, while people may agree on the need for some policy action, they disagree on what action. In fact, several experiments into support for certain redistributive policies show that, on average, having information on rising inequality has little effect on support for any one policy. This is in contrast to the observational evidence reported above which finds a more consistent relation between growing inequality and general demand for redistribution.
Settele (2021[16]) exposes respondents to two different estimates of the gender gap in the United States. One is that the gap is wide, with women earning only 74% of what men do. The other is that the gap is narrow, at 94%. She finds that showing respondents evidence of the gender gap has a very strong effect on both perceptions of inequality and general demand for government action to reduce it. The effects on specific policies are also positive, but slight. Zilinsky (2014[69]) shows that supplying information on the extent of inequality increases demand for redistribution, but not for more progressive taxation.
Disagreement about the right policy mix may be prompted by different beliefs about different aspects of inequality. Evidence from Risks that Matter shows that beliefs about income inequality and intergenerational persistence impact policies differently (Figure 3.23). With regard to the general demand for redistribution, they exert similar impacts. However, beliefs about income disparities exert a much stronger effect on boosting demand for progressive taxation, spending on unemployment, and income support. As for the impact of beliefs about intergenerational persistence, it is strong when it comes to expenditure on education and, even more so, on health and long-term care. The explanation might be that people attribute health inequalities to factors beyond individuals’ control or inherited from earlier generations (OECD, 2018[70]). Both perceptions of higher income disparities and intergenerational persistence increase demand for state spending on pensions, which underlines the importance of public pension systems in protecting poorer retirees, particularly through the first-tier safety net (OECD, 2017[71]).
Figure 3.23. The preferred mix of redistributive policies depends on different perceptions and combinations thereof
Note: Higher perceptions refer to an increase in either perception by 40 percentage points (approximately a shift from the 25th to the 75th percentile). It is estimated as average marginal effects from a probit regression, conditional on equivalised disposable income decile, gender, education, employment status, age, marital status, size of town, housing situation, and country fixed effects. Effects are not the same as in Figure 3.4 because other perceptions are not accounted for; results accounting for them are similar, however. The general question on demand for redistribution is “In your country, do you think the government should do more or less to reduce income differences?” For progressive taxation, it is “Should the government tax the rich more than they currently do in order to support the poor?” For the other categories, the question is: “Thinking about the taxes you might have to pay and the benefits you and your family might receive, would you like to see the government spend less, spend the same, or spend more in each of the following areas?” LTC stands for long-term care.
Source: OECD calculations from the 2020 Risks that Matter Survey.
The general demand for more redistribution masks considerable cross-country differences in the mix of policies which people believe would best reduce income differences. Table 3.7 shows the association between general preferences for more redistribution and the demand for more public spending on different categories of action (relative to the demand for more spending across the board).
On average, across OECD countries, those who demand greater redistribution are most likely to demand more public spending on income support, incapacity benefits, and housing benefits. However, the trend varies from country to country. Although income support is often the category that people associate most readily with redistribution, different countries may also prioritise expenditure on education, unemployment, health, incapacity or pensions.
Table 3.7. People in different countries associate redistribution with different policies
Percentage-point differences in the probability of demanding higher social spending between those who demand more redistribution and those who do not; only top 3 associated categories
Note: The higher the value, the more the demand for that specific public spending category is associated with demand for more redistribution with respect to the average category of public expenditure (public safety excluded). For instance, on average across all countries, people who demand more redistribution – i.e. more government intervention in reducing income differences – are particularly more likely to demand more income-related support, followed by incapacity-related and housing support. In details, for each category in each country, the relative association is the difference in the fraction that demand more spending between those who demand more redistribution and those who do not; it is expressed relatively to the average difference across all categories of public spending.
Source: OECD calculations from the 2020 Risks that Matter Survey.
Kambayashi and Lechevalier (2021[1]) discuss related evidence for France, Japan and the United States. The three countries differ not only with regard to average levels of support for general redistribution and progressive taxation. They also show different shares of respondents who believe that general redistribution is desirable, but progressive taxation not. Evidence as to preferences for specific policies is crucial for understanding the demand for redistribution, but is rather scarce in cross-national surveys (with the exception of Pontusson et al. (2020[72])).
Different preferences for different policies might also depend on different beliefs in the reasons for top and bottom income inequality. Fong and Poutvaara (2019[73]) draw on data from Germany and the United States to show that the belief in poverty caused by bad luck, rather than lack of effort, generates support for transfers to the poor, while the belief that wealth is due to luck drives backing for taxing high incomes individuals. Several of the respondents who content that poverty reflects bad luck do not, conversely, believe that good luck is the only reason for being rich. Cross-country differences in beliefs as to the root causes of poverty and wealth may be wide, but hard evidence to that effect is limited.
The supply of public policies
A final consideration regards the provision of redistribution. Whether higher demand for redistribution translates into policies depends also on the interaction between citizens’ preferences and policy makers. A large body of literature finds that policy makers tend to be more responsive to the opinions of high-income voters (Gilens, 2005[74]; Giger, Rosset and Bernauer, 2012[75]; Bartels, 2017[76]), and are therefore less likely to introduce redistributive reforms. Although analysis of the supply side of redistribution is beyond the scope of this report, it is nevertheless important to highlight that political representation may play a part in shaping people’s confidence in the government and demand for redistribution.
Figure 3.24. Countries where more people believe that income disparities are too large are also those where most people perceive the government to be non-responsive
Note: The relation implies that a 1 percentage points increase in the share of respondents who agree that income disparities are too large is associated with a 0.6 percentage points increase in the share of respondents who believe that people like them do not have any say in what the government does. The relation is statistically significant at the 5% level (p value 0.021 with robust standard errors).
Source: ISSP 2017; the sample is different from Figure 2.1 because not all countries included in Figure 2.1 collected information on perceived political representation.
Rennwald and Pontusson (2021[77]) draw on ISSP data to analyse to what extent people believe that the government responds to their preferences. They show that there are stark differences between social classes in perceived political representation, with the working class perceiving the government as less responsive to their need. And gaps have widened over time. Data from the ISSP 2017 wave (Figure 3.24) show that countries where more people believe that income disparities are too wide are also those where most people see the government as non-responsive to their needs. The combination of strong concern over inequality and perceptions of little political representation fuels social resentment.
Round-up
While both perceived and actual inequality influence preferences for redistribution (Sections 3.1 and 3.3), several factors may weaken or offset their impact:
Rising inequality may increase tolerance for income disparities, therefore lessening demand for redistribution. However, greater tolerance of inequality does not explain why concern over income disparities responds more to growing inequality than redistributive preferences do.
Lack of confidence in the effectiveness of policies to reduce inequality and raise the living standards of the poor is likely to be an important factor. Informing people about the redistributive impact of policies and their effectiveness in addressing inequalities helps raise support.
The perceived efficiency costs of redistributive policies, be they behavioural or macro-economic, may lessen support for more redistribution. Nevertheless, experimental evidence suggests that they are less important than the perceived impact of policies, views of the government, and the demand for greater equality.
Even when people agree that it is the government’s responsibility to take action to reduce income differences, they may disagree on the policy mix. Evidence from questions about preferences for general redistribution – which do not collect information on specific policy options – are not enough for the purpose of investigating the policy issue. And cross-country evidence is still limited as to specific redistributive policies and how they relate to perceptions of inequality and beliefs in reasons for inequality.
References
[83] Aksoy, C. et al. (2021), “Calamities, Common Interests, Shared Identity: What Shapes Altruism and Reciprocity?”, CEPR Discussion Paper, Vol. 16186, https://cepr.org/active/publications/discussion_papers/dp.php?dpno=16186.
[10] Alesina, A. and G. Angeletos (2005), “Fairness and Redistribution”, American Economic Review, Vol. 95/4, pp. 960-980, http://dx.doi.org/10.1257/0002828054825655.
[2] Alesina, A. and P. Giuliano (2011), “Preferences for Redistribution”, in Handbook of Social Economics, Elsevier, http://dx.doi.org/10.1016/b978-0-444-53187-2.00004-8.
[31] Alesina, A., S. Stantcheva and E. Teso (2018), “Intergenerational Mobility and Preferences for Redistribution”, American Economic Review, Vol. 108/2, pp. 521-554, http://dx.doi.org/10.1257/aer.20162015.
[49] Algan, Y., P. Cahuc and M. Sangnier (2016), “Trust and the Welfare State: The Twin Peaks Curve”, The Economic Journal, Vol. 126/593, pp. 861-883, http://dx.doi.org/10.1111/ecoj.12278.
[12] Almås, I. et al. (2020), Fairness Around the World: Preferences and Beliefs, https://www.ifs.org.uk/inequality/wp-content/uploads/2020/11/Fairness-across-the-world-Preferences-and-beliefs.pdf.
[78] Almås, I., A. Cappelen and B. Tungodden (2020), “Cutthroat Capitalism versus Cuddly Socialism: Are Americans More Meritocratic and Efficiency-Seeking than Scandinavians?”, Journal of Political Economy, Vol. 128/5, pp. 1753-1788, http://dx.doi.org/10.1086/705551.
[8] Amiel, Y. et al. (2014), “Do we value mobility?”, Social Choice and Welfare, Vol. 44/2, pp. 231-255, http://dx.doi.org/10.1007/s00355-014-0839-2.
[36] Balestra, C. and G. Cohen (forthcoming), “Income Inequality through People’s Lenses: Evidence from the OECD Compare your Income Web-tool”, OECD Papers on Well-being and Inequalities, OECD publishing, Paris.
[82] Ballard, C. and S. Gupta (2018), “Perceptions and Realities of Average Tax Rates in the Federal Income Tax: Evidence from Michigan”, National Tax Journal, Vol. 71/2, pp. 263-294, http://dx.doi.org/10.17310/ntj.2018.2.03.
[76] Bartels, L. (2017), “Political Inequality in Affluent Democracies: The Social Welfare Deficit”, Centre for the Study of Democratic Institutions, No. 5-2017, Vanderbilt University, https://www.vanderbilt.edu/csdi/includes/Working_Paper_5_2017.pdf.
[50] Bartels, L. (2005), “Homer Gets a Tax Cut: Inequality and Public Policy in the American Mind”, Perspectives on Politics, Vol. 3/01, http://dx.doi.org/10.1017/s1537592705050036.
[48] Bechert, L. and L. Osberg (2016), Social values for equality and preferences for state intervention: Is the USA “Exceptional”?.
[5] Benabou, R. and E. Ok (2001), “Social Mobility and the Demand for Redistribution: The Poum Hypothesis”, The Quarterly Journal of Economics, Vol. 116/2, pp. 447-487, http://dx.doi.org/10.1162/00335530151144078.
[44] Benson, R. et al. (2021), Attituted to inequalities, Institute for Fiscal Studies.
[68] Bonnet, J. et al. (forthcoming), Understanding preferences for redistribution: Evidence from the Trustlab survey.
[41] Bublitz, E. (2020), “Misperceptions of income distributions: Cross-country evidence from a randomized survey experiment”, Socio-Economic Review, https://doi.org/10.1093/ser/mwaa025.
[15] Buser, T. et al. (2020), “Overconfidence and gender gaps in redistributive preferences: Cross-Country experimental evidence”, Journal of Economic Behavior & Organization, Vol. 178, pp. 267-286, http://dx.doi.org/10.1016/j.jebo.2020.07.005.
[21] Bussolo, M. et al. (2019), “I Perceive therefore I Demand: The Formation of Inequality Perceptions and Demand for Redistribution”, Policy Research Working Paper, No. WPS 8929, World Bank Group, Washington, D.C., http://documents.worldbank.org/curated/en/217371561686066298/I-Perceive-Therefore-I-Demand-The-Formation-of-Inequality-Perceptions-and-Demand-for-Redistribution.
[47] Campos-Vazquez, R. et al. (2020), “Perception of inequality and social mobility in Mexico”, AFD Research Papers, No. 124, https://www.afd.fr/en/ressources/perceptions-inequality-and-social-mobility-mexico.
[39] Cansunar, A. (2021), “Who Is High Income, Anyway? Social Comparison, Subjective Group Identification, and Preferences over Progressive Taxation”, The Journal of Politics, Vol. 83/4, pp. 1292-1306, http://dx.doi.org/10.1086/711627.
[84] Cappelen, A. et al. (2021), “Solidarity and fairness in times of crisis”, Journal of Economic Behavior & Organization, Vol. 186, pp. 1-11, http://dx.doi.org/10.1016/j.jebo.2021.03.017.
[43] Ciani, E.; L. Fréget and T.Manfredi (forthcoming), “Learning about inequality and preferences for redistribution: A meta-analysis of in-survey informational experiments”, OECD Papers on Well-being and Inequalities, OECD Publishing, Paris.
[30] Ciani, E. et al. (forthcoming), Perceptions of inequality across OECD and EU countries: Long term trends and recent evidence.
[32] Cingano, F. (2014), “Trends in Income Inequality and its Impact on Economic Growth”, OECD Social, Employment and Migration Working Papers, No. 163, OECD Publishing, Paris, https://dx.doi.org/10.1787/5jxrjncwxv6j-en.
[34] Clark, A. and C. D’Ambrosio (2015), “Attitudes to Income Inequality”, in Handbook of Income Distribution, Elsevier, http://dx.doi.org/10.1016/b978-0-444-59428-0.00014-x.
[25] Colagrossi, M., S. Karagiannis and R. Raab (2019), “The Median Voter Takes it All: Preferences for Redistribution and Income Inequality in the EU-28”, Publications Office of the European Union, No. 2019/6, http://dx.doi.org/doi:10.2760/797251.
[29] Cruces, G., R. Perez-Truglia and M. Tetaz (2013), “Biased perceptions of income distribution and preferences for redistribution: Evidence from a survey experiment”, Journal of Public Economics, Vol. 98, pp. 100-112, http://dx.doi.org/10.1016/j.jpubeco.2012.10.009.
[11] Durante, R., L. Putterman and J. van der Weele (2014), “Preferences for Redistribution and Perception of Fairness: An Experimental Study”, Journal of the European Economic Association, Vol. 12/4, pp. 1059-1086, http://dx.doi.org/10.1111/jeea.12082.
[61] Edlund, J. (2006), “Trust in the Capability of the Welfare State and General Welfare State Support: Sweden 1997-2002”, Acta Sociologica, Vol. 49/4, pp. 395-417, http://dx.doi.org/10.1177/0001699306071681.
[80] Eriksen, K. and L. Fallan (1996), “Tax knowledge and attitudes towards taxation; A report on a quasi-experiment”, Journal of Economic Psychology, Vol. 17/3, pp. 387-402, http://dx.doi.org/10.1016/0167-4870(96)00015-3.
[7] Fong, C. (2001), Social preferences, self-interest, and the demand for redistribution, http://www.elsevier.nl/locate/econbase.
[73] Fong, C. and P. Poutvaara (2019), “Redistributive politics with target-specific beliefs”, ifo Working Papers, No. 297, ifo Institute – Leibniz Institute for Economic Research at the University of Munich, https://www.ifo.de/DocDL/wp-2019-297-fong-poutvaara-redistributive-politcs.pdf.
[6] Gärtner, M., J. Mollerstrom and D. Seim (2017), “Individual risk preferences and the demand for redistribution”, Journal of Public Economics, Vol. 153, pp. 49-55, http://dx.doi.org/10.1016/j.jpubeco.2017.06.009.
[81] Gideon, M. (2016), “Do Individuals Perceive Income Tax Rates Correctly?”, Public Finance Review, Vol. 45/1, pp. 97-117, http://dx.doi.org/10.1177/1091142115615670.
[75] Giger, N., J. Rosset and J. Bernauer (2012), “The Poor Political Representation of the Poor in a Comparative Perspective”, Representation, Vol. 48/1, pp. 47-61, http://dx.doi.org/10.1080/00344893.2012.653238.
[74] Gilens, M. (2005), “Inequality and Democratic Responsiveness”, Public Opinion Quarterly, Vol. 69/5, pp. 778-796, http://dx.doi.org/10.1093/poq/nfi058.
[26] Gimpelson, V. and D. Treisman (2018), “Misperceiving inequality”, Economics and Politics, Vol. 30/1, pp. 27-54, http://dx.doi.org/10.1111/ecpo.12103.
[19] Giuliano, P. and A. Spilimbergo (2013), “Growing up in a Recession”, The Review of Economic Studies, Vol. 81/2, pp. 787-817, http://dx.doi.org/10.1093/restud/rdt040.
[13] Goerres, A. and M. Jæger (2015), “Why Do Women and Men Differ in Their Preferences Towards Redistribution? An Analysis of Sibling-Parent Data in a Most Gender-Egalitarian Context”, SSRN Electronic Journal, http://dx.doi.org/10.2139/ssrn.2665122.
[20] Gualtieri, G., M. Nicolini and F. Sabatini (2019), “Repeated shocks and preferences for redistribution”, Journal of Economic Behavior & Organization, Vol. 167, pp. 53-71, http://dx.doi.org/10.1016/j.jebo.2019.09.016.
[55] Hayes, T. and C. Guay (2020), “Viewing equality as a loss: How highlighting policy costs and benefits influences attitudes”, The Social Science Journal, pp. 1-15, http://dx.doi.org/10.1016/j.soscij.2019.08.006.
[57] Heterington, M. (2006), Why Trust Matters: Declining Political Trust and the Demise of American Liberalism, Princeton University Press.
[38] Hoy, C. and F. Mager (2021), “Why Are Relatively Poor People Not More Supportive of Redistribution? Evidence from a Randomized Survey Experiment across Ten Countries”, American Economic Journal: Economic Policy, Vol. 13/4, pp. 299-328, http://dx.doi.org/10.1257/pol.20190276.
[42] Hvidberg, K., C. Kreiner and S. Stantcheva (2020), Social Position and Fairness Views, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w28099.
[1] Kambayashi, R. and S. Lechevalier (2021), “Why do Redistributive Policies Differ across Countries? Analyzing the Multiple Dimensions of Preferences for Redistribution”, Review of Income and Wealth, http://dx.doi.org/10.1111/roiw.12541.
[28] Karabarbounis, L. (2011), “One Dollar, One Vote”, The Economic Journal, Vol. 121/553, pp. 621-651, http://dx.doi.org/10.1111/j.1468-0297.2010.02406.x.
[40] Karadja, M., J. Mollerstrom and D. Seim (2017), “Richer (and Holier) Than Thou? The Effect of Relative Income Improvements on Demand for Redistribution”, The Review of Economics and Statistics, Vol. 99/2, pp. 201-212, http://dx.doi.org/10.1162/rest_a_00623.
[22] Kerr, W. (2014), “Income inequality and social preferences for redistribution and compensation differentials”, Journal of Monetary Economics, Vol. 66, pp. 62-78, http://dx.doi.org/10.1016/j.jmoneco.2014.03.002.
[18] Klemm, A. and P. Mauro (2021), “Pandemic and Progressivity”, IMF Working Paper, No. 2021/024, https://www.imf.org/en/Publications/WP/Issues/2021/02/06/Pandemic-and-Progressivity-50030.
[24] Kuhn, A. (2019), “The individual (mis-)perception of wage inequality: measurement, correlates and implications”, Empirical Economics, http://dx.doi.org/10.1007/s00181-019-01722-4.
[51] Kuziemko, I. et al. (2015), “How Elastic Are Preferences for Redistribution? Evidence from Randomized Survey Experiments”, American Economic Review, Vol. 105/4, pp. 1478-1508, https://www.jstor.org/stable/43495425.
[53] Lergetporer, P., K. Werner and L. Woessmann (2020), “Educational inequality and public policy preferences: Evidence from representative survey experiments”, Journal of Public Economics, Vol. 188, p. 104226, http://dx.doi.org/10.1016/j.jpubeco.2020.104226.
[14] Luttmer, E. and M. Singhal (2011), “Culture, context, and the taste for redistribution”, American Economic Journal: Economic Policy, Vol. 3/1, pp. 157-179, http://dx.doi.org/10.1257/pol.3.1.157.
[59] Macdonald, D. (2019), “Trust in Government and the American Public’s Responsiveness to Rising Inequality”, Political Research Quarterly, Vol. 73/4, pp. 790-804, http://dx.doi.org/10.1177/1065912919856110.
[3] Meltzer, A. and S. Richard (1981), “A Rational Theory of the Size of Government”, Journal of Political Economy, Vol. 89/5, http://dx.doi.org/10.1086/261013.
[56] Mishagina, N. and C. Montmarquette (2018), “The Demand for Economic Policies, Beliefs, and Willingness-to-Pay: The Case of the Minimum Wage Policy in Quebec”, CIRANO Project Reports, No. 2018rp-14, CIRANO, https://ideas.repec.org/p/cir/cirpro/2018rp-14.html.
[86] Mollerstrom, J., B. Reme and E. Sørensen (2015), “Luck, choice and responsibility — An experimental study of fairness views”, Journal of Public Economics, Vol. 131, pp. 33-40, http://dx.doi.org/10.1016/j.jpubeco.2015.08.010.
[64] Murtin, F. et al. (2018), “Trust and its determinants: Evidence from the Trustlab experiment”, OECD Statistics Working Papers, No. 2018/2, OECD Publishing, Paris, https://dx.doi.org/10.1787/869ef2ec-en.
[66] OECD (2021), Drivers of Trust in Public Institutions in Finland, OECD Publishing, Paris, https://dx.doi.org/10.1787/52600c9e-en.
[17] OECD (2021), Main Findings from the 2020 Risks that Matter Survey, OECD Publishing, Paris, https://doi.org/10.1787/b9e85cf5-en.
[27] OECD (2019), Main Findings from the 2018 Risks that Matter Survey, OECD Publishing, Paris, https://doi.org/10.1787/9266e48a-en.
[37] OECD (2019), Under Pressure: The Squeezed Middle Class, OECD Publishing, Paris, https://dx.doi.org/10.1787/689afed1-en.
[70] OECD (2018), A Broken Social Elevator? How to Promote Social Mobility, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264301085-en.
[71] OECD (2017), Preventing Ageing Unequally, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264279087-en.
[63] OECD (2017), Trust and Public Policy: How Better Governance Can Help Rebuild Public Trust, OECD Public Governance Reviews, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264268920-en.
[65] OECD/Korea Development Institute (2018), Understanding the Drivers of Trust in Government Institutions in Korea, OECD, http://dx.doi.org/10.1787/9789264308992-en.
[23] Olivera, J. (2015), “Preferences for redistribution in Europe”, IZA Journal of European Labor Studies, Vol. 4/1, http://dx.doi.org/10.1186/s40174-015-0037-y.
[45] Osberg, L. and I. Bechert (2016), “Social values for equality and preferences for state intervention: Is the USA “Exceptional”?”, Working Paper, No. 2016-03, Dalhousie University, Department of Economics, https://econpapers.repec.org/paper/dalwpaper/daleconwp2016-03.htm.
[54] Pellicer, M., P. Piraino and E. Wegner (2019), “Perceptions of inevitability and demand for redistribution: Evidence from a survey experiment”, Journal of Economic Behavior & Organization, Vol. 159, pp. 274-288, https://mural.maynoothuniversity.ie/13719/1/MP_perceptions.pdf.
[62] Peyton, K. (2020), “Does Trust in Government Increase Support for Redistribution? Evidence from Randomized Survey Experiments”, American Political Science Review, Vol. 114/2, pp. 596-602, http://dx.doi.org/10.1017/s0003055420000076.
[9] Piketty, T. (1995), “Social Mobility and Redistributive Politics”, The Quarterly Journal of Economics, Vol. 110/3, pp. 551-584, http://dx.doi.org/10.2307/2946692.
[72] Pontusson, J. et al. (2020), Introducing the Inequality and Politics Survey: Preliminary Findings, https://archive-ouverte.unige.ch/unige:135683.
[77] Rennwald, L. and J. Pontusson (2021), A forgotten working class. Perceptions of representation and the role of labour in a comparative perspective.
[58] Rudolph, T. and J. Evans (2005), “Political Trust, Ideology, and Public Support for Government Spending”, American Journal of Political Science, Vol. 49/3, pp. 660-671, http://dx.doi.org/10.1111/j.1540-5907.2005.00148.x.
[35] Rueda, D. and D. Stegmueller (2020), “Preferences that Matter: Inequality, Redistribution and Voting”, mimeo.
[4] Rueda, D. and D. Stegmueller (2019), Who Wants What?, Cambridge University Press, http://dx.doi.org/10.1017/9781108681339.
[33] Rueda, D. and D. Stegmueller (2015), “The Externalities of Inequality: Fear of Crime and Preferences for Redistribution in Western Europe”, American Journal of Political Science, Vol. 60/2, pp. 472-489, http://dx.doi.org/10.1111/ajps.12212.
[67] Scheve, K. and D. Stasavage (2016), Taxing the rich: A history of fiscal fairness in the United States and Europe, http://dx.doi.org/10.1177/0094306117705871nn.
[16] Settele, S. (2021), “How Do Beliefs about the Gender Wage Gap Affect the Demand for Public Policy?”, American Economic Journal: Economic Policy, Vol. forthcoming.
[85] Stantcheva, S. (2020), “Understanding economic policies: What do people know and learn?”, Harvard University Working Paper, Harvard University, https://www.stefanie-stantcheva.com/research/.
[52] Stantcheva, S. (2020), Understanding Tax Policy: How Do People Reason?, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w27699.
[60] Svallfors, S. (1999), “Political trust and attitudes towards redistribution”, European Societies, Vol. 1/2, pp. 241-268, http://dx.doi.org/10.1080/14616696.1999.10749933.
[46] Trump, K. (2018), “Income Inequality Influences Perceptions of Legitimate Income Differences”, British Journal of Political Science, Vol. 48/4, pp. 929-952, http://dx.doi.org/10.1017/S0007123416000326.
[79] VanHeuvelen, T. (2017), “Unequal views of inequality: Cross-national support for redistribution 1985–2011”, Social Science Research, Vol. 64, pp. 43-66, http://dx.doi.org/10.1016/j.ssresearch.2016.10.010.
[69] Zilinsky, J. (2014), “Learning About Income Inequality: What is the Impact of Information on Perceptions of Fairness and Preferences for Redistribution?”, SSRN, http://dx.doi.org/10.2139/ssrn.2485121.
Annex 3.A. Methodological details
Estimates of the effect of inequality on preferences for redistribution by income
The results displayed in Figure 3.16 are predictions from an ordinary probit model estimated on individual microdata, controlling for age, gender, household size, employment status (employed, unemployed, or reference category “other status”), educational level (less than secondary, secondary, or reference category “tertiary”), the log of relative income (with respect to the country average in that point in time), and the Gini index for disposable income from the IDD.
Instead of dropping observations with missing values in the covariates, they are replaced with sample averages (or reference categories) but the covariates include a set of binary indicators, one for each variable, which have the value 1 if that observation was originally a missing value for that covariate. All regressions include country and period (wave) dummies and only countries observed in at least two periods are considered.
The main effects are captured by the interaction between the Gini coefficient and the log of relative income, as in Rueda and Stegmueller (2019[4]). To avoid results driven by spurious patterns over time, relative income (and the missing income indicator) is also interacted with the period dummies. Results are similar if observations with missing income are dropped, or the sample is restricted to the working- age population. Only ISSP is used for consistency in all variables.
Annex Figure 3.A.1. Changes in concern about income disparities and preferences for redistribution in selected countries
Note: The initial year depends on when countries are observed. For countries marked with * it is 1992-93, for the others 1987-88
Source: OECD calculations on ISSP 1987, 1992, 2009, 2019.
Notes
← 1. The finding is robust to excluding outliers, either identified by observations with high leverage or by means of the dfbeta statistic.
← 2. While university students are not representative of the population, they have the numeracy and logical skills needed to express preferences in abstract comparisons of this type.
← 3. Even if people are more willing to compensate inequalities arising because of factors outside the control of individuals (e.g. brute luck), there is evidence that this willingness materializes when those who were negatively affected by these factors had taken some action to prevent them, even if such action could not have changed the outcome (Mollerstrom, Reme and Sørensen, 2015[86]).
← 4. The finding is robust to excluding outliers, either identified by observations with high leverage or by means of the dfbeta statistic.
← 5. One paper that finds different evidence is VanHeuvelen (2017[79]), who, combining data from multiple ISSP and ESS waves, finds no relation with net inequality, although he finds that redistribution intensity, as captured by the proportional distance between the Gini market income and the Gini disposable income, is positively related with redistributive preferences. Another paper is Bussolo et al. (2019[21]), who, combining ISSP data and Gini estimates from the Luxembourg Income Study Database (LIS) and the “All the Ginis” dataset of Milanovic, find no direct relation between the Gini indicator and preferences for redistribution.
← 6. Of course, concern over quality is not the only driver of preferences for redistribution. Concerns and preferences for redistribution might influence each other or they might instead be both related to other unobserved elements, such as general dissatisfaction for the economic conditions. Nevertheless, the results confirm that perceptions of and preferences for inequality are relevant factors in analysing demand for redistribution.
← 7. Apart from preferences for redistribution, there is also evidence that the crisis might have increased people willingness to help others. Aksoy et al. (2021[83]) show that influencing survey respondents into thinking about the COVID-19 pandemic (by asking and telling them about its impact) increases their altruism and reciprocity towards people living in their country or abroad, though the effect is weaker towards non-EU residents. The experiments was conducted in nine European countries: France, Germany, Greece, Hungary, Italy, the Netherlands, Poland, Spain, and Sweden. Also the study from Cappelen et al. (2021[84]) on the United States shows that influencing respondent into thinking about the pandemic increases their solidarity. However, they also show that it increases their acceptance of inequality due to luck.
← 8. The finding is robust to excluding outliers, either identified by observations with high leverage or by means of the dfbeta statistic.
← 9. The evidence is similar, although with larger effects, using lagged inequality and redistribution indicators. However, because of data constraints, using lagged indicators leads to a much smaller, selected sample, and the selection in fact seems to drive the larger results: within this selected sample, even without lagging the indicators, we find similar coefficients, and the one on RS even becomes statistically significant. A fuller evaluation of the relation between actual redistribution and preferences would require disentangling the two side-effects, as well as spelling out the dynamics between policy intervention and changes in preferences. This would require longer time series, which will be available once the next ISSP wave on social inequality is released for a large set of countries. An even more challenging issue is how people gather information about redistributive policies and assess their effectiveness. The evidence in this respect is rather limited, with few exceptions (Eriksen and Fallan, 1996[80]; Gideon, 2016[81]; Ballard and Gupta, 2018[82]; Stantcheva, 2020[85]).
← 10. As discussed in Chapter 2, it is important, when assessing relative income “bias”, not to neglect that what is referred to as “actual” position is an estimate based on methodological choices. These include a specific definition of income, an adjustment based on an equivalence scale that accounts for household size (but not for other needs), and the choice of using the entire population at a specific point in time as a reference group. As discussed by Hvidberg, Kreiner and Stantcheva (2020[42]), all these choices might not reflect what matters for individuals when they formulate their fairness concerns. So providing them with the “actual” estimate might not change their concerns.
← 11. A weak change over time is confirmed by looking at the share who agree or strongly agree, or at the average answer (treating the 5-point Likert scale as cardinal).
← 12. In a similar work, but limited to Norway and the United States, Almås, Cappelen and Tungodden (2020[78]) find that efficiency considerations play a very minor role.