Lisa Keister, Duke University, Gabriel Zucman, London School of Economics, Thomas Shapiro, Brandeis University, Fabian Pfeffer, University of Michigan, Larry Bartels, Vanderbilt University, Michael Sherraden, Washington University, Wojciech Kopczuk, Columbia University, Ken Scheve, Stanford University
On April 24, 2015, a day-long conference on Rising Wealth Inequality was held at the University of Michigan, co-sponsored by the National Poverty Center, the School of Social Work, The Ford School of Public Policy, the Panel Study of Income Dynamics, and the Department of Economics.
The day, which brought together researchers from a number of disciplines, was divided into three sessions: Patterns of Wealth Inequality, with presentations by Lisa Keister (Professor of Sociology at Duke University), Gabriel Zucman (Assistant Professor of Economics at London School of Economics), and Thomas Shapiro (Professor of Law and Social Policy at Heller School for Social Policy and Management, Brandeis University); Consequences of Wealth Inequality, with presentations by Fabian Pfeffer (Research Assistant Professor, Institute for Social Research, University of Michigan), and Larry Bartels (Professor of Political Science, Vanderbilt University); and Responses to Wealth Inequality, with presentations by Michael Sherraden (Professor of Social Work, Washington University), Wojciech Kopczuk (Professor of Economics, Columbia University), and Ken Scheve (Professor of Political Science, Stanford University).
In addition to the over 100 attendees present, the event was live-streamed and included a joint session with a similar gathering of international wealth researchers in Germany. And, building on the conference, the New York-based Russell Sage Foundation plans to publish a special volume of the peer-reviewed Russell Sage Journal of the Social Sciences next year, focusing on new research in the area of wealth inequality. A call for proposals (deadline 6/12/15) can be found at: https://www.russellsage.org/publications/category/current_rfa_rsfjournal/wealth-inequality
Session 1: Patterns of Wealth Inequality
Wealth Inequality in the US: Is the One Percent Permeable?
After providing a quick overview of current income and wealth concentration, Lisa Keister began by arguing that if we want to truly understand whether our social and economic structures are ossified or open, we must understand the degree to which these top positions are accessible. When thinking about the so-called 1%, it is important to study not just the top of the income distribution or the top of the wealth distribution individually, but also those in the top of both distributions.
Based on data from the Survey of Consumer Finances, which is collected by the Federal Reserve, financial profiles of the three groups were presented. The median income of the members in each group (based on pooled data over the 1989-2010 period) varies dramatically: just over $1,000,000 for those in the top 1% of both distributions, about $719,000 for those in the top 1% of the income distribution, but just $249,000 for those in the top 1% of the net worth distribution. Likewise, the median net worth of those in the top 1% of both groups was just over $11,000,000, whereas it was about $2,493,000 for those in the top 1% of income, and about $8,890,000 for those in the top 1% of wealth. Indeed, some of those in the top 1% of income have significant negative net worth, and some of those in the top 1% of wealth have zero income. Looking specifically at the joint group underscores the degree to which income and net worth are concentrated in the US.
The size of the overlap between the groups has been relatively stable, averaging .46% (that is, about 0.5% of the population is in the top 1% in terms of both wealth and income). The fact that the tops of the income and wealth distributions do not overlap completely suggests that the 1% is permeable, but that there are likely different sets of traits that contribute to membership in each group. Of particular interest are the roles of not just inheritance, but also education and self-employment, since if there is permeability, we would expect to see a role for such factors, controlling for inheritance. In fact, an extremely large inheritance is indeed associated with membership in each top group, and the association is stronger for net worth than for income. A high level of education, particularly having a graduate degree, however, also increases likelihood of membership in all three groups, though the effect is stronger for income than wealth. Finally, being self-employed is also strongly associated with membership in all three top groups. Of course these three characteristics do not operate independently, and the combination effect is strong.
Directions for further research: Though the US has high inequality in terms of both income and wealth, other countries are high in one, but not the other; exploring how these patterns vary across the globe could inform public policy here. Another interesting topic is better understanding the source of current US wealth, and the role played by inheritance, occupation, and even national origin.
Wealth Inequality in the United States since 1913
While the increase in income inequality has been well recognized, whether wealth inequality has similarly increased is not as clear, according to Gabriel Zucman. He presented work with Emmanuel Saez in which they develop and test a new method to estimate the distribution of US household wealth over the past century, using capitalized US income tax returns and flow of funds data. This method takes into account the fact that the composition of wealth changes over time (for instance, the importance of agricultural wealth has declined, while financial wealth has increased).
Three trends are clearly visible. First, wealth inequality has followed a U-shaped pattern since 1913: it was high at the beginning of the 20th century, began falling around 1929, and then began to steadily increase again in the 1980s. Notably, the recent rise of wealth inequality is almost entirely due to the rise of the share of wealth held by the top 0.1% – which went from 7% in 1979 to 22% in 2012. (The top 0.1% is currently made up of 160,000 families with over $20 million in net assets, based on 2012 data.)
Second, contrary to widely held perceptions, today’s middle class does not own a significantly greater share of wealth than it did 70 years ago. Rather, the wealth share of the bottom 90% of households has followed an inverted U-shaped pattern: from a low point of 15% of all wealth in the late 1920s it rose to 35% by the mid-1980s (due largely to rising pension and housing wealth), and subsequently dropping to 23% by 2012 (due largely to increased debt, reduced savings, and the housing crash). In terms of wealth share, the entire bottom 90% holds just 22% of all wealth, with average wealth among this group being $84,000. As a comparison, the top 1% hold 42% of all wealth, and of this top group, those in the top 0.1% actually hold 22% of all wealth – the same proportion as the entire bottom 90%.
Third, the increased concentration of wealth at the top is driven by diversified wealth accumulation and surging (top) incomes. This highlights the difficulties faced by the middle class, who have a much more limited asset base, with home ownership representing the main asset for most.
Directions for further research: We need to better understand the structure and dynamic of income inequality by creating distributional national accounts. Also, wealth estimates could be refined by using information available in additional tax and other data bases (such as those that track home prices).
The Widening Racial Wealth Gap
The racial wealth gap in the US is substantial, and widening. According to Thomas Shapiro, this is a result of both historical injustices and on-going policy choices. Currently, the racial income gap between whites and blacks, and whites and Latinos, is roughly $40,000. The wealth gap, however, is over $100,000. The median White household had over $110,000 in wealth holdings in 2011, compared to just over $8,000 for the median Latino household, and just over $7,000 for the median Black household. Several factors are driving the increasing wealth gap. The most important appears to be the number of years of home ownership; at the 50th percentile, it accounts for roughly 28% of the observed racial wealth gap. The next most important factor is household income (explaining 17% of the gap). Having a college education and receiving inheritances are much less significant.
In terms of figuring out why this might be, Shapiro argued that public policies, some intentional and some inadvertent, have played a large role in exacerbating wealth disparities and influencing the ability to grow and maintain wealth. 25% of homeowners today, for example, can trace their ownership to the Homestead Act. In order to evaluate policies’ impact on the racial wealth gap, Shapiro described the Racial Wealth Audit, a tool he helped develop to model and measure the distributional impact of policies. One might ask, for example: what if the returns from home ownership (or education) were the same for Blacks as for Whites? Homeownership is the largest component of wealth for most families. Currently, homeownership rates are 73% for Whites and 45% for Blacks. Using the Audit, if we equalized the rate of ownership, the Black-White wealth gap would be reduced by 31%. Equalizing access/opportunity is not enough, however; we also need to look at the differential in the financial gains, or returns, to investments such as homeownership (or education). By equalizing predicted returns to homeownership, the Black-White gap is reduced a further 16%.
Directions for further research: We need to examine the relationship between rising wealth inequality and any detrimental impacts on families and communities. We need more work examining the tie between rising wealth at the top, and stagnation or de-accumulation of wealth at the middle and bottom. Also, it would be good to see more communication work on attitudes toward the Estate Tax.
Session 2: Consequences of Wealth Inequality
The Intergenerational Consequences of Wealth Inequality
Fabian Pfeffer began his presentation by noting that inequality in wealth among one generation leads to inequality of opportunity for the next. He then went on to discuss several related issues. First, using data from the Panel Study of Income Dynamics, it is now possible to observe two generations of families and thus compare (adult) children’s wealth to that of their parents at the same age. As a result, it is possible to predict a child’s wealth. If you double parents’ wealth, for example, children’s wealth increases 40%. Looking at wealth quintiles, among children growing up in the highest quintile, over half (57%) are in the top quintile themselves as adults, and another third are in the fourth quintile. Among those growing up in the lowest quintile, on the other hand, 26% are in the bottom quintile as adults, and fewer than 15% ever make it to the top fourth or fifth quintile.
These findings raise the question of how wealth is transmitted across generations. Pfeffer reports that 24% of the correlation between parents’ and children’s wealth can be explained by their investment in education, while inheritances and gifts account for just 12%, and marriage 6%. Thus, while we tend to think of intergenerational transfers of wealth occurring later in life, in fact, much of what is transferred occurs when children are relatively young. There are huge gaps in educational attainment by net worth quintile, with wealth playing an important role not just in terms of access, but for graduation as well. Only 2/3 of those growing up in the lowest wealth quintile graduate from high school, just 15% enter college, and only 10% graduate. Among the top quintile, 90% graduate high school, and half enter college, virtually all of whom also graduate. Moreover, the wealth gap in education has been growing – that is, while college attendance has increased overall, it has increased more for those at the top of the wealth distribution.
This raises the question, why? One obvious explanation is the purchasing function of wealth: the wealthy have access to advantageous neighborhoods and schools, and can save for college. Another possibility is the insurance function of wealth. That is, having wealth changes the equation in terms of educational and labor market transition decision-making. If you have your own private safety-net, you can better take on the risk associated with going to college and choosing a career path, since any negative outcomes will be buffered. Moreover, wealth provides a psychological safety-net; if a child believes he or she can go to college, this can have a significant effect.
To further explore these issues, Pfeffer presented findings from cross-national comparisons looking at the US, Germany, and Sweden. Wealth inequality is similar across the three countries, though in the two European countries the purchasing function of wealth with respect to education is little to non-existent. On the other hand, the risk associated with going to college (and failing) may actually be higher in Germany and Sweden than in the US, due to the opportunity cost of college, and the fact that in those countries the returns to a college degree are lower. Based on status attainment models for each country, it appears that while the most important background effect is parental education, parental wealth is at least as important as parental occupation, and more important than family income, in terms of predicting a child’s future. These findings suggest the importance of the insurance role that wealth plays. Finally, to further illustrate this point, empirical evidence was presented indicating that parental wealth is strongly associated with a child choosing an area of study with high earnings uncertainty (such as art history, or sociology).
Directions for further research: We need to better understand the persistence of wealth and asset poverty across multiple generations, and fully account for the factors that contribute to wealth attainment.
Economic Inequality and Political Representation
Focusing on the political consequences of growing wealth inequality, and asking whether the political process can ever consider all citizens as equals, Larry Bartels discussed the impact of wealth inequality on democracy. He began by highlighting the degree to which the voting records of members of Congress are correlated with their constituents’ preferences – that is, Representatives from areas with more conservative constituents tend to have more conservative voting records, and vice versa. While on the one hand this may appear to be an example of perfect democracy, in a period in which the gulf between the two parties is substantial, it is more complicated, especially for moderate districts which get vastly different representation, depending on whether they elect a Democrat or a Republican.
Bartels next presented evidence that since 1948, real income growth, by income level, has looked very different under Democratic and Republican presidents. Those at the 95th percentile tend to do well no matter who is in charge, while those at the bottom do much better under Democrats. Moreover, in terms of Congressional responsiveness to constituents, the preferences of higher-income constituents receive much more attention, especially in the Senate. This has important implications, particularly when the preferences of the wealthy and the general public diverge, as they often do. Based on a survey of (a limited number of) wealthy Americans, conducted by Benjamin Page, Bartels, and Jason Seawright, Bartels reported that the wealthy, Republican and Democrat alike, do not generally support a number of policy issues that the general public favors. For example, while 87% of the general public support the idea of spending whatever is necessary to ensure good public schools for all children, just 35% of the wealthy do; 78% of the general public would like a minimum wage high enough that no family with a full-time worker falls below the poverty line, but just 40% of the wealthy do. In fact, the only thing wealthy and average Americans seem to agree on is that the differences in income in America are too large! (Just over 60% of each group feel this way.)
The policy response to the Wall Street meltdown was presented as a test case of a situation in which the wealthy and the general public had very different desires. The rescue plan that arose was ultimately what the wealthy wanted, and several possible mechanisms for how this came about were discussed. For example, then Treasury secretary Timothy Geithner appears to have spoken numerous times to CEOs of Goldman Sachs, JP Morgan, and Citigroup, all of whom were personal acquaintances, if not friends, but did not represent the most troubled banks or markets. While no suggestion of corruption was implied, this does suggest that political context is shaped by professional/collegial relationships, and that worldviews and preferences are shaped by the company one keeps.
In closing, Bartels, noting that the American public seems largely ok with economic inequality, but wants political equality, suggested that our current system is in fact an oligarchy. Though low voter turnout among the general public and those at the lower ends of the spectrum is often cited as an explanation, disparities in voter turnout actually pale in comparison to disparities in influence.
Directions for further research: Measuring disparities in political responsiveness in other affluent democracies may help identify which systems (if any) produce more egalitarian patterns of political influence. There is also much more to be learned about the political preferences and behavior of wealthy Americans.
Session 3: Responses to Wealth Inequality
Is Universal and Progressive Asset Building Possible? Evidence from a Social Experiment, and Policy Influence
In his talk, Michael Sherraden, presented early results from the SEED for Oklahoma Kids experiment, a statewide randomized policy experiment with automatic Child Development Accounts (CDA). The purpose of the experiment is not so much to test whether individual behavior can be changed, but rather whether a policy structure can be put in place that changes outcomes for an entire population. CDAs are savings and asset building accounts to be used for long-term developmental goals. Ideally they are lifelong (begin at birth), universal (every child is provided with one), progressive (greater subsidies for poor children), automatic (opened without prior permission), and restricted (to be used only for education, for example). The goal of CDAs is to encourage asset building and reduce extreme asset disparities along socioeconomic lines. To the extent this occurs, families may be able to take advantage of opportunities they would not otherwise have been able to afford. Attitudinal and/or behavioral changes, such as developing a college-bound identity, may also result, further improving outcomes for children.
The SEED OK experiment is the first test of a universal CDA model. It began in 2007 with over 2,500 participants (newborns), randomly selected and evenly divided between treatment and control groups. Those in the treatment group were provided with a $1000 initial deposit into a state-owned 529 College Saving Plan. Treatment mothers were also given a time-limited incentive of $100 to open private 529 accounts, and low- and moderate-income families were eligible for a savings match for contributions to their individual accounts, up to $250/year. In addition, treatment mothers received mailings about their accounts and the importance of education. Control mothers received none of this, though like any parent, they were free to open 529 plans for their children if they so desired.
Sherraden presented short-term findings from the experiment. First, virtually 100% of the treatment group children had an account (one mother dropped out for religious reasons), compared to just 2.4% of the control group, indicating it is possible to reach an entire population. Second, SEED OK had positive impacts on both account holding and asset accumulation. The average per-child asset amount across all 529 accounts was $1,130 for the treatment group and $76 for the controls. In terms of opening their own accounts, 17.3% of treatment mothers had opened a private account, with a mean savings amount of $109, compared to the 2.4% of control mothers who had opened private accounts and saved an average of $76.
SEED OK also had an impact on a number of non-financial outcomes. Further analysis of survey results found modest, but statistically significant, positive impacts on parental educational expectations for their children, children’s socio-emotional development, and reduced maternal depressive symptoms (the latter two with effect sizes in line with findings from Head Start). Finally, impacts didn’t generally vary by race, but were often greater for those with lower socio-economic status. In face-to-face interviews, mothers reported that the accounts created hope and confidence for their child’s future, provided a sense of security, and indicated that somebody outside the family (i.e. the government) cared about their child’s future. Moreover, the fact that the money couldn’t be touched was important.
Directions for future research: We need more research asking whether wealth is associated with positive outcomes in outlook and behaviors. Important questions to ask as the SEED children age are: Will the assets accumulated, and any associated changes in attitudes, behaviors and child outcomes be enough to increase rates of college enrollment and completion among disadvantaged youth? Do accumulated assets need to exceed a certain threshold in order to achieve meaningful outcomes?
The Economics of Wealth Taxation
Wojciech Kopczuk focused on the notion that policy responses to growing wealth inequality, including taxation policy, require an understanding of the underlying causes of inequality. To this end, the composition of top wealth has been changing over time. First, since 1986 we have seen a dramatic increase in the share of earnings accruing to the top 1%. Second, the share of the top that is made up of women is decreasing, having peaked in the 1970s. This can be attributed to the decreased role of inheritance, which is relatively egalitarian, and the increased role of self-made wealth, which is largely male.
Turning to the question of how best to tax wealth, there are several options: taxing income from wealth (as the US does); imposing an annual wealth tax (as several European countries do); taxing transfers such as gifts, estates and inheritances; and taxing specific categories of assets, for example real estate. Kopczuk presented economic arguments illustrating the implications of various approaches. Taxing capital income, for example, is different from taxing labor income or consumption, which are similar, because it distorts inter-temporal prices. A tax on capital income is basically a tax on the wealth people accumulate, while a consumption tax taxes initial wealth – though a tax on consumption without a bequest tax will encourage bequests, because bequests are not just altruistic, but rather also provide value to the donor (that is, they are a form of consumption). A wealth tax, on the other hand, implies overly heavy taxation on the normal rate of return. Moreover, there are a number of implementation issues associated with wealth taxes, since many assets are hard to observe and/or value. In the end, Kopczuk argued that a capital income tax is preferable to wealth taxation, though a consumption tax is an even better alternative – particularly if a complementary tax on transfers is also imposed.
Directions for further research: It is important to better understand how the composition of top wealth has changed over time, and how this affects measurements of inequality.
Taxing the Rich: Fairness and Fiscal Sacrifice over Two Centuries
Speaking about when and why societies decide to more heavily tax the rich, Ken Scheve discussed his research with David Stasavage, and argued that the assumed link between democracy and progressive taxation – that democracies tax the rich more heavily, particularly when inequality is high, or when they are not captured by the wealthy – is not so simple. Data on top income and inheritance tax rates from 20 countries, from 1800 to the present, indicate that the average top rate of income and inheritance taxation was extremely low (virtually zero) for all of the 19th century, then rose dramatically in two main spurts between during the first half of the 20th century to a high of about 60%. Top rates then began a gradual decline until about 1990 when a more steep decline began.
Scheve and Stasavage maintain that countries tax the rich when they think the state has failed to treat citizens as equals. There are several competing ways to conceptualize what â€œas equalsâ€ means: everybody pays the same rate (equal treatment argument), everybody faces the same utility loss (ability to pay argument), or taxes compensate for unequal treatment in other – that is, if the rich are getting better treatment elsewhere, they are taxed to make up for it dimensions (compensatory argument). Tax fairness arguments are largely debates about which of these arguments is most compelling. To understand policy changes, the question then becomes, what are the conditions that favor each? Historically, compensatory arguments have come in two forms: offsetting other taxes, and equalizing war sacrifice. Absent political and economic conditions that make compensatory arguments credible, the other two arguments dominate, and there is less overall support for taxing the rich. Looking specifically at when the rich have in fact been more heavily taxed, other possible explanations such as the introduction of universal suffrage, party control by the left, and high rates of inequality, do not seem to explain the dramatic changes in policies observed during the 20th century. Instead, there is a strong correlation with the timing of and mobilization for the two World Wars, and this is particularly the case in countries that were democratic. The explanation is that because of age and exemptions, the rich do not typically supply labor to the war effort to the same degree as the poor and middle class, and so to preserve equal sacrifice in the war effort, they have to be heavily taxed. Both labor and capital are thus conscripted.
In closing, Scheve argued that perhaps the reason we don’t see the poor wanting to soak the rich more than we do is that the general population does not perceive it as fair. The war years were unique periods when compensatory fairness arguments came into play. The implications of modern high-tech warfare, with professional armies rather than mass mobilization, suggest that we are not likely to see a repeat of the 20th century conditions that allowed compensatory arguments to lead to very high top marginal income and inheritance tax rates.
Directions for future research: Fruitful areas for future research would be developing more comparable long-run measures of wealth inequality across countries, and developing better research designs for estimating both the impact of wealth inequality on tax policy and tax policy on wealth.