In this chapter, we use nationally representative March cross-sectional samples of workers from the mid- to late 1990s to address these questions: What has the size of the low-wage working population been since the passage of PRWORA in 1996? Who are low-wage workers, and how do they compare to medium- and high-wage workers? What are the characteristics of jobs that low-wage workers hold? Did the characteristics of low-wage workers and their jobs change between 1996 and 1999?
For most of the analysis, we use a March 1996 cross-sectional sample for several reasons, including the fact that it is the earliest month in the SIPP data that is covered for all sample members (see Appendix A). However, we also conducted some analyses using the March 1997 to March 1999 cross-sectional samples to examine changes in the prevalence and characteristics of low-wage workers over time. To place our findings in perspective, we also present descriptive statistics for all workers and for medium- and high-wage workers.(7) Unless otherwise noted, all figures were calculated using our primary definition of low-wage workers: those with a wage below which a full-time worker would have annual earnings below poverty for a family of four ($7.50 in 1996, $7.72 in 1997, $7.91 in 1998, and $8.03 in 1999). All figures were calculated using the respective calendar year weights. Appendix B contains tables supplemental to those in the text of this chapter.
Because the mid- to late 1990s was a period of strong economic growth with low inflation, our findings must be interpreted carefully. The national unemployment rate decreased from 7.5 percent in 1992 (a period of recession) to 5.4 percent in 1996, and it decreased further to 4.0 percent in 2000, which is low by recent historical standards (see Figure III.1).(8) Thus, the characteristics of low-wage workers during our period of investigation may be somewhat atypical as it may include some workers who were previously unemployed or out of the labor force. Examining trends in the characteristics of low-wage workers and their jobs using earlier SIPP cohorts is beyond the scope of this study. However, we did examine changes in the composition of the low-wage labor market between 1996 and 1999 as the economy improved.
U.S. Unemployment Rate, By Year
|Source: U.S. Bureau of Labor Statistics.
As discussed later, we found that the characteristics of the low-wage population did not change during this period, suggesting that our findings may be representative of low-wage workers in general.
Size of the Low-Wage Population
The share of all workers who were low-wage workers was 28 percent in March 1996 (or nearly 29 million workers, Figure III.2). It decreased slightly to 27 percent in March 1997 and to 25 percent in March 1998 and March 1999. These estimated shares are similar to those found in previous studies (discussed in Chapter II) that used a similar cutoff value to define low-wage workers. For example, using March CPS data, Bernstein and Hartmann (2000) found that 29 percent of all workers in 1997 were low-wage workers, and Mitnik et al. (2002) found a corresponding figure of 25 percent in 2001.
The slight decrease in the size of the low-wage population after 1996 may be due to declines in the unemployment rate during that period, suggesting that some low-wage workers may have been able to find higher-paying jobs because of the tight labor market. This interpretation is consistent with findings in the literature that the real wages of low-skilled male and female workers increased starting in the mid-1990s as a result of the strong economy (Card and Blank 2000; and Mishel et al. 2001).
Most workers in the mid- to late 1990s were medium-wage workers (Figure III.2). These workers are defined as those whose hourly wages were between one and two times the low-wage cutoff (for example, those who earned between $7.50 and $15 per hour in 1996). Roughly equal numbers were low-wage and high-wage workers (Figure III.2). For example, in March 1996, 43 percent of all workers were medium-wage workers (about 44 million workers), 28 percent were low-wage workers (about 29 million workers), and 29 percent were high-wage workers (the 31 million workers who earned at least $15 per hour). Interestingly, the slight decrease in the share of low-wage workers between 1996 and 1999 was offset by small increases in both the medium- and high-wage sectors.
The size of the low-wage labor market differs substantially according to the definition used to identify low-wage workers (Table III.1). These definitions, described in greater detail in Chapter II, include identifying low-wage workers using the minimum wage, the 20th percentile of the wage distribution, annual earnings relative to the poverty level, and those with low education levels. The estimated fraction in the low-wage labor market according to these definitions range from 7 percent to 44 percent of workers. We briefly describe these findings below:
Percentage Of All Workers In March 1996 Who Were Low-Wage Workers,
According To Alternative Definitions Of Low-Wage Workers
||Hourly Wage Cutoff Usedfor the Study($7.50)
||Hourly Wage Below MinimumWage($4.75)
||Hourly Wage Below 20th Percentile ($6.57)
||Annual Earnings Below PovertyCutoff ($15,150)(a)
||Low Education Level(b)
|Percent of All Workers Who Are:
|Source: 1996 SIPP files using a March 1996 cross-sectional sample.
Note: All figures are weighted using the 1996 calendar year weight.
a. Annualized earnings are calculated as total monthly earnings from all jobs and businesses in March 1996 multiplied by 12.
b. Low-wage workers are defined as those with a high school degree or less, medium-wage workers as those who had some college education, and higher-wage workers as those with a B.A. degree or more.
c. Medium-wage workers are defined as those with wages/earnings that are between one and two times the level of the low-wage definition, and higher-wage workers are defined as those with wages/earnings that are greater than twice the level of the low-wage definition.
- Only 7 percent of all workers were in the low-wage labor market if these workers are identified as those earning less than the minimum wage $4.75 per hour. Using the minimum wage threshold, about 58 percent of those employed were high-wage workers (defined as those who earned more than twice the minimum wage). Thus, using the minimum wage sets the bar very low for defining low-wage workers.
- As expected, 20 percent of workers are in the low-wage labor market using the 20th percentile cutoff value of the wage distribution ($6.57 per hour). Thus, using our benchmark $7.50 cutoff value generates a larger estimate of the size of low-wage population than using the 20th percentile of the wage distribution as the cutoff value (28 percent of all workers, compared to 20 percent; Table III.1, columns 2 and 4).(9)
- About 32 percent were low-wage workers using an annual earnings below poverty cutoff. This measure defines a low-wage worker as one whose total monthly earned income from all jobs and businesses multiplied by 12 was below the annual poverty level for a family of four, and takes into account both hours worked and hourly wages.(10) Interestingly, while the 32 percent figure using the annual earnings measure is similar to our 28 percent benchmark measure, a significant number of workers are classified as low-wage workers using one definition but not the other. For example, of all those classified as low-wage workers using either definition, about 42 percent were classified as low-wage using one definition but not the other: 18 percent were classified as low-wage using only our benchmark definition, and 24 percent were classified as low-wage using only the annual earnings measure. These discrepancies suggest that there are (1) many workers with high wages who work only a limited number of hours, and (2) many workers with low wages who work a substantial number of hours. As discussed, we adopt the wage-based measure, because our study focuses on low-wage workers rather than low-income ones.
- About 44 percent of workers were in the low-wage labor market if low-wage workers are defined as those with a high school diploma/GED or less. We believe, however, that the use of this education-based definition does not adequately characterize the low-wage population, because, according to our benchmark wage-based definition, nearly 60 percent of those with a high school credential or less were higher-wage workers. Similarly, under our benchmark definition, about 18 percent of those who attended college are classified as low-wage workers.
Demographic Characteristics of Low-Wage Workers
We examined the characteristics of low-wage workers in two interrelated ways. First, we examined the question: Among all workers within a particular subgroup, what percentage are low-wage workers? For example, we calculated the share of all male workers who are low-wage workers and the share of all workers between ages 30 and 39 who are in the low-wage labor market. Second, we examined the reverse question: Among all low-wage workers, what percentage are in a particular subgroup? For example, we calculated the percentage of all low-wage workers who are male and compared it to the corresponding shares for all workers and for medium- and high-wage workers.
An example can be used to explain the difference between the two analyses and how to reconcile them. As discussed later, in 1996, about 84 percent of workers younger than age 20 were low-wage workers. However, only about four percent of all low-wage workers were younger than age 20, because teenage workers made up only about one percent of the entire labor force. The two sets of findings can be reconciled by using the result that teenage workers were four times more likely to be low-wage than higher-wage workers. Thus, each analysis provides, from a different angle, important descriptive information on the characteristics of those in the low-wage labor market.
We produced summary statistics for each variable one at a time. In addition, we conducted a cluster analysis to identify distinct groups of low-wage workers based on their full set of characteristics. This analysis accounts for the correlation between variables, and hence, provides a concise typology of groups of low-wage workers.
Our results on the characteristics of low-wage workers corroborate findings in the literature that low-wage workers are disproportionately (1) young, (2) female, (3) nonwhite, (4) with a high school credential or less, (5) in single-adult households with children, and (6) in households with incomes below the poverty level. At the same time, however, they are a relatively diverse group they exist in a wide range of subgroups defined by individual and household characteristics.
Job and Overall Employment Characteristics of Low-Wage Workers
SIPP contains some information on job and business characteristics, including usual hours per week worked, hourly wages, monthly earnings, occupation, industry, job tenure, whether health insurance is available on the job, and union membership status. We followed a similar approach for tabulating these characteristics as for tabulating workers' demographic characteristics. Our tables present distributions of job and business characteristics for low-wage workers, and all workers.(17) Unlike the previous section, however, we do not present the reverse figures (that is, the share of low-wage workers among those with a particular job characteristic), because these figures have less policy relevance. We present figures separately for males and females and present selected statistics by age. In addition, we present selected figures for the six (three male and three female) low-wage worker typologies discussed above.
We find that many low-wage workers receive hourly wages substantially below the low-wage cutoff value used in our study. In addition, low-wage workers hold jobs that are markedly less stable and provide fewer benefits than the jobs higher-wage workers hold. Interestingly, however, most report that they usually work at least 35 hours per week (that is, full-time). Low-wage workers are represented in all occupations and industries, but they are disproportionately found in retail trade industries, service occupations, and nonunion jobs. In combination, our results are similar to those found in Acs et al. (2001), Bernstein and Hartmann (1999), Carnevale and Rose (2001), Mishel et al. (2001), and Mitnik et al. (2002).