America's Inequality Trap

My latest solo-authored book is a collaboration with the University of Chicago Press. With a February 2020 release date, this book marks my most recent contribution to the study of economic and political equality. You can find more information about the book, including chapter summaries, short excerpts, online appendix material. and data and replication materials below.

The book can be purchased directly from the press or from Amazon if you prefer.

Summary

The gap between the rich and the poor as big as it's been in the past century. While by most measures the economy has been improving, soaring cost of living and stagnant wages have done little to assuage economic anxieties. Conditions like these seem designed to produce a generation-defining intervention to balance the economic scales and enhance opportunities for those at the middle and bottom of the country’s economic ladder—but we have seen nothing of the sort.

I argue that a key reason for this is politics. Rising inequality changes politics in several key ways that make combating inequality harder, thus creating a difficult to escape trap. Along the way, I address a series of important questions about how economic inequality and American politics shape and respond to each other over time. I ask how public opinion shapes distributional outcomes and how the public responds to rising inequality. I assess the effect of inequality on elections and how who controls government influences the gap between the rich and the poor. And I address the two-way relationship between the policymaking process economic disparity.

I find that, when a small fraction of the people control most of the economic resources, they also hold a disproportionate amount of political power. Among other things, the rich support a broad political campaign that undermines public and policymaker support for policies to reduce inequality. They also take advantage of interest groups to influence Congress and the president, as well as state governments, in ways that stop or slow down reform while at the same time generating policy changes that further contribute to their economic advantage. A key implication of this book is that social policies designed to combat inequality should work hand-in-hand with political reforms that enhance democratic governance as well as efforts to fight racism. A coordinated effort on all of these fronts will be needed to reverse the decades-long trend.

Contents

This chapter discusses the 2016 presidential election contest between Donald Trump and Hillary Clinton as a continuation of a pattern in which rising inequality feeds back into politics in ways that maintain or increase economic inequality. The core arguments of the book are summarized, including a brief description of an inequality trap and the types of feedback between inequality and politics that a trap implies - in elections, public opinion, and public policy. Some of the key findings in the book are quickly summarized, and their implications are briefly discussed. The chapter concludes with an outline of the remaining chapters in the book.

This chapter has four main goals. First, the concept of an inequality trap is introduced and some of the empirical markers of such a trap are discussed. Second, I review previous research on the political underpinnings of economic inequality and discuss several potential theoretical linkages that could create feedback from inequality to the American political system. Third, I discuss in very general terms the analytical strategies used later in the book. Finally, I present preliminary tests establishing the first key requirement of an inequality trap - inequality that feeds back on itself to perpetuate inequality.

This chapter begins with a discussion of thermostatic public opinion and dynamic representation, which I see as the two core components of a modernized median voter model. Discussing these two models and then applying them to income concentration provides the foundation for a potentially self-correcting feedback relationship between public opinion and income inequality. Such a self-correcting relationship would undermine an inequality trap. I then introduce a series of arguments that suggest a breakdown in the self-correcting inequality-opinion relationship. Many of these arguments are connected to anti-pluralist theories and suggest that as income is concentrated in the hands of a few and political power is concentrated in those same hands, mass preferences are likely to shift against the very policies that could level the economic playing field. I then present an analysis that combines macro level time series and individual-level data. I find evidence of a self-reinforcing opinion-inequality link among a subset of Americans. For these people, rising inequality appears to trigger less support for the very policies that would reduce income inequality. One of the key markers of this group is racial resentment. Those who don't respond to rising inequality in a self-reinforcing fashion basically don't respond at all. There is no public backlash to rising inequality, meaning that the public provides at best a flawed backstop against extreme levels of inequality.

The question I turn to in this chapter is whether inequality feeds back into politics by shaping election outcomes. Elections, in this chapter, are another possible pathway for inequality to shape politics in either a self-reinforcing or self-correcting pattern. In the previous chapter the focus was on public opinion. Many of the same factors that connect opinion and inequality might generate a similar feedback effect between inequality and election outcomes. By shifting the focus to election outcomes and voting behavior I ascertain whether the attitudinal patterns discussed earlier also reveal themselves in political behavior. In short, they do. As inequality rises, Republicans perform better electorally, which creates a dynamic that serves to reinforce inequality.

This chapter shows that rising inequality affects the U.S. policymaking process. The shifting electoral fortunes of the parties and (to perhaps some extent) changes in public preferences are only part of the reason that policies exacerbating inequality have been successful even in the face of rising inequality. How America's policymaking institutions aggregate preferences and process input from competing interests also plays a prominent role in the spiral of inequality. I argue that inegalitarian policy change can be at least in part explained by a combination of factors, ranging from unequal political voice to differences in the organizing strategies and success of the left and the right, to the rise of a neoliberal Washington consensus on many aspects of economic policy. Importantly, the factors that drive partisan agreement on inegalitarian policies are more likely to be in place as economic inequality rises. The core message is that U.S. institutions contribute to America's inequality trap.

This chapter explores how status quo bias in the American policy process can contribute to an inequality trap. The core question here is how America's policymaking institutions and the actors situated within those institutions deal with rising inequality. In focusing on the policymaking process, this chapter builds substantively on the analysis from the previous chapter. There, we saw evidence that rising inequality can facilitate inegalitarian policymaking. Thus, rising inequality can help overcome the status quo bias of the American policy process, specifically for a subset of policies that further increase economic inequality. In this chapter we will see that the distributional consequences of status quo bias change as inequality increases. When inequality is low, status quo bias has no effect on income concentration. But when inequality is high, the distributional effects of the policy stagnation resulting from status quo bias tends to be inegalitarian in its consequences. To some extent, the emphasis in this chapter is the institutional setting of the United States and the ways elements of constitutional design such as separation of powers and bicameralism shape the policy response to changing levels of inequality. But the chapter also explores how elite behavior within these institutions is shaped by inequality.

The weight of the findings from the prior chapters leaves us with a pressing question: is there any way out of America's inequality trap? The evidence suggests that breaking the vicious cycle will not be easy. This chapter seeks to illuminate this central question. I begin this effort by comparing the the government response to the 2008 economic crisis and to the Great Depression. Through this discussion I illustrate just how tight the inequality trap's grip is on contemporary U.S. political economy, and I point to strategies for loosening that grip.

Appendix material available here in .pdf format.

Figure Image Files

Figure 1.1: Top .01% Share of Income, 5-year Moving Average, 1913-2014

Figure 1.1: Top .01% Share of Income, 5-year Moving Average, 1913-2014

Source: Author's calculations from World Inequality Database.

Figure 2.1: Describing Feedback Processes

Figure 2.1: Describing Feedback Processes

Figure 2.2: Two Versions of A Political Inequality Trap

Figure 2.2: Two Versions of A Political Inequality Trap

Figure 2.3: A Simple Model of American Oligarchy

Figure 2.3: A Simple Model of American Oligarchy

Figure 2.4: Effect of Income Concentration on Future Values of Inequality

Figure 2.4: Effect of Income Concentration on Future Values of Inequality

Source: Author's calculations

Figure 3.1: Liberalism of Public Opinion

Figure 3.1: Liberalism of Public Opinion

Source: http://stimson.web.unc.edu/data/ accessed 1/21/2016

Figure 3.2: Is There a Reciprocal Relationship Between Inequality and Public Opinion?

Figure 3.2: Is There a Reciprocal Relationship Between Inequality and Public Opinion?

Source: Author's calculations from annual data, 1952 to 2014. Note: Charts plot orthogonalized cumulative impulse response functions based on a vector autoregression including top .01% income share and public mood conservatism. Models also include the top capital gains tax rate, top income tax rate, financial deregulation, and Congressional partisanship. The plot represents the predicted effect of a standard deviation shift in one variable on the other variable over a 20 year period.

Figure 3.3: Attitudes Toward Redistribution and Rising Inequality

Figure 3.3: Attitudes Toward Redistribution and Rising Inequality

Source: Author's calculations from GSS data. Note: Charts plot the predicted marginal effect of an increase in inequality on support for redistribution for those with differing levels of family income. Calculations based on a multi-level logit model including national-level top .01% income share at time of survey, race/ethnicity, sex, age, education, and income along with income interacted with inequality.

Figure 3.4: Racial Bias, Attitudes Toward Redistribution, and Rising Inequality

Figure 3.4: Racial Bias, Attitudes Toward Redistribution, and Rising Inequality

Source: Author's calculations from GSS data. Note: Left chart plots the predicted marginal effect of an increase in inequality on support for redistribution for those with differing levels of racial bias. Right chart plots the predicted marginal effect of an increase in racial bias on support for redistribution for those in increasingly unequal economic contexts. Calculations based on a multi-level logit model including national-level top .01% income share at time of survey, sex, age, education, racial bias, and income along with income and racial bias interacted with inequality.

Figure 3.5: Support for Minimum Wage Increase among the Rich and Poor as Inequality Increases

Figure 3.5: Support for Minimum Wage Increase among the Rich and Poor as Inequality Increases

Source: Author's calculations from CCES data

Figure 4.1: A Self-Reinforcing Pattern of Elections and Inequality

Figure 4.1: A Self-Reinforcing Pattern of Elections and Inequality

Figure 4.2: Is There a Reciprocal Relationship Between Inequality and House Elections?

Figure 4.2: Is There a Reciprocal Relationship Between Inequality and House Elections?

Source: Author's calculations from annual data, 1913 to 2014. Note: Charts plot orthogonalized cumulative impulse response functions based on a vector autoregression including top .01% income share and the percent of Democratic seats in the House of Representatives. Models also include union strength, financial deregulation, and the top capital gains tax rate. The plot represents the predicted effect of a standard deviation shift in one variable on the other variable over a 20 year period.

Figure 4.3: Inequality and Elections in the Senate and Presidency

Figure 4.3: Inequality and Elections in the Senate and Presidency

Source: Author's calculations from annual data, 1913 to 2014. Note: Charts plot orthogonalized cumulative impulse response functions based on VARs. Models include the percent of Democratic seats in the Senate or Democratic control of presidency along with top .01% income share. Models also include union strength, financial deregulation, and the top capital gains tax rate. The plot represents the predicted effect of a standard deviation shift in one variable on the other variable over a 20 year period.

Figure 4.4: Turnout Effects of Economic Inequality

Figure 4.4: Turnout Effects of Economic Inequality

Source: Author's calculations from ANES data, presidential elections 1952-2012. Note: Charts plot the predicted probability of self-reported turnout as inequality increases. Calculations based on a multi-level logit model including national-level top .01% income share at time of election, race/ethnicity, age, education, and income.

Figure 4.5: Individual Level Voting for Democratic Congressional Candidates

Figure 4.5: Individual Level Voting for Democratic Congressional Candidates

Source: Author's calculations from ANES data. Note: Chart plots the predicted probability of support for Democratic House candidate among those voting in the most recent election as inequality increases. Calculations based on a multi-level logit model including national-level top .01% income share at time of election, race/ethnicity, age, education, and income.

Figure 4.6: Marginal Effect of Inequality on Voting Behavior as Trust Increases

Figure 4.6: Marginal Effect of Inequality on Voting Behavior as Trust Increases

Source: Author's calculations from ANES data. Note: Charts plot the predicted marginal effect of an increase in inequality on support for Democratic House candidates for those with differing levels of trust in government. Calculations based on a multi-level logit model including national-level top .01% income share at time of election, race/ethnicity, sex, age, education, and income along with trust in government and trust interacted with inequality.

Figure 4.7: Inequality, Racial Attitudes, and Voting Behavior

Figure 4.7: Inequality, Racial Attitudes, and Voting Behavior

Source: Author's calculations from ANES data. Note: Chart plots the predicted probability of support for Democratic House candidate among those voting in the most recent election as inequality increases. Calculations based on a multi-level logit model including national-level top .01% income share at time of election, age, education, and income. White non-Hispanics only.

Figure 4.8: Predicted Likelihood of Trump Vote as State Income Concentration Increases

Figure 4.8: Predicted Likelihood of Trump Vote as State Income Concentration Increases

Source: Author's calculations from CCES data. Note: Chart plots the predicted probability of support for Donald Trump among those voting as inequality increases. Calculations based on a multi-level logit model including state-level top 1% income share, race/ethnicity, age, education, income, partisanship, and state median income.

Figure 4.9: Inequality and Trump Support by State Income and Individual Education

Figure 4.9: Inequality and Trump Support by State Income and Individual Education

Source: Author's calculations from CCES data. Note: Charts plot the predicted marginal effect of an increase in inequality on support for Trump as state median income and individual level education increases. Calculations based on a multi-level logit model including state-level top 1% income share, race/ethnicity, age, education, income, partisanship, and state median income. For the left chart, an interaction between state median income and state inequality is also included. For the right chart the model includes an interaction between state level income and individual level education.

Figure 5.1: Party Divergence in the House and the Senate, 1913-2014

Figure 5.1: Party Divergence in the House and the Senate, 1913-2014

Source: Calculated by author based on data from the Voteview website. Note: These charts shows the average DW-NOMINATE score for Republicans and Democrats in the House and Senate. DW-NOMINATE scores are calculated for individual legislators based on roll-call votes and provide an indication of the conservatism of each legislator. The 1st dimension of the measure is charted here, which captures an economic dimension of ideology.

Figure 5.2: Campaign Contributions from Non-Labor Sources and Top Income Shares

Figure 5.2: Campaign Contributions from Non-Labor Sources and Top Income Shares

Source: Calculated by author based on data from the World Inequality Database and the Federal Election Commission.

Figure 5.3: Income Concentration and Financial Deregulation

Figure 5.3: Income Concentration and Financial Deregulation

Source: Calculated by author with data from the World Inequality Database and Philippon and Reshef (2013).

Figure 5.4: Is There a Reciprocal Relationship Between Inequality and Financial Deregulation?

Figure 5.4: Is There a Reciprocal Relationship Between Inequality and Financial Deregulation?

Source: Author's calculations from annual data, 1913 to 2014. Note: Charts plot orthogonalized cumulative impulse response functions based on a vector autoregression including financial deregulation and top .01% income share. The plot represents the predicted effect of a standard deviation shift in one variable on the other variable over a 20 year period.

Figure 5.5: The Effect of Partisan Control of Policy Institutions on Deregulation, 1913-2014

Figure 5.5: The Effect of Partisan Control of Policy Institutions on Deregulation, 1913-2014

Source: Author's calculations from annual data, 1913 to 2014. Note: Each dot represents the estimated effect of annual change in each explanatory variable on annual change in financial deregulation. The lines around the dots show a 95% confidence interval.

Figure 5.6: The Effect of Partisanship on Financial Deregulatory Policy Before and After 1982

Figure 5.6: The Effect of Partisanship on Financial Deregulatory Policy Before and After 1982

Source: Author's calculations from annual data, 1913 to 2014. Note: Each dot represents the estimated effect of annual change in each explanatory variable on annual change in financial deregulation. The lines around the dots show a 95% confidence interval.

Figure 5.7: The Conditional Effect of Partisanship on Financial Deregulation

Figure 5.7: The Conditional Effect of Partisanship on Financial Deregulation

Source: Author's calculations from annual data. Note: Each line represents the estimated effect of annual change from Republican to Democratic control of the Senate. The estimated effect is allowed to vary depending on the value of the variable on the x (bottom) axis. Each point on the line, then provides an estimate of the effect of Senate control at a particular value on the x-axis of the conditioning variable. The shaded area around the lines show a 95% confidence interval.

Figure 6.1: Income Concentration and Party Polarization in Congress, 1913-2014

Figure 6.1: Income Concentration and Party Polarization in Congress, 1913-2014

Source: Calculated by author with data from the World Inequality Database and the Voteview website. Note: Top .01% share is plotted as a 5-year moving average.

Figure 6.2: Is There a Reciprocal Relationship Between Inequality and Polarization?

Figure 6.2: Is There a Reciprocal Relationship Between Inequality and Polarization?

Source: Author's calculations from annual data, 1913 to 2014. Note: The plot represents the predicted effect of a standard deviation shift in one variable on the other variable over a 20 year period using orthogonalized cumulative impulse response functions based on two vector autoregressions including top .01% income share, either House or Senate party polarization, and a measure of legislative policy stagnation.

Figure 6.3: The Constant Effect of Party as Inequality Rises

Figure 6.3: The Constant Effect of Party as Inequality Rises

Source: Author's calculations from bi-annual data, 1913 to 2014. Note: Chart plots the marginal effect (with 95% confidence interval) of legislator partisanship on legislator ideology in the House within a Congress as inequality is increasing or decreasing. Values on the left end of the chart show the effect of partisanship when inequality is declining while values on the right end of the chart show the effect of partisanship when inequality is increasing. Results are from a multi-level model with legislators nested within Congresses.

Figure 6.4: The Evidence so Far and Conditional Status Quo Bias

Figure 6.4: The Evidence so Far and Conditional Status Quo Bias

Figure 6.5: Effect of Status Quo Bias at Observed Levels of Top Income Share

Figure 6.5: Effect of Status Quo Bias at Observed Levels of Top Income Share

Source: Author's calculations. Note: Chart plots the marginal effect (with 95% confidence interval) of each variable at increasing values of inequality.

Figure 7.1: Examining the ``Top-Heaviness'' of Income Inequality in Old and New Gilded Ages

Figure 7.1: Examining the ``Top-Heaviness'' of Income Inequality in Old and New Gilded Ages

Source: Author's calculations from World Inequality Database. Note: Values Standardized to Mean=0 and Variance=1, 5-year MA.

Figure 7.2: The Response of Non-Whites to Rising Inequality

Figure 7.2: The Response of Non-Whites to Rising Inequality

Source: Author's calculations from ANES data. Note: Chart plots the predicted marginal effect on support for Democratic House candidate among those voting in the most recent election as inequality increases. Calculations based on a multi-level logit model including national-level top .01% income share at time of election, age, education, racial bias, and income along with racial bias interacted with inequality. Only blacks and Hispanics are included in this analysis.

Data and Code: Coming Soon!