8 How much mobility is there?
8.1 Measuring mobility
To obtain a sound understanding of the degree of mobility from low wage jobs to better jobs, or to unemployment or out of the labour force, it is necessary to have longitudinal data, that follow the same individuals over time. These data are more difficult and expensive to obtain than normal survey data. In consequence, they are less comprehensive across time, countries and questions than are data based on cross section surveys. Nonetheless, there are now sufficient longitudinal surveys (or administrative data that enable individuals to be tracked) that reasonable conclusions can be drawn. Sound empirical data are crucial, since the question is not “is there upward mobility?”, but how much mobility is there and what causes it. In reporting on the degree of labour market mobility, it is sensible to confine attention to people who are not also full-time students.
We have defined as upwardly mobile, people who move from a low wage job in period 1 to a higher paying job in period 2. The emphasis is on movement to a higher hourly wage, rather than to a higher weekly or annual earnings (which is affected by the number of hours worked). People are immobile if in period 2 they are still in a low wage job, have withdrawn from the workforce, or are unemployed.
Mobility is usually expressed in terms of transition probabilities. These describe what proportion of the low wage group in period 1 has moved to a higher wage in the next period (or has dropped out of employment). In most cases the empirical evidence for the US and the UK concludes that there is strong persistence in low pay status. A person who is currently low paid has a much higher chance of being low paid in the next period, than an otherwise similar person who initially had higher pay. (Stewart and Swaffield, 1999;Stewart, 2002; Connolly and Gottschalk, 2001).
We expect that mobility will be higher the longer is the time interval that distinguishes period 1 from period 2, and empirical evidence confirms that this is the case.
8.2 Mobility in the UK
Evidence from Great Britain (Dickens, 2000) shows that hourly wage mobility measured from year to year is low, but increases when measured over longer periods of time. For example, Dickens finds that, for males in the bottom decile of the wage distribution, only between 20 and 34% move up (with 35 to 45% of these only moving as far as the next decile) after one year (1993 to 1994). Between 43 and 48% remain in the tenth decile and the rest fall out of employment. Females’ mobility patterns after one year are similar to those of males. Measured over three years (1991 - 1994), mobility is greater than for just one year, but Dickens’ evidence shows that movement to a higher wage decile is still the minority experience for those who started in the bottom decile. Between 26 and 31% of males in the bottom decile remain there after three years; 25 to 32% move out of employment (about 10% were unemployed), while only 26 to 40% move up: only half of the upward-movers go beyond the second wage decile. Again, females’ wage mobility is similar to that of males, although low wage women are more likely than men to leave the labour force. The evidence on wage mobility over five years shows that it is a little higher than over three years, but many are lost from the sample. Of those who could be traced, fewer than one third of men and women in the bottom decile of the wage distribution at the beginning of the period moved to a higher decile job over five years. Stewart and Swaffield (1999) produce similar findings: they find a high degree of persistence in the earnings distribution, with low wage earners moving frequently between low wage jobs and unemployment.
The mobility estimates of Stewart and Swaffield (1999) illustrate the point that the lower the threshold for the definition of low wage, the greater the mobility. They use three measures of low wage—half the median, half the mean and two thirds of the median full-time gross hourly adult wages. These define between 8 to 22% of men and 24 to 49% of women as low paid (using the 1991 panel of the British Household Panel Survey). They find that for women, 75%, 83% and 87% respectively do not move above each of the low wage thresholds. The comparable figures for men are 60%, 65% and 75%. People who move out of wage jobs (into unemployment, non-employment or self-employment) are included in the percentage who do not move up. (p 27).
The movement between low wage employment and unemployment is examined in detail, for the UK, by Stewart (2002). Stewart seeks to explain why a large proportion of people who move from unemployment to a job then fall back into unemployment again. He describes the experience of unemployment, and of low wage employment, as exhibiting state dependence. That is, the probability of being unemployed, or employed in a low wage job, in period 2 is strongly positively influenced by having been in that same state in period 1. An innovation in Stewart’s work is that he is able empirically to control for a range of observed and unobserved characteristics that of themselves predict unemployment/low wage, such as age and education. He chooses a measure of low pay that classifies about 10% of employees as low paid (quite a low value—which should bias his results towards relatively high mobility). He concludes that those who were low paid in period 1 were about 17 times more likely to be low paid in period 2 than were workers who were paid higher wages in period 1. (p 4). For people who were the same in terms of years of education, possessing a qualification, years of experience, gender, marital status, health status and whether resident in London or the South East, the ratio was reduced, but only to 14. Not only are low paid workers much more likely than higher paid workers to be low paid in the next period, they are also almost three times as likely to be unemployed. He interprets this and other evidence to show that there is a “low pay-no pay cycle” (p 5).
Dickens finds that short-term wage mobility in the UK fell between 1975 and 1994. This fall in mobility mainly occurred in the middle income deciles, while the wage mobility at the top and the bottom of the wage distribution has remained low throughout the period. Nevertheless, Dickens also finds evidence that low-wage earners, the group we are focussing on, are more likely to get stuck between states of unemployment or non-employment and low-paying jobs than they were in the 1970s (see also Stewart and Swaffield, 1999). Dickens concludes that “the low paid are worse off both in terms of the relative wage they receive and in terms of their opportunity to progress out of the low-pay trap.” (p 496)