[ Main page of Report | Contents of Report ]
Our literature review focuses on the small number of empirical studies that have examined the characteristics and labor market experiences of low-wage workers. A much larger body of literature exists on the labor market experiences of all adult workers, but we present results from these studies only when they are pertinent to low-wage workers. Similarly, there is a growing literature on the employment experiences of people who left welfare for work after the passage of PRWORA. The employed welfare population, however, is a narrow segment of the population of all low-wage workers. Therefore, we present findings for the employed welfare population to supplement our main presentation, but we do not provide a complete literature review for this group. Finally, a large body of literature exists on topics tangential to those that our study covers, such as income inequality and the demand for low-skilled workers. These topics are clearly related to those of our study. We do not directly address them in our empirical analysis, however. Thus, to keep our literature review focused, we do not discuss these topics.
The literature review contains three sections. First, we discuss how researchers have defined low-wage workers. Second, we summarize the literature on the characteristics of low-wage workers and their jobs. Finally, we discuss what is known about job turnover and wage progression for those in the low-wage labor market.
[ Go to Contents ]
Researchers have used several definitions of the low-wage labor market. One approach has been to define low-wage workers as those whose hourly wages are below a cutoff value. Some researchers have defined the cutoff value as the hourly wage at which a full-time worker would have annual earnings below the poverty level for a family of three or four (Bernstein and Hartmann 1999; Mitnik et al. 2002; and Ryscavage 1996).(4) The wage cutoff value has also been defined as the minimum wage (Smith and Vavrichek 1992).
Some researchers have defined low-wage workers as those whose annual earnings are below a cutoff value to account both for hourly wages that workers receive and for the amount that they work (that is, to adjust for the possibility that workers may not work enough hours to meet their families' needs). Mishel et al. (2001) define low-wage workers as those who worked full- or part-time involuntarily, but whose annual earnings were not high enough to reach the poverty level for a family of three, which was $15,208 in 1998. Similarly, Carnevale and Rose (2001) use an annual earnings cutoff value of $15,000 a year, and Holzer et al. (2001) use a cutoff value of $12,000 a year for three consecutive years.
Another approach used in the literature is to define low-wage workers as those whose hourly wages are in the bottom percentiles of the wage distribution (that is, a "relative wage" rather than an "absolute wage" approach). For example, Gladden and Taber (2000a) define low-wage workers as those whose hourly wages are below the 20th percentile of the wage distribution. Similarly, Long and Martini (1990) focus on those with earnings below the median value for full-year, full-time workers. This relative wage approach has also been used in studies that have examined changes over time in income inequality (see, for example, Gottschalk 1997).
Still another approach has been to define low-wage workers as those with low education levels or test scores (Gladden and Taber 2000b; and Holzer and LaLonde 2000). This approach does not use wage or earnings information directly; instead, it relies on the fact that low-wage workers usually have only a high school degree or less. One problem with this approach is that a substantial number of higher-wage workers also have low education levels (see below).
Finally, some studies have focused on the working poor. They define a person as a low-income worker if the total annual income of the person's family is below a given level and if the person worked a minimum number of hours during the year. For example, three papers by U.S. Bureau of Labor Statistics (BLS) researchers define a worker as a low-wage worker if his or her family's total income was below the federal poverty level (the official U.S. Census Bureau definition) and if he or she worked or looked for work in at least 27 weeks over the past calendar year (Gardner and Herz 1992; Hale 1997; and Klein and Roens 1989). Similarly, Acs et al. (2001) consider a family to be poor if its total annual income was below twice the federal poverty level and as working if all adult family members worked an average of half-time or more during the year.
[ Go to Contents ]
The literature characterizing low-wage workers and their jobs has focused on three main questions:
The fact that researchers have used several methods to define low-wage workers often makes it difficult to directly compare the findings across studies. In addition, researchers have used a number of data sets, both cross-sectional and longitudinal, to study these issues. The data sets often include different samples and cover different time periods, which further complicates direct comparisons. Despite these differences in definitions and samples, however, the key findings across studies are broadly consistent. This is likely due to the considerable overlap in samples generated using the different definitions of low-wage workers. In this literature review, we draw from research using each of the definitions of low-wage workers described above.
Several recent studies using a variety of data sources and definitions of low-wage workers (that are not based on family income levels) show that about one-quarter to one-third of all workers in the late 1990s and early 2000s were in low-wage jobs. Bernstein and Hartmann (2000) find that 29 percent of all workers in 1997 were low-wage workers, and Mitnik et al. (2002) find a corresponding figure of 25 percent in 2001. Both studies use cross-sectional Current Population Survey (CPS) data and a poverty-level wage cutoff value to define low-wage workers. Using a similar definition of low-wage workers, but Survey of Income and Program Participation (SIPP) data, Ryscavage (1996) estimates that about 25 percent of jobholders in 1993 were in low-wage jobs. Finally, using data from the Panel Study of Income Dynamics (PSID), Carnevale and Rose (2001) find that, of all people who worked in 1998, 32 percent were low earners, whom they define as those with annual earnings below $15,000, which was just above the amount needed to keep a household of three out of poverty.
Although less relevant to our study, estimates of the size of the low-income working population are much lower in studies that have examined workers in low-income working poor families. Several studies show that the poverty rate among working adults was only about six percent in the late 1980s, where a worker is defined as poor if his or her family's total income fell below the federal poverty level and if the person worked or looked for work in at least 27 weeks over the past calendar year (Gardner and Herz 1992; Hale 1997; and Klein and Roens 1989). Schiller (1994) uses a stricter standard to define a worker only those who worked full-time and full-year and finds that the poverty rate among them was only 2.5 percent. Kim (1998) uses a much less stringent standard to define workers: any adult who worked at all in the previous calendar year. She finds that 10 percent of workers were poor. The poverty rates for workers more than double if the family income threshold used to define the working poor is increased from 100 to 150 percent of poverty (Kim 1998; and Schwarz and Volgy 1992).
Why are estimates of the size of the low-income working population much lower in studies that use total family income to define low-wage workers than in those that do not? The explanation is that a significant number of low-wage workers are in families with total incomes above the poverty level. Carnevale and Rose (2001) confirm this they show that 57 percent of workers who earned less than $15,000 a year in 1998 lived in families with incomes above $25,000. Using SIPP data, Long and Martini (1990) find a similar result the lower tail of the earnings distribution coincides only partly with the population in poverty. These results suggest that some low-wage workers are secondary earners and work part-time or take lower-paying jobs. Consequently, they fall in the low-wage group based on their own earnings (but not on their family income).
Has the size of the low-wage working population changed over time? The evidence suggests that it has changed only slightly, although the direction of the change depends on the definition used to identify low-wage workers. Using cross-sectional CPS data from 1973 to 1997, Bernstein and Hartmann (2000) find that the share of workers earning poverty-level wages increased slightly over time, from 24 percent in 1973, to 27 percent in 1987, to 29 percent in 1997. Interestingly, the five percentage point increase between 1973 and 1997 was due entirely to an upward trend for males but not for females. Carnevale and Rose (2001), however, using PSID data, find that the share of the workforce with earnings below $15,000 (in 1998 dollars) decreased slightly over time, from 38 percent in 1979 to 36 percent in 1995 and 32 percent in 1998.
Broad consensus exists among studies that low-wage workers are disproportionately female, minority, young, and without a college education (Bernstein and Hartmann 1999; Carnevale and Rose 2001; Mishel et al. 2001; and Mitnik et al. 2002). Consistent with these findings, low-wage workers are also much more likely to live in households with children, that are headed by single females, that contain fewer adults, and that have fewer secondary workers.
At the same time, the research indicates that low-wage workers are a relatively diverse group. For example, Carnevale and Rose (2001) point out that low earners are a diverse group in terms of their family income among workers whose annual earnings were less than $15,000, more than half lived in families with total incomes above $25,000. Thus, many low-income workers live in families with other earners and with total family incomes above the poverty level.
The research indicates that most changes in the composition of low-wage workers for key characteristics, except for gender, have mirrored those of the total workforce. For example, the share of workers in the low-wage labor market with a high school degree or less decreased substantially during the 25-year period, but the same pattern holds for all workers in the labor force (due to widespread educational upgrading and the long-term wage decline among non-college graduates). Similarly, like the rest of the workforce, the low-wage sector became older and included more minorities. However, studies show that the low-wage workforce became increasingly male between 1973 and 1997, even though the female share of the entire workforce increased.
Several studies examine, in varying detail, the characteristics of jobs held by the population of low-wage workers and their overall employment characteristics (Acs et al. 2001; Bernstein and Hartmann 1999; Carnevale and Rose 2001; Mishel et al. 2001; and Mitnik et al. 2002). These studies focus on such characteristics as annual hours and weeks worked (in the low-wage job and in all jobs), job tenure, number of jobs held, benefits available on the job, and job occupations and industries. However, except for Acs et al. (2001), none of these studies examine the full range of job characteristics.(5)
The studies indicate that most low earners receive low hourly wages and are not full-time, full-year workers. In addition, the jobs that low-wage workers hold provide fewer benefits than the jobs that higher-wage workers hold, and low-wage workers have substantially less job tenure than higher-wage workers. Low-wage workers are represented in all occupations and all industries, but they are found disproportionately in retail trade industries, low-end service and sales occupations, and nonunion jobs (Acs ; Bernstein and Hartmann ; Carnevale and Rose ; Mitnik et al. ; Mishel et al. ; and Osterman ).
A large literature exists demonstrating that real wages of low-skilled workers (especially males) declined between the early 1970s and the mid-1990s, which suggests that the economic circumstances of workers in the low-wage sector worsened during this period (Blank 1994; Card and Blank 2000; Gottschalk 1997; and Mishel et al. 2001). For example, the real wages of males with wages at the 20th percentile of the wage distribution declined by about 20 percent between 1973 and 1994 (Gottschalk 1997). At the same time, real wages rose for workers in the upper tails of the wage distribution; thus, earnings inequality increased during the period.
Since 1994, however, the real wages of low-skilled male and female workers increased as a result of the strong economy (Card and Blank 2000; and Mishel et al. 2001). For example, the real wage of the 10th-percentile worker rose about nine percent between 1995 and 1999.
Finally, some evidence exists of occupational shifts over time within the low-wage sector (Bernstein and Hartmann 2000). For example, low-wage workers became less likely to work in clerical occupations and more likely to work in low-wage sales occupations than higher-wage workers. Similarly, by industry, low-wage workers became less likely to work in manufacturing and more likely to work in low-wage services such as the retail trade.
[ Go to Contents ]
Are low-wage jobs a stepping-stone to higher-paying jobs, or are people in low-wage jobs stuck in them? Despite the policy importance of this issue, little research has been conducted on it. Furthermore, the studies that have examined this issue have focused largely on the period through the early 1990s.
The literature on this topic uses longitudinal data on the same people over time, primarily from the PSID, SIPP, or the National Longitudinal Survey of Youth (NLSY). The Longitudinal Employer Household Dynamics (LEHD) data, which combines administrative data on quarterly employment and earnings for individuals with data on employers, is another good source for examining labor market dynamics over time (see Holzer 2001). Most studies identify low-wage workers or low earners in a base period and examine their labor market outcomes over subsequent periods, ranging anywhere from 1 year to more than 15 years. For example, Carnevale and Rose (2001) used the PSID to identify prime-age workers with earnings less than $15,000 in 1988 and followed them until 1992. Similarly, Gottschalk (1997) used the PSID to categorize workers in 1974 into quintiles of the earnings distribution and examined their earnings quintiles in 1975 and 1991. As another example, Smith and Vavrichek (1992) used the SIPP data to examine the labor market outcomes of minimum-wage workers in 1985 one year later. Another approach taken in the literature to measure the extent of wage growth for low-skilled workers is to use regression methods to estimate the relationship between work experience and hourly wages (Gladden and Taber 2000a, and 2000b).
The evidence in the literature about the extent of wage progression for low-wage workers consistently suggests that some low-wage workers experience wage growth, while others do not. Studies also find that movement out of the low-wage labor market into the higher-wage one increases with time spent in the labor market. Two patterns, however, are noteworthy. First, although there is some increase out of the low-wage labor market with time, the movements are not large. Second, a considerable number of low-wage workers drop out of the labor force over time, so that the group that remains is a somewhat select sample.
Several studies examining the employment experiences of the welfare population also send a mixed message about the extent of wage progression for those in lower-end jobs. Using the NLSY, Rangarajan et al. (1998) show that job retention is a problem for most welfare recipients who find jobs (75 percent of the sample left their jobs within a year). However, on average, welfare recipients who worked steadily experienced considerable increases in earnings over time, primarily as a result of increases in hours and weeks worked; however, wages improved only modestly.(6) Studies by Bartik (1997), Burtless (1995), and Corcoran and Loeb (1999), which focus on the economic returns to work experience, also find modest returns to work for welfare recipients.
An important policy question is the extent to which success in the labor market differs across key subgroups of the low-wage population. The literature on this topic is sparse. The few studies that address this question find that wage progression is lower for females, minorities, and people with low education (Carnevale and Rose 2001; Smith and Vavrichek 1992; and Holtzer et al. 2001). Only limited information exists on the extent of wage progression for low-wage workers by age. The study by Smith and Vavrichek (1992), the only one we found that addresses this matter, finds that wage progression among minimum-wage workers was greater for people ages 25 to 54 than those in other age ranges. Those age 55 or older had the lowest wage gains, followed by teenagers.
In addition, limited information exists on wage growth for subgroups of the low-wage population defined by their initial job characteristics. Rangarajan et al. (1998) examine this matter using NLSY data, but only for the population of welfare recipients who find jobs. They find that initial job characteristics are closely related to employment spell lengths and wage growth, even after controlling for numerous individual characteristics. In particular, wage growth was substantially greater for people in jobs with higher initial wages and with fringe benefits than for people in other jobs. Holzer and Lalonde (2000) use low-skilled youths in the NLSY to study job turnover rates the extent to which workers change jobs by initial job characteristics, although they do not examine wage growth directly. Their results, however, corroborate those of Rangarajan et al. (1998). Specifically, they find that the characteristics of the jobs to which less-educated workers have access, including their starting wages, occupations, and industries, affect their job turnover rates. For example, jobs in construction and service occupations have higher turnover rates than other jobs, whereas jobs in manufacturing (and to a lesser extent, in transportation and utility sectors) have lower turnover rates. Similarly, the starting wage of the job has strong negative effects on job transition rates.
Has it become increasingly difficult for low-wage workers to move out of poverty? Duncan et al. (1995) suggest that the answer to this question is yes, at least during the 1970s and 1980s. Using the 1968-1992 waves of the PSID data, they found that, for all subgroups of 21-year-old men, classified by race, ethnicity, and education level, the time it took them to earn twice the poverty level increased throughout the 1970s and 1980s. Importantly, the worsening of mobility prospects has been particularly severe for workers with low levels of education.
Another salient issue is the role of job retention in achieving wage growth for low-wage workers. Job turnover is common among low-skilled workers (Holzer and Lalonde 2000; Light and Ureta 1992; Royalty 1998; and Topel and Ward 1992). Furthermore, evidence exists that recent declines in employment rates among less-educated people largely reflect increasingly lengthy durations of nonemployment for those who leave their jobs. Consequently, an important policy issue is the labor market consequences of these high job turnover rates. Changing jobs, even with intermittent unemployment spells, might help low-wage workers progress in the labor market. However, it is also possible that workers progress more by staying in the same job.
The evidence on the effect of job turnover on wage progression for low-wage workers is limited. However, the detailed study by Gladden and Taber (2000a) suggests that there is a positive return to some voluntary mobility for those with low levels of education, although the story is complex. Using the NLSY, they show that a voluntary job change was associated with a three percent increase in wage growth for low-skilled workers, although frequent job changes led to earnings losses. In contrast, an involuntary job change led to a five percent decrease in wages. They also find the intuitive result that, when workers moved directly between jobs or were unemployed for a short time, their wages tended to rise with turnover, but when the unemployment spell was longer, their wages fell. They conclude that a substantial amount of wage growth for low-skilled workers comes with job changes.
[ Go to Contents ]
Our review of the literature indicates that a lot is known about the characteristics of recent low-wage workers. About one-quarter to one-third of all workers are in the low-wage labor market, and their share in the full labor force has not changed much over time. Low-wage workers are disproportionately female, minority, young, and with low levels of education. At the same time, however, they are also a relatively diverse group. For example, many low-wage workers are poor, but many also live in families with other earners and with total family incomes above the poverty level.
Consensus also exists on the characteristics of jobs that low-wage workers hold. Most receive low hourly wages, work part-time, and hold jobs that are markedly less stable and provide fewer benefits than those that higher-wage workers hold. Low-wage workers are represented in all occupations and industries, but they are found disproportionately in retail trade industries, low-end service and sales occupations, and nonunion jobs.
Less is known about the employment dynamics and wage growth of low-wage workers, and the available evidence pertains to the pre-PRWORA period only. The literature has identified important patterns, however. First, several studies find that, although there is some movement out of the low-wage labor market over time, the movements are not large. Second, movement out of the low-wage sector increases somewhat with work experience. Third, although some workers escape the low-wage labor market, their wage and earnings growth is modest. Finally, female workers, minority workers, and those with low education levels are less likely than their respective counterparts to move into the higher-wage labor market.
Our study builds on the existing literature in two ways. First, and most important, we use a recent cohort of low-wage workers, a unified data source, and a consistent definition of low-wage workers to address a wide range of topics covered in the literature. We provide a comprehensive profile of recent low-wage workers and their labor market experiences, instead of focusing on narrow issues typically addressed in the literature. Second, we provide a more complete analysis of the employment dynamics and wage progression of low-wage workers than is found in the literature.
[ Go to Contents ]
(4) For example, dividing the 2002 poverty level for a family of four ($18,100) by the number of full-time work hours in a year (2,080) yields a wage cutoff of $8.70 an hour for the low-wage sector.
(5) A larger literature exists on studies that have focused on the characteristics of jobs held by the welfare population only (see, for example, Rangarajan et al. 1998; and Pavetti and Acs 1997).
(6) Rangarajan et al. (1998) showed that a considerable number (nearly 30 percent) also experienced a decrease in wages over time. Recent studies that have examined wage growth among former welfare recipients suggest that those starting at low wages are most likely to experience wage growth, while those starting at relatively high wages are the ones most likely to experience wage reductions over time.
Top of Page | Contents
Main Page of Report | Contents of Report
Human Services Policy (HSP)
Assistant Secretary for Planning and Evaluation (ASPE)
U.S. Department of Health and Human Services (HHS)