Until now, we have known relatively little about the relationships between older people’s incomes and their work history patterns or how other factors, such as gender and ill-health, influence these.
Drawing on data from the British Household Panel Survey, this report fills these gaps. It looks at:
- Changes in employment and income trends during the 1990s
- Income and employment for individuals in their 50s compared with those for individuals of pension age
- How incomes changes in the years around retirement
- The relationship between an individual’s income in old age and his or her employment history.
The report is one of the few UK studies of incomes in later life to use longitudinal data as well as information about work histories.
As working lives shorten and retirement lengthens, there is concern over whether individuals make adequate provision to avoid poverty in old age. A study by Elena Bardasi and Stephen P. Jenkins of the Institute for Social and Economic Research examined the relationship between the risk of having a low income in later life and people's lifetime employment history. The study, based on data from the British Household Panel Survey (BHPS), found that:
- Although employment rates of older men fell sharply in the 1980s, they stabilised during the 1990s. Older women's employment rates increased during the 1990s.
- In the period immediately before retirement, people start to work less and earn less on average, but income still falls sharply in the year of retirement.
- Spending more years in paid work in total between the ages of 20 and 60 did not necessarily lower the risk of having a low income when aged 60+. Instead, the effect depended on the occupational group involved. A reduction in low-income risk was associated with more years of paid work for men in professional, and personal and protective services occupations, and for women in managerial, professional, technical and clerical occupations.
- Holding the number of years spent in an occupation constant, the risk of having a low income when aged 60+ varied according to occupational group. For men, those groups with small low-income risks were professional and clerical occupations, whereas, for women, they were professional, clerical and managerial occupations.
- Working fewer than five years between the ages of 50 and 60 raised the chance of having a low income in later life for men who had spent more of their working life in clerical, craft, personal and protective services, and sales occupations. By contrast, for women, low labour market participation between the ages of 50 and 60 had little association with the risk of having a low income when aged 60+.
- Household type and marital status were more important determinants of low income in later life for women than for men. Women aged 60+ living without a partner had a substantially higher risk of low income than women aged 60+ living with a partner, even if they had worked for much of their life.
Over the last two decades, employment rates have fallen - particularly for older men - reflecting a mixture of voluntary decisions and constrained choices. Pensioner incomes have improved relative to the rest of the population, but inequality among pensioners has also increased and poverty in old age remains a problem. In the political arena, there has been consideration of the extent to which government should increase the basic pension to support living standards in old age or instead make more use of means-testing (as in the Minimum Income Guarantee for pensioners). There is also ongoing discussion of the form that second-tier pensions should take.
There are important family- and gender-related aspects of these issues, including differences in poverty rates between older men and women, and differences in lifetime work attachment. For women, dependence on income from partners or the government remains particularly important, and the conjunction of work career patterns and the possibility of divorce (the risk of which continues to rise) increases the chance that women will fail to secure adequate retirement pension rights.
Against this background, the study aimed to fill some of the gaps in knowledge about the relationships between older people's income, retirement and work history using BHPS data. Three questions were addressed:
- Have the trends in employment and income cited earlier continued during the 1990s, and to what extent are employment and income for individuals in their 50s similar to those for individuals of pension age?
- How does income change in the years around retirement?
- What is the relationship between an individual's income in old age and his or her employment history during his or her working life? Given the growing prevalence of withdrawal from the labour market prior to state retirement age, a question of particular interest under this general heading is the extent to which early exit is associated with having a low income in later life.
Employment and incomes of older people: the 1990s
Taking the 1990s as a whole, trends in employment for older male workers over the decade were not a continuation of earlier ones - instead, the decline levelled out. Among women aged 50-74, there was an increase in employment rates, from around 30 per cent in 1991 to around 40 per cent in 1999. Most of this increase was accounted for by the increasing participation rate among women aged 50-54.
Although older people continued to be over-represented at the bottom of the income distribution for the population as a whole, this was combined with greater income growth among individuals of pension age than for the population as a whole. Among pensioners, income inequality increased over the period and it remained the case that better-off pensioners were most likely to derive significant income from occupational and personal pensions and investment income, while the worse-off pensioners relied mostly on the state retirement pension and other benefits.
Income changes around retirement
Among men aged 60+ and women aged 55+, economic activity rates and work hours declined in the years leading up to retirement, as did household income. There was, however, a marked income fall in the retirement year. This description also characterised the experience of those who retired in their 50s. Thus, despite the greater prevalence of 'bridge job' strategies that several commentators have pointed to, it remains the case that, in the years surrounding retirement, income falls sharply rather than gradually. Large differences between the sexes were not apparent when looking at changes in household incomes around retirement, but there were clear differences between the sexes when looking at changes in the personal incomes of the people who retired. On average, women fared worse. After retirement, fewer women received occupational or personal pension income than men and, among those who did receive some, the amounts were smaller. In addition, on average, women received lower state retirement pensions in their own right than men did. The likely explanations for this are that women were less likely to receive income from the state earnings-related pension scheme and that a significant number of married women had elected to pay National Insurance contributions at a reduced rate when this option was available.
Low income in later life: work history matters
Work history matters for income in later life, most obviously because it is during the working life that entitlements to state, occupational and personal pensions are built up, together with the accumulation of other financial assets.
Spending more time in paid employment between the ages of 20 and 60 was not necessarily associated with a smaller risk of low income for men and women aged 60+. Individuals with low earnings may have had to work longer in order to maintain their already low income. Instead, the risks of low income in old age were closely related to the occupational groups that men and women had worked in, and to the continuity of employment in each occupation. For both men and women, a smaller risk of a low income was associated with having worked more years in occupational groups with higher earnings and likely to offer occupational pension schemes. The specific occupational groups differed between the sexes however. Women benefited from having spent more of their working life in managerial, professional, technical and clerical occupations, whereas, for men, the relevant occupations were the professional and personal and protective services ones. Surprisingly, the number of years spent in self-employment or part-time employment did not increase the chances of having a low income in old age, once a range of other influences had been controlled for.
Holding the number of years spent in an occupational group constant, the risk of having a low income when aged 60+ was lower for some groups than others. For men, those occupational groups with small low income risks were professional and clerical occupations, whereas, for women, they were professional, clerical and managerial occupations.
Men: early exit effects depend on occupation
For men who worked in particular occupational groups, working fewer than five years between the ages of 50 and 60 (early exit) increased the chance of having a low income when aged 60+, whereas for other occupational groups - typically more highly skilled - this was less so. The reduction in low-income risk associated with working more years in clerical, craft, personal and protective services, and sales occupations was completely offset if the man left the labour market early. For example, working 30 years in a craft occupation was associated with a probability of low income of 41 per cent if the individual worked throughout his 50s, but 66 per cent if the individual worked fewer than five years in his 50s. The early exit effect was even larger for personal and protective services occupations: working 30 years in these occupations was associated with a probability of low income of 9 per cent if the individual worked throughout his 50s, but 67 per cent if the individual worked fewer than five years in his 50s. This large effect is likely to reflect two influences: early exit may have hindered the accumulation of savings and pension contributions and thence entitlements; it may also be that the men who exited early were those who were earning less and that lower earners within each occupation were more likely to lose their jobs.
By contrast, for women, early exit was not associated with sizeable differences in the probability of having a low income when aged 60+. An explanation for the lack of association is that low or intermittent labour market attachment was common throughout the women's working lives (not just in their 50s), whereas, for men, its prevalence was relatively high only between ages 50 and 60.
Women: household type and marital status
Particularly relevant for women's low-income risks were her household type and marital status: women living without a partner were far more likely to have a greater risk, other things equal. Older women without a partner - divorced women in particular - had substantially higher probabilities of having a low income when aged 60+ than women living with a partner, even when looking at women who had been continuously in paid employment between the ages of 20 and 60. By contrast, among men in continuous employment, low-income rates did not vary by household type. This result, together with the fact that early labour market exit was not associated with women's low-income risks, suggests that for women having a partner with a good work history (and gaining access to pension entitlements and other financial assets by that route) may be more important for income in later life than what she did during her own working life.
The findings suggest, first, that measures to address the issue of early withdrawal from the labour market are most relevant to the adequacy of retirement income for men rather than women (for whom issues of participation over the whole working life are more relevant).
However, second and importantly, the findings indicate that whether early exit from the labour market raises the risk of low income in later life is contingent on the occupational group(s) in which a man worked. Given the occupational heterogeneity within these broad categories, early exit effects are likely to vary even more than the research found. If policy initiatives are to be formulated to address any adverse consequences of early exit, more disaggregated information is required in order to target them.
Third, by confirming the links between income in later life and work life history, the results underline the relevance of recent initiatives to enable more individuals to make better provision for their old age through the accumulation of assets in the form of pension rights (e.g. stakeholder pensions) and saving more generally.
Fourth, the findings also draw attention to a fundamental policy dilemma, i.e. how to design schemes that provide non-trivial pension entitlements while taking account of intermittent participation in the labour market and low labour market attachment - typically by women - and enabling the low paid to participate. (The new state second pension aims to address this issue.) Part of the policy tension arises because of the different emphases put on different income distribution goals. Measures aiming to increase saving for old age through occupational or personal schemes may raise individuals' retirement incomes in absolute terms - and hence reduce the incidence of low income in old age - but at the same time may also replicate working life differences, to the extent that entitlements are more or less directly related to contributions (funded by earnings etc.). In this case, income inequality in old age may not be reduced. By contrast, a substantial increase in the flat-rate basic retirement pension would both raise incomes and reduce income inequality in old age - but would also be very costly. To the extent that there are constraints on reducing income inequalities in old age through changes in the basic state retirement pension (and associated measures directed at pensioners such as the Minimum Income Guarantee), the issue of reducing income inequalities in later life may really be an issue of reducing inequalities earlier in life in both the labour market and the burden of family responsibilities.
About the project
The analysis was based on samples from waves 1-9 of the BHPS (covering 1991-99). Information about incomes, personal characteristics and work history collected at each annual interview was combined with the lifetime work history data collected at BHPS waves 2 and 3. The analysis of the association between low income in later life and work history focused on men and women aged 60+, and having a low income was defined to mean having an income in the poorest third of the distribution of income among all persons aged 60+. 'Income' was total household income from all sources, adjusted for differences in household size and composition, and averaged over three years to smooth out transitory fluctuations. The analysis of the effects of occupation distinguished between the nine SOC90 'major groups'. A range of other samples and income definitions was used in the other parts of the report.