Population

PopulationData on the civilian population in the "Population" section are compiled from EPI analysis of basic monthly Current Population Survey microdata. Where available, monthly data reflect 12-month moving averages as of the latest month of data.The economically insecure population is defined as those ages 16-64, not enrolled in any type of education, and have income less than 200% of the federal poverty rate. The income cutoffs are calculated using family income values which are compiled from EPI analysis of basic monthly Current Population Survey microdata and IPUMS ASEC microdata.EmploymentData on employment, labor force participation, long-term unemployment, part-time employment, and unemployment in the "Employment" section are compiled from EPI analysis of basic monthly Current Population Survey microdata. Data reflect 12-month moving averages as of the latest month of data.Data on annual hours and weeks worked in the "Employment" section are compiled from EPI analysis of Current Population Survey Annual Social and Economic Supplement microdata.Data on employment by industry in the "Employment" section are compiled from the BLS Current Employment Statistics (CES) for both the national and state levels. National data comes from the CES national database and state data comes from the state and metro area employment, hours, and earnings database.PovertySelf-employed workers are excluded from the sample, with the exception of data for years 1967 and 1973 which are derived from Murphy and Welch (1989) and include the self-employed.Data on official poverty and supplement poverty measure in the "Poverty and annual incomes" section are compiled from EPI analysis of Current Population Survey Annual Social and Economic Supplement microdata. The U.S. Census Bureau releases two poverty measures each September. The first, called the official poverty measure, is based on cash resources. The second, the Supplemental Poverty Measure (SPM), includes both cash and noncash benefits and subtracts necessary expenses (such as taxes and medical expenses). Both measures are adjusted by family size, composition, and age of householder while the supplemental measure also contains a geographic adjustment for differences in housing costs.PricesThe chained Consumer Price Index (C-CPI) from the Bureau of Labor Statistics (BLS) utilizes varying expenditure weights each month. These varying weights are derived using a formula which accounts for how consumers change their spending behavior in response to price changes and is ultimately a closer approximation to a cost-of-living index. The extended C-CPI-U series uses the C-CPI-U since 2000 and extends it retroactively by merging it to the CPI-U-RS (1978-1999), CPI-U-X1 (1967-1977), and CPI-U (1966 and before). The personal consumption expenditures (PCE) index from the Bureau of Economic Analysis (BEA) reflects changes in the prices of goods and services purchased by consumers in the U.S. All of the CPI inflation series used in the SWA data library can be found in the realtalk R package.ProductivityProductivity indexes and levels measure how much income is produced in an hour of work, and the pay indexes and levels measure average hourly compensation (wages and benefits) for production and nonsupervisory workers. We use four data sources to calculate productivity and pay:

Extended Chained CPI-U price seriesThe Productivity–Pay Gap tracker provides a detailed explanation of the underlying methodology and a comparison of productivity and pay trends over time.Productivity is net labor productivity: total GDP net of depreciation from the BEA NIPA Tables divided by economy-wide hours worked from the BLS OPT data. Pay is average hourly wages for production and nonsupervisory workers from the BLS CES multiplied by the BEA NIPA economy-wide ratio of total compensation to wages and salaries. For real levels and indexes, both pay and productivity data are deflated by the same Extended Chained CPI-U price series.The code that generates the productivity and pay data is available here.UnionsData on union membership between 1917 and 1982 are from Freeman (1997). Data since 1983 are from Bureau of Labor Statistics CPS data on union affiliation (BLS CPS). Data prior to 1983 report union density as a share of non-agricultural employment while more recent data report union density as a share of overall employment to remain consistent with government reporting. The small difference based on alternative methods has no bearing on the economic findings shown.The union wage premium is based on average wages and controls for gender, race and ethnicity, citizenship, education, marital status, age, industry, occupation, class of worker, and state. The log of the hourly wage is the dependent variable.Wage disparitiesThe regression-based gender and racial/ethnic wage gaps are based on average wages and controls for gender, race and ethnicity, education, age, and includes marital status and state fixed effects. The log of the hourly wage is the dependent variable.The college wage premium is based on average wages and controls for gender, race and ethnicity, age, and includes marital status and state fixed effects. The log of the hourly wage is the dependent variable.The teacher pay gap comes from Allegretto (2024). The gap is based on average weekly wages and controls for public school teacher, private school teacher, gender, and married, along with indicator sets on education (M.A., professional degree, Ph.D.) and race/ethnicity (Black, Hispanic, other); also included are age as a quartic and state fixed effects. The log of the weekly wage is the dependent variable.The CEO-to-worker pay ratio uses data from S&P Compustat ExecuComp database and the Wall Street Journal. More information about the CEO-to-worker pay ratio can be found in Bivens, Gould, and Kandra (2024).WagesData on hourly wage levels, percentile ratios, and percentiles from the "Wages" section are compiled from EPI analysis of Current Population Survey Outgoing Rotation Group (CPS ORG) microdata. Wages from the CPS ORG are the hourly wages of a population subsample ages 16 and older that includes all wage and salary workers (not self-employed) with valid wage and hour data, whether paid weekly or by the hour.For those who met these criteria, an hourly wage was calculated in the following manner: If a valid hourly wage was reported, that wage was used throughout our analysis. For salaried workers (those who report only a weekly wage), the hourly wage was their weekly wage divided by their hours worked. For more information on our hourly wage data, including top-code adjustments, imputations, outliers, see https://microdata.epi.org/methodology/wagevariables/. Real wages are inflation-adjusted using the Chained Consumer Price Index.The construction of hourly wage percentiles is often complicated by the fact that reported wages in the data tend to "clump," particularly around round numbers (e.g. $20 an hour). This becomes a problem if say the 50th percentile wage—or the wage at the very middle of the wage distribution—happens to fall right into such a clump (you can find that more than half of workers are earning the 50th percentile wage or less). Alternatively, it can become a problem when a percentile (say the 50th) lands right on a large clump of workers, which mechanically can cause that percentile to no growth for a couple of years.For these reasons, we enforce some smoothness in the wage data using the averaged percentile smoothing method recommended by Tedeschi (2024). A given percentile is calculated using a weighted average of 9 neighboring percentiles in the distribution. For example, the median wage (50th percentile) is calculated using the weighted average of the 46th through 54th percentiles with weights 1/25, 2/25, 3/25, 4/25, 5/25, 4/25, 3/25, 2/25, 1/25, respectively.Data on annual wages for the top 1% and bottom 90% are taken from Social Security Administration (SSA) wage statistics (1991 - present) and Kopczuk, Saez, and Song (2007), which includes all workers based on reports by employer on Forms W-2. For data years 1991 forward, we use the SSA earnings data provided in wage groups to build each desired wage percentile bracket. We linearly interpolate within wage groups as required to estimate the bottom 90%, next 5%, and so on to achieve the share of wages in each wage percentile bracket. To extend this series prior to the SSA reports in 1991, we use the wage shares provided in Kopczuk, Saez, and Song (2007). We apply the calculated shares of wages in each percentile group to total wages and employment from SSA Table 4.B2 (2022) to obtain total and average wages for each group in 2022 dollars. For a more detailed explanation, see the methodology from Gould and Kandra (2022).Data on payroll earnings by industry in the "Wages" section are compiled from EPI analysis of the Bureau of Labor Statistics Current Employment Statistics (CES). State-level data comes from the CES state and metro area employment, hours, and earnings page.IncomeMedian household incomes are compiled from published Census Bureau American Community Survey (ACS) tables. The tables are as follows: B19013 (overall), B19013H (white only), B19013B (Black only), and B19013I (Hispanic only). The 2020 (ACS) data collection process was severely disrupted by the COVID-19 pandemic, so data is not available for that year.