8.3  Mobility in the US

British workers are not alone in their experience of reduced wage mobility. Buchinsky and Hunt (1999) report similar patterns of declining wage mobility in the US, especially at the bottom end of the wage ladder. They use data from the US National Longitudinal Survey of Youth (NLSY) for the period 1979-91, for young civilian wage earners who have ceased formal schooling. Their sample was aged from 14 to 24 in 1979. They find that wage mobility over a four-year period reduced wage inequality by 12 to 26%. This mobility predominantly occurs within groups of people with the same observable characteristics (ie, the same age, sex, education, race and level of experience at the start of the time horizon). In addition, they calculate the probability of moving or staying in the same wage quintile. Here they find, for all education and experience groups, that the probability of staying within the same quintile is high for all quintiles, but much higher for the low quintiles. Furthermore, this staying probability has sharply increased over time, meaning a rising inequality and a rapidly falling mobility. For example, the unconditional probability of staying in the first earnings quintile has increased from 33% in 1980 to 56% in 1990. Similarly, the probability of staying in the first wages quintile has risen from about 27% to 55% during the same period. Increases by about 20 - 30% in the staying probabilities are common for the first to the fourth quintiles of both the earnings and wages distribution, indicating a general rise in inequality. Overall, for the years 1990-91, wage inequality over the two years was 7% lower than the average of inequality for each of the two years.

Buchinsky and Hunt (1999) draw the important conclusion that the rise in cross-section wage inequality in the US, between 1979 and 1990 in their data, “reflects a severe widening of gaps between the same individuals.” (p 361). The rising cross-section inequality has been accompanied by a sharp fall in mobility, across all skill groups but especially for the less skilled. There is no comfort here from the hope that high levels of wage mobility mean that cross-section inequality does not translate into worrying levels of lifetime inequality.

As expected, the upward wage mobility of youth is relatively high. In the US, it is mainly teenagers who are paid at or very close to the minimum rate. Of course, people do not stay teenagers for long, so there is considerable upward wage mobility for youth. Even so, Carrington and Fallick (2001) find that “more than 8% of workers spend at least 50% of their first 10 post-school years working in jobs paying less than the minimum wage plus $1.” These are predominantly black and women and less educated.

In contrast, Long (1999) finds evidence of substantial wage mobility among the low-paid US workers. Using data from the 1991 and 1992 Survey of Income and Program Participation (SIPP), Long analyses the earnings and labour force status of respondents one and two years after they were in jobs paying the minimum wage. He finds that after one year of employment, seven in ten minimum wage workers stay employed on an hourly wage, just under 6% are either in a salaried job or self-employed, 5% are unemployed, and 20% have left the labour force. Long estimates that about 64% of minimum wage workers had a real wage increase, averaging 30% in one year. After two years, two-thirds of minimum wage workers were still in hourly paid jobs (down from seven in ten), with 68% of them reporting real wage increases (up from 64%). Minimum wage workers who moved to salaried positions reported earnings growth of more than double. These figures suggest mixed outcomes in term of employment status, and significant (given the short period of two years) upward wage mobility for US minimum wage workers who stay employed. However, Long’s sample has a high proportion of youth: About 46% of the respondents were aged between 15 and 19. Again, Long’s findings reinforce the expectation that youth have higher upward wage mobility than other low-wage earners. Recall also that the US minimum wage is low compared with the more common measure of low wage, namely two thirds of the median wage.

Table 5, from Carrington and Fallick (2001), shows the movement into and out of minimum wage jobs in the US, for young people who had been in the full-time labour force for up to 10 years (ending in 1995). The data come from the National Longitudinal Study of Youth, 1979 panel. The measure of low wages is the minimum wage plus $0.25. This is a very severe measure of low wage. But the transition probabilities show exactly the sort of information that is needed to obtain a good understanding of the extent to which people get stuck in low wage (in this case, minimum wage) jobs. For this reason, it is worth reporting, as an upper bound to mobility.

Notes: A job in year t is a minimum wage job if it pays less than the minimum wage plus $.25 in year t, where years are indexed by their position within a person's career.

Source: Carrington and Fallick (2001), from the US National Longitudinal Survey of Youth 1979.

The table refers to people who are no more than 10 years out of full-time education. It shows the proportion who make the transition into or out of a minimum wage job in year 2, contingent on having been/not been in a minimum wage job in year 1. This is shown for people with varying levels of workforce experience. It shows that once a person leaves a minimum wage job for a higher paying one, she/he rarely falls back into a minimum wage job. People who start the year in a minimum wage job have about a 50:50 chance of moving to a higher paying job by the next year, independent of how long they have been in the labour force. As Buchinsky and Hunt (1999) show, for many the move is not very far up the wage scale. Indeed, Carrington and Fallick themselves conclude that “- a substantial proportion of most workers early careers is spent on minimum or near minimum wage jobs” (p 23). Specifically, the average worker spent 29% of his/her first 6 years of employment in jobs paying no more than the minimum wage plus $2.00.

In an important study, Gottschalk (2001) examines the dispersion of wage outcomes for people who stay in the one job, for people who move straight from one job to another, and for people who experience a period of non-employment before finding their next job. He uses US data from the Survey of Income and Program Participation (SIPP, 1986-93). Each person (aged 18-55) in the data set is followed for a period of 24-40 months. Job changes and wages can be tracked. While on average, real wages grow from one year to the next, there is a wide dispersion of wage changes around this average. For many people, real wages fall, both for those who stay in the one job and even for some of those who move direct from one job to another. Specifically, for people who stayed with the one employer, on average real wages grew by 3.2% (female) or 2.1% (male). This growth rose to 3.7% and 4.0% respectively if people moved directly to a new job. However, if there was a spell of non-employment before obtaining a new job, real wages on average fell, by 2.6 and 2.9% respectively.

These averages conceal considerable diversity. Seventy% of people with less than high school education experienced a fall in their real wage if they stayed in the same job, as did 56% of those with college degrees. These percentages were less for people with longer tenure. The comparable figures for those who moved directly to a new job are 52 and 47%. For those who went to a new job via a spell of non-employment, the comparable figures are 58 and 54%. In each case, gender did not make a difference. For all levels of education, both average and median wage growth for those who moved via a spell of non-employment was negative. For job stayers, the median, if not the average, was also negative. (p 19-21).

The results produced by Gottshalk imply that individuals face considerable instability in their real wages, even if they stay in the same job. For the majority of employees over the period examined, the movement in their real wage was down rather than up. The least skilled were the ones most likely to have a fall in their real wage, whether they stayed in the same job or moved. There was clearly a large range of wage movements experienced by US workers in the late 1980s and early 1990s. While the majority suffered a fall in their real wage, a minority had increases that were large enough to ensure that average wages rose. This combination of outcomes reminds us how important it is to look behind the averages when seeking to understand mobility and wage movements. It also shows how difficult it was for low wage US workers in the late 1980s and early 1990s to gain rises in their real wages: it clearly is wrong to assume that work experience and willingness to stick with a job routinely lead to rising real wages. We note that over the last 5 years low wage workers in the US have on average begun to obtain some rise in their real wages: conclusions based on the earlier period may have less force in the contemporary environment.

The evidence of limited mobility for low wage workers is reinforced by the evidence presented by Connolly and Gottschalk (2001). Their sample includes people aged 18-55 who had no more than secondary school education. It is taken from the Survey of Income and Program Participation 1986-88 and 1990-93 panels. They formally model the possibility that workers will quit their current job for one that may even pay less, if it is expected to improve future wage offers. They draw several relevant conclusions. First, people employed in jobs that have low wages and low prospects of wage growth are much more likely to have short tenure in the job than people in high wage/high wage growth jobs (only around 12% of such jobs lasted more than 28 months, compared with about 55% for the better jobs: p20). Second, “People in jobs with low starting wages or low wage growth are most likely to obtain offers of similar jobs. Thus, even forward looking agents are likely to remain in jobs with poor prospects.” (p 31). They were unable to measure with sufficient precision whether the role played by stepping stone jobs was quantitatively substantial, so they remain agnostic on this point.