Does Information about Income Inequality Shape Demand for Redistribution? Evidence from an On-line Experiment

Transcription

1 Does Information about Income Inequality Shape Demand for Redistribution? Evidence from an On-line Experiment Jan Zilinsky University of Chicago Booth School of Business February 28, 2013 Preliminary and incomplete Abstract In the face of rising income inequality, why has demand for redistribution not increased? One possibility is that voters simply underestimate the growth in income disparity. By experimentally varying information exposure of more than 1,300 subjects in an on-line experiment, this paper shows that lack of information does not explain preferences for redistribution, although reading how much the top 0.1% earns makes subjects more likely to say that the government should do something to address income inequality. Provision of information about current income distribution decreases beliefs in a just world (hard work leads to success) and increases the perceived importance of personal connections for success, but it does not change the perceived importance of education and it does not engender support for the Occupy Wall Street movement. Overall, beliefs about mobility, gender, and perceived sources of inequality all predict demand for government action and people who report feeling uncertain about the future are more likely to report that the government should do more to equalize incomes. I am grateful to Steven Davis, John List, Loukas Karabarbounis and Richard Van Weelden for essential feedback. I have also benefited from discussions with Marianne Bertrand, Erik Hurst, David Laibson, Richard Thaler, and Brian Weller. I also thank seminar participants at the National Bank of Slovakia and at the University of Chicago for helpful comments. Please address your correspondence to zilinsky@chicagobooth.edu. 1

2 1 Introduction In the aftermath of the Great Recession, the proportion of Americans who believe that today s youth will have a better life than their parents decreased by 22 percentage points (Table 1). Theories of voting for redistribution (e.g. Benabou and Ok, 2001) would predict that this trend (falling perceived prospects of income mobility) should be followed by an increase in demand for some form of insurance, but the demand for welfare spending or insurance programs more broadly has not increased. In fact, between 2008 and 2010, the percentage of Americans who believe that the government should reduce income differences fell by almost 8 percentage points (Table 2). In recent years, several economists and political scientists have suggested that taxes remain low in the United States because voters fail to connect inequality and public policy (Bartels, 2005), because the tax policy is deliberately crafted to distort public perceptions (Hacker and Pierson, 2005) or because Americans drastically underestimate the current level of wealth inequality (Norton and Ariely, 2011). While there is no agreement about the reasons for limited information political scientists on average emphasize the supply channel, and economists are more likely to underscore the demand side 1 both groups agree that limited information is one reason for low redistribution in the United States compared to Europe. This paper builds on previous work in political economy and asks whether the lack of demand for redistribution can be explained by limited information of the electorate about the extent of income inequality. The answer is no. In a sample of 1,303 individuals who live in the United States, exogenous exposure to information about the income distribution does not cause subjects to report that they would be open to higher taxes, if the revenue was used to improve social insurance (or for entitlements or for redistribution ). Although the subjects in the information treatment are more likely to agree that the government should do something about income inequality, they are not inclined to support additional taxation to achieve this goal. Moreover, those subjects who are exposed to a brief summary of how much different income brackets earn in the United States today are less likely to say that hard work is sufficient to get ahead in life and they are more likely to say that one needs personal connections to succeed in life. On the other hand views about the importance of education do not differ between the subjects in the information exposure treatments and the control group. 2 These results are consistent with the just world of theory formalized by Benabou and Tirole (2006) and with the model of Alesina and Angeletos (2005) which explains why citizens are less likely to demand redistribution if effort (instead of luck) is believed to be the key source of income, and why beliefs in a fair society can become self-fulfilling. By combining a large on-line experiment with a detailed survey, this paper attempts to disentangle competing channels that drive demand for redistribution. One plausible factor information (as suggested by Norton and Ariely, 2011, and others) is manipulated experimentally, and follow- 1 This includes the view that low demand for policy-relevant information falls under rational ignorance. Other explanations include deliberate information avoidance by voters (Benabou and Tirole, 2006) and motivated and self-serving beliefs about the income process Di Tella et al. (2004). 2 The information treatment has no effect on support for the Occupy Wall Street movement. 2

3 up questions are designed to test previous theories (such the centrality of the prospects of upward income mobility), controlling for the previously documented drivers of demand for social insurance (e.g. convictions that the United States has superior income mobility Alesina and Glaeser, 2004; Alesina and Angeletos, 2005)) and for personal characteristics: age, gender, race and socioeconomic status. 3 The survey portion also asks subjects about personal values (freedom, family, and professional success), about uncertainty about the future, and about the probable reasons for income inequality. Most subjects believe that the weak economy is the main culprit, although many people also believe that people s lack of necessary skills is also a key factor. When attributing blame to big banks, the Bush tax cuts, Obama bailouts and the claim that the rich pay too little in taxes there is a large ideological split between subjects along the expected lines. Beyond ideology, other observables also predict how people think about the phenomena that may shape the income distribution. For example, the more educated respondents are less likely to blame the economy, big banks and the rich. Women are significantly more likely to blame the weak economy, and regular viewers of Fox News are less likely to blame big banks whereas those who often watch CNN are more likely to blame financial institutions. The paper also briefly explores attitudes to inequalities that reach beyond income. Subjects are asked not only whether they believe that making incomes more equal is important, but also about making access to health care equal and about equalizing exposure to environmental risks and toxins. The median respondent is more likely to in favor equalizing access to health care and environmental risks, as opposed to flattening the income distribution. In addition to the information treatments, an additional experimental element is the variation in salience of indirect costs of taxation. Since a lack of appreciation for the incentive effects of taxation could be responsible for excessive demand for taxation (and insurance provision) one half of the subjects is randomly assigned to a salience treatment where subjects are asked to answer whether they believe that taxes discourage effort, lower job creation, or lead to government waste. Exposure to questions that (implicitly) question the wisdom of raising taxes does not have a significant impact reported preferences for higher taxes. Do reported preferences depend on how the purpose for taxation is framed? No. Changing the wording of the taxation question by randomly switching the goal for taxation ( entitlements vs. redistribution vs. social insurance ) does not have an impact on stated preferences for taxation. The remained of the paper offers a brief window to prior research, describes the experiment, and outlines the main results. 2 Background The evolution of beliefs about expected prosperity for American children by representative population sample from the General Social Survey are outlined in Table 1. While we do not know whether the anticipation of lower prosperity and social mobility will persist, the post-crisis jump in beliefs is striking. Reported preferences for reduction of income inequality (carried out by the 3 Described in detail in Alesina and Giuliano (2011); Fong (2001); Alesina et al. (2004); Aristei and Perugini (2010); Chow and Galak (2012); Cruces et al. (2013) and others. 3

4 Today s youth will have a better life than their parents Very/somewhat likely Very/somewhat unlikely Table 1: Source: CBS News/New York Times Poll (1998, 2003), USA Today/Gallup poll (2008, 2011). The full question answered by the respondents is: In America, each generation has tried to have a better life than their parents, with a better living standard, better homes, a better education, and so on. How likely do you think it is that today s youth will have a better life than their parents very likely, somewhat likely, somewhat unlikely, or very unlikely? The 1998, 2003 and 2008 surveys were carried out in January. The 2011 survey was carried out in April. The government should reduce income differences Strongly agree / Agree Neither Strongly disagree / Disagree Table 2: Respondents were asked to express their (dis)agreement with the statement It is the responsibility of the government to reduce the differences in income between people with high. Source: General Social Surveys, N= 1041 (2008), 1430 (2010). government) are shown in Table 2: while the changes in views about the role of the government are smaller, they go in the opposite direction compared to a standard insurance model. 2.1 Beliefs Douglass North (the recipient of the 1993 Nobel Memorial Prize in Economic Sciences) has argued that stability comes from the moral and ethical codes of a society and that knowledge or ideologies can be seens as intellectual efforts to rationalize the behavior (North, 1982). From this perspective, too much ideological disparity is costly as it increases the risk of instability. 4 North identified geographic location and occupational specialization as sources of ideological differences, and additional factors have been shown to shape ideology since North s seminal work. His key observation is that [t]he costs of maintenance of an existing order are inversely related to the perceived legitimacy of the existing system. The questions in the present survey that ask for opinions and policy preferences are partly an attempt to measure the perceived legitimacy of economic outcomes today. 4 North also explains why having a world-view simplifies decision-making in general. Income share of the top 0.5% Pre-tax income share Pre-tax income share Change in 1970 in 2000s (in perc. points) France United Kingdom United States Table 3: Change in pre-tax income shares (in percent) of the P99.5-P100 income group in France, United Kingdom and the United States. In the second column, the data for France is from 2005, the data for the U.K is from 2000, and the data for the U.S. is from Source: Calculations based on Piketty and Saez (2007). 4

5 Income share of the top 0.5% Post-tax income share Post-tax income share Change in 1970 in 2000s (in perc. points) France United Kingdom United States Table 4: Change in post-tax income shares (in percent) of the P99.5-P100 income group in France, United Kingdom and the United States. In the second column, the data for France is from 2005, the data for the U.K is from 2000, and the data for the U.S. is from Source: Calculations based on Piketty and Saez (2007). Income differences are too large Agree or strongly agree China France United Kingdom United States Table 5: Sources: ISSP 197, ISSP 1999 and ISSP In 2009, the sample sizes are 2,989, 2,772, 932 and 1,512 respectively. In 1999, the samples sizes in France, U.K, and the U.S. was 1,889, 804 and 1,272. In 1987, the samples sizes are (1,177) U.K and 1,490 (U.S.). Differences in income are too large China France United Kingdom United States Percent Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree Figure 1: Responses to the statement Differences in income in are too large. Source: International Social Survey Programme,

6 Perceived amount of federal income tax paid Too high About right Too low Table 6: Source: General Social Surveys, N= 1,394; 869; 1,777; 1,952 respectively. Preferred choice Reduce taxes Spend more on social programs Men 47.9% 52.1% Women 33.5% 66.5% Table 7: Source: General Social Survey ( ), N=1, Some costs of inequality: efficiency concerns and incentive effects Economic research in general recognizes the trade-off between efficiency and equality of outcomes (e.g. Krueger, 2002, writes that societies must strike a balance between the beneficial incentive effects of inequality and the harmful welfare-decreasing effects of inequality ) but several papers have also document that inequality can lower efficiency. Chong and Gradstein (2007) present a theory backed by empirics that the size of the informal sector grows with income inequality. Using individual-level data, Gustavsson and Jordahl (2008) show that inequality in disposable income is negatively associated with trust (but inequalities in gross income are generally insignificant). Petrova (2008) shows evidence that media freedom is lower in countries with large income inequality. Shiller (2012) quotes the former French president Sarkozy as saying that [p]urely financial capitalism has perverted the logic of capitalism. Financial capitalism is a system of irresponsibility and is... amoral. It is a system where the logic of the market excuses everything. (note that Sarkozy was a right-wing politician). Frank (2008) advocated progressive taxes for efficiency reasons: If positional externalities influence spending patterns in the ways suggested by available evidence, higher marginal tax rates on top earners would appear justified not only on grounds of equity, but also on grounds of narrow economic efficiency. For a model of optimal progressive taxation of capital, see Saez (2013). On the other hand, Friedman (2006) warns that romantic socialism, combining strains of Marx and Rousseau, continues to attract adherents. So do fundamentalist movements that celebrate the presumed purity of pre-industrial society. Other relevant work includes Gravelle and Sutton (2009), Oshio et al. (2010), Chen (2009), Lo Prete (2013), Chow and Galak (2012) and Höchtl et al. (2012). 3 Data In December 2012, 1350 subjects were recruited through Amazon MTurk to answer a survey. After accepting the task, subjects click on a link that moved them to a website designed in Qualtrics, which allowed for randomization of subjects into distinct groups. 6

7 In the whole sample, 62.2% of subjects thought that the government should do something about income inequality, This fraction is consistent with a recent ABC News/Washington Post poll in which 60% of Americans supported government action to reduce income inequality. ) 5 Subject takes a "survey" Control group (No information) Information treatment Figure 2: Experimental design 3.1 Exposure information Shortly after starting the survey, approximately 2/3 of the subjects was placed into the information treatment (see Figure 2). Of the subjects in the information treatment group, about 1/2 was asked to read and comment on a table that showed the average income of U.S. households in the bottom quintile (using the more informal description of lowest 20% ), the averages for the remaining quintiles and then the average income of richest 0.1% of American households. 405 subjects were allocated to this treatment. The same information was represented visually to a different group of subjects (N=424). This treatment is also shown in the figure. 5 The results are summarized at accessed on 02/27/2013. The report adds that 75 percent in an ABC/Post poll...supported raising taxes on millionaires. On the other, the public in general favors smaller government that provides fewer services. 7

8 Treatment group 1: Table with income groups shown (N=405) Mean SD Median Trimmed Female 44% % Age Education group Voted of Obama 53% 50% 1 53% Voted for Romney 16% 37% 0 8% Income group Rich pay their fair share Treatment group 2: Graph with income groups shown (N=424) Mean SD Median Trimmed Female 44% % Age Education group Voted for Obama 54% % Voted for Romney 14% % Income group Rich pay their fair share Control group: (N=468) Mean SD Median Trimmed Female 43% % Age Education group Voted for Obama 56% 50% 1 58% Voted for Romney 12% 33% 0 3% Income group Rich pay their fair share Answered prior to treatment assignment. (0=Far too litte; 100=Far too much). Table 8: Summary statistics of demographics by treatment. trimmed by the top and bottom 10%. The last column shows means for sub-samples that The remainder of the subjects was assigned to a control group. These subjects observed no information related to income distribution (and were asked the same questions in the part of the survey that followed). Selected summary statistics for all three groups are shown in Table 8. Balance was achieved through computerized randomization. (Summary statistics for the entire sample are shown in Table 9.) 4 Data Analysis 4.1 Impact of exogenous information Does showing a summary of incomes earned by different income groups in the United States impact reported preferences for government activity to address income inequality? Yes: Table 10 shows 8

9 Female 43.4% Voted for Obama 54.2% White 76.1% Black 6.2% Hispanic 5.1% Asian 10.6% Mean age 33.0 Median education Some college, no degree 10th percentile education High school graduate 25th percentile education Some college, no degree 75th percentile education College degree Median household income group $37,500 - $49,999 10th percentile household income group Less than $12,500 25th percentile household income group $12,500 - $24,999 75th percentile household income group $62,500 - $74,999 Employment status Employed, full-time 35.5% Employed, part-time 18.5% Unemployed 16.1% Student 25.4% Other 4.5% Table 9: Summary statistics (N=1303) the regression results, and the information treatment effect is shown visually in Figure??. The impact of the viewing the graph that depicts income groups in the U.S. is slightly less than 1/3 the size of the coefficient on voting for Obama. Voters for the Democratic president are more open to government activity (controlling for treatment group assignment) and the opposite is true for those who voted for Mitt Romney in High income and more educated individuals are significantly less likely to say that the government should be involved in reducing income disparities. Women in general are more open to government activity, but in the most expentsive specification the significance of the coefficient on gender fall to 10%. Reactions the the just world statement ( people who want to get ahead could make it if they were willing to work hard ) is predictive of agreement with the proposition that the government should do something about income inequality (again, controlling for treatment assignment). 4.2 Do sources of information matter? Watching MSNBC and reading political blogs frequently are activities generally associated with openness to government activity (Table 11) and reading political blogs is associated with different beliefs about the sources of income inequality. For example, those who read political blogs more frequently are more likely to that a key source of inequality is that people lack necessary skills and they are less likely to assign blame to low taxes paid by the wealthy. Higher frequency of reading about politics is also correlated with a weaker belief that Obama bailouts are responsible 9

10 Income Graph/Table shown Do you think that the government should do something about income inequality? No Yes No opinion No information provided Percent (by treatment) Figure 3: Information treatment effects (general views about government activity) Income Graph/Table shown No information shown Romney voters only No Yes No opinion Income Graph/Table shown No information shown Obama voters only Figure 4: Information treatment effects (by ideology) 10

11 Dependent variable: The government should do something about inequality (1=yes, 0=no) (1) (2) Treatment: inequality shown (table only) (0.0864) (0.0942) Treatment: inequality shown (graph only) (0.0852) (0.0921) Voted for Obama (0.0862) Voted for Romney (0.124) People who want to get ahead could make it if they were willing to work hard ( ) I have a good chance to improve my standard of living ( ) Female (0.0800) Income group Education group (0.0157) (0.0475) Age (0.0210) Age ( ) Observations Probit regressions. Standard errors in parentheses + p <.1, p <.05, p <.01 Table 10: The government should do something about inequality (1=yes, 0=no). provision remains significant only when provided in a graphical form. Information for inequality. The differences in beliefs about reasons for inequality are explored in more detail in Section Beyond information Mobility Regressions without treatment controls are reported in Table 11. In this specification, ideology and beliefs about mobility are the best predictors of demand for government action. Those said earlier in the survey that the right pay their fair share (or too much) in taxes are less likely to say that the government should address income inequality. Belief in a just world, high income, and high education remain significant predictors of a lower propensity to demand government activity. Although reading political blogs remains positively correlated with the view that the government should so something, the significance of the coefficient on regularly viewing MSNBC fall to 10%. 11

12 4.4 Perceived reasons for inequality Broad patterns Table 14 lists the mean value for each of the 6 reasons by ideology. Among those who voted for Obama, low taxes paid by the rich are the most common perceived reason for inequality, followed by big banks and Bush tax cuts. For a general summary of reported beliefs, see Figure 5. Among those subjects who voted for Mitt Romney, weak economy is on average viewed as the most important reason for income inequality, followed by the fact that many people lack necessary skills. Obama bailouts are the third most popular reason among Romney voters, although they are the last reason among Obama voters. Inspection of mean values shows that there is some disagreement between Obama and Romney voters about the responsibility of Bush tax cuts and Obama bailouts (as one would expect), but the largest gap in attribution between these voters is about the possibility that the rich pay too little in taxes. Among those who voted for a third-party candidate, who did not vote, or who preferred not disclose their choice of candidate, weak economy is the most highly scoring reason. On the possibility that the low taxes on the rich a reason for inequality, independents and non-voters are closer to Obama voters than to Romney voters. Table 15 shows mean values and t-tests by gender. Among women, the most popular reason is weak economy, followed immediately by low taxes on the rich and big banks. For each of the 6 reasons, men report lower attribution values than women. Not controlling for ideology, men report low taxes, big banks and weak economy as the most important reasons for inequality Regression results In some cases, ideology predicts how people think of reasons for income inequality. Belief in a just world correlates with assessments of the sources of inequality in all cases (see Table 12). In some cases, uncertainty is positively correlated with a belief in a particular source of inequality. This is the case for reasons three, four, and six: people who are more uncertain about the future assign a higher blame to Obama bailouts, low taxes paid by the rich, and to big banks. 6 Education in negatively correlated with viewing weak economy, low taxes by the rich, or the big banks as the culprits for inequality (the other 3 choices are uncorrelated with education). High income individuals are significantly less likely to assign blame to the weak economy and Obama bailout. Belief in other reasons is not correlated with income. There is some agreement between the viewers of CNN and Fox News. Those who watch either or these channels more often are more likely to blame the weak economy (on average). In general, however, coefficients on CNN and Fox News have the opposite sign. For example, those who watch CNN are significantly more likely to blame the big banks, while the opposite is true for the regular viewers of Fox News. These viewers are also significantly more likely to see Obama bailouts as a 6 No causal interpretation is made here. 12

13 Fraction.of.subjects.who.believe.the.reason.is.relatively important Fraction.of.subjects.who.believe.the.reason.is.relatively unimportant Reason.for.inequality:.weak economy Voted.for Obama Voted.for Romney Voted.for Did.not. someone vote else (or.answer) Reason.for.inequality:.lack of skills by people Voted.for Obama Voted.for Romney Voted.for Did.not. someone vote else (or.answer) Reason.for.inequality:.the rich pay low taxes Voted.for Obama Voted.for Romney Voted.for Did.not. someone vote else (or.answer) Reason.for.inequality:.big banks Voted.for Obama Voted.for Romney Voted.for Did.not. someone vote else (or.answer) Reason.for.inequality:.Bush tax cuts Reason.for.inequality:.Obama bailouts Voted.for Obama Voted.for Romney Voted.for Did.not. someone vote else (or.answer) Voted.for Obama Voted.for Romney Voted.for Did.not. someone vote else (or.answer) Figure 5: Which reasons for income inequality are classified as relatively important (value > 50)? Fractions of respondents by ideology. 13

14 culprit (the coefficient on CNN in the regression is negative, but insignificant). Ideology is a very strong predictor of views on the taxation of the wealthy, Bush tax cuts, and Obama bailouts (all in the expected direction), but ideology does not predict beliefs about the importance of skills, the weak economy, or the big banks. Lastly, those who believe in a just world are more likely to see the weak economy, the lack of skills, or Obama bailouts as sources of inequality, and they are significantly less likely than the rest of the sample to see the big banks, the Bush tax cuts, or the taxation of the wealthy as an important variable Dispersion of beliefs Figure 12 shows the attribution to the weak economy by ideology. Republicans are slightly more likely to blame the economy for inequality than the average Obama voter, but the mode is the same for both groups. Figure 13 shows the perceived importance of the fact that many people lack necessary skills. This explanation might be the closest to an answer that an average economist would provide to someone who asks for helicopter perspective on changing fortunes of different groups in the society. Figure 14 outlines the believed importance of low taxes paid by the rich. The subjective perception of the blame of the big banks is show in Figure 15. Figure 16 shows the results for Bush tax cuts. Figure 17 show the believed importance of Obama bailouts. 4.5 Do the rich pay their fair share in taxes? (Survey results) Holding a collection of observables constant (see Table 25), higher income income is not associated with a belief that the rich pay more than their fair share of taxes. While the coefficient has the expected sign (it is positive, and the Moreover, people who are uncertain about the future are more likely on average to say that the rich do not pay their fair share. Those who voted for Obama, and people who watch MSNBC more often are also more likely to believe that the rich pay too little in taxes. The ideology effect of voting for Obama is approximately nine times stronger than a 10-point movement (out of 100) on the uncertainty scale. The Romney coefficient almost the exact inverse of the Obama coefficient. People who recently voted for the Republican generally believe that the share of taxes paid by the rich is about fair. While the correlation with education and belief is weaker, education is associated with beliefs that the rich pay either their fair share or too much. Moving from one education group to a higher group is roughly one fifth of the magnitude of the Romney coefficient. Controlling for all other observable, watching Fox News mor frequently and believing that people who want to get ahead could make it if they were willing to work hard is also associated with a higher propensity to say that taxes paid by the rich are either fair or too high. Lastly, whites 14

15 are more likely to evaluate taxes on the rich as too high, while the coefficient on other race groups are insignificant. Other specifications are reported in Tables 28 and 29 in the appendix as robustness checks. The coefficient sizes remain similar. 4.6 Uncertainty and demand for redistribution In an ideal society, would making incomes more equal be important? The correlations exploring the association between this sentiment and and uncertainty are reported in Table 24. (In the appendix, associations with optimism (the belief that things will get better for me ) are reported in Table 27 and correlation with optimism about the future of the country ( things will get better for the country ) are reported in Table 26.) 4.7 Does it come down to beliefs about mobility and ideology? Subjects who say that a poor person is most likely to escape poverty in the United States (out of a menu of three countries: U.S, France, and Russia) on average want less government involvement, regardless of treatment assignment. 5 Conclusion These are the preliminary results from the experiment: Exposure to information about income distribution does not make subjects more willing to support higher taxation. But information provision does make subjects more open to government activity in general if the stated goal is reduction of income inequality. Attitudes on equalizing incomes are sharply divided along ideological lines. Assignment to the treatment that makes salient the indirect costs of taxation does not lead subjects to change their prior views on the desirability of additional taxation. People who feel uncertain about the future on average demand more government activity. Belief is a just world is negatively associated with demand for redistribution. Framing of the tax questions does not push subjects to change their answers. Naturally, the results in this paper do not imply that citizens do have accurate information, or that they do not make interpersonal comparisons (a vast literature in economics and psychology documents the comparisons are basically automatic). While the results suggest that information asymmetries are very unlikely to drive the low demand for redistribution in the United States, they do not show that Americans are indifferent to growing income disparities (in fact, some answers in 15

16 the survey suggest the opposite). The emerging pattern is consistent with the view that Americans wish there was less income inequality, but they are unwilling to employ redistribution as a tool to address the problem. The challenge for researchers who try to shed light on voters motivations is that measuring what exactly people know about their standing in the income distribution is tricky, and obtaining truthful answers to personal feelings about their standing on the income ladder may be even harder. 7 For this reason, provision of information to a randomly chosen subset of subjects who agree to answer a survey is a useful to way to work around measurement difficulties. 8 While the experimental part of this paper captures the causal effect of exposure to information (in a diverse sample of Americans), it would be best to interpreted it in conjunction with evidence collected by other researches. Given that the sample is not representative of the United States (men, Obama supporters, and people not currently in the labor force are over-represented), these limitations should lead the reader to interpret the correlations gleaned from the survey part of the experiment with caution. The message of this paper can be combined with the lessons from prior (generally survey-based) prior work. 7 In the United States, displays of envy would not generate much social approval, but perhaps measuring expressions of envy through voting and other means could be accurate in a different setting. 8 Moreover, even if some subjects are not able to understand the table or the graph that offers them an overview of incomes earned by different groups, the ability to interpret charts and tables is orthogonal to treatment. 16

17 Would you be open to higher taxes? (Revenue used for entitlements) Subject viewed a table or a graph Density Subject exposed to no information (Not at all) (Definitely in favor) Figure 6: Reported openness to higher taxes, by information treatment. Tax revenue would be used for entitlements. Would you be open to higher taxes? (Revenue used to make incomes more equal) Density Subject viewed a table or a graph Subject exposed to no information (Not at all) (Definitely in favor) Figure 7: Reported openness to higher taxes, by information treatment. Tax revenue would be used to make incomes more equal. 17

18 Would you be open to higher taxes? (Revenue used for social insurance) Density Subject exposed to no information Subject viewed a table or a graph (Not at all) (Definitely in favor) Figure 8: Reported openness to higher taxes, by information treatment. Tax revenue would be used for social insurance (such as unemployment benefits) Income table shown In which country is a poor person most likely to escape poverty? U.S. France Russia Income graph shown No information shown Percent (by treatment) Figure 9: Perceived social mobility (by information treatments) 18

19 In which country is a poor person most likely to escape poverty? U.S. France Russia Voted for someone else Did not vote/say Obama voters Romney voters Percent (by vote in 2012) Figure 10: Perceived social mobility (by political preferences) Would you be open to higher taxes if revenues were used for... Making incomes more equal Social insurance Entitlements Not at all Maybe Indifferent Mostly in favor Definitely in favor In 2012, voted for... Obama Romney Other None/NA Percent (by vote in 2012) Figure 11: Openness to higher taxes (by political preferences) 19

20 Perceivedlimportanceloflthelweak economy aslalreasonlforlincomelinequalityltoday Votedl forlobama Voted forlromney Density Figure 12: Reported importance of the weak economy, by ideology PerceivedDimportanceDofDa lack of skills by many people asdadreasondfordincomedinequalitydtoday Density VotedD fordobama Voted fordromney Figure 13: Reported importance of the fact that many people lack necessary skills, by ideology 20

21 Density Perceived.importance.of.low taxes paid by the rich as.a.reason.for.income.inequality Did.not.vote or.prefers.not.to.say Voted for.romney Voted. for.obama Figure 14: Reported importance of the rich pay too little in taxes, by ideology PerceivedDimportanceDofDbig banks asdadreasondfordincomedinequalitydtoday Density Voted fordromney VotedD fordobama Figure 15: Reported importance of big banks, by ideology 21

22 Perceived5importance5of5Bush tax cuts as5a5reason5for5income5inequality5today Density Voted for5romney Voted5 for5obama Figure 16: Reported importance of the Bush tax cuts, by ideology Perceived5importance5of5Obama bailouts as5a5reason5for5income5inequality5today Density Voted5 for5obama Voted for5romney Figure 17: Reported importance of Obama bailouts, by ideology 22

23 Reason for inequality: weak economy Reason for inequality: lack of skills by people Does not view OWS favorably Views OWS favorably Does not view OWS favorably Views OWS favorably Reason for inequality: the rich pay low taxes Reason for inequality: big banks Does not view OWS favorably Views OWS favorably Does not view OWS favorably Views OWS favorably Reason for inequality: Bush tax cuts Reason for inequality: Obama bailouts Does not view OWS favorably Views OWS favorably Does not view OWS favorably Views OWS favorably Figure 18: Boxplots of reported importance of the six competing reasons for income inequality, by views towards the Occupy Wall Street movement. For each of the six phenomena, subject expressed on a scale from 0 to 100 whether they thought that it was a reason for income inequality (0=Not a reason; 100=Key reason). 23

24 References Alesina, Alberto, A Devleeschauwer, W Easterly, S Kurlat, and R Wacziarg, Fractionalization, Journal of Economic Growth, 2003, 8, and Edward Glaeser, Fighting Poverty in the US and Europe: A World of Difference, Oxford University Press, and George-Marios Angeletos, Fairness and Redistribution, American Economic Review, March 2005, 95 (4), , 3 and Paola Giuliano, Chapter 4 - Preferences for Redistribution, Vol. 1, Elsevier B.V., , Rafael Di Tella, and Robert MacCulloch, Inequality and happiness: are Europeans and Americans different?, Journal of Public Economics, August 2004, 88 (9-10), Appleby, J, The intellectual underpinnings of American democracy, Daedalus, 2007, 136 (3), Aristei, David and Cristiano Perugini, Preferences for redistribution and inequality in well-being across Europe, Journal of Policy Modeling, April 2010, 32 (2), Benabou, Roland and Efe Ok, Social mobility and the demand for redistribution: The poum hypothesis*, Quarterly Journal of Economics, 2001, (May), and J. Tirole, Belief in a just world and redistributive politics, Quarterly Journal of Economics, 2006, 121 (2), Card, David, Alexandre Mas, Enrico Moretti, and Emmanuel Saez, Inequality at Work: The Effect of Peer Salaries on Job Satisfaction, American Economic Review, October 2012, 102 (6), Cervellati, Matteo, Joan Esteban, and Laurence Kranich, Work values, endogenous sentiments redistribution, Journal of Public Economics, October 2010, 94 (9-10), Chen, W H, Cross-national Differences in Income Mobility: Evidence from Canada, the United States, Great Britain and Germany, Review of Income and Wealth, 2009, 55 (1), Chong, Alberto and Mark Gradstein, Inequality and informality, Journal of Public Economics, February 2007, 91 (1-2), Chow, R M and J Galak, The Effect of Inequality Frames on Support for Redistributive Tax Policies, Psychological Science, December 2012, 23 (12), , 6 Corneo, Giacomo and Christina M Fong, What s the monetary value of distributive justice?, Journal of Public Economics, February 2008, 92 (1-2), Cruces, Guillermo, Ricardo Perez-Truglia, and Martin Tetaz, Biased perceptions of income distribution and preferences for redistribution: Evidence from a survey experiment, Journal of Public Economics, March 2013, 98, Dehejia, Rajeev, Thomas DeLeire, and Erzo F P Luttmer, Insuring consumption and happiness through religious organizations, Journal of Public Economics, February 2007, 91 (1-2), Faravelli, Marco, How context matters: A survey based experiment on distributive justice, Journal of Public Economics, August 2007, 91 (7-8),

25 Dependent variable: The government should do something about inequality (1=yes, 0=no) (1) (2) (3) (4) (5) (6) People who want to get ahead could make it if they were willing to work hard ( ) ( ) ( ) ( ) ( ) ( ) To succeed in life, it is important to have good education ( ) ( ) ( ) ( ) ( ) ( ) To succeed in life, it is important to have personal connections ( ) ( ) ( ) ( ) ( ) ( ) Most mobile: U.S (0.0775) (0.0805) (0.0823) (0.0879) (0.0941) Voted for Obama (0.0835) (0.0868) (0.0937) (0.102) Voted for Romney (0.121) (0.124) (0.136) (0.149) I have a good chance to improve my standard of living ( ) ( ) ( ) Female (0.0815) (0.0878) (0.0945) Income group (0.0158) (0.0171) (0.0180) Education group (0.0479) (0.0524) (0.0556) Age (0.0219) (0.0248) (0.0265) Age ( ) ( ) ( ) Fair share paid by the rich ( ) ( ) Employed (0.0956) (0.0991) White (0.130) (0.134) Black (0.221) (0.236) Hispanic (0.219) (0.242) Trust (0.0883) (0.0937) TV: Fox News (frequency) ( ) TV: CNN (frequency) ( ) TV: MSNBC (frequency) ( ) Political blogs (frequency) ( ) Constant (0.210) (0.215) (0.221) (0.449) (0.523) (0.554) Observations R 2 Standard errors in parentheses + p <.1, p <.05, p <.01 Table 11: How important are beliefs about mobility? 25

26 In your view, what are the reasons for income inequality today? (Each option 0-100) (1) (2) (3) (4) (5) (6) Weak economy Many people lack skills The rich pay too little in taxes Big banks Bush tax cuts Obama bailouts Voted for Obama (1.574) (1.676) (1.681) (1.545) (1.619) (1.742) Voted for Romney (2.298) (2.209) (2.779) (2.363) (2.204) (2.650) Female (1.320) (1.476) (1.441) (1.364) (1.390) (1.486) I am uncertain about the future (0.0319) (0.0359) (0.0341) (0.0315) (0.0342) (0.0344) I have a good chance to improve my standard of living (0.0372) (0.0436) (0.0428) (0.0350) (0.0393) (0.0422) People who want to get ahead could make it if they were willing to work hard (0.0306) (0.0365) (0.0314) (0.0317) (0.0347) (0.0355) The Occupy Wall Street movement is worth supporting (0.0283) (0.0295) (0.0293) (0.0272) (0.0287) (0.0307) Income group (0.265) (0.302) (0.299) (0.277) (0.277) (0.308) Education group (0.811) (0.929) (0.832) (0.763) (0.793) (0.889) Employed (1.501) (1.637) (1.603) (1.510) (1.562) (1.699) Age (0.367) (0.411) (0.512) (0.416) (0.439) (0.450) Age ( ) ( ) ( ) ( ) ( ) ( ) TV: Fox News (frequency) (0.0310) (0.0335) (0.0355) (0.0309) (0.0330) (0.0359) TV: CNN (frequency) (0.0305) (0.0321) (0.0346) (0.0320) (0.0344) (0.0335) TV: MSNBC (frequency) (0.0329) (0.0377) (0.0368) (0.0332) (0.0370) (0.0357) Political blogs (frequency) (0.0259) (0.0279) (0.0264) (0.0259) (0.0264) (0.0282) White (1.933) (2.057) (2.173) (1.918) (1.885) (2.045) Black (3.113) (3.675) (3.414) (3.068) (3.296) (3.506) Hispanic (3.043) (3.686) (3.526) (3.231) (3.324) (3.921) Trust (1.364) (1.519) (1.465) (1.346) (1.387) (1.526) Constant (8.175) (8.984) (10.14) (8.759) (9.216) (9.433) Observations R The order of options offered to subjects was randomized. Standard errors in parentheses + p <.1, p <.05, p <.01 Table 12: What predicts attributions of different reasons for income inequality? 26

27 Fong, C, Social preferences, self-interest, and the demand for redistribution, Journal of Public Economics, 2001, 82 (2), Frank, Robert H, Should public policy respond to positional externalities?, Journal of Public Economics, August 2008, 92 (8-9), Freeland, Chrystia, Plutocrats: The Rise of the New Global Super-Rich and the Fall of Everyone Else, Penguin Press, Friedman, Benjamin M., The Moral Consequences of Economic Growth, Vintage, Gravelle, Hugh and Matt Sutton, Income, relative income, and self-reported health in Britain , Health Economics, February 2009, 18 (2), Gustavsson, Magnus and Henrik Jordahl, Inequality and trust in Sweden: Some inequalities are more harmful than others, Journal of Public Economics, February 2008, 92 (1-2), Hines, J R, Taxing consumption and other sins, Journal of Economic Perspectives, 2007, 21 (1), Höchtl, Wolfgang, Rupert Sausgruber, and Jean-Robert Tyran, Inequality aversion and voting on redistribution, European economic review, October 2012, 56 (7), Hodler, Roland, Redistribution and Inequality in a Heterogeneous Society, Economica, October 2009, 76 (304), Krueger, A B, Inequality, too much of a good thing, World Bank Policy Research Working Paper, Lusardi, A., Michaud, P.C., and O S Mitchell, Optimal Financial Knowledge and Wealth Inequality, NBER Working Paper, Mollerstrom, Johanna and David Seim, Does the Demand for Redistribution Rise or Fall with Cognitive Ability?, Mimeo, June 2012, pp Mullainathan, S., M. Nöth, and A. Schoar, The market for financial advice: Unpublished Working Paper, An audit study, North, Douglass, Structure and Change in Economic History, W. W. Norton & Company, Norton, M. I and Dan Ariely, Building a Better America One Wealth Quintile at a Time, Perspectives on Psychological Science, February 2011, 6 (1), Oshio, Takashi, Kayo Nozaki, and Miki Kobayashi, Relative Income and Happiness in Asia: Evidence from Nationwide Surveys in China, Japan, and Korea, Social Indicators Research, November 2010, 104 (3), Petrova, Maria, Inequality and media capture, Journal of Public Economics, February 2008, 92 (1-2), Piketty, T and E Saez, How progressive is the US federal tax system? A historical and international perspective, The Journal of Economic Perspectives, 2007, 21 (1), , 5 Prete, Anna Lo, Economic literacy, inequality, and financial development, Economics Letters, January 2013, 118 (1), Saez, Emmanuel, Optimal progressive capital income taxes in the infinite horizon model, Journal of Public Economics, January 2013, 97 (C),

28 Saint-Paul, G, Cognitive Ability and Paternalism, IDEI Working Paper, Shiller, Robert J., Finance and the Good Society, Princeton University Press, Solt, Frederick, Philip Habel, and J Tobin Grant, Economic Inequality, Relative Power, and Religiosity*, Social Science Quarterly, April 2011, 92 (2), Spitzer, Eliot, Government s Place in the Market, The MIT Press, Wang, Xiao Yu, Endogenous Insurance and Informal Relationships, MIT Job Market Paper, January 2013, pp

29 Weak economy People lack necessary skills The rich pay too little in taxes Relatively Relatively Relatively Relatively Relatively Relatively unimportant important unimportant important unimportant important Voted for Obama 22.7% 77.3% 39.3% 60.7% 11.0% 89.0% Voted for Romney 19.6% 80.4% 19.1% 80.9% 62.0% 38.0% Other candidate 25.2% 74.8% 36.4% 63.6% 30.6% 69.4% Did not vote* 20.7% 79.3% 32.0% 68.0% 29.9% 70.1% Big banks Bush tax cuts Obama bailouts Relatively Relatively Relatively Relatively Relatively Relatively unimportant important unimportant important unimportant important Voted for Obama 23.0% 77.0% 27.7% 72.3% 77.0% 23.0% Voted for Romney 52.2% 47.8% 77.7% 22.3% 41.4% 58.6% Other candidate 30.3% 69.7% 39.3% 60.7% 59.5% 40.5% Did not vote* 27.8% 72.2% 47.7% 52.3% 60.7% 39.3% Table 13: Attribution of reasons for income inequality by ideology Mean value Reason Voted for Obama Voted for Romney Diff. candidate Did not vote/say Weak Economy Many people lack necessary skills The rich pay too little in taxes Big banks Bush tax cuts Obama bailouts Share 54.2% 14.2% 8.6% 23.0% Table 14: Mean attribution score (0=Not a reason; 100=Key reason) of perceived reasons for inequality, by ideology 29