Long term trends in earnings inequality: what the CPS can tell us
Introduction
The Census Bureau's mission is to provide policymakers and the research community with consistent information on important policy success indicators, like earnings inequality. The Current Population Survey's Annual March Demographic Supplement (CPS) is the most frequently used source of annual information on earnings and income of American households. The Census Bureau calculates official levels and trends in earnings and income inequality using these data (see Jones and Weinberg, 2000). In addition, the Census Bureau disseminates CPS data files for public-use, in which earnings and income values are topcoded. Fig. 1 reports Census Bureau derived Gini coefficients for the earnings of full-time year-round workers based on internal CPS data files, as well as our Gini coefficients based on the topcoded, public-use data files between 1975 and 2001 (survey years 1976 and 2002).
Section snippets
The problem
Improvements in CPS data collection methods and changes in the richness of the public-use data, limit the ability of researchers within and outside the Census Bureau to consistently estimate trends in earnings inequality using these data.
Over its history, the CPS has inconsistently collected and reported the total earnings of workers. For instance, between 1979 and 1984 the maximum amount that could be recorded in the internal data files for earnings in the longest job was $99,999. Beginning in
The solution
We use the public-use CPS data to derive time-consistent data for 1975–2001. We calculate the percentage of individuals subject to topcoding in every year. We determine the year in which the greatest percentage of the population is affected by the topcode and then topcode all years to yield this percentage. This procedure ensures that the same percentage of the upper tail distribution is affected in each year. In doing so we adjust the topcodes used for each of the subcomponents of earnings in
Results
Our time-consistent topcode affects 1.3% of wage and salary earners, 3.6% of the self-employed and 2.0% of farmers in each year. (Since the effective topcodes have been less restrictive in the internal CPS data, our procedure would affect an even smaller share of the population.) Fig. 1 shows Gini coefficients derived from our time-consistent public-use data. In this series, there is no spike in earnings inequality in either 1993 or 1995.
We use regression analysis to test whether our
Conclusions
Researchers must be cautious when making inferences about trends in earnings inequality derived from public-use CPS data. The same caution holds for inferences about income inequality trends, since earnings comprise the largest component of income. Important changes in the effective topcodes in both the internal and public-use CPS data series have resulted in yearly data that are not comparable over the 1990s. We impose a time-consistent topcoding rule on the public-use data and find trends in
Acknowledgements
Partial funding for the work reported in this paper came from the United States Department of Education, National Institute of Disability and Rehabilitation Research, cooperative agreement No. 1331390038. This paper does not necessarily reflect the views of the National Institute of Disability and Rehabilitation Research. We thank Mary Daly for comments on this work.
References (5)
- Burkhauser, R.V., Butler, J.S., Feng, S., Houtenville, A.J. (2003). Data Appendix: Long Term Trends in Earnings...
- et al.
The changing shape of the nation's income distribution
Current Population Reports, June
(2000)
Cited by (29)
The effect of social security, health, demography and technology on retirement
2013, Review of Economic DynamicsEarnings volatility in America: Evidence from matched CPS
2011, Labour EconomicsCitation Excerpt :These observations are dropped prior to matching. Second, in a series of papers, Richard Burkhauser and co-authors (Burkhauser et al., 2004, 2007; Larrimore et al., 2008) have raised concerns about trends in income inequality because of changes in the way the Census top-codes income data for public release. Prior to 1995 the Census assigned top-coded data a common value (though this value varied across income sources, and at times, years), but starting in 1995 they assigned top-coded data the mean values of actual income based on broad demographic groupings (age, race, gender, education).
Effects of federal programs on children: Absolute poverty, relative poverty, and income inequality
2011, Children and Youth Services ReviewCitation Excerpt :The second limitation is that total personal earnings were top-coded at $200,000 in the public CPS files due to confidentiality issues (U.S. Department of Labor, 2006). This income measure may affect the results on analyses of income distribution and inequality although its impact will be limited when obtaining changes in the levels of inequality (Burkhauser, Butler, Feng, & Houtenville, 2004). The study sample includes a total of 125,956 children under 18 years old who were children, grandchildren, relatives, or unrelated persons (e.g., foster children) of adult respondents of the ASEC supplement survey (47,109 children for 1995 and 78,847 children for 2007).
Earnings Inequality and Market Work in Husband-Wife Families
2007, Research in Labor EconomicsSkills, Computerization and Income Inequality in the Postwar U.S. Economy
2006, Research on Economic InequalityThe atlas of inequality aversion: theory and empirical evidence on 55 countries from the Luxembourg Income Study database
2022, Equilibrium. Quarterly Journal of Economics and Economic Policy