The discussion so far makes it clear that the probability of being a low-wage worker has increased, an increase that has been driven by an increasing share of men in the low-wage sector. In addition, the low-wage workforce has become more highly educated but has seen falling real wages. This section addresses some of the reasons behind these changes. The findings discussed are based on a statistical technique called regression analysis, which allows the contribution of different causal factors to be distinguished from one another.13 Thus, the impact of changes in the economic returns to education, experience, occupation and industry can be separated from the impact of changes in the characteristics of the workforce, and from the impact of the changes in the likelihood of being in the low-wage market.14
Disentangling Impacts. Using education as an example helps to explain why it is important to disentangle the contributions of these three separate sets of factors, all of which have caused the low-wage sector of the labor market to grow. If the likelihood of being a low-wage worker falls over time, it might be because the education premium increases (that is, the extra earnings that come with additional education, say, being a high school graduate rather than a high school dropout, increase). In this case, the share of low-wage workers would fall even if the educational characteristics of the low-wage workforce remained unchanged. But what if the economic returns to education remained the same but the characteristics of the workforce improved (that is, a larger share of the workforce went to college)? In this case, the low-wage share of the workforce would also fall, but the cause would be different. It is also possible for the low-wage share to fall, even if the returns to education and the characteristics of the workforce remained unchanged, because of structural changes in the economy and/or institutional changes in the labor market. The next section discusses the relative importance of changes in the economic returns to work, in the characteristics of the workforce, and in the structural and institutional factors in explaining trends in the low-wage labor market over the past 25 years.
The Findings. Changing returns to work, other things equal, have lowered the probability of being in the low-wage sectorfor both men and women. Worker characteristics — educational and experience upgrading for men and women and occupational upgrading for women generally — have also lowered the probability of low-wage work, other things equal. Yet, the probability of low-wage work increased for the overall workforce. How can we explain this apparent riddle?
The answer is that the probability of low-wage work increased within groups of workers narrowly defined by age, race, education, occupation, and industry.15 Even after controlling for changes in all of these characteristics and their returns, we are still left with the secular increase in the probability of low-wage work.
What would have happened if there had been no changes in the economic returns to work or workforce characteristics? The probability of being in the low-wage sector would still have increased for both sexes, but the increase would have been much larger for men than for women.16
Various changes in the structure of the U.S. economy and labor market account for the increased probability of a worker being in the low-wage sector, irrespective of changes in the economic returns to work and the characteristics of the workforce. Explanations fall into two general camps: (1) a shift in labor demand against low-wage workers, driven by globalization and technological change, and (2) erosion of the institutions that used to bolster the economic conditions of the low-wage sector.17
Labor Demand Shift. The demand-shift argument interprets the sharp increase in educational returns over the 1980s as evidence that the demand for labor has shifted against the low-skilled worker. This shift, in turn, is due to an increasing mismatch between the skills these workers bring to the labor market and the skills employers seek, according to this argument. There is some evidence to support the demand-shift argument, particularly for workers at the very lowest skill level.18 But this cannot be the complete explanation, for several reasons. First, it fails to account for the increasing share of the workforce in the low-wage sector, irrespective of educational change. Second, wages have fallen for some groups of workers at all levels of education (even though the fall has been sharpest for the least educated).19 Third, if a labor demand shift were the complete explanation, one would expect to find persistent increases in the returns to education combined with declining employment opportunities for low-wage workers. But neither is true in the current economic recovery. Returns to education have significantly slowed for both men and women. Furthermore, low-wage employment, particularly at the low end of the service sector, has been growing quickly over this economic recovery, and the Bureau of Labor Statistics projects that some of the largest sources of future job growth will be in the low-wage categories (such as cashiers, retail sales workers, and low-wage clerical workers).20
Eroded Labor Market Institutions. This argument holds that a set of labor market institutions that have historically bolstered the wages of those in the low-wage sector have weakened in their protective role. Such institutions include minimum wage laws, unions, monetary policy, and trading regimes that protected jobs in favored industries. The real value of the minimum wage has fallen. Unbalanced trade in manufactured goods has expanded. The Federal Reserve has kept the unemployment rate at specific targets set to keep inflation down rather than employment up. All these factors have, indeed, reduced the ability of low-wage workers to keep their wages ahead of inflation.