Copyright Sociological Research Online, 2002


Jay Ginn and Sara Arber (2002) 'Degrees of Freedom: Do Graduate Women escape the Motherhood Gap in Pensions?'
Sociological Research Online, vol. 7, no. 2, <>

To cite articles published in Sociological Research Online, please reference the above information and include paragraph numbers if necessary

Received: 7/2/2002      Accepted: 31/8/2002      Published: 31/8/2002


British women's increasing levels of educational attainment have led to expectations of gender convergence in employment patterns and hence in lifetime earnings and pension income. However, it is not clear how far losses due to motherhood vary with educational qualifications. A polarisation in mothers' employment is evident, according to whether women have high levels of educational and occupational capital and some writers have suggested that a young graduate mother is likely to maintain almost continuous full time employment, with minimal loss of lifetime earnings and no loss of pension income. This paper uses data from the British General Household Surveys from 1994-1996 to examine how the impact of childrearing on women's full and part time employment, earnings and private pension coverage varies according to educational level. Less than half of women with dependent children were employed full time in all educational groups, including graduates. Even among women graduates, only a third of those with a pre- school child were in full time employment. Motherhood substantially reduced women's earnings and private pension coverage at all educational levels, indicating the scale of losses in lifetime earnings and hence in private pension entitlements. The motherhood gap in private pension coverage was least for graduates and greatest for mid-skilled women but in view of the amount of the motherhood gap among graduates it is concluded that the pension protective effect of a degree for mothers has been overstated.

Private Pensions; Motherhood; Education


Despite recent rises in British women's employment participation, a substantial gender gap in earnings and in full time employment remains, especially for women who ever have children (Harkness and Waldfogel 1999). However educational level mediates the effect of motherhood on employment participation and hence pension acquisition.

According to an important and influential Cabinet Office publication (Rake et al. 2000), the 'motherhood gap' in pensions - the difference between a childless woman and a comparable mother - will be negligible for young women with a degree or equivalent qualification. This conclusion echoes the finding of a computer-simulation exercise by Davies et al. (2000), which indicated that for women with high educational qualifications motherhood would typically incur no pension loss, as their employment would be almost continuously full time across their reproductive years. These authors suggest that for a typical graduate mother; 'employment is hardly perturbed by bearing two children' and that 'Part time employment is limited for the mid-educated and negligible for graduates' (Davies et al. 2000: 297). This study also concluded that for mid-skilled women, the effect of motherhood on full time employment is less than for low-skilled women. For the typical graduate woman, the loss of lifetime earnings due to motherhood is estimated to be trivial and the pension loss zero: 'The graduate mother of two forgoes ..... just 4 per cent of her potential earnings' (: 300) and she 'suffers no cut in pension relative to the woman with no (or one) child' (: 303). The earnings losses are estimated as rather more for mid-skilled mothers (47 per cent) and substantial for low-skilled (58 per cent), with correspondingly greater adverse effects on pensions. Rake et al. (2000: 85) drawing on the same computer simulations, conclude that there are minimal effects of childbearing for graduate women: 'high-skilled mothers of two are estimated to remain continuously employed, with one year of part-time work following the birth of their second child'.

The simulation exercise by Davies et al (2000) is valuable in illustrating the consequences of motherhood for women's incomes in later life and how these are likely to vary with educational level. The simulations were not intended to reflect average or representative women, but to provide typical scenarios. Nevertheless, the findings have been widely reported and influential, implying that a diminishing proportion of women will experience pension penalties due to motherhood in the future, as the proportion of graduate women rises from the current 20 per cent.

The article is organised as follows. First we consider relevant research on motherhood and employment and second, we outline how mothers tend to lose out in private pensions and why these are increasingly necessary to avoid personal poverty in later life. The methods and data used in this article are set out. The claims made by Davies et al. (2000) and Rake et al. (2000) concerning graduate mothers' employment, earnings and pensions are then examined through three hypotheses. In the conclusions, we consider possible reasons why our findings do not support the claims made.

Mothers' Employment And Educational Level

While the economic activity rate of British working age men (16-64) declined from 96 per cent in 1901 to 84 per cent in 1996, that of working age women (16-59) has risen from 38 per cent to 72 per cent over the same period (Hakim 1996: Table 3.1). However, much of the increase was in part time employment. Women's full time employment rose between 1984 and 1990, from 35 to 40 per cent, but was still at this level in 1998 (OECD 1999: 276-7). In 2001, 37 per cent of working age women were full time employees (ONS 2002a).

Family care commitments, combined with the lack of affordable high quality care services in Britain, help to explain why the increase in women's employment has been mainly in part time employment of married women (Joshi 1989; Hakim 1993). In spite of changing attitudes towards gender roles, the bulk of domestic and unpaid caring work is still performed by women (Brannen et al. 1994; Gershuny et al. 1994; Murgatroyd and Neuburger 1997; ONS 2002b) constraining their employment opportunities while boosting those of married men (Gershuny 1997). The disruptive effect of domestic and other caring responsibilities on women's employment histories, combined with gender discrimination in employment, have led to women having different kinds of jobs from men, lower pay and working shorter hours (Joshi 1991). Childcare responsibilities have a long-lasting effect on women's employment. On return to the labour market after having children, women tend to experience occupational downgrading (Jacobs 1999; Dex 1987). This is especially marked when the return is to a different employer and to part time work, a pattern reflecting women's need to accommodate dual roles. Thus there can be irreversible handicaps in employment opportunities even after children have left home.

Among all full time employees, women's hourly earnings are still only 82 per cent of men's, and among non-manual employees only 70 per cent. Hourly earnings of women employed part time were even lower, only 61 per cent of the hourly earnings of men employed full time (ONS 2001). Even among full timers, women tend to work shorter hours than men and their weekly pay was 74 per cent of men's in 1999 (Rake et al. 2000). The earnings of married women, in particular, remain far below men's, both nationally and within couples (Arber and Ginn, 1995). Among British employees, 13 per cent of men but as many as a third of women earned less than two thirds of the median gross hourly wage (Millar et al., 1997). A fifth of employed women work less than 8 hours per week, most earning wages too low to allow contributions to either state or private pensions. Those who are low paid tend to remain low paid and there is a high turnover between low pay and non-employment (McKnight et al., 1998). Although the gender gap in full time hourly wages among those in their 20s and 30s has diminished since 1986 to about 10 per cent, it remains larger (around 25 per cent) among those over age 40 (Rake et al. 2000: 46-47). It is recognised that the years after age 40 are key years for pension building and are especially important in final salary pensions.

Older women's pension income is related to their earlier domestic roles as well as to their occupational class and ethnicity (Ginn and Arber 1999; 2000a). Therefore the persistence of a motherhood gap in terms of employment patterns and lifetime earnings has serious implications for women's retirement income in the future.

However, the impact of motherhood varies according to employment rights such as maternity leave and return arrangements, which in turn are associated with educational level and occupational class (Glover and Arber 1995) as well other factors such as household income, availability and cost of childcare. Thus for highly qualified young mothers, return to employment after childbearing is more rapid than for the less qualified (McRae 1993; Dex et al. 1996). Glover and Arber (1995) confirmed that in terms of gaps in employment, a polarisation according to occupational class was evident, especially when children were young. Moreover, the rate of full time employment rose more rapidly with the age of the youngest pre-school child among women in higher non- manual occupations than among women in manual occupations. Nearly half of professional mothers were employed full time when their youngest child was aged 5- 11 years compared with a fifth of manual mothers (Glover and Arber 1995). Since occupational class is closely associated with educational level, high qualifications can be expected to reduce the motherhood gap.

Developments since the 1970s in women's educational attainment and employment participation have led to expectations of gender convergence in working patterns and hence in lifetime earnings. But without such convergence, women's periods of non-employment, part time employment and low earnings mean their disadvantage in building an adequate retirement income will persist, exacerbated by the current shift in the balance of pension provision towards the private sector.

Pension implications of low earnings and part time employment

An independent pension entitlement is important for women since reliance on a husband's pension (as wife, then widow) is a hazardous strategy. Lifelong marriage is declining, with 40 per cent of marriages expected to end in divorce (Shaw 1999; Haskey 2002). Legislation on pension sharing at divorce, operative from December 2000 for both state and private second tier pensions, is a welcome recognition of wives' contributions to a marriage, yet it is uncertain how adequate wives' shares will be (Ginn and Price 2002). Even where a marriage lasts into retirement, there is no guarantee that a husband will be willing to share his private pension. Widows may benefit from a deceased husband's pension, but a third of men have no private pension to 'bequeath' and many more have only small amounts (Ginn and Arber 1999). Cohabitees, like women without partners, cannot rely on a survivor's pension since cohabitation confers no legal right to a partner's state or private pension.

Private pensions are increasingly important to later life income in Britain. Pension policy under both the Conservatives and the post-1997 Labour government has been to promote the private sector and curb the role of National Insurance pensions. The value of the basic pension (BP) for a non- married pensioner fell from 20 per cent of average male earnings in 1980 to about 16 per cent in 1999 and if price-linking continues, is estimated to be about 7.5 per cent by 2050, worth only 32 per week in 1999 prices (Falkingham and Rake 2001). The full basic pension (BP) for a lone (non-married) pensioner in 2002 was 75.50 per week, which was about 23 below the Minimum Income Guarantee (MIG) and over 30 below the minimum income considered necessary for a lone pensioner woman to avoid poverty (Parker 2000). The amount of the full BP is also at least 30 per week less than it would have been if earnings-linked since 1980. The State Earnings Related Pension Scheme (SERPS) has also been substantially reduced. Further, women's state pension age is to be raised from 60 to 65 for those born after 1950, phased in from 2010 to 2020. Women's loss in state pensions, if they are not employed up to age 65, will be considerable (Hutton et al. 1995). In addition many employers have already raised the pension age for women in their occupational pension scheme to 65, reducing the pension payable at 60.

An indication of the extent of state pension cuts is given by projections to 2050. The cost of BP and SERPS together has been projected to fall from 34bn to 26bn (in 1997 earnings terms) and the replacement rate of the combined BP and SERPS/S2P to fall from 37 to 20 per cent of average male earnings (PPG 1998). Spending on state pensions was projected to fall from 4.4 to 3.4 per cent of GDP (DSS 1998). Above-inflation rises in the BP in 2001 (+7.4%) and 2002 (+4.1%) slightly modify these projections.

The low level of state pensions will make it increasingly difficult to build a decent retirement income without having a substantial amount of private pension or other private income - which is less likely for women than for men. Between the mid-1980s and the mid-1990s, the contribution of private pensions to older people's income rose from 11 per cent to 17 per cent. The rise was greater for men (17 to 25 per cent) than for women (7 to 11 per cent) increasing the gender gap in private pension income (Ginn and Arber 1999). Employees have been encouraged to opt out of the State Earnings Related Pension Scheme (SERPS) and contribute instead to private pensions. These may be either occupational pensions or personal (individual money purchase) pensions, which since 2001 have included stakeholder pensions.

Research has shown that it is full time employment that is crucial for building private pension entitlements (Ginn and Arber 1996; 1998) and therefore financial well-being in later life. Part time employment is associated with low membership of occupational pension schemes, mainly because in the private sector part timers tend to work for employers who do not provide an occupational pension (Ginn and Arber 1993). Moreover, part timers who do contribute to a final salary scheme will receive a reduced pension. For example, two years employed at half time hours generally count as one year's scheme membership in calculating an individual's pension.

The lower hourly wages earned by part timers usually reduce the amount of second tier pension they can earn. Although low earnings early in the life course have no direct effect on final salary occupational pensions, employers are increasingly switching to defined contribution (or money purchase) schemes. In such schemes, the pension is reduced by periods of low earnings, disproportionately so if these occur early in the working life. Low earnings will have a similarly detrimental effect on personal pensions (including stakeholder pensions), compounded by the fact that employers generally make no contributions to these pensions. Charges in stakeholder pension schemes, although lower than in earlier forms of personal pension, will still absorb a substantial part of the fund at retirement. Charges, of course, apply to all contributors, but are more likely to result in a pension insufficient to avoid means-testing for those with interrupted employment and periods of low earnings (Davies and Ward 1992).

Because of the projected continuing decline in the basic pension (DSS 1998), the amount of second tier pensions will need to be increasingly large to lift the recipient above the level of means tested benefits. Estimates of stakeholder pension amounts in 2050 by Falkingham and Rake (2001) show that even hypothetical women with full time continuous employment on average female wages are likely to receive a pension income which barely exceeds the level of means testing (if pension policy does not change). Thus on present policies women who have periods of low earnings due to motherhood are unlikely to escape poverty in later life in terms of their own pension income. Individuals who are unable to build substantial private pension entitlements face living on the margins of poverty, dependent on family members or on means tested benefits.

In view of the decline in the value of state pensions in the UK, it is urgent to consider the extent to which working age women who have children will lose out in private pensions and to what extent obtaining qualifications protects women from the motherhood gap in pensions.

Aims and Methods

The aim of this paper is to assess the suggestion of Davies et al. (2000) and Rake et al (2001) that, for women who are graduates, motherhood has little effect on employment, earnings and private pension acquisition. In this article we question whether this optimistic scenario concerning graduate women's immunity from the effects of motherhood on full time employment and pensions is justified, by analysing a larger and more representative sample of women.

Data from the General Household Surveys (GHS) from 1994-1996 was used to examine women's employment, earnings and private pension coverage according to maternal status and educational level. The GHS, with a response rate of just over 80 per cent in the mid-1990s (ONS 1996), provides high quality information about a nationally representative sample of nearly 17,000 women aged between 20-59 living in private households in Britain, including 3212 graduate women. This compares with under 4,000 women, of whom 253 were graduates, in the 1994 British Household Panel Survey (BHPS) used by Davies et al. (2000) (see Appendix 2). The analysis is restricted to the prime working age range from 20-59 and excludes full time students.

Three hypotheses concerning graduate women are examined:

  1. The effect of motherhood on participation in full time employment is minimal.
  2. The effect of motherhood on earnings is minimal.
  3. The effect of motherhood on private pension membership is minimal.

Women were grouped according to their occupational class (current or last job) into six socioeconomic categories and according to their highest educational qualification into five categories: Degree/equivalent or above, A levels, O levels/GCSE, other qualifications and none. Full time employment includes all those whose usual hours of work were at least 31 per week, whether employees or self-employed. Membership of private pensions includes occupational and personal pensions (APPs and pensions for the self-employed).

A lifecourse variable was constructed for women, with six categories:

  1. those aged under 35 who had never borne a child
  2. those with a youngest child aged 0-4 in the family
  3. those with a youngest child aged 5-9 in the family
  4. those with a youngest child aged 10-15 in the family
  5. those who had borne a child but had no child aged under 16 in the family
  6. those aged over 35 who had never borne a child.

This variable is designed to capture successive stages in the lifecourse. Among women under 35 who had never had a child, roughly three quarters will pass through the four stages of motherhood, while a quarter can be expected to remain childless, moving straight from category 1 to 6. Age 35 is used as a marker to distinguish those for whom childbearing has become relatively unlikely.

Since the GHS data is cross sectional, part of the difference found between women in different lifecourse categories may reflect cohort differences. However, conducting GHS analyses according to lifecourse category provides some control for age and hence for any such changes. Women aged under 35 will have entered the labour market between 1980 and 1995 so their employment and pension circumstances reflect the momentous changes of the 1970s in policy, legislation, expanded educational opportunities and attitudes towards sex equality and women's roles. To this extent, comparing women with similar educational attainment aged under 35 in the first two lifecourse stages (childless and with a youngest child aged under 5) can be treated as indicating the effects of motherhood. The mean ages of these two groups were 26.3 and 28.5 years old.

Although the duration of employment or pension scheme membership cannot be directly inferred from this GHS cross-sectional data, we have nevertheless found a reasonably close correspondence between the cross-sectional gender difference in occupational pension scheme membership, where women's coverage rate is 63 per cent of men's (Ginn and Arber 2000b) and the gender difference in years of membership, where women have 67 per cent of men's membership duration (Walker et al. 2000). This gives some confidence that cross-sectional pension coverage data can indicate relative duration of coverage.

Lifecourse Stage and Women's Employment

Women's rates of employment and of full time employment differ according to maternal status and age group (analysing all childless women together) (see Figure 1). Women's total employment rates were little affected by maternal responsibilities, probably because part time employment can often be accommodated (Figure 1a). The employment rate of childless women and women with children aged over 10 was high among those in their 40s but fell to only 50 per cent among all women in their late 50s, a key age for pension building. Reduced employment due to having young children was more marked for women aged under 30 than for those in their 30s (Figure 1a).

Figure 1a Percentage of women employed by maternal status and age group

Figure 1b Percentage of women employed full time by maternal status and age group

Full time employment, in contrast, was reduced considerably by women's childcare responsibilities (Figure 1b). Among mothers of children aged under 10, under a fifth were employed full time at all ages. The full time employment rate was still under a third for all age groups of women with children aged 10-15 and was under half where women had children aged 16 and over. Full time employment of childless women peaked among those in their late twenties at 80 per cent, declining thereafter, more steeply from the late forties to only 40 per cent in the early 50s and under 30 per cent in the late 50s. This suggests that few women in their 50s are in an employment position that is conducive to maximising their pension entitlement.

In order to assess the effect of motherhood on the employment of women with different levels of qualifications, women's employment was analysed by lifecourse category, controlling for educational level. Since accumulating adequate private pension entitlements depends mainly on full time employment, this is the main focus of our discussion. Table 1 and Figure 2b show a dramatic reduction in full time employment among mothers in each educational group. In each of the four stages of motherhood, under half of graduate women were employed full time and among those with children aged under 5 less than a third were in full time employment. This casts doubt on the models of lifetime earnings and pensions of graduate mothers estimated by Davies et al. (2000), suggesting graduate mothers' losses may have been seriously underestimated.

Table 1: (a) Percentage employed and (b) percentage employed full time* by lifecourse category and educational level. Women aged 20-59
(a) % employed (b) % employed full time
Degree+A levelsO/GCSE OtherNone Degree+ A levelsO/GCSEOtherNone
All women81767265535550393322
Nev child, <3592919081638482796952
Child 0-4705451422629211512 6
Child 5-981726958434021192110
Child 10-1587847877634733302920
Child 16+75767674584647403725
Nev child, 35+81818367576971635137
Cramer's V0.210.330.310.300.240.410.540.490.370.24
Under age 35:
Never child92919081638482796952
Child 0-4695350402531211510 5

p<.001 at each educational level
* Full time employment is defined as 31+ hours per week usually worked
Full time students excluded from the analysis
For base numbers see
Appendix 1
Source: General Household Surveys 1994-6 (authors' analysis)

Figure 2a Percentage of women employed by life course category and educational level

Figure 2b Percentage of women employed by life course stage and educational level

The impact of having children varied according to educational level. The strongest association between lifecourse category and full time employment was for women with A levels (Cramer's V 0.54), followed by those with O level/GCSE (Cramer's V 0.49) (Table 1b). The effect was less for those with a degree (Cramer's V 0.41) or 'other' qualifications (Cramer's V 0.37) and was least for those lacking any qualifications (Cramer's V 0.24).

Comparing the first and second lifecourse stages only for women aged under 35 (those who had never had a child and those with a child aged 0-4, see last three rows of Table 1b), the association between motherhood and declining rates of full time employment was greatest for women with O levels/GCSE (full time employment reduced from 79 to 15 per cent, Phi 0.65). Women with A levels/equivalent or 'other' qualifications showed a stronger association between motherhood and reduced full time employment (Phi 0.60 and 0.62 respectively) than among women who were graduates (Phi 0.52). For graduates the reduction in full time employment was nevertheless substantial, from 84 per cent to 31 per cent. Thus although the effect of motherhood on full time employment was least for graduate women, the first hypothesis, that the effect of motherhood on full time employment is minimal for graduate women, is not supported.

Lifecourse Stage and Women's Earnings

Both lack of earnings and the level of earnings are relevant to private pension building. Measuring the average earnings of all working age women, including those not employed, can provide an indication of the effect of motherhood on the eventual amount of a woman's private pension, if any. Mothers had substantially lower median earnings compared with childless women (Table 2, last column). Median weekly earnings were zero for those with a youngest child aged under 5 and 42 for those with a youngest child aged 5-9, 80 for those with a youngest child aged 10-15, 69 for those with no children under 16 at home, compared with 196 and 162 for the two age groups of childless women (under 35 and over 35 respectively) (see Table 2). Thus median earnings of mothers of children aged 5-9 were only 21 per cent of those of childless women aged under 35, while the proportion rose to 41 per cent for mothers of older dependent children. Compared with the median earnings of all men aged 20-59 (250 per week), mothers of children aged 5-9 earned on average only 17 per cent. The earnings of mothers of the youngest children, aged 0-4, were clearly an even smaller proportion of the earnings of childless women or men.

Table 2: Median gross earnings in per week, by lifecourse category and educational level. Women aged 20-59, including those not employed*
Degree+A levels O/GCSEOtherNoneAll
All women22013185601477
Nev child, <35 27219117915692196
Child 0-4120 15 0 0 0 0
Child 5-9150 75 49 29 0 42
Child 10-15 231115 88 7140 80
Child 16+ 191134100 9135 69
Nev child, 35+ 30021519212953162

*Full time students excluded from the analysis
Source: General Household Surveys 1994-6 (authors' analysis)

Figure 3 Median earnings by life course category and educational level

Whatever women's level of educational qualifications, median earnings of mothers of the youngest children were dramatically reduced compared with childless women aged under 35 (see Table 2, first two lifecourse stages and Figure 3). Graduates maintained higher earnings across all lifestage categories, as would be expected, yet even graduate mothers experienced an initial fall in earnings to less than half that of childless graduates, on average. The median weekly wage of childless graduates aged under 35 was 272, compared with 120 for graduate mothers of children aged 0-4, reducing potential current pension contributions. The earnings of graduate mothers 'recovered' after the first two stages of motherhood, on average to 231 per week when children were aged 10-15 but 'fell' again to 191 among those whose children were all aged over 16; this fall may be an age cohort effect, since many graduates with children aged over 16 were in their 50s, an age group with a lower employment rate (see Figure 1). The highest median earnings were received by childless graduates aged over 35, 300 per week. Thus even for graduates, the lifetime earnings of mothers are likely to be substantially less than for childless women. We conclude that the second hypothesis, that the effect of motherhood on earnings is minimal for graduate women, is not supported.

The analysis so far has shown that obtaining a degree, while increasing women's chance of full time employment and high earnings, does not allow women to escape the adverse effects of motherhood on full time employment and earnings. In the next section, the effect of motherhood on private pension coverage is analysed according to educational level.

Private pension coverage

Before turning to the effects of maternal status on pension building, the proportions of men and women with occupational and personal pension coverage are shown (Table 3). Among adults aged 20-59, including those not employed (except for full time students), 64 per cent of men and 38 per cent of women contributed to a private pension, mainly to occupational schemes (Table 3a). Among employees, 80 per cent of men and 56 per cent of women contributed to a private pension. The lower proportion of women with private pension coverage mainly reflects lower coverage for women working part time; only just over a third of part time women employees contributed to a private pension (Table 3b).

Table 3: Percentage contributing to a private (occupational or personal) pension scheme. Men and women aged 20-59
a) All adultsb) Employees
Has private pension:5164386880567137
Employee, occupational pension3340275159425526
Employee, personal pension1215 91721141611
Self-empld, personal pension 59 2
No private pension:4936623220442963
Self-employed 5 63
Not employed 231631
Col %100100100100100100100100
N=3485516817180382266311403112606325 4935

* Full time employment defined as 31+ hours per week usually worked
Full time students excluded from the analysis
Source: General Household Surveys 1994-6 (authors' analysis)

Among employees, private pension coverage varies according to educational level and to socioeconomic category based on the individual's current or last main occupation. Table 4 and Figure 4 show the proportions of men and women employees with a private pension by a) educational level and b) socioeconomic category (SEC), distinguishing between occupational and personal pensions in Figure 4a and 4b. As expected, gender differences among full time employees were much reduced within each SEC and educational level. Graduate women showed a clear advantage over less qualified women. There was little gender difference in pension coverage among full time employed graduates but a larger gender difference at lower educational levels (Table 4 and Figure 4a). However, women graduates employed part time were no more likely to have private pension cover (57 per cent) than full timers with no qualifications (59 per cent). This highlights the fact that even for the most highly qualified women working part time severely reduces the chance of contributing to a private pension scheme. Since the proportion of graduate women contributing to a personal pension was actually slightly higher for part timers than full timers, this may reflect graduate part timers' desire to contribute to a private pension in the context of the lesser availability of occupational pension schemes to part timers noted above.

Table 4: Percentage of employees contributing to a private pension by a) educational level and b) socioeconomic category. Men and women 20- 59
a) Educational attainment:
1. Degree/equivalent and above88768457
2. A level/equivalent81616845
3. O level/GCSE79577239
4. Other qualifications75506534
5. No qualifications73395926
b) Socioeconomic category (SEC):
1. Professionals/managers in large organisations93848671
2. Intermed non-manual/managers in small orgs84727957
3. Routine non-manual79547038
4. Skilled manual77526134
5. Semi-skilled manual66375126
6. Unskilled manual54214218

* Full time employment defined as 31+ hours per week usually worked
Full time students excluded from the analysis
Base numbers for education and SEC differ because of missing data on educational qualifications; among women employees 518 had missing data for educational qualifications
Source: General Household Surveys 1994-6 (authors' analysis)

Figure 4a Percentage of employees contributing to a private pension by education level, gender and hours of work

Figure 4b Percentage contributing to a private pension by socioeconomic category, gender and hours of work

Comparing Figures 4a and 4b shows that occupational class (SEC) is more important than educational level in predicting private pension coverage among employees. The stark class gradient in pension coverage can be seen in Table 4b, reducing from 84 per cent of women who were professionals or managers (SEC 1) to only 21 per cent of women who were unskilled manual workers (SEC 6).

The stronger association of private pension coverage with class (SEC), compared with educational level, indicates the extent to which the occupational downgrading that typically accompanies interrupted employment and periods of part time work can reduce a woman's initial educational advantage. Maintaining high occupational status and the associated fringe benefits is not guaranteed by possession of degree level qualifications. On the other hand, women who were able to work as professionals and managers in large organisations (SEC 1), despite working part time, retained much of their advantage, with a pension coverage rate (71 per cent) exceeding that of women employed full time in routine non-manual occupations (SEC 3) or in manual occupations (SECs 4, 5 and 6). However, maintaining a high occupational status is rare among part timers. While half of women aged 20-59 employed in routine non-manual occupations (SEC 3) worked part time, only 13 per cent of women employed as professionals and managers (SEC 1) did so. This group (part timers in SEC 1 occupations) represents only 3 per cent of all women part time employees, 1.3 per cent of all women employees and less than 1 per cent of all women aged 20-59. Thus only a tiny minority of women can accommodate the demands of motherhood by working part time in a high status occupational group. Moreover, even where part timers contribute to a private pension, their lower earnings compared with full timers will reduce the pension they obtain; the amount of the loss will depend on the duration of their part time working. Differences in private pension coverage by class and educational level arose almost entirely from membership of an occupational pension scheme (Figure 4). As discussed earlier, this type of pension is more advantageous than a personal pension.

Private pension coverage of working age adults represents the outcome of many labour market factors, including their employment participation, hours of work and employment status (employee or self-employed). Figure 5 shows private pension coverage according to age group and maternal status for all women aged 20-59, including those not currently employed. Among women who had ever had a child, pension coverage only exceeded a third among those aged over 40 whose youngest child was aged over 10, whereas childless women had much higher coverage. However, childless women had declining pension coverage rates over age 40, reflecting their falling full time employment rates with age (see Figure 1b).

Figure 5 Percentage of women contributing to a private pension by maternal status and age group

The relationship between women's lifecourse category and private pension cover is shown for each educational group in Table 5 and Figure 6. As expected, in each lifecourse category those with the highest qualifications were most likely to contribute to a private pension and at each educational level childless women had higher coverage rates than women who had ever had children. For example, among graduate women over two thirds of those who were childless (65 per cent of those under 35, 72 per cent over 35) contributed to a private pension, compared with about 55 per cent of mothers of young children. The reduction in pension coverage for mothers of young children was much greater for women without a degree, as indicated by values of Cramer's V and Phi in Table 5. The impact of lifecourse category was least for graduates (Cramer's V 0.11) and greatest for those with mid-level qualifications, O level/GCSE (Cramer's V 0.27), while the effect was intermediate for the remaining three educational groups.

Table 5: Percentage contributing to a private pension, by lifecourse category and educational level. All women aged 20-59
LifecourseDegree+A levelsO/GCSEOtherNone
All women6148413422
Never child, <356554583926
Child 0-455382822 7
Child 5-95438262310
Child 10-156246413519
Child 16+6048454027
Never child, 35+7270635135
Cramer's V0.
Under age 35:
Never child6554583926
Child 0-454372719 6

p<.001 at each educational level
Full time students excluded from the analysis
Source: General Household Surveys 1994-6 (authors' analysis)

Figure 6 Percentage of women contributing to a private pension by lifecourse stage and educational level

Restricting the analysis to a comparison of women aged under 35 with and without young children (last three rows of Table 5), the effect of on private pension coverage of having a child aged under 5 was least for graduates and those with A levels (Phi=0.11 and 0.16 respectively). Motherhood had the greatest effect among mid-skilled women, reducing pension coverage by over half, from 58 percent to 27 per cent (Phi 0.32). For lower skilled women the effect was intermediate (Phi 0.22 and 0.28).

This analysis does not support the third hypothesis, that the effect of motherhood on private pension coverage is minimal for graduates. However, the impact of motherhood is much greater for less qualified women than for graduates.

Summary and Conclusions

The analysis in this paper provides a cross-sectional picture of women's employment, earnings and private pension coverage in the mid-1990s. It cannot predict lifetime accumulation of pension entitlements with certainty since many of those not currently contributing to a private pension may have done so in the past and may do so in the future. Nevertheless, gaps in pension contributions, or contributions made from part time earnings, reduce the eventual amount of pension. The analysis thus provides an indication of likely differentials in pension outcomes between childless women and those who ever had children.

Contrary to the conclusions of Davies et al. (2000) and Rake et al. (2000), there is no support for the expectation that graduate mothers will maintain almost continuous full time employment throughout their lifecourse. Full time employment was under a third for graduate mothers of pre- school children and remained below half for those with older children. This indicates that even graduate mothers take several years, on average, out of full time employment. Since only a third of graduate mothers of children aged under 10 were employed full time, this implies that on average women with a degree lose nearly seven years of full time employment while their children are young.

Women employed part time had lower private pension coverage than full timers irrespective of educational level. Educational qualifications on their own were less important than occupational class in predicting private pension cover for part timers. Socioeconomic inequalities in pension coverage were mainly due to the unequal distribution of occupational pension scheme membership, with personal pensions spread more evenly. Mid-skilled women (with GCSEs or O levels) experienced a greater impact of motherhood on private pension coverage than less qualified women. Thus the intuitively appealing assumption of a linear relationship between the pension penalties of motherhood and level of qualifications was not borne out.

Graduate women, as well as having the expected advantage in pension coverage, were also the group whose pensions were least affected by motherhood. Nevertheless, pension coverage was reduced among graduates with a child aged under 10, from 65 to 55 per cent. Among employed graduate mothers of a child aged under 10, the majority worked part time, reducing the amounts of pension contributions they could make even if they belonged to a private pension scheme. The halving of median earnings among graduate mothers of young children indicates the extent of the pension contribution loss for this group due to a combination of non-employment, part time hours and occupational downgrading, a loss which will be reflected in lower pension entitlements at retirement. In all, the analysis suggests that even graduate mothers will experience a substantial loss of private pension entitlements compared with their childless counterparts.

The difference between the findings reported in this paper and the conclusions of Davies et al. (2000) and Rake et al. (2000) is likely to be due to several factors. First, the GHS provides a sample of working age women that is much larger than the BHPS (17,000 compared with 3,700). This allows the maintenance of adequate cell sizes when analysing by lifecourse stage and educational level (see Appendix 2 for a comparison of sample sizes of graduate women, which in the BHPS falls to only 21 graduate women with a youngest child aged 5-9). Second, while the GHS analysis has followed standard practice of defining employment as full time if weekly hours are 31 or more, the BHPS analysis may have used a lower (30 hour) threshold for full time employment, boosting the proportion defined as working full time. Similar levels of full time employment by lifecourse stage for graduate women are obtained in the GHS and BHPS if full time is defined as 31+ hours (see comparison in Appendix 2). Third, the GHS analysis used 'hours usually worked' per week, while the BHPS analysis may have included irregular overtime hours, which could increase the proportions classified as full time. In terms of fringe benefits, such as access to an occupational pension scheme, it is contractual hours rather than overtime that is relevant. Fourth, the computer simulation technique used by Davies et al. (2000) may have magnified any over-estimation of the probability of full time employment in the BHPS.

While recognising the value of simulation of lifetime incomes of women with different levels of education and maternal history, we believe it is important to raise questions concerning the conclusions of such an exercise where these appear to be contrary to findings from analysis of large representative cross-sectional surveys. Our analysis suggests that the pension-protective power of a degree for women has been exaggerated.

The persistence of a motherhood gap in private pensions for the majority of working age British women, including those with high qualifications, has implications for pension policy. Increasing emphasis on private pensions reinforces these pension penalties of motherhood. On the other hand, a basic state pension set at an adequate level can minimise such losses for working age women, as well as lifting out of poverty those older women whose pension building has been restricted by family caring commitments.


Click here for Excel file of Figures shown above

Appendix 1: Base numbers of women by lifecourse category and educational qualifications
Degree+A levelsO/GCSE OtherNoneAll% in each lifecourse category
Lifecourse Category:
Never child, <35831586921294171280316%
Youngest child 0-46244031204547687346520%
Youngest child 5-9375224659354557216913%
Youngest child 10-15363154565324742214813%
Youngest child 16+65025310117332479512630%
Never child, 35+3711173391963991422 8%
All aged 20-593214173746992448503517133100%
% in each qualifications group19%10%27%14%29%100%
All aged <351303102123219549886587
% in each qualifications group20%16%35%14%15%100%

Source: General Household Surveys 1994-6 (authors' analysis)

Appendix 2: Percentage of graduate women employed and employed full time by lifecourse stage, comparing GHS 1994-6 and BHPS 1994
% employed% employed FTN
Never child <35929684928183189
Youngest ch 0-4706829403162457
Youngest ch 5-9818140624637521
Youngest ch 10-15878547675236327
Youngest ch 16+759046695765029
Never child 35+818469836937130
All graduate women81865572573214253

" FT defined as 30+ hours per week
* FT defined as 31+ hours per week
Sources: BHPS 1994; GHS 1994-6

Thanks are due to Dr Kim Perren for the analysis of the 1994 BHPS used here


We are grateful to the Office for National Statistics for permission to use the General Household Survey data and to the Data Archive, University of Essex and Manchester Computing Centre for access to the data. The analysis is the responsibility of the authors alone. Thanks are due to Dr Kim Perren for analysis of graduate women in the British Household Panel Survey. The research was part of a project funded by the Economic and Social Research Council (Grant ref: R000271002).


ARBER, S. and GINN, J. (1995) 'The mirage of gender equality: Occupational success in the labour market and within marriage', British Journal of Sociology. 46,1, 21-43.

BRANNEN, J., MESZAROS, G., MOSS, P. and POLAND, G. (1994) Employment and Family Life. A review of research in the UK, London: Department of Employment.

DAVIES, B. and WARD, S. (1992) Women and Personal Pensions. London: EOC.

DAVIES, H., JOSHI, H. and PERONACI, R. (2000) 'Forgone income and motherhood: What do recent British data tell us?' Population Studies. 54: 293-305.

DEPARTMENT OF SOCIAL SECURITY (1998) Partnership in Pensions, Cm 4179. London: The Stationary Office.

DEX, S. (1987) Women's Occupational Mobility: A lifetime perspective. Basingstoke: Macmillan.

DEX, S., JOSHI, H. and MACRAN, S. (1996) 'A widening gulf among British mothers', Oxford Review of Economic Policy, 12,1, 65-75.

FALKINGHAM, J. and RAKE, K. (2001) 'Modelling the gender impact of British pension reforms', chapter 5 in J. Ginn, D. Street and S. Arber (eds) Women, Work and Pensions: International issues and prospects. Buckingham: Open University Press. Also at <>

GENERAL HOUSEHOLD SURVEYS (1994-6) Computer files held at the Data Archive, Essex University.

GERSHUNY, J., GODWIN, M. and JONES, S. (1994) 'The domestic labour revolution: a process of lagged adaptation?' in M. Anderson, F. Bechhofer and J. Gershuny (eds) The Social and Political Economy of the Household. Oxford: Oxford University Press.

GERSHUNY, J. (1997) 'Sexual divisions and the distribution of work in the household', chapter 13 in G. Dench, (ed) Rewriting the Sexual Contract. London: Institute of Community Studies.

GINN, J. and ARBER, S. (1993) 'Pension Penalties: the gendered division of occupational welfare' Work, Employment and Society, 7,1, 47-70.

GINN, J. and ARBER, S. (1996) 'Patterns of employment, pensions and gender: the effect of work history on older women's non-state pensions', Work Employment and Society, 10,3, 469-90.

GINN, J. and ARBER, S. (1998) 'How does part-time work lead to low pension income?' chapter 8 in J. O'Reilly and C. Fagan (eds) Part time Prospects. London: Routledge.

GINN, J. and ARBER, S. (1999) 'Changing patterns of pension inequality: the shift from state to private sources', Ageing and Society, 19, 3, 319-342.

GINN, J. and ARBER, S. (2000a) 'Ethnic inequality in later life: Variation in financial circumstances by gender and ethnic group', Education and Ageing, 15,1, 65-83.

GINN, J. and ARBER, S. (2000b) 'Personal pension take-up in the 1990s in relation to position in the labour market', Journal of Social Policy, 29,2, 205-28.

GINN, J. and PRICE, D. (2002) 'Can divorced women catch up in pension building?' Child and Family Law Quarterly, 14(2): 157-73.

GLOVER, J. and ARBER, S. (1995) 'Polarisation in mothers' employment', Gender, Work and Organisation, 2,4, 165-79.

HAKIM, C. (1993) 'The myth of rising female employment', Work, Employment and Society, 7, 97-120.

HAKIM, C. (1996) Key Issues in Women's Work. London: The Athlone Press.

HASKEY, J. (2002) Personal communication.

HUTTON, S., KENNEDY, S. and WHITEFORD, P. (1995) Equalization of state pension ages: the gender impact. Manchester EOC.

JACOBS, S. (1999) 'Trends in women's career patterns and in gender occupational mobility in Britain', Gender, Work and Organisation, 6,1, 32-46.

JOSHI, H. (1989) 'The changing form of women's economic dependency', in H. Joshi (ed) The Changing Population of Britain. Oxford: Blackwell.

JOSHI, H. (1991) 'Sex and motherhood as handicaps in the labour market', pp. 179-93 in M. McLean and D. Groves (eds) Women's Issues in Social Policy. London: Routledge.

MCKNIGHT, A., ELIAS, P. and WILSON, R. (1998) Low Pay and the National Insurance System: a statistical picture. Manchester: EOC.

MCRAE, S. (1993) 'Returning to work after childbearing: Opportunities and inequalities', European Sociological Review, 9,2, 125-37.

MILLAR, J., WEBB, S. and KEMP, M. (1997) Combining Work and Welfare. York: Joseph Rowntree Foundation.

MURGATROYD, L. and NEUBERGER, H. (1997) A Household Satellite Account for the UK. London: The Stationery Office.

OECD (1999) Labour Force Statistics 1978-1998. Paris: OECD.

OFFICE FOR NATIONAL STATISTICS (1996) Living in Britain. Results from the 1994 General Household Survey. London: HMSO.

OFFICE FOR NATIONAL STATISTICS (2001) New Earnings Survey 1975 and 2000. London: The Stationery Office.

OFFICE FOR NATIONAL STATISTICS (2002a) Social Trends, no. 32, London: The Stationery Office. Also at <>

OFFICE FOR NATIONAL STATISTICS (2002b) The UK 2000 Time Use Survey, <>

PARKER, H. (ed.) (2000) Low Cost But Acceptable Incomes for Older People. A minimum income standard for households aged 65-74 years in the UK. Bristol: The Policy Press.

PENSIONS PROVISION GROUP (PPG) (1998) We All Need Pensions - the prospects for pension provision. London: The Stationery Office.

RAKE, K. with DAVIES, H., JOSHI, H. and ALAMI, R. (2000) Women's Incomes over the Lifetime. London: The Stationery Office.

SHAW, C. (1999) '1996-based population projections by legal marital status for England and Wales', Population Trends, London: Office for National Statistics.

WALKER, R., HEAVER, C. and MCKAY, S. (2000) Building Up Pension Rights, Research Report No. 114, London: DSS.

Copyright Sociological Research Online, 2002