Movements in and out of low pay

Amanda Gosling, Paul Johnson, Julian McCrae, Gillian Paull

This analysis examines how long people are likely to stay on very low wages...

This analysis examines how long people are likely to stay on very low wages and finds that - due to the level of movement in and out of low pay - any minimum wage legislation is likely to affect far more people than a single annual ‘snapshot’ of those on low pay might imply.

Summary

Summary

Over the last 20 years there has been a large increase in the gap between average earnings and earnings of the lowest paid workers. Recent research by Amanda Gosling, Paul Johnson, Julian McCrae and Gillian Paull of the Institute for Fiscal Studies has looked at how long people are likely to stay on these very low wages and at the link between low wages and unemployment in the 1990s. The study used British Household Panel Study data for waves between 1991 and 1994. The researchers found:

  • There is considerable mobility in the wage distribution. Twenty-eight per cent of men had real wage rises of 20 per cent or more over the period. Twelve per cent saw their wages fall by 20 per cent or more.
  • In the period between 1991 and 1994 a third of men between the ages of 18 and 60 spent some time out of work. Fewer than 9 per cent spent the whole period out of work.
  • There is a clear relationship between the current relative pay of workers and the chance that they will move out of work in the future. Ninety per cent of men who started in the top quarter of the earnings distribution remained in work for all the next three years, compared with 70 per cent of those who started in the bottom quarter of the earnings distribution.
  • Because people move in and out of work and in and out of low pay, over just a few years a minimum wage will affect many more people than at any one point in time. Taking into account the dynamics of low pay, the number of people affected by a minimum wage set at half male median earnings increases from 7 per cent of men and 28 per cent of women to 12.5 per cent and 42 per cent respectively.
  • Half of the differences in experiences of low pay in the 1990s can be explained by observable characteristics such as age, gender and educational qualifications. Controlling for these factors, however, knowing someone was low paid last year doubles the chance that they will be low paid the next.
  • Women with low educational qualifications face an unusually high probability of low pay over their entire working life.
  • Time spent with a particular employer, in addition to general employment experience, is important for wage growth. This is especially true for women. 

Introduction 

Concern about those on low pay has moved higher up the political agenda in recent years as the gap between the wages of the lowest paid workers and those on average earnings has grown. This new work confirms earlier findings that the low paid tend to be young and poorly educated. For example, men without any formal qualifications are 1.7 times as likely as the average to be in the bottom fifth of the wage distribution.

The main focus of this work, however, is on the dynamics of the labour market. Are the low paid stuck at the bottom of the earnings distribution or do they move up? For how long do people stay low paid? What is the relationship between pay and employment? In this new study, data from the first four waves of the British Household Panel Survey (BHPS) have been used to look at wage and employment mobility. 

Pay and employment 

Only about two-thirds of men remained in continuous employment over the whole period from 1991 to 1994, while fewer than 9 per cent were out of work all the time. The rest spent some of the time in work and some of the time in unemployment.

Importantly, lower relative wages of individuals are associated with a higher probability that they will move out of work in the future (see Table 1). Among men, 30 per cent of those starting in the bottom quarter of the wage distribution spent some time out of work in the next two-and-a-half years. This was true of just 12 per cent of those starting in the top quarter of the distribution. Similarly, looking at transitions into work, 56 per cent of the men moving out of unemployment moved in to a job with wages in the bottom quarter of the distribution.

table 1: Probability of moving out of work within 30 months by position in the wage distribution
Quartile proportion leaving work
Men  
Bottom 30%
2nd 19%
3rd 15%
Top 12%
   
Women  
Bottom 33%
2nd 25%
3rd 15%
Top 17%

Employment changes and wage changes 

It is also the case that movements in and out of work are associated with substantial wage changes. Table 2 shows that of those men who moved out of work and then back in again, over 40 per cent experienced a significant reduction in wages, compared with fewer than 20 per cent of those who continued in work. On the other hand, three in ten of them experienced wage increases of 20 per cent or more.

Table 2: Percentage change in wages according to whether any time was spent out of work between waves one and four
Percentage change in wage proportion of those not spending time out of work proportion of those who do spend time out of work
Down more than 20% 9.5% 33.7%
Down 10-20% 8.8% 10.1%
Between 10% down and 10% up 37.0% 20.4%
Up 10-20% 16.4% 5.9%
Up more than 20% 28.3% 29.9%

The clear, and unsurprising, message is that movements in and out of work are overwhelmingly experienced by those on, or who can only obtain, rather low wages. 

Mobility and the minimum wage 

One important issue associated with wage mobility is how it alters the number of people affected by a minimum wage. We know how many people would be affected by a minimum wage at any point in time, but the numbers affected over a longer period will depend on how many people move onto and off low pay.

Taking half male median earnings as a possible level for a minimum wage would mean something like 7 per cent of employed men and 28 per cent of employed women would be paid the minimum wage at any point in time. But over the period between 1991 and 1994, 12.5 per cent of men and 42 per cent of women, who were in work at some point, would have been affected at one time or another. The numbers vary according to the exact level at which a minimum is placed but the pattern is clear. Substantially more people are affected over just a three-year period than are affected at any one time as a direct consequence of the degree of wage and employment mobility.

Movement between low pay and unemployment means that a minimum wage will appear to be more redistributive in terms of income measured over a few years than in terms of income measured at one point in time. Those currently unemployed may benefit from a minimum wage at some time in the future. This, of course, is providing a minimum wage is set at such a level that it does not affect the probability of the unemployed getting a job in the future. 

Persistence of low pay 

Of men who were in the bottom quarter of the earnings distribution in 1991, half were still there in 1994 and a further 13 per cent were out of work altogether. Just 6 per cent had made it into the top half of the earnings distribution. Among women there was slightly more movement into the top half of the distribution and much more movement out of work. The figures for men and women who started in the bottom two quarters of the age distribution are summarised in Table 3.

Table 3: Position of low paid in 1994, by quartile in 1991
  Position in wave 4  
Quartile Wave 1 Bottom quartile Second quartile Top half Self-employed Out of work All
Men            
Bottom 52 23 6 6 13 100
Second 14 44 27 5 10 100
             
Women            
Bottom 44 22 10 2 22 100
Second 20 45 22 1 11 100

Continuing effects of low pay and unemployment 

Using usual statistical techniques would give a prediction of about 31 per cent of those men who were low paid in 1991 still being low paid in 1992 (defining low pay as earning less than £5 per hour). But taking into account the fact that they were low paid in 1991 leads to a prediction of 68 per cent of them being low paid in 1992.

This means that for men about half the persistence in low pay is explained by the fact that certain sorts of people - those poorly educated and/or young - tend to be low paid. But half of the persistence seems to be down to other things. Perhaps being in low-paid jobs itself traps you in low pay. Or possibly there are other characteristics that we can't measure that trap people in low pay. In either case this is an important finding: people on low wages at one point in time are likely to remain there in the future.

The most important determinant of movement out of low pay seems to be job tenure. Low-paid men who have been in their current job for five and ten years are over 80 per cent more likely to move out of low pay than are those who have only been in their current job for less than two years.

Table 4
Group   10 years of labour market experience
    Direct job changes Out of work between jobs
  Entering labour market for first time Move at 5 years, 5 years tenure Move at 10 years, 0 years tenure Move at 5 years, 5 years tenure Move at 10 years, 0 years tenure
Men          
No qualifications 41.8 3.8 3.8 6.8 10.0
School only qualifications 33.3 4.0 4.2 7.2 6.6
Other further qualifications 15.7 1.6 1.6 3.3 3.7
College qualifications 3.4 1.0 1.0 1.7 1.7
           
Women          
No qualifications 35.6 13.7 19.4 20.9 21.2
School only qualifications 35.7 6.3 8.3 17.6 13.2
Other further qualifications 21.4 3.8 9.6 7.8 13.7
College qualifications 8.2 0.7 1.3 3.3 4.4

Lifetime wage growth and job tenure 

The effects of length of job tenure are exemplified in Table 4 which is based on statistical simulations to look at the effects of work experience and job tenure on pay. In the table low pay is defined as a wage level below half median earnings. Low pay is quite prevalent across the working population in the sense that a substantial proportion of workers will be in low pay at the start of their working lives. However, some groups of workers also face an unusually high probability of being in low pay over their entire working lives, particularly low qualified women.

Qualification level is important for determining both the prevalence and continued likelihood of low pay. For example, the simulations suggest that over 40 per cent of men with no qualifications will be in low pay at the start of their working lives, while only just over 3 per cent of college-educated men will be low paid.

Men are slightly more likely than women to experience low pay upon entering the labour market for the first time, but are far less likely to be in the lowest part of the wage distribution as experience increases. Men are also less likely to experience low pay when moving jobs. For example, while the simulations show that only 3.8 per cent of men with no qualifications will be in low pay following a job-to-job move at 10 years of labour market experience, the corresponding figure for women is 19.4 per cent. This indicates much lower returns to general employment experience for women than men, which may be due to differences in skill growth with general employment experience or in limited possibilities for finding 'better' jobs. It may also reflect the deterioration of general employment skills for women who spend time out of employment.

In addition, the differences in the proportions in low pay between the columns showing 5 and 0 years of tenure highlight the greater importance of accumulating job tenure for women. This may reflect differences in the acquisition of job-specific skills or in the types of job contract offered to men and women.

Finally, the simulations show that the proportions in low pay are consistently higher for those who were out of work between jobs than for those who moved directly from job to job. This was found to be true both for job starting wages (shown in the zero tenure numbers) and throughout the life of the job (illustrated in the 5 years of tenure numbers). This suggests a detrimental effect on wages of time out of employment and highlights the importance of finding a 'good' job for the unemployed. 

About the study 

The data used in this study come from the first four waves of the British Household Panel Study (1991-1994). The sample includes men and women of working age (18 to 60) who were not in full time education.

Downloads

Downloads