Income, Inequality, and Participation

As is now well known, U.S. income inequality has grown dramatically over the past forty years. Less well appreciated is that there have been two distinct phases in the evolution of the income distribution. The first phase from the 1970s through the mid to late 1990s involved widening income gaps between skilled and unskilled workers and an ever-increasing education premium. Since the 1990s, however, inequality has been driven primarily by the increasing concentration of income in the top 1 percent, .1 percent, .01 percent, and so on, of the population.

The political implications of these changes is an active area of recent research. Much of this research has focused on the extent to which politicians and policy differentially respond to the views of upper and lower income constituents. One of the most important recent contributions is that of my colleague Marty Gilens (2012). Gilens collected information from more than 1,900 national polls from 1981 to 2002 that asked voters whether or not they support or oppose some policy change. In each of the polls used in his study, voters were also asked about their family income, allowing Gilens to estimate policy responsiveness—the likelihood of a policy change as a function of the level of polling support for the change from different income groups. When Gilens examines policy responsiveness to voters in different income groups, he finds only a modest difference between the responsiveness to public opinions for low- and high-income voters. But Gilens argues that this finding of equality is due to the fact that rich and poor voters often have similar preferences on many policy questions. What is relevant, Gilens argues, is what happens when the rich and poor disagree. Here his findings are disquieting. When the rich and poor (ninetieth and tenth percentile incomes) disagree by more than ten percentage points on a poll, the odds show a marked lack of responsiveness to the views of the poorer voters. Middle-income voters fair no better when pitted against high-income opposition; when the ninetieth percentile voters disagree with the fiftieth percentile voters, Gilens’s finding are the same: policymaking is responsive to the higher income group, but not the middle income group.

Gilens’s findings, however, do not always corroborate the claim that increasing levels of income inequality have exacerbated political inequality. When he considers trends in responsiveness to different income groups, his findings are surprising. He does not find that the political system was more responsive to middle- and low-income voices before the era of increasing inequality. While Gilens does find a trend toward greater responsiveness to high-income views, there are parallel if somewhat weaker trends in responsiveness to lower-income voters.

An obvious suspect for the mechanisms behind Gilens’s finding is the extent to which political participation is stratified by income. The bias in responsiveness may result if higher-income citizens are more active in politics than the poor. To address these issues, Kay Schlozman, Sidney Verba, and Henry Brady (2012) have updated and extended their classic work on political participation and civic engagement to examine how economic inequality contributes to inequality in citizen involvement in politics.

Schlozman et al.’s main focus is on gaps in participation and engagement across groups with different “socioeconomic status” (SES). Their measure of SES combines both income and education. This methodological decision helps the authors draw comparisons with their earlier work, but it unfortunately obscures comparisons with other work focusing on income. Moreover, their measure is also not entirely appropriate for contemporary discussions of inequality, since increases in income inequalities among groups with a given educational attainment have grown tremendously. Nevertheless, their findings are striking. Compared to voters in the lowest socio­economic quintile, the top quintile votes more often, discusses politics more, works on a campaign more often, contacts more office holders, and makes more campaign contributions. These gaps range from about thirty percentage points for contacting public officials or making campaign contributions to about ten percentage points for working on a campaign. The only activity in which low socioeconomic status citizens are more active than high socioeconomic status citizens is protest. But in contemporary America, neither group engages in a whole lot of that (so far).

But these authors’ data also do not tell a story that clearly matches the trends in economic inequality with trends in political behavior. Growing economic inequality has dramatically increased the material advantages of the top group relative to the bottom and has been associated with the institutional decline of lower- and middle-class mobilization platforms such as labor unions. But the authors argue that the gaps in political participation between groups have not grown over the past thirty years. In the 1960s, the gap between high and low groups in terms of the percentage engaging in at least one political act other than voting was about thirty-five points. It rose to forty-five in 1990, but settled back to thirty-five by 2008. As a summary of inequality in political activities, the authors often use a ratio of participation rates of the top group to the participation rates for the bottom group. Higher scores indicate more inequality. The specific trends are telling. For voting, the ratio declined from 1.84 in 1952 (high socioeconomic status citizens were 84 percent more likely to vote) to 1.36 in 1964 before climbing to over 2.0 in 1984. But this measure has been declining ever since, and was at only 1.54 in 2008. Other trends are similar. The ratio for a broad composite measure of political activity has been generally declining since the early 1980s. So any simple story of how a demobilization of the middle and lower classes might have created inequality—or an equally simple story of how inequality rather caused such a demobilization—is not consistent with their evidence.

As it happens, the findings for one political activity constitute an important exception to these trends: financial donations to campaigns. The gap in campaign contribution behavior across socioeconomic statuses is currently somewhat higher by historical standards, and the gap is magnified further if one looks at the top 10 percent of the scale. Yet this trend does not pack the punch of the well-known displays of increasing economic inequality. Thus, the findings do not suggest that the large increases in economic inequality have created more political inequality. Rather, it appears that the causality might run in the opposite direction. Persistent political inequalities permitted the increased concentration of income and wealth by impeding government responses to it. From the perspective of democratic theory, as well as the common­sense standards of the American ethos, this conclusion is, of course, much worse.

Because Schlozman et al. focus on inequalities among U.S. citizens, they do not address an important source of increasing participatory inequality among its long-term residents. In my previous work with Keith Poole and Howard Rosenthal (2006), we document the trends in voting participation accounting for the changing size and economic profile of noncitizen residents.

Figure 1 shows, for each income category, the 2000 distribution of respondents by citizenship and voting (using the Current Population Survey, or CPS). The figure illustrates the strong relationship between income and voting. In the lowest income category used by the CPS, nearly two-thirds of the respondents are either noncitizens or nonvoters. In contrast, of those respondents who report themselves to be in the highest income category, over three-fourths also report having voted. The figure also indicates that noncitizens are far more likely to appear in the lower income categories than in the higher ones.

The upshot my work with Poole and Rosenthal is that while the level of participatory inequality among citizens has been fairly constant (albeit high), the level of inequality in participation among residents has gone up dramatically. Consider Figure 2. The solid line shows the ratio of the median voter’s income to that of the overall average family income.1 Other than a decrease in the early 1970s, this ratio has been remarkably constant despite large increases in overall income inequality over the period. Over the entire time period the ratio has fallen .0017 points per year. But there is no statistically significant effect from 1986 onward. But if we look at the ratio of the median family and the median citizen to the overall mean, the story is somewhat different. Both of these ratios declined substantially throughout the period, but the decline for all families is considerably steeper than that for citizens. The median citizen income has declined .0019 per year, about the same as the decline for the voter median-mean ratio. This indicates that nonvoting by citizens has consistently led to a lower median-mean ratio by about .1. The ratio of the median family income to mean income has fallen at a faster rate, however. The difference, of course, is due to increasing numbers of poor noncitizens.

So what accounts for the wedge between the median-mean ratios for citizens and residents? Increased immigration, lower incomes for immigrants, or both could lead to this result. The answer is all of the above. Based on the November CPS sample, we find that the percentage of noncitizens went from 2.63 percent in 1972 to 8.6 percent in 2006. This almost tripling of the number of noncitizens drives a wedge between the citizen and resident ratios. But the more important effect is the decline in immigrant incomes relative to voters and nonvoting citizens.

As Figure 3 shows, the median incomes of immigrants have fallen dramatically relative to that of voters. In 1972, the median immigrant family had an income that was 78 percent as high as the median voter and higher than the median nonvoter. The ratio of the immigrant median to the voter median has fallen to 58 percent of the voter median and lies well below the median of non­voters. Alternatively, the ratio of the income of nonvoters to voters has not trended downward to the same extent.


So to sum up, the recent literature on the politics of inequality has demonstrated that political inequalities across income groups have been a persistent and consequential feature of American politics. But there is no strong evidence that inequalities of political participation among citizens have worsened as economic inequality has spiraled. But this small piece of good news is tempered by the role of immigration and naturalization policies in enhancing political inequality among America’s residents.



Brady, Henry E., Kay Lehman Schlozman, and Sidney Verba. 2012. The Unholy Chorus: Unequal Political voice and the Broken Promise of American Democracy. Princeton, NJ: Princeton University Press.

Gilens, Martin. 2012. Affluence and Influence: Economic Inequality and Political Power in America. Princeton, NJ: Princeton University Press.

McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2006. Polarized America: The Dance of Ideology and Unequal Riches. Cambridge, MA: MIT Press.

McCarty Concepts 1


  1. The ratio of median to mean incomes is a standard measure of inequality. If this ratio falls, inequality increases.
About the Author
Nolan McCarty is the Susan Dod Brown Professor of Politics and Public Affairs at Princeton University.