Gender and Generational Continuity: Breadwinners, Caregivers and Pension Provision in the UK

The UK is considered a ‘male breadwinner/female part-time carer’ state due to men and women conforming to stereotypical gender roles within partnerships, and welfare policies reflecting and reinforcing this gender division. Using data from the General Household Surveys 2001 and 2002, this article shows that mothers continue to be markedly disadvantaged in participating in the accumulation of pensions compared to women who have never had children. This is mostly because they take on caring roles at the expense of paid work, but also because where women earn much less than their partners, they are more likely to depend on them for pension provision. Female breadwinners are likely to be low earners, and so, in contrast with men, their status as ‘breadwinner’ does not usually imply pension accumulation. Consideration of the impact of the institutional framework of pension provision requires an understanding of inequalities within couples and societal expectations of mothers’ caring responsibilities.


Introduction
Widespread population ageing and the imminent retirement of babyboomer cohorts in the developed world has led to an explosion of interest in pension systems and the financial and fiscal stresses that they are predicted to encounter.While receiving far less attention, social changes also contribute to the need for pension reform.These include changing family structures and gender relations, increases in relationship breakdown, and growth of female participation in the paid labour force (Pierson 2001).In the UK, the pension system was originally conceived as providing a pension to a household where men worked continuously in the paid labour force with a financially dependent wife (Beveridge 1942;Harris 2006).This has shaped the way that current pension benefits -whether public or private -are distributed in later life, with women over 65 having on average only 58 per cent of the income of men over 65 (Arber & Ginn 2004).Women who undertake care work during their adult lives are less able to be secure in their anticipation of sufficient income in old age, regardless of class or partnership status (Ginn 2003; Ginn & Arber 2002).
In this article, first the conceptualisation of the UK pension system as part of a 'liberal' welfare regime will be considered in the light of feminist work highlighting the importance of the organisation of work within the private household.The gendered impact of the current UK pension system will be reviewed.Then the extent to which the post-war breadwinner/caregiver paradigm persists among those currently of working age and the implications for pension provision will be investigated.While gender relations have not been and are not static in the UK (Crompton 1999;Williams 2004), change in the extent of male earnings' dominance and female provision of household labour has been very slow (Arber & Ginn 1995).
The UK: a Liberal, Gendered Regime As motherhood and marriage become more distinct, with increased risk of marital and partnership breakdown and an increase in the proportion of women living without partners for periods of the life course, there is an increased need for women to have independent pension entitlements.The structure and ideology of the UK pension system, designed to benefit those who have high and continuous earnings, becomes increasingly problematic (Ginn 2003).The government recognises that an ideological shift is necessary in response to social change (DWP, 2005;2006b;Pensions Commission 2005), but despite proposed reforms to the UK pension system which give more recognition to care work in the form of better credits to state pensions and fewer years in the labour market qualifying for a full basic state pension, narrowing the gender gap in pension provision will still depend mostly on increased labour market participation and higher earnings for women (Price, forthcoming 2007a).
Even among OECD countries, pension systems and their outcomes for different sectors of the population vary widely.In his analysis of these, as well as other differences in the provision of welfare, Esping Andersen (1990) famously classified nation states into three welfare regime clusters: 'corporatist', 'liberal' and 'social democratic', viewing the clusters as the product of political ideology, working class mobilisation, class alliances and institutional history.He perceived that in corporatist countries such as Germany, the pension system serves to maintain crucial class and status distinctions, with women encouraged to be homemakers, caring for children and families, and relying on marriage for financial support before and during retirement.Liberal countries, such as the United States, provide a residual system of state pension to prevent absolute poverty, and encourage market provision of pensions through tax incentives and subsidies.He considered that in such countries, social policy would not be specifically concerned with gender roles, in the belief that the market would determine labour supply and demand.Finally, in the social democratic countries such as Sweden, since full employment has been a major goal, the state purposefully facilitates women's participation into the full time paid labour force, ensuring the least gender inequality in pension provision.
Although Esping-Andersen conceived of 'welfare' as the result of a complex interaction between the state, the market and the family, the part of his thesis relating to the role of the family was seriously underde-veloped 1 .Since then, the articulation of the state/market/family nexus has been explicated by a number of writers, showing how the failure to think about the cultural norms and institutional structures that determine household relationships, social care, and particularly gender relations, deprived the theory of a crucial dimension (Daly 2000;Lewis 1992;1997;O'Connor 1996;O'Connor et al. 1999;Orloff 1993;Sainsbury 1994).
While not the only gender issue, essential to this dimension is the extent to which within the family unit women have traditionally been and remain dependent on men financially, sometimes characterised as the extent of the breadwinner/homemaker or breadwinner/carer model (Daly 2000;Lewis 1992;Pfau-Effinger 1999;Sainsbury 1994).This is not mere typologising.If social policies (e.g. for the provision of income in old age) are designed around assumptions of financial dependency particularly within heterosexual marriage, then the impact on those who do not conform or, who conform at a given time but are at risk of marital or partnership breakdown in the future, merits particular attention.
In this respect, 'dependency' need not refer to the inability, in an absolute sense, to form an independent household -the meaning attributed in the term 'welfare dependency' or when talking of women who have no ability whatever to be autonomous (Millar 2003).In a typical modern household in the UK, where a woman works part-time and her partner works full-time (Warren 2000), the couple are likely to adjust their standard of living to their joint incomes.Women thus become dependent on men for their lifestyle.The dependency is associated with the inequality of contribution -the lower the proportion of contribution, the more a woman has to lose if the relationship were to end (Ward et al. 1996a;1996b).Arber (1999) has argued that gender inequality in earnings within households 'forms the basis of fundamental inequalities in economic power between husbands and wives ' (1999:175), and that this pattern of inequality is resilient to change, varying little regardless of labour market 1 Esping-Andersen later accepted some of the feminist critique of his work, and reconsidered his thesis with much more attention paid to household structures and gender relations (Esping-Andersen, 1999).
participation, age or whether a woman has children.Apart from being a determinant of power in gender relations, a 'vicious circle is created which connects women's lack of economic power within marriage and their disadvantaged position in the labour market' (Arber 1996).Using data from the late 1980s, Arber and Ginn (1995) showed that despite increases in women's labour market participation and some convergence in gender pay gaps, it was rare for women in any occupational stratum to earn more than their husbands or partners, and that gender inequality in earnings was more pronounced within the household than in society generally.
The UK: a Liberal, Gendered Regime In Esping-Andersen's (1990) early classification the UK occupied a somewhat ambiguous position between liberal and social democratic regimes 2 (and see Daly 2000:10).However, as the impact of neo-liberal policies of the Conservatives (1979Conservatives ( to 1997) ) have been felt, many analysts now consider it a 'liberal' regime without much hesitation, with an ideological commitment to free-market economics, reliance on means testing, and a reduction of universalism that has continued under Labour since 1997.Adopting this categorisation, and much of the underlying welfare regime theory, O'Connor et al. (1999) undertook a detailed gender analysis of policy in four liberal regimes: the United States, Australia, the UK and Canada.The authors take issue with the notion that liberalism is not overly concerned with gender relations, arguing that liberalism privatises welfare either in the form of the market or in the form of the family, both of which have implications for gender relations, and that the specifics vary between liberal states.Fundamental to their conceptualisation is the liberal tradition of separation of state and family, reflected in minimal state intervention in family affairs.Paradoxically for women, this means that liberal tradition can result in policies that support traditional gender roles within the family, rather than being an instrument for their dis-2 Although he never considered that the UK was in any sense a 'social democratic' regime.In later work he characterised the UK as liberal (1999).
mantling.O'Connor et al (1999) emphasise the relative resilience of the UK gendered division of labour in policy logic and in the way that social policy drives provision through state, market and family: The shift in liberal ideology from gender difference to gender sameness is variously represented in the policy regimes of the four countries, with Britain holding more determinedly to the breadwinner-carer family model than the other three countries … The United States and to some extent Canada again show a clear and distinct pattern of encouraging families to have recourse to the market for support services, while Britain shows a clear pattern of encouraging the privatisation of need within the family, including continued dependence on former spouses (1999: 233) In earlier work, Lewis (1992) argued in an influential analysis that Britain was a 'strong' male breadwinner state meaning that relationships tended towards a male breadwinner/female caregiver paradigm, which both explains and is the result of women's low rates of participating in the labour market and their low pay, their tendency to work part time, the lack of child-care provision by state or market, and inequalities in access to social security rights.Sainsbury (1994) constructed two ideal types: the breadwinner/homemaker model with strong gender divisions of labour and the individual model where earning and caring are gender neutral.She too characterised the UK as approximating the male breadwinner type, albeit with some state recognition of the care work of mothers.And Daly, while eschewing the 'current fashion' for typologising (Daly 2000:12), after a careful and detailed comparison of the British and German welfare states, recognises among British women a high level of dependence on their husbands for income (Daly 2000: Chapter 7).
Reasons given for the strength of the male breadwinner model in the UK include an historical perspective whereby idealised versions of the male breadwinner/female carer family predominated at the time that the welfare state was forming (Lewis 1992), and the powerful separation of state and family, public and private, within liberal ideology (Lewis 1992;O'Connor et al. 1999).Liberal states provide meagre services for families and children, targeting those defined as most in need, reinforcing the norm of female caring in the private domain for most families (Sainsbury, 1994).
More recently, the debate has moved from considering whether policies and policy discourse support a 'male breadwinner' to considering the extent to which policy frames individuals as 'adult workers' -i.e.normatively supporting an 'adult worker' model: for couples a dualearning household (Lewis & Giullari 2005).Changes in this direction have been observed in many countries, including the UK.Many policies, however, remain ambiguous in their gendered aims and effects, and the synchronicity between social behaviour and policy logic is important.Policy logics that support 'adult workers' and the individualisation of benefits may leave women more disadvantaged than before if patterns of unpaid work within the household do not change.
Reductions in the amount of pension that can be inherited by a wife, altered incentives for private pensions to provide for dependents automatically, abolition of credits into the state pension for those caring for teenage children, and the creation of new incentives for low paid workers to enter private pension schemes are all recent or planned policy developments in the UK which frame adults as independent workers, but will leave women who continue in patterns of care and financial dependency within their partnerships relatively worse off.Similarly, lone mothers who are not able to work full time because of lack of affordable child-care or a scarcity of good local jobs, or do not see full time work as compatible with bringing up their children, are disadvantaged by pensions policy that assumes individual accrual of pensions through full-time labour force participation.

Engendering Pensions Analysis in the UK
The gender structure of pension provision in the UK is largely a result of the very strong explicit assumption in the Beveridge Report (1942) that the average household would consist of a breadwinning husband who would, if fit, be in full time employment for the whole of his working life, and his non-working wife who would provide unpaid care for children, for her husband, for the sick and for elderly relatives.The prevailing ideology indicated that pensions needed to be provided for the family unit rather than the individuals within it.This was so through the state system, and through a system of private occupational welfare schemes, where pension provision (with working men as the principal beneficiaries, and their wives on widowhood deriving widow's benefits through marriage) can be thought of as an extension of the 'family wage'.
The subsequent impact of neo-liberalism in the UK is reflected in two contradictory patterns.First, pensions underwent many reforms in the latter part of the twentieth century, the most abiding of which involve a rolling back of the direct involvement of the state in favour of the state's encouragement and subsidy of a larger and larger private market (Ginn & Arber 1994;1999).This means that labour force participation and, as importantly, sufficiency of earnings for those participating, became essential pre-requisites for a stable and reasonable income in old age.Secondly, with the marked separation of state and family that accompanies neo-liberalist ideology, and policies essentially of 'non-interference' in the way that parents and couples live, there is minimal pension provision for women who take on caring roles.Proposed pension reforms to be legislated in 2007 will provide more formal recognition for care, but in keeping with the underlying ideology will continue to emphasise a residual role for state provision, and the importance of private pension contributions linked to wages in providing income security in old age.In any event, reforms will only be effective for cohorts retiring after 2010, and it will take decades for the benefits to be felt among older generations (DWP 2006b;PPI 2006).
Various elements of the overall system in the UK accord with liberal philosophy.The most significant of these are paucity of pensions paid directly by the state, state encouragement of the market in pensions, and the non-interference by the state in the division of labour within families.For example, there is very little state provided child-care (that which is provided is mostly aimed at 'poor' and 'problem' children), and the market has failed to deliver affordable childcare to low and moderate earners (O'Connor et al. 1999;Paull et al. 2002).Even with recently proposed reforms, state pension transfers and pension benefits for carers remain relatively small and their relative value will continue to steadily decrease to 2012 or 2015.Until then the gap between the basic state pension and the minimum income for means testing will grow 3 .This means that as long as well paid work and financial provision for the family is largely a male domain while part-time work (or non-employment) due to caring for the family and household is largely a female domain, the pension system will serve women poorly.Pension welfare through the state and private systems is therefore still modelled on a partnership in which marriage is for life and resources within it are shared equally.
Leitner ( 2001) proposed a framework for the consideration of the extent of sex and gender discrimination in European state pension schemes, along three dimensions: the biological (sex), the work behaviour dimension (gender), and the care dimension (gender).The mechanisms which she considered included access to schemes, basis of benefit calculation, number of coverage years, credits for unpaid care work (for children and other family members), the extent to which benefits derive from marriage alone, and universalism versus means testing.She concluded that whilst the UK basic state pension was fairly gender neutral (and here, she possibly underestimated its gender bias, see Evandrou and Glaser (2003)), the UK state earnings related scheme was highly discriminatory, compared with other countries.Ginn (2004) extended this framework to consider the role of private pensions in the system.She argues that private pension welfare entrenches and exacerbates gender inequality through its reliance on long-term secure employment and size of financial contributions to pension schemes, and that it is difficult to imagine a benignly designed private system that would not have this effect.
Even with proposed pension reforms, private pensions will continue to be the driver of gender inequality in later life, since basic state pension level will remain well below the poverty line, and additional pension (whether through the state second pension or private pension saving) will continue to be related to number of years in the labour market and heavily dependent on earnings levels (Arber & Ginn 2004; Pensions 3 The basic state pension increases only at the rate of inflation.The government proposes to link it to earnings growth, but only from 2012 or 2015 at the latest. Commission , 2006).The implication is that women's pension disadvantage in the UK depends largely on the extent to which gender roles remain divided4 .

Data and Methods
The analysis which follows combines data from the General Household Surveys 2001 and 2002.The General Household Survey is a multi-purpose cross-sectional continuous survey carried out by the Office for National Statistics, collecting information on a range of topics from people living in private households in Great Britain.As well as reasonably detailed information about current pension scheme participation, full marital and cohabitation histories are collected from those aged 59 and under, as well as parenthood histories from women in this age group.A clustered probability sample of about 13,000 households is selected, and each adult individual in the household is interviewed.The survey achieves response rates of approximately 70 per cent.
The combined dataset yields 10,314 men and 11,087 women aged between 20 and 59.Included in the data is information about the earnings of 5,772 partnered men and 6,141 partnered women aged 20 to 59.Many of these men and women are of course partnered with each other as this is a household survey.Detailed information about each person's partner has been matched with his or her own individual data, allowing for the construction of couple-level variables and couple-level analysis 5 .
Those aged between 20 and 59 have been classified into the four groups described below according to whether they are (i) not in the paid labour force, (ii) working but earning below the lower earnings limit for national insurance contributions, or, if they are earning above that limit, whether they participate (iii) only in the mandatory second tier pension scheme (either state or private) or (iv) have further private pension (an occupational pension and/or additional private pension).Women have been grouped into those who have never had children, those whose youngest child is under 5 or under, whose youngest is between 6 and 15, and those whose youngest is aged 16 or over.In the latter case, the child may still live at home but in most circumstances can be considered as not needing daytime care.Where partnered men and women of working age (20 to 59) are considered, spouses may be older or younger.
All those in paid work have been divided into earnings quintiles for each year of the survey.Partnerships have been classified according to their combined earnings.Those partnerships in receipt of any income from earnings have been divided into joint earnings quintiles.The cut points for earnings quintiles and relative median and mean contributions of partnered men and women to them are provided in Appendix 1.

Stratification in the UK Pension System
The UK system has been described as the most complex in the world (Pensions Commission 2004), and it is not intended to examine the detail of it here.Rather, a very brief overview follows of the components that result in large gender and class differentials.The description is of the pension system prior to anticipated legislation for reforms that will take effect variously from 2010 for future cohorts only.Some of the gendered effects described should be ameliorated by the reforms, and reforms are designed to incentivise low paid workers into joining private pension schemes which may well assist women.However several million women will be unaffected by these reforms due to their prospective nature; for younger women effects are anticipated in the long term, but even so they remain heavily gendered.For a discussion of gender implications of the reforms see Price (forthcoming 2007b).
There are two principle mechanisms by which retirement income becomes stratified in the UK system: the design of the system, and differences in earnings from paid work.There are essentially four hierarchical categories of pension provision: No pension: At its lowest, it is possible in the UK to accrue no pension provision at all by participating in neither the state system of national insurance, nor in any private alternative.This will apply to low paid workers who do not qualify for caring relief (mostly women), as well as wholly dependent women who are not carers, and whose male partners do not purchase pension provision on their behalf.It also applies to those who do not work for sufficient numbers of years at high enough pay to qualify for the state system, again, mostly women 6 .At best, women with no or low basic state pension provision can, if they remain married into retirement, claim entitlement to 60 per cent of their husband's basic state pension (maximum £50.55 per week in 2006) and a widow's pension.
The basic state pension: Although the basic state pension system is well developed, it is not universal, and even maximum benefits accrued are so low (maximum £84.25 per week in 2006) as to leave recipients eligible for means tested benefits (minimum level £114.05 per week for a single person) if they do not have a wealthier partner to rely on for income or other independent income.In 2005/6, only 30 per cent of women at retirement age 60 were entitled to a full basic state pension (DWP, 2005: 73).As well as being unlikely to have accrued the full basic state pension (Evandrou & Glaser, 2003), women are likely to spend part or much of their retirement alone: about 60 per cent of women over 65 are not married, and 80 per cent of women over 65 do not have a cohabiting partner at death (ONS, 2002;2004).
A mandatory second tier: Since 1978, a second tier of pension provided either through a state scheme or a private alternative is mandatory for workers earnings above the lower earnings limit 7 .The majority of employed individuals are contracted out of the state second tier scheme, with a national insurance rebate paid into a private (occupational or personal) scheme instead.Whether state or private, the second tier has historically provided a very low level of benefit in practice, and has been insufficient of itself (i.e.without a wealthier partner or other independent 6 It will become less likely after reforms that anyone accrues no pension.Minimum numbers of years for contribution will be abolished, and the number of years of either work or care credits that will be needed for a full basic state pension will be reduced to 30.Work will still need to be at rates above the lower earnings limit.7 The lower earnings limit in 2005/6 is £84 per week. means) to ensure a retirement without means testing for the majority of older people8 (Falkingham & Rake, 2001;Hills, 2004).This is especially the case for women, due to low earnings and gaps in participation in the paid labour force.
If pension participation is at the level of (i), (ii) and/or (iii), without more, retired individuals for many years into the future are likely to be dependent on means testing providing income barely above the official poverty line for the remainder of their lives.Women are most likely to fall into these categories.Seventy-two percent of those benefiting from means testing in retirement are women (1.2 million women) -more than one in five women, compared with about one in ten men9 .Approximately half of pensioners are currently entitled to means tested benefits but government estimates are that between 30 and 40 per cent of entitled pensioners do not claim, and means testing for the very poorest, known as Guarantee Credit, is not claimed by a fifth to a quarter of entitled pensioners -disproportionately women (DWP, 2006a).Also, entitlement to means tested benefits is jointly assessed.There is an explicit assumption that income within partnerships will be shared equally -the government has no interest in individual poverty behind the household door 10 .
Additional private pension: The private system, over and above the mandatory provisions of (iii) above, is the single largest determinant of inequality among pensioners in the UK.Until new reforms are in place, it is and has been the case that only those with sufficient earnings and in good employment have the option to belong to good occupational pension schemes with generous employers contributions, or the 'choice' to buy into other types of private schemes lacking any employer contribution 11 .Gender stratification occurs because of women's low levels of participation in full-time work, their low average earnings, and their higher risk of being in employment with poor pension provision.
Apart from these institutional features, the second major determinant of stratification is earnings.Even if earnings are sufficient to contribute to a private scheme, contributions will in general be proportionate to earnings, and so higher earners will accrue a much greater benefit.This advantage to high earners is accentuated by tax relief on private pension contributions, a subsidy at the expense of other taxpayers.The system of tax relief is not transparent and the distributive impact is difficult to quantify.Agulnik and Le Grand (1998) estimated that half of tax relief goes to the top 10 per cent of earners (over 80 per cent of whom are men) and a quarter to the top 2.5 per cent, a highly regressive distribution 12 .
Individuals' pension depends not only on how much they can contribute, but also on the tax relief allowed, which is greatest for high earners.
The gendered elements of these mechanisms are illustrated by the pension outcomes shown in Figure 1 and Figure 2. Figure 1 shows the proportion of men and women in the UK between the ages of 20 and 59 participating in the four tiers of pension scheme outlined.About 30 per cent of women in this age range are not in paid work, compared with under 20 per cent of men.A far greater proportion of women than men (10 per cent compared with 5 per cent) are doing paid work, but earning 11 Reforms take the form of auto-enrolment into a national scheme for those earning above (in 2006) £5,000 per annum.If employees do not opt out, employers will be forced to contribute 3% of wages, and 1% will be credited via tax relief on contributions.This policy development is likely to benefit millions of women, if they do not opt out of the national scheme.12 Since this work was done, the taxation of pension funds has changed so that the regressive effect might now be a little less extreme.below the lower earnings limit for national insurance contributions.While similar proportions (around 22 per cent) are participating either in the state second pension scheme or its mandatory private equivalent, a much higher proportion of men (54 per cent) has some form of additional private pension, either from their employer or an individual scheme, than women (37 per cent).But as Figure 2 shows, this inferior coverage is not the whole extent of the gender disadvantage: even if covered, women earn far less than men.In this graph, each box-plot shows the earnings distribution for men and women respectively within each category of pension scheme participation.The dark line shows median earnings for the category, while the top of the box shows the 75 th percentile and the bottom the 25 th percentile.The important items are those to the right of the graph -the mandatory level of second tier pensions, and then those with additional private pension over and above this mandatory tier.In both, the 75 th percentile of women's earnings is well below the median earnings for men in the same category of pension provision.Women by necessary implication are amassing much lower pensions.There is little difference in the earnings distribution of women with additional private pensions and that of men with only second tier pensions.

Pension Provision for Future Retirees
These data raise the issue of the pension prospects for future generations of retirees.It is well known that among the elderly population, women are much poorer than men with little independent pension and high dependence on means-tested welfare benefits (DWP, 2005).Are things changing for younger generations of women? Figure 3 and Figure 4 suggest not, and that for those who undertake care work of children, and particularly for the increasing proportions of lone mothers, there are real difficulties in accumulating private pension provision.First, in Figure 3(a), the median weekly earnings of mothers are shown as at 2001-2.There is a substantial difference in median income as between those women who have never had a child and those who have ever had a child, with more than half of lone mothers with children under five having no earnings, and those in couples with median earnings well under £100 per week.Lone mothers do not average over £100 a week until their youngest child is over 16 and even partnered mothers at no stage average more than £150 per week.
At these earnings levels, it is not surprising that the proportion of mothers making additional private pension provision shown in Figure 3(b) is very low, and much lower than those who have never had children.Among those with a very young child, barely 10 per cent of lone mothers and just 30 per cent of partnered mothers are making additional pension provision.As the youngest child grows older, the proportions increase, but for partnered women, the increases are not substantial, and for lone mothers, the proportion barely reaches above 30 per cent at its highest -for mothers with children who are no longer dependent.
Figure 4 illustrates that these differentials are largely related to childcare rather than age, and, although the data are cross-sectional, strongly suggests that things are not improving for younger cohorts of women.If they were, we would expect that younger women with children would be earning more and participating to a greater extent in private pension schemes than older cohorts, holding the age of the youngest child constant.. Figure 4(a) illustrates that this is not so.In each category of motherhood in each age group, younger women fare worse.Indeed, the median income for mothers under the age of 30 never exceeds £100 per   groups, compared with the much higher participating levels of those women who have never had children.These results are highly suggestive of the continued dominance of the breadwinner/caregiver model of partnerships, with its structural implications for the employment and pensions of all mothers, whether partnered or lone.In order to assess whether this is so, partnered households are examined in more detail.

Earnings Inequality Among Partners
That the contemporary breadwinner is as decidedly male as his predecessor is suggested by the cross-sectional results shown in Figure 5 and Figure 6.In Figure 5, the mean ratio of own earnings to joint earnings among partnered men and women has been calculated for each five-year age group.Men earn about two thirds of joint earnings across all age groups, varying from 64 per cent in their late 20s and late 50s to more than 72 per cent in their early 40s.The late 20s represents the most (although still not a very) equal period in men's and women's lives -before gender pay gaps are much in evidence and before many women have begun having children.The late 50s are more likely to represent a time of early or semiretirement for men while their (often younger) partners are still working.Their partners are also quite likely to be working full time, as children have grown up or left home.
The pattern for partnered women in Figure 5 as expected largely shows the converse, although not a precise mirror image because women are likely to be younger than their partners.Even in their twenties, women are not equal earners, providing on average only a third of joint earnings.This drops to less than 30 per cent in the late thirties.The highest average contribution, of just over 40 per cent, is evident only among those in their late 50s, as stated above, likely to be as a result of men's ill health or early or semi-retirement for other reasons, or having a husband over 65 who has retired at state pension age.There is little difference in the share of earnings between partnered women in their 20s and in their 40s, suggesting that there is no structural change towards greater equality in couples' earnings.The U-shaped pattern implies strongly that the dip from age 30-44 is related to child rearing and not cohort changes.Women age 30-35 for example are likely to be fairly similar to those aged 25-29 in terms of education, training, attitudes to motherhood and partnering norms.
In order to explore this issue further, Figure 6 controls for maternal history, but also controls for position on the joint earnings distribution, because it could be said that differences might be expected as between wealthier couples and those less well off, with dual earner families predominating at the top of the distribution.
Earnings differences within couples will reflect the gender pay gap and gender segregated labour markets.But as Figure 6 indicates by comparing childless women with mothers, these patterns of earnings' inequality between partners are to a large extent the result of child care and allied household responsibilities which women of all ages still take on to such a substantial degree in the UK.The solid line in each graph denotes partnered women who have never had a child.The next two lines represent the age of the youngest child in the family unit: either 0 -5 or 6-15.The final line shows that the youngest child is either still at home but over 16, or has left home -i.e., these are women who have had children,  but none of their children, save in exceptional circumstances, could be considered as in need of daytime care.
The impact of motherhood on inequality within couples is clearly shown.In each graph the solid line floats well above the others.In each graph, women who have never had children (which will of course be a diminishing group with age) are contributing between 40 and 50 per cent of joint earnings on average within each age group (an even higher percentage in the lowest joint earnings quintile).Whilst clearly not representing equality between the sexes, it is, in historical context, a reasonably close approximation.Interestingly, there is no really noticeable dip among childless couples with age, which would suggest that changes in society in the sense of expectations of dual earning among couples are impacting as much on the financial relationships between mid-life couples as younger couples.
The effect of having children, and particularly the different effect of children on inequality dependent on the age of the youngest child, depends to some extent on which earnings bracket a couple falls into, although the broad picture is similar across joint earnings' quintiles.Women with children under 16 generally contribute less than a third to joint earnings, and in some cases much less than that.The lowest quintile apart, women who have ever had a child do not in any age group contribute more than 35 per cent to joint earnings, and there is no evidence here to suggest change for younger cohorts.The lasting impact of childbearing on the financial dependence of partnered women appears to be pervasive.
The differences in proportions among the women in households at different levels of joint earnings are surprisingly slight.An exceptional pattern is shown only among those in the lowest quintile, where women without children are on average equal or better contributors to joint earnings in each age group shown here, and those with children over the age of six also contribute substantially to joint earnings.For couples in the lowest earnings quintile, there is greater reliance on the woman's contribution.This is most likely a 'needs must' situation -with relatively low earning women partnered with men who are not earning or are themselves low earners.

Inequality and Pension Provision
It is not simply descriptive to describe a relationship as one of 'breadwinner/caregiver' or 'breadwinner/part-time-worker-caregiver'.The description reflects gender relations and gender culture which in turn are influenced by and influence gender identity.Personal identity in this sense is forged within a normative framework of gender relations.In this context, decision making (for example about child-care options or the extent of paid work) reflects personal identity and preference but also reflects external constraints, for example financial constraints (Duncan et al. 2003;Himmelweit & Sigala 2004;Williams 2004: Chapter 4).
Although largely unexplored in sociological analysis, it is reasonable to suppose that the construction of the 'breadwinner' identity carries with it a sense of obligation and/or preference to acquire pensions for the future provision of self and dependants.Conversely, economic dependence within such a relationship may carry the implication that a partner will provide a pension in the same way as living expenses are provided for.If this is so, then inequality within relationships will influence pension provision, with breadwinners more likely to make pension provision and dependents less likely to do so, simply by reason of the gendered relationship.But any such norms will be constrained by external circumstances -for example, if there is insufficient income to contribute to a pension.
To elucidate the relationship between inequality within partnerships and the making of provision for additional private pension during the working life, Table 1 and Table 2 present the results of a multivariate logistic regression analysis for partnered men and women aged 20 to 59.The model investigates who is most likely to have additional private pension coverage over and above the mandatory tier -that is the odds of being in category (iv) of pension provision defined earlier in this article as opposed to being in categories (i), (ii) or (iii).The results show the odds ratios for being in this privileged category relative to a reference category for which the odds have been defined as 1.In Model 1, the odds ratios are shown according to the extent and direction of earnings inequality within a relationship, with reference to the category '40% -60%', who for these purposes are considered 'equal' earners, while age group is controlled.Model 2 also controls for individual earnings quintile and Model 3 for joint earnings quintile.In Model 4, maternal status for women and the presence of children in the family for men are controlled, together with educational qualifications.This is to test whether, apart from their influence on individual earnings, joint earnings and degree of inequality within partnerships, these variables also have any independent impact on participation in pension provision, and whether they affect the association between inequality and pension scheme participation.
The primary question of interest here is the effect of inequality of earnings within couples on participation in voluntary additional private pension schemes.Although these models contain much other information, this analysis will comment only on the coefficients across models for the first category, that is the percentage of joint earnings (i.e. the degree of earnings inequality).As indicated on these tables, only 11 per cent of men contribute less than 40 per cent to joint earnings, compared with 63 per cent of women.While over a third of men contribute more than 80 per cent to joint earnings, only 9 per cent of women provide this degree of 'breadwinning'.
Men in this highest category of breadwinning are more likely to be high earners -this is shown in Table 1 by the lowering of the odds ratio in Model 2 once personal earnings are controlled from 0.97 (ns) to 0.73 (p<0.01).This is not the case for women, as shown in Table 2. Female breadwinners are much more likely to be low earners.This is shown by the almost doubling of the odds ratio for women providing over 80 per cent of joint earnings from 0.4 in Model 1 to 0.76 in Model 2, when the strong association between low earnings and low pension scheme participation is taken into account.
Men who provide more than 80 per cent of earnings find themselves (as a result of low or no contributions from their partners) slightly lower on the joint earnings distribution than men whose partners contribute more to the household, with the odds ratio for this group for contributing to additional private pensions changing from 0.73 (p<0.05) in Model 2 to 1.21 (ns) in Model 3 once joint earnings are controlled (Table 1).But women providing 80 per cent plus of earnings as well as being low earners themselves are also much more likely to be in the low end of the joint earnings distribution.Thus it is only when both personal earnings and joint earnings are controlled that we see the odds ratio for the pension contributions of these 'breadwinning' women jump from 0.76 to 1.63, relative to an equal earner (Table 2).
In Table 2, Model 3 shows that holding personal earnings and joint earnings constant, the degree of earnings inequality in a relationship has a significant impact on whether women contribute to pensions.Women who are more financially dependent are less likely to contribute to additional private pensions.For women earning 0-20% of joint earnings, the odds are only a third of the odds for equal earners, and for those earnings between 20% and 40% of joint earnings, the odds ratio is 0.8.All other things being equal, women who have assumed the role of major breadwinner are more likely than equal earners to make additional pension provision.But all other things are not equal -the starting odds ratio for women in couples where they are the major breadwinners is only 0.4 because these effects are counterbalanced by their higher probability of low earnings and low joint earnings where pension provision is very unlikely.The additional statistical controls of maternal history and educational qualifications do not affect these relationships.
For men, shown in Table 1, once personal and joint earnings are controlled, the effects of inequality within partnerships on pension provision are similar to women, but less pronounced.Breadwinners are more likely to make pension provision than financially dependent men, but there is much less variation than among women.The differences between men no longer register as statistically significant at 5 per cent 13 .
The results shown indicate that all other things equal, for both men and women financial dependence of itself implies less participation in pension provision.For men, if they are equal earners or better, then they are more likely to accumulate private pension.Breadwinning men tend 13 The similar pattern to the effects on women after the major gender differences have been controlled suggests that they probably do represent underlying patterns in society.For the coefficient for men earning between 60% and 80% of joint earnings, p=0.069, and for men earning between 0% and 20% p=0.15. to be relatively high earners.The most advantageous position for them is if their partner contributes between 20% and 40% of joint earnings, presumably freeing up some of their own earnings for pension provision while maintaining their role as breadwinner.The few women who are breadwinners are most likely to be low earners themselves and/or in low earning households.Thus neither they nor their partners are likely to be accumulating private pension.
These results suggest that breadwinning is normatively associated with pension accumulation for both men and women, but subject to different financial constraints, which are gendered.The additional income into the household from a relatively low earning second earner acts as a financially enabling mechanism for breadwinning men to contribute to private pensions.For women, the financial constraints associated with being a breadwinner due to low earnings, and partners having low earnings, are so severe that few 'breadwinning' women can overcome them.

Conclusions
The breadwinner/homemaker or breadwinner/carer relationship has always been problematic in terms of access to financial resources because of doubts about the extent to which money is in fact shared within couples.While legal marriage provided some protections in terms of derived social security, pension and widow's benefits, the growth of partnering outside legal marriage, increases in those living alone, and a rise in the incidence of separation and divorce all make gender issues of financial dependence during the life course more problematic than before.It is becoming increasingly important, in personal and policy terms, for women to be able to provide for their own futures.
In the UK, with a policy regime of heavy reliance on the market for the provision of financial welfare in old age, this means that women must be able to participate in private pension savings.Private pension saving is optimised for those in relatively stable, long term employment at high wages.The underlying cause of female pensioner poverty in the UK, a country which privileges private market solutions, is the pervasiveness and the resilience of the gender division of market and household labour.
In the examination of the patterns of earning within cohorts currently of working age, gender difference is much more clearly associated with the birth of children than with partnership as such.For those on very low incomes, where joint earnings are in the lowest quintile of the distribution, there is much more equality in terms of earnings for men and women.For all other strata, the breadwinner/carer or breadwinner/part-time-carer household is evident in Britain, with the earnings of men dominating partnerships with children.This cross-sectional data suggests that the resulting financial dependency lasts well beyond the time when the children have grown up.
For men, being the major breadwinner is associated with making pension provision for their retirement.This is likely to have cultural roots (man-as-provider) as well as institutional parameters such as working conditions and pay.Women breadwinners tend to be poor and generally cannot afford private pensions.Women who are carers or part-time carers on the whole cannot afford to save enough for an adequate pension and this is especially so for lone mothers, whether never married, divorced or separated.
These social patterns are a result of complex interactions of culture, institution and politics.The structure of these models -with poor quality part-time work and low pay for women, and long hours associated with high pay for men -and the normative reproduction of gender dependency within partnerships with children, are mutually reinforcing.The consequence, in pension terms, is that women will continue to struggle to provide a comfortable pension provision for themselves within that normative and institutional framework.
There are two possible ways to reduce marked gender differentials in the accumulation of provision for old age.Either the reliance on the market must be substituted with a system which substantially redistributes financial benefits from workers to carers, or the underlying cultural norms about the gendered division of labour must change.Both are radical solutions that require considerable political intervention.The staunch neo-liberalism of current UK politics and the persistence of the breadwinner/part-time carer model in the UK in gender relations present a considerable challenge to the feminist project.

Figure 1 .
Figure 1.Participation in State and Private Pension Schemes in the UK: men and women aged 20 to 59

Figure 2 .
Figure 2. Distribution of earnings (£ per week) according to sex and type of pension contribution: men and women aged 20 to 59

Figure 3 .
Figure 3. Median weekly earnings and percentage of women making pension contributions by motherhood and partnership status (women aged 20 -59) a)

Figure 4 :
Figure 4: Median weekly earnings and percentage of women making pension contributions by motherhood status and age group (women aged 20 -59) a)

Figure 5 .
Figure 5. Mean ratio of own to joint earnings across age groups, partnered men and women aged 20 to 59 The impact of motherhood on earnings' equality within couples.Women aged 20 to 59 living with a spouse or cohabiting Source: General Household Survey 2000/1 and 2001/2, author's analysis.Excludes couples where neither earns.

Table 1 .
Odds ratios for contributing to additional pensions, partnered men aged 20 to 59 Odds ratios for educational qualifications and maternal status are shown only after all other variables are controlled.

Table 2 .
Odds ratios for contributing to additional pensions: partnered women aged 20 to 59