sign in sign up
<<
Home , Income distribution

ECONOMICS

Measuring Trends in Inequality

By Michael T. Owyang and Hannah G. Shell

©THINKSTOCK/ISTOCK

central issue in concerns how which split the into five even the primary source for census data on Aoutput (equivalent to income) is distrib- buckets (the bottom quintile is the 20th per- income and poverty. The CPS data are uted across economic agents (e.g., workers, centile); quintiles are commonly used percen- reported in money income—market income entrepreneurs). A first step in addressing tiles for studying inequality except at the top plus other cash income, excluding noncash this issue is understanding how output (or of the income distribution, where the income benefits, such as employer-provided health income) is distributed in the United States difference between 98th and 99th percentiles insurance. While the CPS provides quality and understanding how the distribution has is large. To summarize inequality across the low- and middle-income data, changed over time. entire distribution, economists use the Gini above a certain threshold are not reported to Measuring income inequality, however, coefficient. The measures protect individual privacy. This makes it less is not a trivial endeavor. Multiple sources income concentration at each percentile of ideal for high-income estimates. of income—salary, gains income, the population and ranges from 0 (perfectly The Congressional Budget Office (CBO) employer-provided health insurance and equal) to 1 (perfectly unequal). also constructed a dataset that merges the other non-salaried compensation, etc.—make In order to study income inequality, one CPS and SOI and draws on each source’s simply measuring income itself problematic. needs income at an individual level. While strengths—the CPS for low income and the Nonetheless, using a number of different is the usual aggregate SOI for high income. The CBO reports mar- definitions of income and employing various indicator for income, there are many defini- ket income, both before-tax (market income metrics, researchers have attempted to quan- tions of income and many data sources avail- plus government transfers) and after-tax tify income inequality in the U.S. able at the individual level. Economists often income (before-tax income less federal taxes). Economists have identified two broad peri- use the Internal Revenue Service’s Statistics Most studies find that more equality is seen ods in income inequality over the post-World of Income program (SOI) or the Census in after-tax income, followed by before-tax War II period—first in the 1970s and then, Bureau’s Current Population Survey (CPS). income and then market income.4 Moreover, more recently, prior to the Great Recession. Studies using different data sources reach it is generally accepted that the U.S. economy In the sections that follow, we describe how various conclusions on income inequality, is similar to other developed nations’ in income inequality is measured and then how depending on the definition used for income. terms of pretax and transfer income inequal- it changed over these two periods. For example, economists ity. In other words, U.S. income inequality is and compiled a dataset using not intrinsically different from what is seen Income Inequality SOI data back to 1913. They focused on the in other countries, and any differences are and How It’s Measured share of income earned by the top percen- mainly driven by the lack of income- Assessing income inequality boils down in tiles to avoid poor data quality in the lower redistributing fiscal policies in the U.S. effect to measuring the income gaps between percentiles.1 The SOI definition of income is high and low earners. Income inequality implies market income, the cash income reported Trends in Income Inequality that the lower-income population receives on tax forms.2 The SOI data more accurately From the end of World War II to the early disproportionately less income than the higher- measure the top of the income distribution, 1970s, income inequality in the U.S. was rela- income population: The larger the disparity, the but less accurately measure low-income tively low. The graph shows that from 1947 to greater the degree of income inequality. statistics because low-income households are 1970, the Gini coefficient was flat or declin- To measure inequality, economists often not always required to file income taxes.3 ing.5 Piketty and Saez, using SOI data with a sort the population by income percentiles and Another source of individual income longer history, found that income inequality measure the difference across these percen- data is the CPS. Every March, the CPS—a peaked in the 1920s, then decreased after the tiles. For example, the top 10 percent of earn- monthly survey of 75,000 households—pro- Great Depression, when top capital incomes ers would be the 90th percentile. A related vides the information used in the Annual fell and were unable to recover. Although way of dividing the population is quintiles, Social and Economic Supplement, which is the U.S. economy rebounded during World

4 The Regional Economist | April 2016 War II, wage controls prevented growth in period from 2002 through 2007, for example, ENDNOTES top incomes. Once the war ended, a progres- the top 1 percent captured about two-thirds 1 Piketty and Saez also estimate the portion of lower sive tax structure and reforms such as Social of overall income growth, Piketty and Saez income tax units that are excluded in the SOI data Security and unionization kept low- and estimated. Further, even though top incomes and add these estimated values into their measure of total income. middle-income growth strong. fell 36.3 percent in the 2007-09 recession, 2 Market income consists of before-tax income from Starting in the 1970s, wage growth at the the incomes of the bottom 99 percent also wages and salaries; profits from businesses; capital income, such as dividends, interest and rents; real- top of the income distribution outpaced the decreased 11.6 percent. This decrease is the ized capital gains; and income from past services. rest of the distribution, and inequality began largest two-year fall in the incomes of the bot- Other forms of income include cash and in-kind to rise. The Gini coefficient grew from 0.394 in tom 99 percent since the Great Depression. payments from programs like Social Security, food stamps and private benefits (e.g., health insurance). 1970 to 0.482 in 2013. The CBO estimates that So far, the top 1 percent has captured 3 The SOI data also exclude noncash benefits like between 1979 and 2011 market income grew 58 percent of income gains from 2009 to health insurance, which are a growing portion of middle-class income. 56 percent in the 81st through 99th percen- 2014. The newest data on income show that 4 The differences in inequality by income concept tiles and 174 percent in the 99th percentile. growth from 2013 to 2014 was more equal. are largely due to a progressive tax structure and In contrast, market income growth averaged The incomes of the bottom 99 percent grew social safety nets, such as food stamps, that benefit individuals at the bottom of the distribution. 16 percent in the bottom four quintiles. 3.3 percent, the best rate in more than 5 Family income is defined as that of two or more Government transfers and federal taxes 10 years, and the Gini coefficient on house- related persons living in a household. It may did have a redistributive effect during this hold income decreased slightly, marking the exclude single-person households and households with multiple residents who are all not related. period, but income inequality in after- first nonrecession decrease since 1998. Family income is available in the CPS from 1947 tax income grew substantially. The 1970s to 2011, while household income was not collected until 1967. increase in inequality was different from the Conclusion increase during the 1920s. During the period Economists use Gini coefficients, percen- REFERENCES from 1940 to 1970, top-income composition tiles and detailed survey data to study trends DeNavas-Walt, Carmen; and Proctor, Bernadette D. shifted from capital income to wage income. in income inequality. They find that inequality “Income and Poverty in the United States: 2014.” In the top 0.01 percent, the total income share has been rising in the U.S. since World War Current Population Reports. September 2015. See www.census.gov/content/dam/Census/library/ from capital income fell from 70 percent in II, reaching its highest level in 2013 since the publications/2015/demo/p60-252.pdf. 1929 to just above 20 percent in 1998. Wage 1920s. This result is robust for the definition of “The Distribution of Household Income and Federal income rose over the same period, from income and the chosen measure of inequality. Taxes, 2011.” Congress of the United States: Congressional Budget Office. November 2014. 10 percent to about 45 percent. High growth Understanding the facts about inequal- See www.cbo.gov/sites/default/files/113th-con- in top wages is partly explained by the Tax ity is the first step in assessing what can and gress-2013-2014/reports/49440-Distribution-of- Income-and-Taxes.pdf. Reform Act of 1986, which lowered the top should be done. While there is a general Piketty, Thomas; and Saez, Emmanuel. “Income marginal-income tax rates. The short-term consensus that some reallocative transfers Inequality in the United States, 1913-1998.” The impact of tax reform is circled in red on the from the top of the income distribution to the Quarterly Journal of Economics, Vol. 118, No. 1, 2003, pp. 1-39. See http://eml.berkeley.edu/~saez/ graph. Longer-lasting wage growth came bottom are desirable, the optimal amount of pikettyqje.pdf. from the reporting of stock options and other these redistributions is still up in the air. Saez, Emmanuel. “Striking It Richer: The Evolution forms of income as wages on tax returns. of Top Incomes in the United States,” updated with 2014 preliminary estimates. University of After the increase in the 1970s, inequality California, Berkeley. June 2015. See http://eml. continued to rise. In the 2001 and 2007-09 Michael T. Owyang is an economist, and Han- berkeley.edu/~saez/saez-UStopincomes-2014.pdf. Stone, Chad; Trisi, Danilo; Sherman, Arloc; and recessions, top incomes fell sharply as stock nah G. Shell is a senior research associate, both at the Federal Reserve Bank of St. Louis. For DeBot, Brandon. “A Guide to Statistics on His- market crashes decreased the value of capital torical Trends in Income Inequality.” Center on more on Owyang’s work, see https://research. gains and stock options. However, losses to top Budget and Policy Priorities. October 2015. stlouisfed.org/econ/owyang. See www.cbpp.org/sites/default/files/atoms/ incomes were temporary. During the recovery files/11-28-11pov_0.pdf.

Gini Coefficient for Family and Household Income 0.50 SOURCES: Gini coefficients calculated by the Bureau of Labor 0.48 Statistics using Current Population Survey data, accessed via 0.46 Haver Analytics. 0.44 0.42 NOTE: The figure to the left shows Gini coefficients calculated 0.40 from Current Population Survey data for family and household 0.38 income. Only family income is available from 1947 to 1967, but 0.36 this measure is less ideal than household income because the Gini Coefficient, Family Income 0.34 census defines a family as two or more related individuals living Gini Coefficient, Household Income 0.32 in the same house. Roommates or single-person households are 0.30 excluded. The red circles mark the temporary increase in income inequality from the Tax Reform Act of 1986, which lowered the

1947 1949 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 top marginal tax rate. Gray bars indicate recessions.

The Regional Economist | www.stlouisfed.org 5