The One Percent

Recent protest movements brought attention to the one percent, a segment of the population that is critical to understanding inequality and social mobility but that attracts relatively little research attention. In this article, I survey current research on the one percent in the United States. I distinguish income from wealth and show that both are very concentrated but that the concentration of wealth, particularly financial wealth, is extremely high. I describe the demographic traits and finances of households who are in the one percent and discuss how these have changed in the past decade. I review literature that explains rising top incomes, and I propose that future research will usefully concentrate more on top wealth owners and on the demographic and life course processes that underlie income and wealth concentration. I conclude by speculating about why Americans are so tolerant of resource concentration. 347 A nn u. R ev . S oc io l. 20 14 .4 0: 34 736 7. D ow nl oa de d fr om w w w .a nn ua lr ev ie w s. or g A cc es s pr ov id ed b y D uk e U ni ve rs ity o n 03 /0 8/ 17 . F or p er so na l u se o nl y. SO40CH16-Keister ARI 27 June 2014 12:52 Occupy Wall Street (OWS): a protest movement that started in September 2011 and is focused on raising awareness about inequality, the wealthy, and related issues Income: the flow of funds into a household over time from wages or salaries, government payments, investments, gifts, etc. Wealth or net worth: total property owned by a household at one time; usually measured as net worth or total household assets less total debts Survey of Consumer Finances (SCF): household survey data collected by the Board of Governors of the Federal Reserve System triennially since 1983; includes oversamples of high-income households INTRODUCTION OccupyWall Street (OWS) and related protest movements that started in September 2011 generated popular interest in income and wealth concentration by drawing attention to the gap between the financial status of the top one percent and that of other Americans (the other 99%). The movement began in Zuccotti Park in New York’s Wall Street financial district and quickly spread throughout the country, raising awareness about the power and advantages enjoyed by top income and wealth holders and related issues regarding the influence and regulation of corporations. The movement’s slogan—“we are the 99%”—gained popularity by suggesting that top income earners and wealth owners are both extraordinarily privileged and insulated from the negative effects of financial downturns of the sort that shocked the US economy starting in 2007. Indeed, during the 2007–2009 recession, the United States lost 8.5 million jobs (Hout et al. 2011), and unemployment reached a 26-year high of 10% (Bur. Labor Stat. 2012); wage and salary growth for private industry workers slowed to a 20-year low (Bur. Labor Stat. 2012); home values fell by about onethird (Fligstein&Goldstein 2011, Grusky et al. 2011a); foreclosures and personal bankruptcies rose to record levels (Treas 2010); and large numbers of middle-class families had to borrow from their retirement savings to pay basic living expenses (Wolff et al. 2011). Yet, as OWS supporters publicized, the share of total income going to top earners had risen continuously since the 1980s and was higher in 2007 than it had been since the Great Depression (Congr. Budget Off. 2011). In this review, I assess the current state of knowledge about the one percent, a segment of the population that is central to understanding social stratification and mobility but that attracts relatively little research attention. I start by defining income and wealth, two concepts that are important for identifying and understanding the one percent but that are often not distinguished carefully in academic writing or public discourse. I discuss the data challenges involved in conducting research on high income and wealth, and I summarize and evaluate the current state of knowledge about two distinct groups: (a) the top one percent of income earners and (b) the top one percent of wealth owners. I provide an overview of contemporary levels of income and wealth concentration, and I situate these trends historically by describing long-term patterns in income and wealth ownership. I then describe the demographic traits, income, and wealth of the households in the one percent in the past decade. Finally, I discuss explanations for the growing concentration of income and propose that future research should focus more on top wealth owners and on the demographic and life course patterns that contribute to resource concentration. I conclude by briefly speculating about why Americans are relatively tolerant of income and wealth concentration. Throughout the article, I review published research, but I also present original empirical estimates from the 2001–2010 Surveys of Consumer Finances (SCF) to summarize patterns in resource concentration and to highlight issues that future research might address. My focus is on sociological research, but I also include economic research, where the one percent attracts more attention. I focus on the United States in order to provide sufficient depth regarding a single case. Because income andwealth concentration are not perfectly correlated with general patterns of inequality over time (Hacker & Pierson 2010) and because there are other excellent reviews available on broader trends in income inequality (Lemieux 2008, McCall & Percheski 2010), I do not provide a thorough review of the inequality literature. Similarly, I do not assess literature on elite studies because it has been reviewed elsewhere (Khan 2011, 2012). DEFINING THE ONE PERCENT: INCOME AND WEALTH The one percent can be defined by either their income or their wealth. The two terms are 348 Keister A nn u. R ev . S oc io l. 20 14 .4 0: 34 736 7. D ow nl oa de d fr om w w w .a nn ua lr ev ie w s. or g A cc es s pr ov id ed b y D uk e U ni ve rs ity o n 03 /0 8/ 17 . F or p er so na l u se o nl y. SO40CH16-Keister ARI 27 June 2014 12:52 Financial wealth: total financial assets (e.g., stocks, bonds) but excluding real assets (e.g., housing, other real estate) often used interchangeably, but this can be misleading because they have very different meanings and potentially different implications for understanding resource concentration. Income is a flow of funds into the household over time fromwages or salaries, businesses, investments (i.e., interest and dividend income), capital gains, government transfer payments, gifts, and other sources. Income can be measured at either the individual or household level, and income from various sources can have different implications for well-being. For example, wage/salary income involves work and time commitments that are very different from those required to manage the investments that produce interest/dividend income or to run a business that produces business income. In contrast, wealth refers to the things people own at a single point in time and is usually measured as net worth (total household assets less total liabilities or debts). Assets include real assets (e.g., the home or primary residence, other real estate, business equity, vehicles) and financial assets (e.g., transaction accounts, certificates of deposit, bonds, stocks, mutual funds, retirement accounts). Debts include mortgages, consumer debt, student loans, and other liabilities. Financial wealth is total financial assets, a measure of relatively liquid assets such as stocks and bonds that, for most households, refers to nonhousing wealth. Financial wealth is particularly significant for understanding resource concentration because ownership of financial assets tends to be even more highly concentrated than ownership of real assets.Networth andfinancial wealth are both usually measured at the household level because many assets, such as the family home, tend to be jointly owned. Both income and wealth have important advantages, and although they are related, the correlation between the two is relatively low, suggesting that considering them separately tells only part of the financial story. Income is essential for paying for current needs and desires, and it can provide a degree of social and political power. Income becomes wealth only when it is saved, and the advantages of wealth ownership are even more far-reaching. The family home, for instance, has both current use value and investment value. Similarly, a business can provide current income and long-term investment advantages. Wealth can enhance educational attainment, occupational opportunities, political power, and social influence. It provides a buffer against income interruptions; medical emergencies; and other crises, such as accidents and natural disasters (Keister & Moller 2000, Shapiro 2004, Wolff 2002). Wealth can create more wealth when it is reinvested, and it can generate income in the form of interest or dividends. Perhaps most significantly, wealth can be passed to future generations to extend these benefits indefinitely. Total household income and total household net worth have been correlated at about 0.50 to 0.60 since 2001 (my estimate from the SCF), a pattern that reflects extremes and that underscores the importance of defining income and wealth clearly. At one extreme are households that have high income from current work but have relatively low savings and, as a result, low wealth; for instance, some top executives, surgeons, and professional athletes have high salaries but relatively low savings rates and thus low wealth. At the other extreme are households with high net worth but low income; for example, a person who inherited high wealth or a retiree who saved consistently over theworking yearsmayhavehigh levels of assets but low income from current work. The correlation is further complicated by the fact that those with high wealth are also likely to receive interest/dividend income, highlighting the importance of specifying income and wealth sources. DATA CHALLENGES AND STRATEGIES Studying top incomes and wealth creates unique data challenges that account for at least some of the gap in current knowledge about these households. By definition, the one percent is a small portion of the population, making it unlikely that they will be adequately represented in most social and economic surveys unless they are deliberately oversampled. Yet www.annualreviews.org • The One Percent 349 A nn u. R ev . S oc io l. 20 14 .4 0: 34 736 7. D ow nl oa de d fr om w w w .a nn ua lr ev ie w s. or g A cc es s pr ov id ed b y D uk e U ni ve rs ity o n 03 /0 8/ 17 . F or p er so na l u se o nl y. SO40CH16-Keister ARI 27 June 2014 12:52 because this group receives large proportions of total household income and owns large proportions of household wealth, including some assets that other households are unlikely to own (e.g., businesses, bonds), their absence from survey samples biases distributional estimates. In addition, high-income and highwealth households are more reluctant to report financial information than other households, or theymay provide inaccurate information in surveys because they do not know details about their income and wealth. These problems can lead to additional bias in survey data that can be resolved only through careful efforts to ensure reporting accuracy and sample retention. Longitudinal data can be useful for understanding income and wealth concentration, but collecting data on high-income and high-wealth households over time creates additional problems. For instance, sample attrition and declining sample representativeness over time are both inevitable in longitudinal data but disproportionately affect high-income/high-wealth households and lead to bias that compounds over time. Similarly, studying high income and wealth over long historic periods can be useful for understanding income and wealth concentration, but the need for data over long stretches of time creates its own problems. In particular, survey data on income and wealth did not exist prior to the 1950s for income and the 1980s for wealth, and the available data are often limited in detail and not comparable to each other over time (Atkinson et al. 2011; Moore et al. 2000; Piketty & Saez 2003, 2006). One exception was the Survey of the Financial Characteristics of Consumers (SFCC) from 1962, which includes household income and wealth data (Projector & Weiss 1966); however, the SFCC was conducted once and thus cannot be used to study patterns over time. Researchers use three data sources and strategies to deal with these challenges. First, the SCF, a unique data set that provides highly accurate information on cross sections of top income earners and wealth owners, has become standard in research on resource concentration. The SCF is a triennial survey of US households collected by the Board ofGovernors of the Federal Reserve System since 1983; the survey also includes two panels in 1986–1989 and 2007– 2009. The survey contains detailed information about household income, assets, debts, balance sheets, demographics, attitudes toward risk, relationships with financial institutions, and related information ( Johnson & Moore 2005). Particularly noteworthy is that the SCF uses a dual-frame sample design to adequately represent all households: (a) a multistage national area probability sample and (b) a sample of high-income households identified with Internal Revenue Service data ( Johnson & Moore 2005, Kennickell 2008). The unique sample design improves the likelihood that top incomes andwealth are represented and that bothwidely held assets (e.g., homes, cars) and those held more narrowly by wealthy households (e.g., businesses, bonds) are included ( Johnson & Moore 2005, Kennickell 2009a). Unique efforts to improve data accuracy and imputemissing values make the SCF particularly useful for studying households with high income and wealth (Kennickell 2008, 2011). Second, to study issues regarding changes in income and wealth over time, researchers have relied on several data sets that contain repeated cross sections or longitudinal data on the same households (Gouskova&Stafford 2009;Keister 2000a,b; Kopczuk et al. 2010). For example, the decennial US Census and the Current Population Survey (CPS) are useful for understanding trends in income ownership and concentration because of their large sample sizes and detailed content on household financial status (Burkhauser et al. 2012, Kopczuk et al. 2010). The Census is an important source of information on income sources for large samples at regular intervals. Similarly, the CPS, a monthly household survey conducted by the US Census Bureau for the Bureau of Labor Statistics, can be used effectively to understand changes in income over time. Because the CPS includes detaileddata on income, someassets, debts, demographic traits, labor force characteristics, and related issues, it is widely used to study labor earnings and income inequality. The National 350 Keister A nn u. R ev . S oc io l. 20 14 .4 0: 34 736 7. D ow nl oa de d fr om w w w .a nn ua lr ev ie w s. or g A cc es s pr ov id ed b y D uk e U ni ve rs ity o n 03 /0 8/ 17 . F or p er so na l u se o nl y. SO40CH16-Keister ARI 27 June 2014 12:52 Longitudinal Survey of Youth, the Panel Study of Income Dynamics, the Health and Retirement Survey, and the Survey of Income and Program Participation all include both income and household wealth information and are longitudinal. Each of these data sets includes very detailed information on the same households over long periods, allowing researchers to study lifetime incomes and wealth accumulation rates that are not possible with cross-sectional data sets. Unfortunately, these longitudinal data sets include only small numbers of high-income or high-wealth households, making it difficult to generalize about the one percent. Third, to study patterns in resource concentration over long historic periods, researchers have begun to use government tax records to estimate the holdings of the one percent by income and wealth (Atkinson & Piketty 2007, Atkinson et al. 2011, Piketty 2005, Piketty & Saez 2006). The US government has collected and published income tax data since 1913, when they established a progressive income tax system, and these data tend to be relatively similar over time (Atkinson et al. 2011, Piketty 2005). Income tax records allow researchers to estimate both total household income and income sources (e.g., wage/salary income, capital gains income, business income) for top earners and to examine how these have changed over time in response to business cycles, social change, and government policies. Because other Western (non-US) governments have also collected and published income tax data since the early twentieth century, similar estimates can be generated for other countries, allowing comparisons of patterns cross-nationally to study the effects of structural variation and public policy on income and wealth concentration (Atkinson et al. 2011,Piketty 2005).Estate tax data are also available for the United States, and researchers are effectively using these to estimate wealth holdings over long periods for top wealth holders (Kopczuk & Saez 2004). These data have their own limitations: They include only top incomes and wealth, the unit of observation is the individual rather than the household, and they might be biased because of tax evasion. However, researchers have used them to study long-term patterns in a way that is not possible with survey data. In the remainder of this article, I provide basic empirical estimates of income and wealth concentration to summarize current knowledge and highlight gaps in the literature that future research might address. I use the SCF to produce these estimates because I want to accurately represent top income earners and wealth holders and because I focus on contemporary, cross-sectional patterns rather than longitudinal or historic trends. I report values for 2001, 2004, 2007, and 2010, the most recent years for which SCF data are available. I use median values because income, net worth, and financial assets are highly skewed; I show values for the top one percent, the next nine percent, and the remaining 90% to highlight the unique characteristics of those in the top of the income and wealth distributions. The SCF contains five imputed cases for each observed household as part of its effort to accurately represent income and wealth (Kennickell 2008, 2011). Consistent with the strategy used by other researchers (e.g., Kennickell 2003, 2008, 2009b), I use the five imputations as independent observations to take into account uncertainties in imputation, and I use sample weights to correct for oversampling. When I report demographic traits, individual characteristics (e.g., gender, race/ethnicity) refer to the respondent. I use the consumer price index (CPI-U) to deflate income and wealth values; I report all values in 2010 dollars. I define other terms in the text below. CONTEMPORARY INCOME AND WEALTH CONCENTRATION Figures 1 and 2 use the SCF to summarize what we know about income and wealth concentration. Figure 1 shows the percentage of total income going to various segments of the distribution, including income from earnings, investments, transfer payments, and other sources. Figure 2 shows median household income; the thresholds defining the top of the www.annualreviews.org • The One Percent 351 A nn u. R ev . S oc io l. 20 14 .4 0: 34 736 7. D ow nl oa de d fr om w w w .a nn ua lr ev ie w s. or g A cc es s pr ov id ed b y D uk e U ni ve rs ity o n 03 /0 8/ 17 . F or p er so na l u se o nl y. SO40CH16-Keister ARI 27 June 2014 12:52 30.4% 30.6% 28.6% 25.6% 37.4% 36.2% 37.8% 40.3% 32.3% 33.2% 33.6% 34.1% 54.9% 57.4% 52.9% 55.6% 25.4% 25.6% 25.8% 27.2% 19.7% 17.0% 21.3% 17.2% 2001 2004 0 20 40 60 80 100