Moving in parallel? Economic inequality and public demand for redistribution in unequal societies

https://doi.org/10.1177/01925121241309929 International Political Science Review 1 –17 © The Author(s) 2025 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/01925121241309929 journals.sagepub.com/home/ips Moving in parallel? Economic inequality and public demand for redistribution in unequal societies Cristian Márquez Romo Institute of Sociology, Goethe University Frankfurt, Germany Abstract Although extensive research indicates that economic inequality drives public demand for redistribution, longitudinal evidence of this association in unequal contexts remains scarce. Using pooled cross-sections of surveys from over 140,000 individuals consistently observed between 2008 and 2019, this study tests the inequality-redistribution nexus in Latin America. I examine both the general association between inequality and public demand for redistribution as well as the conditional effect of individual-level income. Main results suggest that public preferences over redistribution systematically react to rising inequality. Findings further indicate that this effect is consistent across income groups. In line with a growing body of work, public demand for state-led redistribution increases as inequality grows, holding household income constant, suggesting that individuals tend to update their redistributive preferences in parallel and the gap in support for redistribution among income groups is small given the region’s sharp levels of economic inequality. Keywords Inequality, redistribution, public opinion, political economy, Latin America Introduction Extant research has emphasised the importance of examining the existence of a self-regulating mechanism that prevents income inequality from rising to too high a level in a society. From a selfinterest framework, the Romer–Meltzer–Richard (RMR) model (Meltzer and Richard, 1981; Romer, 1975) expects that citizens demand more redistribution as inequality increases. Although some studies have offered evidence consistent with this hypothesis (e.g., Finseraas, 2009; Franko, 2016; Kenworthy and Pontusson, 2005), scholarship yields conflicting results. Some scholars argue that since citizens are unlikely to know where they fall in the income distribution, people’s understanding of the level of inequality tends to be biased (e.g., Becker, 2024; Bobzien, 2020; Cruces et al., 2013; Gimpelson and Treisman, 2018). Others show, however, that exposure to unequal contexts can increase the awareness of objective levels of inequality (e.g., Minkoff and Lyons, Corresponding author: Cristian Márquez Romo, Institute of Sociology, Goethe University Frankfurt, Theodor-W.-Adorno-Platz 6, Frankfurt am Main, Hesse 60323, Germany. Email: [email protected] 1309929IPS0010.1177/01925121241309929International Political Science ReviewMárquez Romo review-article2025 Review 2 International Political Science Review 00(0) 2019; Sprong et al., 2019) and that people can react to macroeconomic changes by updating their opinion and signalling their relative preference towards a desired policy (e.g., Enns and Kellstedt, 2008; Wlezien, 1995, 2004; Wlezien and Soroka 2011, 2012). Given that recent evidence suggests that income inequality is on the rise in most countries (Piketty and Sáez, 2014; Zucman, 2019) and that regions including Latin America have comparatively high levels of income inequality (Alvaredo and Gasparini, 2015; Amarante et al., 2016; Sánchez-Ancochea, 2020), testing the proposition underlying the RMR model remains crucial. Yet, to do so, most studies employ either non-random samples or samples from countries with comparatively low levels of inequality (e.g., Breznau and Hommerich, 2019; Hillen and Steiner, 2024; Schmidt-Catran, 2016), or test their expectations by relying on cross-sectional or aggregate longitudinal data (e.g., Dallinger, 2010; Finseraas, 2009; Lübker, 2007), which makes it harder to validate whether this relationship is spurious or not (Fairbrother, 2014). To contribute to filling this gap, this article explores the inequality-redistribution nexus in Latin America, using pooled cross-sections of surveys from over 140,000 individuals consistently observed in 18 countries during a period spanning over a decade (2008–2019). In order to test the general relationship between income inequality and public demand for redistribution, I first fit fixed effects (FE) and random effects within and between (REWB) models. To further assess the extent to which this relationship is conditioned by individuals’ positions in the income distribution, I test the interaction effect between country-level inequality and individual-level income. Main results indicate a strong positive and sizeable longitudinal effect of income inequality on public demand for redistribution. Firstly, echoing Wlezien’s (1995, 2004) thermostatic model, findings suggest that public preferences regarding redistribution systematically react to rising levels of income inequality. In line with a growing scholarship (e.g., Andersen et al., 2021; Hillen and Steiner, 2024; Schmidt-Catran, 2016), this result suggests that when inequality is comparatively high in a country, public demand for redistribution tends to increase. Secondly, the conditional relationship suggests that this positive within effect is consistent across income groups. That is, people tend to demand more state-led economic redistribution as inequality increases, regardless of where they fall within the income distribution. These results provide evidence that in highly unequal contexts, the gap in support for redistribution among income groups tends to be small. In line with previous research indicating that income groups often change their redistributive preferences in parallel (e.g., Enns and Kellstedt, 2008; Gonthier, 2017; Soroka and Wlezien, 2009), findings suggest that all income groups update their preferences towards more redistribution as inequality rises, with a slightly weaker effect among the well-off. Implications and limitations of these findings are further discussed in the conclusions section. Income inequality and public demand for redistribution: The Romer–Meltzer–Richard model and beyond An abundant body of research analyses the existence of a self-regulating mechanism that prevents income inequality from rising too high in a society. The RMR model, a pioneering theory in the study of redistributive preferences (originally proposed by Romer (1975) and developed by Meltzer and Richard (1981)), expects public demand for redistribution to be stronger in countries with higher levels of economic inequality. Under a basic tax assumption, this theory expects the median voter to demand more redistribution as long as that voter’s income is smaller than the average income. From a self-interest approach, preferences over redistribution will depend on the effect of redistribution on an individual’s net income. As inequality increases, the median voter will have more incentives to benefit from, and thus support, redistribution (Franko, 2016; Kevins et al., 2018; Meltzer and Richard, 1981). Márquez Romo 3 Scholars have offered some evidence in support of the RMR model. Using cross-sectional individual-level data from the European Social Survey (ESS) and the International Social Survey Program (ISSP), Dallinger (2010) and Finseraas (2009) find a positive effect of economic inequality on public demand for redistribution. Using panel data, Jæger (2013) finds that economic growth generates a lower demand for redistribution, but the opposite is true for income inequality. Analysing both differences across countries and over time, Schmidt-Catran (2016) finds a strong positive longitudinal effect of inequality on public demand for redistribution in 27 European countries. Despite its importance as a pioneering theory in the study of redistributive preferences, an important body of research has questioned the RMR model, yielding conflicting results. Analysing the USA, Bénabou (2000) initially found that income inequality could actually depress support for redistribution across all income groups, showing that the relationship can be negative depending largely on welfare-enhancing benefits. This finding has been supported by most American scholarship (e.g., Erikson et al., 2002; Kelly, 2009; see Romero Vidal, 2021, for a detailed explanation), suggesting that increasing inequality can trigger conservative preferences among both the rich and the poor (Luttig, 2013). Several cross-national studies also claim that public demand for redistribution can largely depend on persistent differences in values and cultural understandings or beliefs about inequality (e.g., Breznau and Hommerich, 2019; Gimpelson and Treisman, 2018). For example, support for redistribution can depend on which is social group is considered the main beneficiary of these redistributive policies (i.e., redistribution is conceived of as either ‘taking’ from the rich or ‘giving’ to the poor; see Cavaillé and Trump, 2015), or simply whether people are more or less averse to inequality (Fehr and Schmidt, 1999). Furthermore, against the expectations of the RMR model, some studies find a positive interaction effect between individualand country-level income inequality (e.g., Dion and Birchfield, 2010; Finseraas, 2009), indicating that the negative effect of individual-level income can become weaker in contexts of sharp inequality. Thus, although demand for redistribution is expected to decline for individuals located at the highest income deciles, this effect can be highly dependent on the macrolevel of inequality. Under certain circumstances (e.g., when they acknowledge the negative externalities of inequality, such as violent crime; see Rueda and Stegmueller, 2016), even affluent individuals can be prone to supporting redistribution. Hence, the structure of inequality (i.e., the relative distance between the rich and the poor and thus the level of stratification of a society) can be a more important predictor of demand for redistribution than the absolute level of inequality (Lupu and Pontusson, 2011). One of the main shortcomings of the RMR model is that it assumes individuals are aware of their exact positions relative to the median income. Yet, knowing where one falls within the income distribution requires both access to information and abilities to process it (i.e., to compare their current situation with that of someone earning the median income). Since this information can be costly to acquire—and the advantages of doing so not always evident—people tend to develop biased perceptions of the overall income distribution (Cruces et al., 2013). Recent research shows that, since people are not aware of their exact position in the income distribution, this makes them more prone to underestimating the objective levels of inequality (e.g., Becker, 2024; Bobzien, 2020; Cruces et al., 2013; Gimpelson and Treisman, 2018). Going beyond the expectations of the RMR model, an important body of work suggests that demand for redistribution can depend not only on the net tax benefits or disadvantages of redistribution, but also on the extent to which people are uncertain about their future income (Drazen, 2000; Rehm, 2009). This approach echoes Rawls’ (1971) ‘veil of ignorance’, suggesting that so long as people are uncertain about their societal position in the future, they will have incentives to support policies in favour of the most disadvantaged. Given that people can be risk averse, demand 4 International Political Science Review 00(0) for redistribution can largely depend on peoples’ risk exposure, that is, the extent to which they think that they will need redistributive support given the possibility of being poor in the future (Moene and Wallerstein, 2001). The expectation here is straightforward: The higher the risk exposure, the more individuals will be in favour of redistribution (Rehm, 2009). The underlying expectation of the risk aversion framework is that both rich and poor citizens can support redistribution given that it ‘smooths the income stream of individuals and shares the risk of income shocks across society’ (Rehm, 2009: 858). In this vein, an alternative salient explanation of why macroeconomic changes can be a driving force of changes in people’s support of redistribution builds upon Wlezien and Soroka’s (2011, 2012) ‘thermostatic feedback’. The thermostatic feedback model illustrates the evolution of public preferences by measuring how policy and public preferences adjust to each other. For instance, people can support government intervention when unemployment is rising, update their preferences towards a desired level of taxation or react to incumbents by shifting ideologically (Bartle et al., 2011, 2020; Weiss, 2012). In a nutshell, the ‘thermostat’ effect measures how members of the public signal their position towards a desired policy direction in respect to current policy (Romero-Vidal, 2020). Thus, while people might not be aware of their exact position in the income distribution, nor of the exact amount of governmental spending required, they can react and signal their relative preferences about the extent to which they believe government should implement policies to reduce the gap between the rich and the poor. Echoing the long-standing argument that claims macroeconomic fundamentals affect individual preferences and behaviour, I expect economic inequality to be a driving force of changes in mass-policy attitudes and, more specifically, that public preferences react to rising levels of income inequality. To examine the association between income inequality and support for redistribution, scholarship relies either on cross-sectional or aggregate-level longitudinal data (e.g., Dallinger, 2010; Dion and Birchfield, 2010; Finseraas, 2009; Jæger, 2013; Kenworthy and Pontusson, 2005; Lübker, 2007) or tests expectations using non-random samples or samples from economically developed countries (e.g., Breznau and Hommerich, 2019; Hillen and Steiner, 2024; Schmidt-Catran, 2016) with comparatively low levels of income inequality (Theyson and Heller, 2015). Yet, static survey data makes it more difficult to validate whether a relationship is spurious or note and survey data considered in aggregated form are exposed to the risk of committing an ecological fallacy (Fairbrother, 2014). Furthermore, considering that demand for redistribution can depend largely on the macrolevel of inequality and thus the degree of stratification in a society, income differences can be less relevant in explaining support for redistribution in high-inequality contexts (Dion and Birchfield, 2010; Rueda and Stegmueller, 2016). An exception to this research agenda is Franetovic and Castillo (2022). The authors assess the longitudinal effect of inequality on support for redistribution in 17 Latin American countries.1 However, the authors do not find statistically significant associations between income inequality and economic redistribution, concluding that ‘in contrast to the evidence from studies conducted in other regions, the results reveal that in Latin America it is not possible to detect a clear association between income and redistributive preferences at specific times’ (Franetovic and Castillo, 2022: 1). As I will show below, the results presented here differ substantively from those presented in their study. Building on this literature, I establish three hypotheses. Firstly, I expect a positive cross-sectional relationship between income inequality and public demand for economic redistribution (H1). Secondly, since not only persistent levels but also changes in macrolevels of inequality affect redistributive preferences, I also expect a positive longitudinal relationship between income inequality and public demand for economic redistribution (H2). Acknowledging that, particularly in highly unequal contexts, exposure to inequality can affect individuals from different income groups (Dimick et al., 2018; Minkoff and Lyons, 2019; Rueda and Stegmueller, 2016), I expect the Márquez Romo 5 positive association between income inequality and public demand for economic redistribution to be consistent across income levels. In other words, the redistributive preferences gap among income groups should be smaller when and where inequality is comparatively higher (H3). Data This study uses data from six survey waves of the Americas Barometer from the Latin American Public Opinion Project (LAPOP Lab, 2021),2 which has gathered data on the policy preferences of citizens within the region every two years since 2004. Each survey wave typically includes between 25,000 and 30,000 respondents from all Latin American countries. These data are gathered in faceto-face interviews and the final sample is representative at the national level.3 LAPOP data allows us to study the evolution of citizens’ public demand for economic redistribution, using pooled cross-sections of surveys (non-repeated observations on a large random sample of micro-level units, nested in a repeated set of observations from a non-random sample of macro-level units; see Fairbrother, 2014), consistently collected every two years for over a decade (2008–2019). Combining data from six survey waves strongly increases the number of cases, making results less dependent on specific survey-wave peculiarities (Duijndam and van Beukering, 2021). After listwise deletion of missing values, pooling six survey waves in which citizens were asked about their preferences towards economic redistribution results in a sample of 140,001 respondents, 101 country-years and 18 countries.4 The dependent variable in my analysis is measured with a question asking citizens the extent to which they believe the government should implement policies to reduce income inequality on a 7-point Likert scale (1—strongly disagree, 7—strongly agree).5 This survey item has been validated and used in previous studies that measure public demand for economic redistribution (e.g., Finseraas, 2009; Luttmer and Singhal, 2011; Schmidt-Catran, 2016). For descriptive statistics of the dependent variable, see the online appendix (Table A2, supplementary materials). The independent variables of my analysis are country-level economic inequality and individuallevel income. To measure country-level economic inequality I rely on the Gini index, based on disposable (post-tax, post-transfer) household income distributions, from the Standardized World Income Inequality Database (SWIID) (Solt, 2020). To account for potential confounding, considering that disposable income partially measures how much the government is currently redistributing via policies and outlays, I also present additional models, including the pre-tax Gini measure (see Table A5, supplementary materials). Individual-level income is measured using the item from the Americas Barometer: ‘And into which of the following ranges does the total monthly income of this household fit, including remittances from abroad and the income of all the working adults and children?’. Given that the LAPOP’s income measure was introduced with a scale that ranges between 0 and 10 in waves 2008 and 2010, yet between 0 and 16 afterwards, I recoded the 17-point scale into an 11-point scale, following Franetovic and Castillo (2022: 5). At the country level, previous research suggests that it is necessary to control for a country’s level of economic prosperity to ensure that the effect of the Gini index is not spurious (e.g., Finseraas, 2009; Heston et al., 2002; Schmidt-Catran, 2016). Although the relationship might not always be linear and it may be more pervasive in some countries than others, economic prosperity can be a potential confounder when growth produces more inequality. Although some studies have found this association to be positive (e.g., Forbes, 2000; Li and Fu Zou, 1998) and others negative (e.g., Alesina and Rodrik, 1994; Persson and Tabellini, 1994), an important body of work has offered evidence of this relationship (see Van der Weide and Milanovic, 2018, for a discussion). To account for this, I include the logged annual national real gross domestic product (GDP) per capita, 6 International Political Science Review 00(0) drawn from the Penn World Table (PWT) (Feenstra et al., 2015). Finally, at the individual level, I control for a set of standard socio-demographic variables: gender (1 = female), age (a continuous variable with a mean of 39), years of schooling (19-point scale, from 0—none to 18—university or more), location (1 = urban) and employment (1 = employed). For descriptive statistics of all variables, see the online appendix (Table A3, supplementary materials). Method To examine whether public demand for redistribution is affected by a country’s level of economic inequality, I fit FE and REWB models. Firstly, since economic inequality is a property of the context in which individuals are socially embedded, not accounting statistically for this dependency between observations would violate the independent errors assumption (Bell and Jones, 2015; Moulton, 1986). Secondly, given that including the Gini index in a random effects framework without decomposing it into within and between components would provide an uninterpreted weighted average of both (see Schmidt-Catran and Fairbrother, 2016), I calculate and separately introduce the group-mean of economic inequality for each country, pooling across all available years and then subtracting each overall average from each country-year. The latter procedure, also known as ‘demeaning’, has the important advantage of allowing researchers to relax the assumption of omitted variable bias caused by any time-invariant, unit-specific differences (Jordan and Philips, 2023). This is the standard procedure used in FE or within-group models. Thus, the socalled ‘within transformation’ should make the resulting coefficient and standard error similar in the within portion of both the REWB and the FE models (Bell et al., 2019). The model can be specified as follows: yxtrendZZZ ue itk itk tk WE tk kBEk ktkitk        0112 The model treats respondents (i) as nested in country-years ( t ), which are in turn nested in countries ( k). Predictors can thus be included at any of these levels. While the coefficient vector γ WE provides the within effects and the coefficient vector γ BE the between effects, the ZZ tk k   term is equivalent to the FE transformation. This estimation will not be identical if the transformation is not applied to all time-varying predictors and the panel is unbalanced, but the corresponding coefficients should be similar (Andreß et al., 2013). Finally, to ensure that the relationship is not spurious given the fact that both the outcome and predictor variables are trending due to unrelated reasons, I control for time by including a time trend. I believe that the most appropriate functional form for time given the data generating process is linear, considering that the evolution of the outcome variable appears to be trending linearly (see Figure A1, supplementary materials). The next section presents the results. Results: Descriptive At the individual level, the grand-mean of citizens’ support for redistribution is 5.57, indicating that Latin Americans tend to endorse the implementation of policies to reduce income inequality between the rich and the poor.6 However, demand for redistribution in the region has declined during the last decade (see Figure A1, supplementary materials). As Figure 1 shows, this pattern is consistent overall within countries. Márquez Romo 7 At the contextual level, preliminary results indicate a positive association between citizens’ demand for redistribution and the level of income inequality in their countries. Yet, the correlation between the cross-sectional portion of the Gini index and the average levels of public demand for redistribution across all years is only 0.09 (p < 0.001) (Figure 2(a)). Instead, the longitudinal component suggests a stronger association between income inequality and public demand for redistribution within countries, with a correlation of 0.54 (p < 0.001) (Figure 2(b)). To assess whether this this association holds when both individualand country-level controls are included, the next section presents the multivariate results. Results: Multivariate Table 1 presents the results. Models 1–3 introduce the REWB specification, and Models 4 and 5 the FE specification. Model 1 includes the Gini index, without decomposing into its cross-sectional and longitudinal components. Model 2 introduces separate cross-sectional and longitudinal effects of economic inequality on public demand for redistribution. Models 3 and 5 include interaction terms between inequality and individual-level income. Model 1 suggests that the Gini index is positively and significantly associated with public demand for redistribution, indicating that citizens’ support for redistribution is stronger in countries with higher inequality. Nevertheless, Model 2 shows that when decomposing into the within and between portions of inequality, the association only reaches statistical significance within countries. This clear positive association is statistically significant both in the main ( β = 0.099, p < 0.001) (Model 2) and the interactive model ( β = 0.140, p < 0.001) (Model 3). As Models 2 and 4 show, the coefficient is similar in the REWB and the FE specifications. Figure 1. Public demand for redistribution in Latin America. Source: Author’s own elaboration based on Latin American Public Opinion Project data (2008–2019). Note: The figure shows individuals’ mean levels of support for redistribution (black line) and level of income inequality (red line) for each country-year. Both variables have been rescaled for substantive interpretation. Colour online only. 8 International Political Science Review 00(0) These results suggest that, opposite to what was expected in H1, there is no evidence of a crosssectional association between economic inequality and public demand for redistribution. Instead, as hypothesised in H2, these results indicate a clear positive longitudinal relationship between income inequality and public demand for redistribution. Thus, while we cannot reject the null hypothesis that countries with greater inequality tend to have higher levels of support for redistribution, results are consistent with the hypothesis that, as inequality grows within a country, public preferences regarding redistribution tend to increase. The coefficient indicates than a 1-unit increase on the Gini index produces an average increase of 0.10 points on the 7-point public support for redistribution scale. In other words, support for redistribution should increase (or decrease) by 1 percent for each 1-unit change in the Gini index. For example, in Argentina, where inequality declined from 41.6 to 37.9 between 2008 and 2019, the model predicts a decrease in support for redistribution of about 4 percentage points. In Bolivia, where inequality declined from 48.8 to 40.5 during the same period, the model predicts a decrease in demand for redistribution of about 8 percentage points. In order to test H3, Figure 3 presents the association between economic inequality and demand for redistribution over time, conditional on individual-level income differences. The figure shows an overall positive relationship between income inequality and public demand for redistribution, which holds across income groups. As hypothesised in H3, individuals tend to support income redistribution as inequality increases, regardless of where they are located within the income distribution. The slope, however, is less steep for individuals located at the highest income levels, indicating that although there are no significant differences between income groups (i.e., the association remains positive for both more and less advantaged individuals), the relationship between inequality and demand for redistribution becomes slightly weaker as household income increases. In other words, these results suggest that all income groups tend to update their preferences towards more redistribution with rising inequality, with a marginally weaker effect among the well-off. The positive association, however, is consistent across income levels overall, within yet not between countries (see also Figure A2, supplementary materials). This result suggests that the redistributive preferences gap between income groups is consistently low when but not where income inequality is comparatively higher. Figure 2. (a) Income inequality and public demand for redistribution in Latin America. (b) Income inequality and public demand for redistribution in Latin America. Source: Author’s own elaboration based on Latin American Public Opinion Project data (2008–2019). Note: The figures show the mean level of demand for redistribution versus income inequality (a) or time (b), with a linear regression line. BE: between; WE: within. Márquez Romo 9 Table 1. The effect of income inequality on Latin Americans’ redistributive preferences. Predictors REWB specification FE specification (0) (1) (2) (3) (4) (5) Income 0.001 (0.002) 0.001 (0.002) –0.180* (0.072) –0.002 (0.002) –0.065** (0.021) Economic inequality 0.064*** (0.016) 0.097*** (0.003) 0.091*** (0.004) GDP/capita (logged) 0.057 (0.109) 0.004 (0.026) –0.004 (0.027) Economic inequality (BE) 0.008 (0.018) –0.008 (0.018) Economic inequality (WE) 0.099*** (0.020) 0.140*** (0.021) GDP/capita (logged) (BE) 0.259* (0.124) 0.285* (0.123) GDP/capita (logged) (WE) 0.000 (0.141) –0.022 (0.141) Cross-level interactions Economic inequality (BE)*Income 0.004* (0.002) Economic inequality (WE)*Income –0.010*** (0.001) Economic inequality*Income 0.001** (0.000) Constant 5.580*** (0.073) 2.792** (0.869) 4.781*** (0.918) 5.467*** (0.913) 0.868*** (0.173) 1.187*** (0.206) Var (countries) 0.073 0.105 0.049 0.047 Var (country-years) 0.124 0.091 0.087 0.088 Var (individuals) 2.619 2.570 2.570 2.564 Countries Yes (18) Yes (18) Source: Author’s calculations, Latin American Public Opinion Project 2008–2019. REWB: random effects within and between; FE: fixed effects; GDP: gross domestic product; BE: between; WE: within. Notes: ***p < 0.001, **p < 0.01, *p < 0.05. Standard errors in parentheses. All models are based on 18 countries, 101 country-years and 140,001 individual observations, and control for gender, age, level of education, location and labour market status. Model 3 includes both random intercept and random slope for the income variable. FE models include robust standard errors, and Bolivia, the country with the widest sample of interviewees (12,789), is the reference category. Full results in Table A4. 16 International Political Science Review 00(0) Rawls, John (1971) A Theory of Justice. Cambridge: Cambridge University Press. Rehm, Philipp (2009) Risks and Redistribution: An individual-level analysis. Comparative Political Studies 42(7): 855–881. DOI: 10.1177/0010414008330595. Romer, Thomas (1975) Individual Welfare, Majority Voting, and the Properties of a Linear Income Tax. Journal of Public Economics 14: 163–185. DOI: 10.1016/0047-2727(75)90016-X. Romero-Vidal, Xavier (2020) Two Temperatures for One Thermostat: The evolution of policy attitudes and support for independence in Catalonia (1991–2018). Nations and Nationalism 26: 960–978. DOI: 10.1111/nana.12559. Romero-Vidal, Xavier (2021) The Macro Polity Revisited: The Interplay Between Public Opinion, Policy and Economics in European Democracies. PhD Thesis, Leuphana University of Lüneburg, Germany. Rueda, David and Daniel Stegmueller (2016) The Externalities of Inequality: Fear of crime and preferences for redistribution in Western Europe. American Journal of Political Science 60: 472–489. DOI: 10.1111/ ajps.12212. Sánchez-Ancochea, Diego (2020) The Costs of Inequality in Latin America: Lessons and Warnings for the Rest of the World. London: I. B. Tauris. Schmidt-Catran, Alexander W (2016) Economic Inequality and Public Demand for Redistribution: Combining cross-sectional and longitudinal evidence. Socio-Economic Review 14(1): 119–140. DOI: 10.1093/ser/ mwu030. Schmidt-Catran, Alexander W and Malcolm Fairbrother (2016) The Random Effects in Multilevel Models: Getting Them Wrong and Getting Them Right. European Sociological Review 32(1): 23–38. DOI: 10.1093/esr/jcv090. Schmidt-Catran, Alexander W, Malcolm Fairbrother and Hans-Jürgen Andreß (2019) Multilevel Models for the Analysis of Comparative Survey Data: Common problems and some solutions. KZfSS Kölner Zeitschrift für Soziologie und Sozialpsychologie 71: 99–128. DOI: 10.1007/s11577-019-00607-9. Solt, Frederick (2020) Measuring Income Inequality Across Countries and Over Time: The standardized world income inequality database. Social Science Quarterly 101(3): 1183–1199. Soroka, Stuart N and Christopher Wlezien (2009) Degrees of Democracy: Politics, Public Opinion and Policy. New York, NY: Cambridge University Press. Sprong, Stefanie, Jolanda Jetten, Zhechen Wang, Kim Peters, Frank Mols, Maykel Verkuyten, Brock Bastian, Amarina Ariyanto, Frédérique Autin, Nadia Ayub, Constantina Badea, Tomasz Besta, Fabrizio Butera, Rui Costa-Lopes, Lijuan Cui, Carole Fantini, Gillian Finchilescu, Lowell Gaertner, Mario Gollwitzer, Ángel Gómez, Roberto González, Ying-Yi Hong, Dorthe Høj Jensen, Inga Jasinskaja-Lahti, Minoru Karasawa, Thomas Kessler, Olivier Klein, Marcus Lima, Laura Mégevand, Thomas Morton, Paola Paladino, Tibor Polya, Tuuli Anna Renvik, Aleksejs Ruza, Wan Shahrazad, Sushama Shama, Heather J Smith, Ana Raquel Torres, Anne Marthe van der Bles, and Michael JA Wohl (2019) “Our Country Needs a Strong Leader Right Now”: Economic inequality enhances the wish for a strong leader. Psychological Science 30(11): 1625–1637. DOI: 10.1177/0956797619875472. Theyson, Katherine C and Heller, Lauren R (2015) Development and income inequality: A new specification of the Kuznets hypothesis. Journal of Developing Areas 49(3): 103–118. https://www.jstor.org/ stable/24737310 Van der Weide, Roy and Branko Milanovic (2018) Inequality is Bad for Growth of the Poor (but Not for That of the Rich). The World Bank Economic Review 32(3): 507–530. DOI: 10.1093/wber/lhy023. Weihua, An and Maoliang Ye (2017) Mind the Gap: Disparity in redistributive preference between political elites and the public in China. European Journal of Political Economy 50: 75–91. Weiss, Steffen (2012) The German Macro Polity. Doctoral Thesis, University of Essex, Essex. Wlezien, Christopher (1995) The Public as Thermostat: Dynamics of preferences for spending. American Journal of Political Science 39(4): 981–1000. DOI: 10.2307/2111666. Wlezien, Christopher (2004) Patterns of Representation: Dynamics of public preferences and policy. The Journal of Politics 66(1): 1–24. DOI: 10.1046/j.1468-2508.2004.00139.x. Wlezien, Christopher and Stuart N Soroka (2011) Federalism and Public Responsiveness to Policy. Publius 41(1): 31–52. DOI: 10.1093/publius/pjq025. Márquez Romo 17 Wlezien, Christopher and Stuart N Soroka (2012) Political Institutions and the Opinion—Policy Link. West European Politics 35(6): 1407–1432. DOI: 10.1080/01402382.2012.713752. Zucman, Gabriel (2019) Global Wealth Inequality. Annual Review of Economics 11(1): 109–138. DOI: 10.1146/annurev-economics-080218-025852. Author biography Cristian Márquez Romo is a postdoctoral fellow at the Institute of Sociology at Goethe University Frankfurt. He is currently a researcher in the European Research Council project POLAR (‘Polarization and its discontents: does rising economic inequality undermine the foundations of liberal societies?’) and the chair for Social Stratification and Social Policy. His research, comparative in nature, lies in the intersection between social stratification and political behaviour. ([email protected])