The set of definitions more typically seen in contemporary research on the low-wage labor market focuses on the characteristics of workers (or potential workers) themselves — such as wage level, earnings and hours worked, or skill level.
Wage Level. Defining the low-wage labor market by the wages of its workers is clearly tautological. Even so, it is certainly reasonable to define, or at least discuss, the low-wage labor market by referring to the wage level itself. After all, the definition one chooses will correlate strongly with low levels of compensation if it is to be useful. And looking at the other characteristics of persons who work for low wages can tell us quite a bit about who the low-wage worker is likely to be.
The disadvantage of a simple wage-based approach is that it treats all workers who happen to be receiving low wages at a given point in time as similar, which covers up the important issue of differential labor mobility. A college student in a low-wage job, for example, is typically of much less policy concern than a single mother stuck indefinitely in a dead-end job.
This is a less-severe problem than it first appears, however, and can be alleviated by taking snapshots of the wage distribution at different points in time. Such snapshots allow comparisons of the different characteristics of workers at different wage levels. Unless the rate of mobility (that is, the speed with which workers move up the wage distribution) changes, comparing snapshots over time presents a useful description of the conditions of the low-wage sector and the characteristics of those who work there. For example, as discussed further below, declining real wages in the low-wage sector have led to increasing shares of the workforce being in low-wage jobs over time, and these workers are older and more highly educated than their predecessors. Unless older, better-educated workers that start out earning low wages are jumping ahead more quickly than in the past — and there is actually evidence to the contrary — these findings imply that the low-wage sector has truly expanded and includes older and better-educated workers.
The Absolute Wage Approach. The easiest wage-based definition to understand and interpret is an absolute measure. Analysts look at the share of the workforce in the same real wage range in different years and observe both the characteristics of workers and the proportions of workers within those ranges at particular points in time. A common way of determining such wage ranges is to use the U.S. standard poverty level as a reference and dividing the wage distribution by multiples of the wage rate derived from that level. Table 2, for example, divides the poverty level for a family of four ($16,400 in 1997) by 2,080 hours (52 weeks of work at 40 hours a week) to derive an absolute wage-level cutoff for the low-wage sector of $7.89 an hour. Using the poverty level for a family of three would yield a correspondingly lower wage-level cutoff for that sector.5 The table then shows comparisons between that sector and two higher wage ranges defined, respectively, as wage levels between the poverty-level wage and twice that level, and wage levels above twice the poverty-level wage.
Besides being easy to interpret, the absolute measure has the advantage of facilitating comparisons of absolute living standards (that is, real consumption opportunities) between low-wage and other sectors. It is, of course, sensitive to how price changes over time are measured. Any bias in the Consumer Price Index, for example, will be reflected in a corresponding bias in absolute wage rate comparisons. Any absolute measure is also unavoidably arbitrary. This weakness can be alleviated by doing sensitivity tests. These tests, by replicating the calculation for wage levels around the central choice, show how sensitive the results are to the particular wage level chosen.
The Relative Wage Approach. The danger of bias from measured prices failing to accurately reflect changes in real living standards is removed if the wage-based definition uses a relative approach, for example, by referring to the bottom 20 percent of the wage distribution. This definition has intuitive appeal because all would agree the bottom 20 percent are worse off relative to the top 20 percent, for example. The downside of relative measures is that they are not as rigorously tied to changes in living standards as absolute measures. Thus, the living standards of relatively low-wage workers — those in the bottom 20 percent, say — could rise markedly if real wages rose throughout the distribution, yet they would still be classified as low-wage workers. In other words, this approach allows for no change in the proportion of the workforce that is defined as low wage.
One way to solve this problem within the relative framework is to define low earnings as a fraction of the median wage.6 This measure will move with the median (a relative measure), but it allows the share of workers who fall into the low-wage category to vary over time. Again, however, an increasing median (implying increasing multiples of the median) would mean that at least some in the low-wage category will increase their standards of living over time.
Another limitation of the hourly wage level definition is that it fails to account for the possibility that workers may not work enough hours to meet their families' economic needs. Even if the wage structure were to rise high enough for workers at all wage levels to be able to support their families if they worked full-time/full-year, there is still the issue of whether enough hours of work are available in the low-wage sector.
Earnings/Hours Worked/Time Employed. The problem of work availability is very real for the low-wage sector. There is considerable evidence that disadvantaged workers (for example, workers whose personal characteristics are correlated with low earnings or incomes) experience higher levels of unemployment or underemployment than those with characteristics associated with higher earnings, even when the economy is strong. Furthermore, the share of persons with low-wage characteristics (such as young, less-educated minorities) who fail to participate in the labor market has increased over time.
To take this factor into account, analysts use definitions based on a variety of measures of time working. Blank (1994), for example, looks at the unemployment of family heads and finds that in 1991 (a business cycle trough), 40 percent of the reported weeks of unemployment by family heads occurred in the bottom 20 percent of the income distribution. Another approach looks at weeks unemployed and weeks out of the labor market altogether.7 This approach has found that between the late 1960s and the late 1980s, the largest deterioration occurred among workers in the bottom 10 percent of the wage distribution. Yet another variant looks at employment, unemployment, and underemployment.8 This approach reveals, for example, that in 1996-97, when the national unemployment rate was 5.2 percent, it was 19.7 percent for young (ages 16 to 25) African American women with a high school degree.
Results like these imply that, even in periods when wage levels are rising, low-wage workers will often not be able to work enough hours to fully meet their economic needs. Furthermore, they suggest that comparisons of the share of the workforce that is in the low-wage sector over time will progressively underestimate the size of the low-wage workforce. This is because the share of the potential low-wage workforce that is out of the labor market is not counted in the comparison because they earn no wage at all.
Education. A measure of the low-wage sector that does not depend on actual wage or earnings levels, although highly correlated with them, is education level. Low-wage workers are often those with a high school degree or less.
This approach has some intuitive appeal, particularly since the wages of those with college degrees increased sharply over the 1980s9 while the wages of those with a high school degree or less fell steeply. But it has two limitations. First, as of 1997, the "high school or less" definition included 44.3 percent of the workforce.10 Even avid critics of the U.S. labor market might be hard pressed to argue that such a large share of the workforce was "low-wage" or "low-skilled." Second, as with the relative wage approach, the sector of workers defined as low wage by the education approach will not change, even if rising real-wage levels in fact increase their standards of living.
Thus, there is a range of worker-based definitions of the low-wage labor market, each with its own strengths and limitations. Together they provide a portrait of the low-wage labor market and the workers in it, from which potentially useful policy conclusions can be drawn. The next section of the paper lays out the characteristics of low-wage workers, defined as workers who earn poverty-level hourly wages or less. This is followed by a discussion of how the low-wage labor market sector, variously defined, has been changing and the factors that have led to these changes.