October 2016

How should we define “low-wage” work? An analysis using the Current Population Survey

Low-wage work is a central concept in considerable research, yet it lacks an agreed-upon definition. Using data from the Current Population Survey’s Annual Social and Economic Supplement, the analysis presented in this article suggests that defining low-wage work on the basis of alternative hourly wage cutoffs changes the size of the low-wage population, but does not noticeably alter time trends in the rate of change. The analysis also indicates that different definitions capture groups of workers with substantively different demographic, social, and economic characteristics. Although the individuals in any of the categories examined might reasonably be considered low-wage workers, a single definition obscures these distinctions.

Low-wage labor is a central topic in policy, scholarly, and popular discussions concerning the economic well-being of low-income Americans. Prominent U.S. social welfare programs, such as the Temporary Assistance for Needy Families (TANF) program (popularly termed “welfare”) and the Earned Income Tax Credit program, are designed to facilitate and incentivize labor force participation among the economically disadvantaged, linking antipoverty policy to the low-wage labor market. Municipalities such as Seattle and Chicago are experimenting with higher minimum wages in an effort to promote the well-being of low-wage workers. Low-wage labor has also moved to the forefront of public attention as a result of the 2014 fast-food worker movement, which sought to improve wages and working conditions for food service workers.1 Although the debates and developments surrounding low-wage work highlight the importance of understanding its scope and identifying the characteristics of, and the economic conditions experienced by, low-wage workers, no consensus exists regarding what constitutes low-wage work. How should low-wage work be defined? What is the incidence of low-wage work under different definitions, and do different definitions capture distinct pools of workers?

In this article, we examine three different hourly wage cutoffs for defining low-wage work. These cutoffs are based on what it would take a full-time worker to earn an income lifting his or her family above the federal poverty threshold for (1) a family of two; (2) a family of three; and (3) a family of three, with the cutoff set at 125 percent of the poverty threshold. We first estimate the proportion of workers, among the hourly workforce, considered low-wage workers by each definition, both cross-sectionally for the year 2013 and over time for the period 1990–2013. Next, to understand changes in the composition of the population of low-wage workers under different definitions, we use the wage cutoffs to “bracket” discrete groups of workers in 2013 and to describe and compare their characteristics. Overall, we find that raising the wage threshold increases the population of individuals considered low-wage workers, but does not create markedly different trends over time. In addition, altering the definition adds qualitatively different workers to the population. Workers in the discrete wage groups differ in terms of demographic characteristics, human capital factors (e.g., education), health insurance coverage, economic hardship, and use of public economic supports.


Research suggests that, since the 1970s, low-wage jobs have undergone marked growth as a proportion of the U.S. labor market.2 These jobs include food service; housekeeping; low-level healthcare positions, such as nursing assistants; and low-level retail positions, such as cashiers. Women tend to be disproportionately represented in these occupations.3 Men are also found in low-wage jobs, and declines in unionization and the associated loss of the union wage premium in recent decades have particularly affected less educated men.4

For some workers, a low-wage job is merely a transitional phase on the path to higher earnings and better working conditions. For others, it is more permanent and reflects fewer opportunities for upward mobility.5 Race, gender, and education are related to wage mobility within the low-wage labor market, with African Americans, women, and those with limited education having a lower chance of advancement.6 At least part of the lower advancement opportunity for low-wage African American and Latino workers is labor market discrimination, which limits the jobs available to a given candidate.7 Some workers in low-wage jobs are economically vulnerable not just because of limited income, but also because of employer business practices. Part-time and unpredictable schedules are common in the low-wage labor market. In the hotel industry, for example, employers frequently use flexible scheduling based on occupancy rates,8 and this practice could lead to inconsistent income. Low-end earners are also less likely to have employer-provided benefits such as health insurance.9

Scholars do not share the same definition of low-wage work. Sheldon Danziger and David Ratner, for example, set a threshold of $9 per hour (in 2007 dollars) to identify low-wage workers.10 In the same vein, Gerhard Bosch uses the Organisation for Economic Co-operation and Development cutoff of two-thirds of the median hourly wage.11 Colin Campbell sets the line at $12 per hour (in 2008 dollars).12 Each of these thresholds may capture subgroups of workers that differ across a number of characteristics, and treating low-wage work as a binary variable—one which indicates whether or not a worker is in a low-wage job—may mask these distinctions. Workers and jobs closer to the upper cutoff could differ in important ways from workers and jobs closer to the minimum wage. The following analysis evaluates the implications of varying the definition of low-wage work.

Data and methods

The data for this article are drawn from the Annual Social and Economic Supplement to the Current Population Survey (CPS) and the CPS Earner Study.13 The CPS, conducted by the U.S. Census Bureau for the U.S. Bureau of Labor Statistics, is a key source of economic and labor force statistics, such as unemployment rates, in the United States. The core survey instrument is administered monthly and collects a range of demographic and labor force data. Supplemental modules periodically gather additional data on specific topics. The Annual Social and Economic Supplement covers areas such as family characteristics, health insurance, sources of income, and participation in public support programs, while the Earner Study documents hourly wage rates, hours worked per week, and union membership.14 The CPS uses a probability sample of 60,000 U.S. households in any given panel, although the survey’s administration is staggered such that the exact sample changes monthly. Typically, one respondent reports for all members in a household. Using weighting to adjust for sampling procedures, the CPS provides a representative sample of the U.S. civilian noninstitutional population.15

For this study, the unit of analysis was the individual. The sample was restricted to people ages 18 to 64 who reported being currently employed. This subsample allowed us to estimate the proportion of currently employed workers falling into various low-wage categories. After defining three such categories, we estimated the prevalence of low-wage work under each category for 2013, the ending year of the study period. Trends over time were then examined with the use of CPS samples from 1990 to 2013. The social and economic characteristics of hourly workers within each category were studied cross-sectionally for the year 2013. Because all data were categorical, F-tests were conducted to determine the probability that the observed differences were due to chance. F-tests are based on χ2 statistics adjusted to account for weighting procedures. These tests are only descriptive and do not imply a causal relationship between worker characteristics and wages.

Some factors examined in the descriptive analysis, such as poverty status and participation in the Supplemental Nutrition Assistance Program (SNAP; formerly known as the Food Stamp Program), are household-level variables in the CPS. Since the unit of analysis is the individual worker, these household characteristics are treated as individual characteristics (e.g., to capture whether or not an individual in a household is receiving SNAP assistance). Household characteristics, such as the number of workers in a household and their combined earnings, could influence individual behavior in the labor market. However, because the purpose of this analysis is to describe low-wage workers under different definitions of low-wage work, such relationships are not examined.

Definitions of low-wage work

To define low-wage work, we use three different hourly wage cutoffs based on the federal poverty guidelines: a wage lifting a family of two (one adult and one child) above the official poverty threshold, a wage lifting a family of three (one adult and two children) above the threshold, and a wage bringing a family of three to 125 percent of the threshold.16 For the year 2013, these cutoffs are associated with wages of $9.25 per hour, $10.75 per hour, and $13.50 per hour. These wage levels are the basis for determining, both cross-sectionally and longitudinally, the prevalence of low-wage work within the larger workforce, defined here as all wage and salary workers. The cutoffs are also used to define three mutually exclusive categories of low-wage workers: those paid $9.25 per hour or less; those paid $9.26 per hour, up to $10.75 per hour; and those paid $10.76 per hour, up to $13.50 per hour (inclusive). The discrete categories allow us to identify how the characteristics of low-wage workers change under each possible cutoff.

Because our approach is based on the federal poverty guidelines, all cutoffs use an absolute definition of low-wage work. Although the cutoffs change over time as the poverty thresholds are adjusted, they are consistently anchored to the official U.S. definition of poverty in any given year. A plausible alternative approach would be to adopt a relative definition of low-wage work, using a measure based on hourly wage quantiles. For our analysis, however, an absolute definition provides several advantages over a relative definition. First, the poverty guidelines—the official indicators of economic hardship in the United States—are themselves based on an absolute cutoff calculated with the use of the cost of an emergency food budget. This cutoff, in turn, is used to determine eligibility for public programs such as SNAP. In addition, policies related to low-wage labor provisions, such as minimum wages and refundable tax credits, tend to use absolute dollar figures rather than relative measures. Finally, because wage quantiles are constant by construction, using them makes it difficult to examine the prevalence of low-wage labor and its change, if any, over time. The lowest quartile, for example, is always 25 percent of the population.

Using the U.S. poverty guidelines to set the wage cutoffs does not imply that all low-wage workers are officially living in poverty. Determining which individual workers are in such status requires examination of the household overall, in terms of both income and composition. A low-wage worker may, for instance, be a secondary earner in a household in which the primary earner’s income brings the household above the poverty threshold. The poverty guidelines, as used here, provide a convenient means of subdividing low-wage workers by wage level.

Prevalence of low-wage work

Using each of the previously defined thresholds as the top cutoff for defining low-wage work produces substantively different estimates of the incidence of such work within the larger workforce. An estimation based on the lowest cutoff in 2013 suggests that only 12.63 percent of American workers (approximately 15 million) are low-wage workers. (See table 1.) Including the middle and highest wage brackets, however, brings the cumulative percentage of low-wage workers to 29.54 percent of the workforce (approximately 36 million workers). Depending on the cutoff used, anywhere from 1 in 10 to 3 in 10 American workers are in a low-wage job.

Table 1. Weighted percentage and population count of low-wage workers among all wage and salary workers ages 18 to 64, 2013
StatisticWage bracket
Up to $9.25 per hour$9.26 to $10.75 per hour$10.76 to $13.50 per hourUp to $13.50 per hour

Weighted percentage (observed count)

12.63 (1,455)7.11 (859)9.79 (1,242)29.54 (3,556)

Estimated population count


Source: U.S. Census Bureau, Current Population Survey, Annual Social and Economic Supplement and Earner Study.

The overall patterns revealed in table 1 persist over the 1990–2013 period. (See figure 1.) Elevating the wage cutoff for defining low-wage work consistently increases the proportion of workers falling into the low-wage population, and the magnitude of the shift is relatively constant. On average, in any given year, using the middle rather than the lowest cutoff increases the population of low-wage workers within the larger workforce by approximately 7 percentage points (the increase ranges from 5 to 9 percentage points). Switching from the middle to the highest cutoff further increases the population by approximately 10 percentage points (the increase ranges from 9 to 12 percentage points). Although the estimated size of the population changes under different definitions of low-wage work, the stability in the magnitude of the change suggests that trends over time are similar regardless of the cutoff used. Under all three definitions, the proportion of low-wage workers in the workforce decreases slightly but noticeably in the mid-to-late 1990s, generally levels off through the 2000s, and then increases slightly in the 2010s. Although these trends exhibit small periodic discrepancies across definitions (e.g., the spike in low-wage work in 2008 is more evident for the middle cutoff), they generally parallel one another.

Descriptive characteristics

Table 2 presents the demographic, educational, household-composition, and employment characteristics of low-wage workers in 2013, both for each of the three cutoff categories and overall. The data highlight some key differences among workers at each wage level and, in turn, changes in the composition of the low-wage workforce under different definitions. Although a majority of workers in the middle and highest wage categories are White non-Hispanic, a slight majority (50.82 percent) of workers in the lowest category are non-White. Individuals identifying themselves as Hispanic are found with the greatest relative frequency (28.46 percent) in the lowest wage bracket, whereas individuals identifying themselves as Black are roughly equally found in the lowest (14.63 percent) and middle (15.59 percent) brackets. The presence of both groups is markedly lower in the highest wage category than in the lowest category. No statistically significant difference by gender was found across categories, but women are a majority in all three brackets. In contrast, the distribution by age differs across the wage categories, with the lowest category clearly dominated by younger workers and the middle and highest categories represented by workers more evenly distributed across the age range.

Table 2. Characteristics of low-wage workers, 2013 (percent)
CharacteristicWage bracket
Up to $9.25 per hour$9.26 to $10.75 per hour$10.75 to $13.50 per hourUp to $13.50 per hour

Race or ethnicity


White non-Hispanic


Black non-Hispanic


Other non-Hispanic




F(5.95, 21,141.40) = 4.97(1)







F(2.00, 7,106.08) = 0.07














F(9.94, 35,345.18) = 16.45(1)



Less than high school


High school diploma or equivalent


Some college or 2-year degree


Bachelor’s degree or higher


F(5.99, 21,308.85) = 19.18(2)

Currently a high school or college student


F(2.00, 7,109.48) = 41.03(1)

Household composition


In a household with children


F(2.00, 7,106.72) = 0.54

Job characteristics


Part time (< 35 hours per week)


F(2.00, 7,104.91) = 114.45(1)

Involuntary part time


F(2.00, 7,104.51) = 41.61(1)

Union member or covered by union


F(2.00, 7,105.08) = 23.31(1)


(1) p < 0.001

(2) p < 0.01

Source: U.S. Census Bureau, Current Population Survey, Annual Social and Economic Supplement and Earner Study.

Contrasts among workers at different wage levels also emerge by human-capital factor and job characteristic. Educational attainment shifts higher as the wage cutoff for defining low-wage work increases. In particular, the proportion of workers with less than a high school diploma declines with each successive category, whereas the proportion of workers with a bachelor’s degree or higher increases. Slightly less than one-quarter (22.45 percent) of workers in the lowest category do not possess a high school diploma or equivalent, compared with 11.61 percent of workers in the highest category. Conversely, only 6.52 percent of workers in the lowest category possess a bachelor’s degree or higher, compared with 17.80 percent in the highest category. The proportion of workers who are high school or college students declines in each category, from slightly less than one-quarter (22.53 percent) of workers in the lowest wage bracket to only 8.74 percent of workers in the highest.

Strikingly, a majority (52.16 percent) of workers in the lowest bracket and approximately one-third (32.30 percent) of workers in the middle bracket work part time, defined as less than 35 hours of work per week. Many of these part-time workers are in such status involuntarily, desiring full-time work but not working full time (in the CPS, these workers are coded as “working part time for economic reasons”). Workers in the lowest bracket are the most likely to work part time involuntarily (18.77 percent), whereas workers in the highest bracket are the least likely to do so (6.18 percent). Finally, while the rate of unionization does increase with each successive wage category, only 3.63 percent and 5.18 percent of workers in the lower two categories are members of, or otherwise covered by, a union. These percentages compare with 10.80 percent of workers in the highest category.

Economic hardship and public program participation

Rates of economic hardship differ among the three categories of workers. All individuals included in the cross-sectional analysis are considered low-wage workers by some definition, but not all of them are in a household experiencing “official” poverty (i.e., poverty as defined by the official federal formula). The proportion of workers in a household experiencing official poverty declines with each successive wage category: the rate of poverty among those in the highest category (9.13 percent) is less than half that of workers in the lowest category (20.89 percent). (See table 3.) It is frequently argued that the federal definition of poverty is limited, because it is based on an emergency food budget in an era when food is less expensive than it was when the definition was first introduced.17 In addition, income eligibility for some state welfare programs, such as SNAP, is set higher than the poverty threshold. Despite these concerns, using an indicator of economic adversity set at 150 percent of poverty still shows that the proportion of workers experiencing economic hardship declines with increasing wages, from 36.38 percent in the lowest wage category to 19.35 percent in the highest.

Table 3. Economic hardship and program participation among low-wage workers, 2013 (percent)
CharacteristicWage bracket
Up to $9.25 per hour$9.26 to $10.75 per hour$10.75 to $13.50 per hourUp to $13.50 per hour



Family at or below poverty


F(2.00, 7,109.24) = 28.64(1)

Family at or below 150 percent of poverty


F(2.00, 7,108.44) = 35.53(2)



F(2.00, 7,107.78) = 26.41(1)

Program participation


SNAP (received by household), past year


F(2.00, 7,108.57) = 12.08(1)

TANF, past year

F(1.99, 7,084.22) = 1.71

Public health insurance (Medicaid, Medicare, military insurance, Indian Health Service)


F(2.00, 7,102.23) = 10.76(1)

Private health insurance (employer provided or own insurance)


F(2.00, 7,107.37) = 54.41(1)


(1) p < 0.001

(2) p < 0.01

Note: SNAP = Supplemental Nutrition Assistance Program; TANF = Temporary Assistance for Needy Families.

Source: U.S. Census Bureau, Current Population Survey, Annual Social and Economic Supplement and Earner Study.

Patterns of health insurance coverage and public program participation differ across wage categories.18 Nearly 40 percent of workers in the lowest category report not having had health insurance in the past year. This rate decreases to 33.36 percent for the middle category and 24.19 percent for the highest. Despite this decline, many workers in all three categories lack health insurance. In addition, the three groups of workers differ by source of health insurance. Workers in the lowest wage category (17.55 percent) are more likely than their peers in the middle and highest categories (12.94 percent and 10.65 percent, respectively) to have been covered by some type of public health insurance in the past year. The rate of private health insurance coverage, whether purchased or employer provided, increases from 46.91 percent in the lowest wage bracket to 69.79 percent (a clear majority of workers) in the highest. With regards to other public support programs, SNAP participation is most common among workers in the lowest wage category (18.04 percent) and steadily declines with successively higher wage brackets. Finally, only a handful of workers in any wage group report receiving benefits from TANF in the past year.

Discussion and conclusion

The findings presented in this article have implications for the operational definition of low-wage work. First, simply raising the wage cutoff used to identify low-wage workers substantially changes the estimated size of the low-wage population, from slightly more than 10 percent of American workers to 30 percent of the workforce in 2013. Trends over time, however, are similar under each definition. Second, examining workers within the three discrete wage brackets created by the cutoffs suggests that these brackets sometimes capture groups with quite different economic, demographic, and social characteristics. In particular, workers in the lowest wage bracket are more likely to work less than full time, are more likely to identify as members of a racial or ethnic minority group, are younger, are more likely to experience poverty and to use public support programs, and are less likely to be covered by health insurance. Although these differences are expected, fitting intuitive patterns, our analysis provides an indicator of the composition of different segments of the workforce and, by extension, the different groups added to the population of low-wage workers by more expansive wage criteria.

Comparing workers in successive wage brackets indicates that how one defines low-wage work in research matters substantively. On the one hand, pooling diverse groups of workers in a single definition of low-wage work obscures important distinctions among them. On the other hand, using different definitions sidelines the fact that some worker characteristics, such as lack of health insurance coverage, are common across the low-wage spectrum. There are tradeoffs, then, in setting a threshold for defining low-wage work: setting the threshold higher washes out the characteristics and experiences of workers facing the greatest challenges, whereas setting the threshold closer to the minimum wage ignores the prevalence of issues experienced across the wage range. Therefore, an approach that considers subgroups of earners within the low-wage labor market or explicitly evaluates the tradeoffs inherent in different definitions of low-wage work could facilitate more nuanced research and public policy.

Our study has some limitations. It does not distinguish between primary and secondary earners in a household, and these earners could differ in terms of demographics, economic hardship, and program participation.19 In addition, the analysis does not account for associations between worker characteristics, so no claims can be made about the independent relationship between any single characteristic and a particular wage level. The study does, however, provide a representative descriptive overview of low-wage workers and allows for a comparison across various wage levels.

Suggested citation:

Vincent Fusaro and H. Shaefer, "How should we define “low-wage” work? An analysis using the Current Population Survey," Monthly Labor Review, U.S. Bureau of Labor Statistics, October 2016,


1 See, for example, Steven Greenhouse, “Hundreds of fast-food workers striking for higher wages are arrested,” The New York Times, September 4, 2014,

2 David H. Autor and David Dorn, “The growth of low-skill service jobs and the polarization of the US labor market,” American Economic Review, vol. 103, no. 5, August 2013, pp. 1553–1597; and Sheldon Danziger and David Ratner, “Labor market outcomes and the transition to adulthood,” The Future of Children, vol. 20, no. 1, spring 2010, pp. 138–155.

3 Eileen Appelbaum and John Schmitt, “Review article: low-wage work in high-income countries: labor-market institutions and business strategy in the US and Europe,” Human Relations, vol. 62, no. 12, December 2009, pp. 1907–1934.

4 Bruce Western and Jake Rosenfeld, “Unions, norms, and the rise in U.S. wage inequality,” American Sociological Review, vol. 74, no. 4, August 2011, pp. 513–537.

5 Gerhard Bosch, “Low-wage work in five European countries and the United States,” International Labour Review, vol. 148, no. 4, December 2009, pp. 337–356; and Katherine S. Newman, Chutes and ladders: navigating the low-wage labor market (New York: Russell Sage Foundation, 2006).

6 Colin Campbell, “Low-wage mobility during the early career,” Research in Social Stratification and Mobility, vol. 30, no. 2, June 2012, pp. 175–185.

7 Devah Pager, Bruce Western, and Bart Bonikowski, “Discrimination in a low-wage labor market: a field experiment,” American Sociological Review, vol. 74, October 2009, pp. 777–799.

8 Appelbaum and Schmitt, “Review article: low-wage work in high-income countries,” pp. 1919–1920.

9 Jeffrey R. Harris, Yi Huang, Peggy A. Hannon, and Barbara Williams, “Low-socioeconomic status workers: their health risks and how to reach them,” Journal of Occupational and Environmental Medicine, vol. 53, no. 2, February 2011, pp. 132–138.

10 Danziger and Ratner, “Labor market outcomes.”

11 Bosch, “Low-wage work.”

12 Campbell, “Low-wage mobility.”

13 Data files were retrieved from the Integrated Public Use Microdata Series (IPUMS), a project of the Minnesota Population Center that harmonizes data from the entire history of the CPS. For more information on the IPUMS project and its data, see

14 “Supplemental surveys,” Current Population Survey (U.S. Census Bureau),

15 “Methodology,” Current Population Survey (U.S. Census Bureau), The data for this article were weighted with the use of weights from the Earner Study.

16 See “Poverty thresholds” (U.S. Census Bureau, 2015), The cutoffs were created with dollar values rounded to the nearest quarter. Full-time employment was defined as working 35 hours per week, 50 weeks per year.

17 Rebecca M. Blank, “Presidential address: how to improve poverty measurement in the United States,” Journal of Policy Analysis and Management, vol. 27, no. 2, spring 2008, pp. 233–254; and Bruce D. Meyer and James X. Sullivan, “Identifying the disadvantaged: official poverty, consumption poverty, and the new supplemental poverty measure,” The Journal of Economic Perspectives, vol. 26, no. 3, summer 2012, pp. 111–136.

18 Because the full implementation of the Affordable Care Act occurred after 2012—the reference year for gauging insurance coverage in the 2013 CPS—the present study cannot determine whether and how the expansion of Medicaid coverage and subsidies for the purchase of private insurance could affect the insurance status of low-wage workers.

19 H. Luke Shaefer, “Part-time workers: some key differences between primary and secondary earners,” Monthly Labor Review, October 2009, pp. 3–15,

About the Author

Vincent A. Fusaro

Vincent A. Fusaro is a doctoral candidate at the School of Social Work, University of Michigan, Ann Arbor, MI.

H. Luke Shaefer

H. Luke Shaefer is an associate professor at the School of Social Work and the Gerald R. Ford School of Public Policy, University of Michigan, Ann Arbor, MI.

close or Esc Key