Skip to main content

Work-rich and work-poor: three decades of change

An analysis of trends in the distribution of jobs between social groups, this report tracks the number of non-working families in Britain over the last three decades.

Written by:
Richard Berthoud
Date published:

An understanding of why the number of people with no earned income has increased substantially could provide directions for future social policy. Tracking the rise of non-working families in Britain from 1974 to 2003, this report:

  • analyses the distribution of employment, particularly between men and women, disabled and non-disabled people and older and younger people; and
  • compares changes in group-specific employment probabilities with the overall employment rate, to find underlying trends.

The report concludes that the problem is about ‘traditional’ inequality just as much as changing family patterns – the polarisation between two-earner and no-earner couples may be because it is now acceptable for a woman to contribute earnings alongside her partner, but not for her to replace them. Inequality between men and women within couples has been reduced, but inequality between families has been increased by the two-earner/no-earner polarisation.


The number of people in employment is higher now than in the mid-1970s. There have, though, been substantial changes in the distribution of jobs between social groups. These trends have had important consequences for equality and inequality, and for the number of non-working families in Britain. This study, by Richard Berthoud of the Institute for Social and Economic Research at the University of Essex, is a detailed analysis of the trends over the past 30 years. The research found:

  • Around two million adults (aged 20-59) who are in work today would probably not have had a job in the mid-1970s.
  • Those whose job prospects have improved most are mothers, especially those with adequate qualifications, good health and a working partner.
  • This means that the number of couples who both have a job has increased. They are 'work-rich'.
  • On the other hand, there are another two million adults who would have been likely to have had a job thirty years ago, but are now out of work.
  • Those whose chances have deteriorated most are disabled men with poor educational qualifications and no working partner.
  • There has been a steep increase, too, in the number of non-working adults without a partner, or whose partner does not have a job. The proportion has doubled from 7 per cent to 14 per cent over 30 years. Most of these 'work-poor' families live on social security benefits, and have very low incomes.
  • These trends have not mainly been associated with changes in the demand for labour in the economy as a whole, but there are some signs that the underlying growth in the number of non-working families may have levelled off over the past few years.


In 1974, 7 per cent of adults in Britain between the ages of 20 and 59 were dependent on income, mostly social security benefits, from outside their immediate family. They were not in employment, and they did not have an employed partner either. In 2003, the rate was 14 per cent – double the figure from the early 1970s.

This research tracks the growth in the number of families who have no income from employment. (The 'family' is defined here as either a single adult, or a couple, treated as a unit – plus any dependent children of the adult or couple.) Most such families depend on social security benefits, and they have a high risk of poverty, particularly if children are involved. Family non-employment and the huge rise in the number of working-age adults depending on social security benefits are central to the analysis of poverty and disadvantage. Reducing the number of non-working families in Britain is at the top of the Government's social policy agenda.

The increase in the number of people living in 'work-poor' families has occurred over a period when the overall number of people in employment has been rising, not falling. So the issues addressed by this research are concerned with the distribution of jobs between different social groups, rather than with any weakness in the economy as a whole.

The personal distribution of employment

Men and women

It is widely known that the number of women in employment has increased over the years, but that they are still less likely to have a job than men. In fact, this applies mainly to mothers, rather than to women as a whole. Childless women's labour market participation has not changed very much, and they are now in a similar position to single childless men. So it is convenient to treat women without children as the middle case, and to compare changes affecting men (with or without children) with those affecting mothers.

Men, especially those with a partner, have traditionally had very high employment rates, and this is still true. On the other hand, their prospects have been declining: the proportion of men without a job has nearly trebled (from 5 per cent to 14 per cent) over thirty years. Men's prospects are more affected by disability than women's, though less affected by advancing age, but both disability and age are increasingly important factors. Men's prospects are much more sensitive to variations in the demand for labour (as measured by unemployment rates), both at the regional level and from year to year over the business cycle. Once the cyclical pattern has been ironed out, there has been a steady trend for men's employment to decline across the three decades under review.

Mothers were very unlikely to have a job in 1974 – just 36 per cent were in employment. (It should be noted that the proportion of women who are mothers has declined over the period analysed and the proportion of mothers living without a partner has risen.) Their employment rate had increased hugely by 2003, though it was still much lower than that of other groups, at 58 per cent. Initially, it was mothers in couples who improved their labour market contribution, but lone parents have been catching up more recently. Having young children, or many children, are both disadvantageous to mothers' employment, but the first of these factors is less important now than it used to be, while the second has become more important. Mothers' job chances were only weakly affected by disability, or education, at the start of the period, but these issues have become increasingly important for them. The demand for labour (locally, and over time) makes little difference to mothers' employment rates. Unlike men, the underlying trend in mothers' employment rate has not been continuous: the rate fell between 1974 and 1979, but has improved strongly ever since.


Like many general purpose surveys, the GHS data analysed for this research (see 'About the project') does not have strictly defined questions about disability and impairment. The analysis uses 'limiting long-standing illness' as a broad indicator of disability, although this has been shown to understate the extent of disadvantage experienced by people with serious impairments.

Disabled men have always had poorer job chances than other men, and this disadvantage widened somewhat over the period. A rapidly increasing proportion of disabled men describe their market position as 'long-term sick and disabled', rather than 'unemployed'. Women – with or without children – were less affected by disability than men in the 1970s, but this has become an increasingly important disadvantage for women, relative to their generally improving position. Disabled people with poor qualifications are much more disadvantaged by their impairments than those with a good educational background, and this gap has also widened. Disabled people seem to be highly sensitive to regional variations in overall employment rates, but are little affected by cyclical labour market changes from year to year. The trend was continuously adverse to disabled people between 1974 and 1996, but flattened out at that point.


It may be a surprise to find that older men were not disadvantaged in the 1970s, once other factors (such as disability and education) have been allowed for. However, men over 50 have faced an increasing penalty since then. Older women have had lower employment rates than their younger counterparts throughout the period, but the age penalty for women has reduced – so the trends are in opposite directions for men and women.

Families and employment

The changes outlined so far are all concerned with the employment position of individuals. The primary aim of this research was to account for the substantial increase in the proportion of people in non-working families over the decades – that is, people (below pension age) who do not have a job themselves and do not have a working partner either.

Since the overall employment rate increased slightly, the growth of work-poor (no-earner) families is not primarily due to a scarcity of jobs. It is a consequence of the redistribution of jobs, especially between men and mothers. The number of work-rich (two-earner) families has been increasing too, so that the number of one-earner families has been squeezed. In detail, the rise in two-earner families (after taking account of cyclical trends) was mainly between 1984 and 1994; the rise in no-earner families was continuous from 1974 to 1998, but has now flattened off.

The processes affecting men and women have been different. For men, the main reason for the increase in family-worklessness has been the reduction in their own employment rates. For women, changes in their own employment rates might have been expected to reduce the problem at the family level. But the reduction in partnership rates, and the deteriorating labour market position of partnered men, both tended to increase women's risk of living in a family with neither direct nor indirect access to earnings. So the 'family' outcome for men and women has been very similar.

It might have been expected that couples could adopt a policy of 'substitution' – one partner taking paid work, but not the other. In practice, the evidence supports the idea of 'combination' – both of them take paid work, or neither of them. So both men and women are more likely to have a job if their partner has one, and less likely if their partner does not have one. This tendency towards polarisation between couples ('both' or 'neither'), has increased over the years. Most of the rise in partnered mothers' employment has occurred in couples where the man had a job, and most of the decrease in partnered men's employment has occurred in couples where the woman did not have a job.


The issue of workless families (measured at that time in terms of households) first came to public attention in the early 1990s, and was much discussed by the Labour Party's Commission on Social Justice. It was conceptualised at that time as the uneven impact of the very high rates of unemployment at that period. The new research shows that the problem is still important, even after ten years of increasing economic prosperity, and should be conceptualised as an indirect exchange of jobs. About two million people who would have been outside the labour force in 1974 were in work in 2003 – mainly mothers with good health, good qualifications and a working partner. Another two million people who would have been in work at the start of the period were out of work at the end – mainly disabled men with poor qualifications and no working partner.

Although much of the redistribution of jobs has been between men and women, the outcomes are still closely associated with social disadvantage: more than half of disabled people without qualifications are in workless families, ten times the rate among non-disabled people with degrees.

In general, the trends have been smooth across the period rather than showing sharp rises or falls in group-specific employment rates at particular points in time. Where there has been a kink in a trend, it has usually been difficult to link that to any particular policy change – with the important exception that the deterioration for disabled people seems to have flattened out after the twin policy reforms (Incapacity Benefit and the Disability Discrimination Act) of the mid-1990s.

Some of the forces contributing to the changing pattern of employment can be found in the standard economic theories of the labour market. These forces include:

  • the changing structure of the economy;
  • the fall in demand for occupations requiring (male) strength; and
  • the technology-led reduction of the amount of time needed to care for a home and children.

Economists commonly attribute the low rates of employment among disabled people, or among the wives of non-employed men, to the incentive structures built-in to the benefit system, although this factor seems to fall short of a full explanation for the observed trends. Other possible economic influences, which would require further research to substantiate, could include a direct link between the increase in women looking for work and the reducing demand for poorly qualified disabled men; or perhaps the increasingly competitive global economy requires a more rigorous analysis of productivity, so that firms are more reluctant nowadays to find jobs for poorly qualified workers with failing health or impaired capacities. Other components of the redistribution of jobs require sociological, rather than economic, explanations. The decision to work, or not to work, is strongly affected by an individual's sense of identity in the context of standard values. The redistribution between men on the one hand and mothers on the other has coincided with a massive change in social conventions: it is no longer considered appropriate for men to exercise all of the economic power in couples and women to undertake all the domestic duties. It may also be that 'disability' is more commonly accepted as an explanation for not working. These changes in social convention are far from complete. The findings show that while the 'male breadwinner' is increasingly rare, he has not yet been replaced by the 'female breadwinner'. The polarisation between two-earner and no-earner couples may simply reflect the fact that it is now acceptable for a woman to contribute earnings alongside those of her partner, but not for her to replace them.

Inequality (between men and women) within couple families has undoubtedly been reduced. Yet inequality between couple families has been increased by the two-earner/no-earner polarisation. If part of the greater equality between husbands and wives consists of both of them having a job, another part consists of neither of them having a job. By the same token, the reduction in partnering rates may leave many women (and men too) worse off in relation to the prosperity of the now-dominant two-earner family. So inequality among women, and among men, may have increased.

About the project

The research was based on new analysis of the General Household Survey (GHS), covering 26 of the 30 years between 1974 and 2003. A 'personal employment equation' was developed to estimate adults' probability of being in work, taking account of their gender and family position, disability, age, education, regional unemployment rate and ethnic group. Analysis was confined to men and women aged between 20 and 59, with samples totalling 337,000 in that age range across the 26 years. 'Personal employment' was defined as having a job of at least 16 hours per week, or being a student. Someone was identified as being in a 'non-working family' if s/he was not in employment (as defined), and did not have a partner in employment either. Year by year analysis was used to show how each group's employment pattern was affected by cyclical trends in the economy as a whole, and then to reveal the underlying trends.


Music producer in a music studio sat on a chair.

This report is part of the work topic.

Find out more about our work in this area.

Discover more about work