Mobility Optimism in an Age of Rising Inequality

ABSTRACT

Four decades of rising incomes at the top, combined with income stagnation and declining intergenerational social mobility for the majority of American households, have combined to produce what has been widely described as a “new gilded age” in the United States. Yet analyses of the best available survey data reveal little evidence of a proportional increase in policy demands for redistribution over time. What explains this puzzling pattern of non-responsiveness? One classical explanation, revived in recent economic literature, postulates that high and persisting levels of optimism about the chances for advancement and social mobility reduce Americans’ willingness to support redistributive public policies. Although seemingly paradoxical in the current economic environment, the “prospect of upward mobility” (POUM) hypothesis is consistent with cross-national survey evidence revealing relatively high levels of support for such beliefs. What about trends over-time? This raises an important and largely unexamined question: Have underlying beliefs about POUM also shaped Americans’ attitudes toward redistributive policies during the era of rising inequality? In this paper, we examine the POUM-policy preference link, and how it has changed in recent decades. We find that POUM beliefs have shaped how individuals form policy attitudes toward inequality and taxes, net of partisanship, income, and confidence in government. Study findings provide new and provocative evidence in support of the POUM hypothesis, and we discuss implications for models of inequality attitudes and more generally scholarship on the politics of rising inequality.

KEYWORDS:

• Political sociology
• inequality
• poverty and mobility
• social psychology

Acknowledgments

We thank Leslie McCall, Delia Baldassarri, David Grusky, Michael Hout, and audiences at Stanford, Temple, NYU, Yale, and the CIFAR Working Group on Successful Societies, chaired by Michele Lamont and Peter Hall, for vigorous discussions of these issues. We welcome comments and reactions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. See below (and also Table 1) for additional details about these items.

2. We note that most tax and inequality items were fielded in subsets of the GSS data collections. Figure 2’s charts display all years for which each item was fielded using identical instrumentation. For instance, Figure 2’s fourth item asking respondents whether taxes are too low for those with high incomes was fielded in four surveys (1987, 1996, 2006, and 2008). It was modified between the 1996 and 2006 surveys, and once more between the 2006 and 2008 surveys. These changes in instrumentation leave us with a pair of surveys (1987 and 1996) with which to consider item-specific opinion trends, and this is what is summarized in Figure 2’s fourth chart.

3. The tax and inequality scale combines (factor-weighted) responses to the first, seventh, eighth, and tenth items in Figure 2 (see Appendix A for question wordings). These items are the ones most commonly fielded across surveys, including with respect to the requirement of standardized instrumentation and item availability on the GSS’ rotating ballots. For Figure 2’s trend analysis, the tax and inequality scale spans the years 1987, 1996, 2000, and 2008; inclusion of any additional items requires deletion of two (or more) of these survey years’ worth of data.

4. Could this result be consistent with the longstanding distinction between political abstractions versus more concrete evaluations of policies (e.g., Free and Cantril 1968), or what is sometimes termed the “principle/policy” phenomenon? If we interpret the GSS’ items 5-11 (see Figure 2 and Appendix A) as variously making reference to abstractions/principles, one item shows a significant inegalitarian trend, four display significant egalitarian trends, and the final pair of items show no significant pattern. Of the four non-principle items (items 1-4 in Figure 2), three are characterized by significant egalitarian trends, with the fourth shows an inegalitarian pattern of change. The principle/policy distinction thus appears to cross-cut the pattern of trends observed in the GSS tax and inequality items.

5. Among the policies with respect to which Kuziemko et al. (2015:1490-1491) report small-to-nil treatment effects are respondents’ preferred tax rate on the richest 1% of Americans and on millionaires, government provision of food stamps, minimum wage laws, and the Earned Income Tax Credit.

6. The reducing income differences item is one of the only two of the eleven tax and inequality items that have been fielded in standardized fashion across more than four surveys; the other item is respondents’ assessments of their own income tax levels (see Appendix A, item 1). It is, in principle, possible to combine responses to these two items into a scale, but their lack of a statistical relationship (r=.003) argues against this. For the record, we do permit the respondents’ income tax item to factor into the scale comprising our second dependent variable. As discussed below, its estimated weight is notably low, again suggesting the relatively peripheral status of the income tax item to attitudes toward taxes and inequality.

7. The taxes and inequality scale spans the four GSS surveys conducted in 1987, 1996, 2000, and 2008. But in the multivariate analysis, the more limited fielding of key independent variables limits our estimation sample to the three, following years: 1996, 2000, and 2008. The four items comprising our tax and inequality scale have identical instrumentation for years in the analysis. Items with the highest factor loadings ask respondents whether the “government in Washington ought to reduce … income differences” (.53) and whether “inequality … benefits the rich” (.51), followed by items asking respondents “large differences in income are [not] necessary for America’s prosperity” (.14) and whether respondents’ federal income tax is [not] “too high” (.09).

8. Scoring coefficients for the two POUM items are .277, and with a negative eigenvalue for the second factor, the principal factor algorithm returns evidence of a single factor underlying responses to the two POUM items.

9. This measure asks respondents about “confidence” in leaders rather than “trust” in government. This traditional wording is a constraint of the GSS data, but we are aware of no viable alternative for analysis (e.g., the American National Election Studies surveys ask about trust in the federal government, but lack suitable POUM measures).

10. Throughout the analyses, we use survey estimation to take into account the sampling framework employed by the GSS. We also employ the recommended GSS weight (wtssnr) to take into account variation in the number of adults in a household, subsampling of nonrespondents, and nonresponse patterns stemming from introduction of the GSS panel starting with the 2006 survey.

11. A second point of relevance concerning government confidence is the possibility that this factor moderates the impact of POUM beliefs. If so, we should observe a significant interaction between these factors. We find no evidence for an interaction between POUM beliefs and government confidence: F(1, 254) = 1.43, Prob > F =.233.

12. Similarly, the fully standardized coefficient for household income is –.069 in the analysis of reducing income differences versus –.127 in the analysis of the tax and inequality scale.

Additional information

Notes on contributors

Jeff Manza

Jeff Manza is Professor of Sociology at New York University. With Clem Brooks, he is the co-author of Whose Rights? Counterrorism and the Dark Side of American Public Opinion (Russell Sage 2013) and the forthcoming The Two Inequalities: Rights and Redistribution in a New Gilded Age (Oxford University Press).

Clem Brooks

Clem Brooks is Rudy Professor of Sociology at Indiana University. His research interests are in the areas of political sociology, political psychology, and quantitative methods.