Living Longer on Less

This article examines the paid work experiences of women aged more than 55 with less than Aus$100,000 in superannuation living in Victoria, Australia. In Australia, superannuation is a compulsory retirement scheme in which employers make contributions to employees’ superannuation accounts. This research attempts to discover how participants’ experiences of paid work were linked to their positions of low superannuation. We interviewed 32 participants, 18 from a metropolitan and 14 from a regional setting, and analyzed the results qualitatively. We found that although the participants experienced considerable discrimination in the paid workforce, which contributed to their low superannuation balances, the more important factors were outside the workforce, preventing them from accessing well-paid work. These factors included the deprioritization of women’s paid work compared with men’s, unpaid care, disability, ill health and older age, and violence.


Retirement Policy in Australia
Australia's retirement savings system has three components. These include government transfers, particularly the age pension; compulsory superannuation; and voluntary contributions or private savings (Jefferson, 2005;Sadiq, 2010). In the context of an aging population and the expectation that health and welfare expenditure will rise, the Australian government has implemented a range of strategies to encourage older workers to remain in the labor market for longer, including increasing the qualifying age for the age pension (Hodgkin, 2014;Mackenzie, Keuskamp, Ziersch, & Baum, 2011).
The "compulsory superannuation" scheme was introduced in 1992. Under this scheme as it currently operates, employers are required to make contributions to employees' individual superannuation accounts equal to 9.5% of base salary (as of July 2014), rising to 12% by 2025. Withdrawal restrictions apply to these accounts until certain age and employment criteria are met. Superannuation represents a privatization of the risks of retirement saving (Jefferson & Preston, 2005) in contrast to the pension, a system of intergenerational transfer in which the younger generation supports the older generation's retirement years through taxation (Hamilton & Hamilton, 2006). The compulsory superannuation system disadvantages those whose paid work falls outside the compulsory superannuation system (such as low-paid, casual workers), and those who do not perform paid work for a sufficient length of time (Brown, Brosnan, & Gallery, 2000;Jefferson, 2005;Sadiq, 2010). Women are disproportionately represented in these groups (Chomik, 2012;Jefferson, 2005).
The superannuation structure further contributes to the inequity in women's superannuation levels, as it is based on a model of a "normal taxpayer" (i.e., a male) employed fulltime for an uninterrupted 35-year period. Under these circumstances, such an individual should be able to accumulate sufficient superannuation to support "himself" in retirement (Chomik, 2012;Sadiq, 2010). However, women are much less likely to follow this pattern of work. Even when contributing 15% of her salary to superannuation, a woman working full-time for 40 years at average female wage levels is likely to outlive her superannuation (Hodgson & Medd, 2013).

Women, Paid Work, and Retirement Savings
A range of different models aiming to describe women's experiences of paid work exists. Some of these, such as that developed by British sociologist Catherine Hakim (2000), focus on individual choice, assuming women are free to choose what paid and unpaid care they perform, and differences between working patterns of men and women are attributed to innate biological differences. Other theories focus on the effects of socially constructed normative and moral frameworks in shaping ideas of what defines a "mother," a "woman," and a "worker" (e.g., Duncan, Edwards, Reynolds, & Alldred, 2003). A third type of model focuses on institutional factors that shape the context in which individual preferences can be expressed and choices made, for example, through the embedding of traditional gender ideology in government policy (Crompton, 2006;Hill, 2007).
Debates about women and superannuation frequently examine factors such as the time spent in the paid workforce and the gender pay gap, while divorce, single parenthood, and disability are less often considered (Jefferson, 2005). Furthermore, application of equal pay and anti-discrimination laws has been slow and erratic, so women's work histories can vary dramatically. This is especially true for older women, who were likely to have more children and less likely to do paid work (de Silva & Harnath, 2011). Given the long-term nature of retirement saving, these factors are highly important to current and older generations of retirees but are relatively little recognized in the academic literature (Jefferson, 2005).
There are many other factors affecting women's paid work and retirement savings, of which this article can give only a brief overview. 1 A particularly important factor is that of the patriarchal structure of families and society that prioritizes men's work over women's, constructing the male role as that of the "breadwinner." Although this notion may seem outdated as many Australian women expect to earn an independent living, it is still an underlying assumption of many societal institutions and policies (Barns & Preston, 2010;Hill, 2007;Jefferson, 2005).
When women enter the workforce, they are more likely to enter female-dominated professions, which are devalued and pay lower wages (Cortis & Meagher, 2012). Once entering a profession, women have often faced considerable discrimination, particularly before legislative changes made such practices illegal (Jefferson, 2005). Nonetheless, discrimination remains in areas such as hiring and promotion (Australian Human Rights Commission [AHRC], 2014;Wilson, Marks, Noone, & Hamilton-Mackenzie, 2010). Such discrimination has a significant impact on women's ability to save for retirement by limiting their income. Consequently, women, much more than men, tend to rely on pooled financial resources in a household, meaning that women who divorce are frequently left with significantly reduced financial resources (Jefferson, 2005;Sheehan, 2002;Weston & Smyth, 2000).
Gender norms constructing men as providers also construct women as carers, leaving to them the majority of the responsibility for unpaid care of children, the sick, and the elderly. In 2013 in Australia, women comprised 72.5% of primary carers aged 15 to 64 (AHRC, 2013). Responsibility for unpaid care work affects women's paid work: For example, in 2012, female parents' labor force participation rate was 39% lower than male parents'. Furthermore, 66% of employed females with children under 6 worked part-time compared with 7% of males, and this reduction in hours continued as children grew older (Australian Bureau of Statistics [ABS], 2012). It has been estimated that a seven-year career break costs a woman Aus$70 000 in superannuation (de Silva & Harnath, 2011). These Australian findings echo those of international research on unpaid carers in the labor market (see Lilly, Laporte, & Coyte, 2007).
Health problems and disability are major causes of early exit from the workforce, both in Australia and internationally, for women and men (ABS, 2013;Mackenzie et al., 2011;Pit & Byles, 2012;Topa, Moriano, Depolo, Alcover, & Morales, 2009). One third of the increase in the recipiency rate of the disability support pension over the past three decades has been found to be due to the increased retirement age for women (McVicar & Wilkins, 2013), suggesting that health problems prevent many women from working until the pension qualifying age. Workers with disabilities, including mental health problems, generally have lower earnings throughout their working lives (Brazenor, 2002;Schofield et al., 2011). This fact, combined with their tendency for earlier retirement, means that women with health problems and disabilities have reduced retirement savings.
Violence against women is experienced by one in three Australian women and was estimated to cost the economy Aus$13.6 billion in 2009 (Office for Women, 2012). Such violence typically has a significant effect on an individual woman's ability to work and manage her finances (Murray & Powell, 2008;Zufferey, Franzway, Chung, & Moulding, 2010). Lack of financial independence prevents women from leaving violent relationships, and women who do leave face great financial challenges and may suffer from financial discrimination (Braaf & Meyering, 2011;Parkinson, Burns, & Zara, 2000). This financial stress limits the ability of women victims of violence to accumulate retirement savings.
Finally, many women experience additional forms of discrimination and barriers to work if they belong to other groups such as lesbians, migrants or refugees, Aboriginals, and people with disabilities, and this further reduces retirement savings (see, for example, Abdelkarim & Grace, 2012;Frohmader, 2008;Gray, Hunter, & Lohoar, 2012;Smith, Oades, & McCarthy, 2013).
To these authors' knowledge, only one recent qualitative study has explicitly addressed the topic of superannuation and women in the context of these broader factors that influence women's paid work and retirement savings. This important study, "Super"-Poor but Surviving: Australian Women in Retirement (de Silva & Harnath, 2011), combined quantitative data from a telephone survey of 800 women aged 50 to 80 with a small number (12) of qualitative interviews. The sample did not include women with diverse backgrounds (J. de Silva, personal communication, 12 February 2013)-a gap that has been addressed in the current study-but reached a similar conclusion about the detrimental effect of women taking time out of the workforce for child raising.
Although superannuation is unique to Australia, the results of this study have considerable application to other industrialized countries with similar gender norms. The institutionalization of gender inequality in societal institutions is common to many countries (e.g., Crompton, 2006;Duncan et al., 2003), where women experience similar forms of discrimination in the workforce and similar barriers to accessing well-paid work allowing them to accumulate sufficient retirement savings (see, for example, Amato, 2010;Cohen, 2013;Cotter, Hermsen, & Vanneman, 2004;Crompton & Harris, 2006;Gazso, 2005;Gibson, Le, & Scobie, 2006;Hodgson & Marriott, 2013;Jacobs, Laporte, Van Houtven, & Coyte, 2014;Yeandle, 2006). Superannuation-poor women are an important group to policymakers as they rely heavily on the age pension, and, increasingly, the Disability Support Pension (McVicar & Wilkins, 2013). Policy is formulated to reduce reliance on government support (Australian Government Treasury, 2013), but little is known about the compounding effects of different types of disadvantage on these women. This study aims to address this gap by examining the ways in which issues relating to paid work have led to low superannuation balances among Australian women older than 55. This information is necessary to generate policy driving structural change that will allow women to support themselves in their older age. It will address this question by looking first at issues directly related to paid work and second at wider structural barriers to accessing well-paid work.

Study Design and Sample
The data used in this article were from a larger study called Living Longer on Less: Women Planning and Paying for Their Older Age (see Parkinson, Weiss, Zara, Duncan, & Judd, 2013). A qualitative approach was chosen to examine how barriers to paid work experienced at younger ages lead to disadvantage at retirement, and the way in which multiple types of disadvantage interact to compound poverty in older age. Data were collected in December 2012 and January 2013 from a sample consisting of 32 participants (18 from a metropolitan and 14 from a regional area) ranging in age from 55 to 78 and with less than Aus$100,000 in superannuation. Participants were selected using purposive sampling (Patton, 2002) to achieve a diverse sample including married and unmarried women, migrants, Aboriginal women, lesbians, and women with disabilities. Researcher bias in selection of participants was reduced by discussion between three researchers and input from contacts with community service organizations (see below). North East Health Human Research Ethics Committee gave ethics approval for all study procedures.

Recruitment
Advertisements and articles in local newspapers combined with contacts with community service organizations were used to recruit participants. Potential respondents contacted the researchers by phone or email (those participants who were nominated by contacts were first contacted by the researchers), and the researchers conducted a short telephone conversation to establish the outline of their life history. Following this initial conversation, a written consent form was posted to the participants. Participation was strictly voluntary.

Data Collection
At the time of the interview, the consent form was discussed a second time and participants were asked to sign it if they still agreed to participate. For those participants who did not speak English, the consent form and procedures were explained by an interpreter. A semi-structured interview process was used to capture the complexity of the participants' stories (McKague & Verhoef, 2003). Interviews took place in the offices of one of the parent agencies or the home of the participant.
Interviews began by discussing why the participants had volunteered to take part, followed by a discussion of what was important for them in their older age and what plans they had made. Subsequently, the researchers asked how the participant came to be in this position and what they saw as critical events in their lives that had brought them here. At the end, the participants were offered the opportunity to focus on what they considered important to this topic. Interviews averaged 90 min in length and were transcribed in full, using pseudonyms to maintain confidentiality. Upon completion, participants were given an Aus$30 Coles Myer voucher and a list of free and low-cost economic and psychological counselors operating in their region.

Data Analysis
Researchers worked together to develop a thematic framework guided by the interview transcripts and the aims of the research project: This included topics such as paid work, unpaid work, violence, and discrimination. Researchers coded several transcripts together to develop consistency, before beginning coding separately using QSR International NVivo 10. Coding techniques were compared regularly, inconsistencies discussed, and coding processes amended.

Results and Discussion
Overview Eight of the 32 participants were migrants: 1 each from England and New Zealand, and 6 from culturally and linguistically diverse (CALD) backgrounds-2 from Iraq and 1 from each of South Africa, Sri Lanka, Italy, and Argentina. A further 5 women who were born or grew up in Australia reported some influence from the culture of their overseasborn parents. Four of the regional interviewees were Aboriginal. Only 4 participants were currently partnered, with 19 separated or divorced, 4 not mentioning a long-term heterosexual relationship, and 3 widowed. One (unpartnered) was a lesbian.
The participants' experiences of paid work varied considerably. Almost all the participants (31) worked in paid employment for at least part of their life, and 14 were still working. Past and present employment sectors ranged from administration, nursing, aged care, and community services. More than half (18) mentioned working in the social and community sector, one of the most feminized sectors of the Australian workforce. Two (both public servants) worked in relatively high profile, highly paid jobs, yet even they did not experience the increasing seniority and pay of the "normal taxpayer" presupposed by the superannuation structure (see the introduction). Some participants worked in a large number of jobs, sometimes up to three simultaneously, over their lives, while others engaged mostly in unpaid work, including activism and/or community work. All participants engaged in unpaid care work. This heterogeneity made it difficult to identify participants with a single profession, so the profession following their pseudonym should be regarded as indicative.
Importantly, many of the participants mentioned that they had always expected to rely on the age pension in their retirement: With a significant portion of their working life occurring before the introduction of superannuation in 1992, they were unaccustomed to the neoliberal logic underlying superannuation (Jefferson, 2005). Consequently, the need to accumulate a significant amount of superannuation was not a factor in their decisions about paid work. One participant said, The pension age eligibility has changed enormously just in my lifetime-it's gone from 55 to 67.5 years . . . In my early working years there wasn't any super. I then went into my own business as an artist for 15 years and of course again I didn't have super. So now I'm left with literally no super . . . the aged pension not being available till 67.5 years makes me very anxious. (Michelle,59,metropolitan,artist) It is important to note that all the participants experienced more than one form of disadvantage. This was particularly the case for unpaid care work and disability, ill health and older age, factors that affected almost every participant. In every case, different forms of disadvantage interacted in complex ways over the course of the women's lives to bring them to their current position of low superannuation. These multiple constraints invalidates theories of women's work patterns that focus on women's preferences (e.g., Hakim, 2000), being better explained by the interaction of gender norms (Duncan et al., 2003) and the institutionalization of gender inequality in the structure of the workplace (Crompton, 2006), government policy (Hill, 2007), and the household (Jefferson, 2005).
A number of participants mentioned experiencing disadvantage directly related to the workforce. Three themes emerged as most relevant: the type of work performed, discrimination in the workforce, and non-payment of superannuation. What was striking, however, was the extent to which obstacles outside the workforce contributed to their low superannuation balances. Outside paid work, four major themes emerged: deprioritization of women's paid work; unpaid care work; disability, ill health, and older age; and violence.

Factors Directly Related to Paid Work
Type of work performed. The majority of participants had spent most or all of their working lives in feminized professions; only three did not mention having worked in such a profession. Working in feminized professions reduced the participants' superannuation as they tend to be lower paid than male-dominated professions as they are seen as less important because performed largely by women (Cortis & Meagher, 2012).
Many participants who worked in the community sector mentioned that they enjoyed this work and that it accorded with their value system. For instance, Elizabeth (55, regional, community manager) explained that her work was a continuation of her beliefs in the importance of community, and Peta (60, metropolitan, migrant, self-employed community worker) described forgoing a well-paid job in academia to work in the community sector. Although it seems likely that the women chose to enter these professions, it is probable that the socialization of women into caring roles affected these decisions.
Participants working in other feminized professions generally did not freely choose to enter them. One said, In most other cases, the participants did not clearly articulate why they entered feminized professions. Instead, the need to quickly find a source of income combined with other disadvantage (such as lack of education and qualifications, having to look after children, lack of English, health problems, or disability) pushed them toward feminized professions, which offered the path of least resistance. For instance, when asked why she worked in administration, Adele (62, metropolitan, admin worker) responded "I don't have a uni degree" and Bronwyn (57, regional, community worker) explained that she did community work because "you can moderate it," which allowed her to work with a disability. Two recent migrants from Iraq found that almost all paid work was impossible because of their lack of English. This meant they could only access informal, feminized work such as child care, sewing, and cooking. Being outside the formal economy, this work was low paid and they did not receive superannuation.
The influence of compounding forms of disadvantage on pushing women into feminized work was perhaps best illustrated by Anu (55, metropolitan, migrant, catering assistant). As a recent migrant from Sri Lanka, Anu was unable to find work in banking (her previous profession) due to her strongly accented English. Furthermore, her need to care for a new baby led her to seek part-time work, which she eventually found as a caterer in aged care, a low-paid, feminized profession dominated by casual shift work.
Discrimination in the workforce. Because of their age, the participants were exposed to forms of discrimination in the workforce that have now been reduced or eliminated by changes to the law. This discrimination affected their superannuation in various ways. The most significant form of discrimination in this respect was the fact (mentioned by three participants) that when they began working, women were unable to access superannuation at all, even if they sought it out.
The most common form of discrimination experienced was unequal pay. The fact that women received lower pay meant that they were less able to save for their retirement, and if they were receiving superannuation it would have been lower than that received by men doing the same job. The participants frequently expressed emotional reactions to unequal pay, whether of indignation or sadness, suggesting that it affected their sense of self-worth and of the value of their work.
I got a bit sick and fed up of training young men who got paid more the day they came in. (Esther, 78, regional, Aboriginal, public servant) The lower pay received by the participants again demonstrates the formalization of gender inequality in societal institutions (Crompton, 2006): Workplaces could justify paying women employees less than men based on the assumption that women were not "breadwinners." This was untrue for most of the participants, as only four were currently partnered.
Women were discriminated against for marrying and having children. Three participants were made to resign when they married and, upon return to the workforce, were reemployed under worse conditions or not re-employed at all. Women who went back to work after having children found their progress up the promotional ranks interrupted: two participants (both public servants) reported that when they returned to work after childbirth, they had to start again at the lowest level of pay. Measures based on continuous years of service, such as redundancy payments, also disadvantaged some participants because every break they had taken to have children counted against them.
Although more difficult to identify a causal link, general sexism in male-dominated work environments seemed to have affected the superannuation levels of some participants by discouraging them from entering or pursuing a career in the male-dominated (and often higher paid) professions. For example, Anna (58, metropolitan, musician, teacher and composer) explained that the sexism she experienced in her work as a composer (including being patronized by men younger than her, exclusion from decision making, and exposure to pornographic language) affected her confidence, preventing her from pursuing this career as much as she would have liked. Another participant changed jobs due to sexism: I was the first female graduate clerk and a joke, so I transferred to the library.

Was it hard being the only woman there?
Oh it was a joke. I used to pour the tea and coffee. (Esther, 78, regional, Aboriginal, senior-level public servant) Non-payment of superannuation. Some participants did not receive their entitled superannuation. Two of these were migrants, one a cleaner and the other a catering assistant in aged care, both low-paid jobs in industries with reputations for exploiting migrant workers. As a recent migrant, Anu (55, metropolitan, migrant, catering assistant) was unaware of superannuation and was not paid it for 14 years. Her attempts to recuperate it were unsuccessful.
Six participants worked for their partner at some point in their lives, five of whom are now separated or divorced. Although their partner would otherwise have had to hire someone to do this work, only one participant was paid and none received superannuation, meaning that none were able to make contributions to their retirement savings over this period. Although these participants benefited from the contributions of the business's income to their household, it did not benefit their retirement savings as paid work would have. Furthermore, this unpaid work occupied time, preventing them from doing paid work that could have contributed to their retirement savings.
Stefania was able to work full-time while running her husband's cleaning business from home and raising four children, but her testimony illustrates the difficulty of maintaining a paid job while doing unpaid work for a partner: I'd come back, pick up the kids . . . and while I was fixing them, feeding them, doing everything else, I had the phone, I usually had about 10 or 12 messages on the answering machine and I'd have to ring back and make appointments for him . . . Put the kids down, we'd sit there, fold nappies, blah, blah. Then I'd book all his jobs. (Stefania, 59, metropolitan, admin worker) The work of running her husband's business was simply expected of these women, part of their duties as "wives"; those who are now separated receive no gains from the unpaid work they performed earlier, while their ex-husbands live on the prosperity of the business they helped to establish.
These discriminatory practices operating within the workforce demonstrate the institutionalization of gender inequality in the legal system and the workplace (Crompton, 2006;Hill, 2007). These practices were applied uniformly despite the frequency of divorce, and were applied equally to single women and lesbians despite the fact that they could not rely on a male breadwinner.

Factors Outside the Paid Workforce
Deprioritization of women's paid work. The force of normative ideas about appropriate behavior for women (Duncan et al., 2003) was prominently expressed by the deprioritization of women's paid work within heterosexual relationships. This included the assumption of a sexual division of labor in which the man would provide financially for the family while the woman would perform unpaid housework and child care (see also "Unpaid care work" section). When it later came to separation, divorce, or widowhood, as in the cases of 23 participants, participants were invariably left with lesser monetary resources, including superannuation, as a result of the systematic deprioritization of their paid work over their married lives (Sheehan, 2002;Weston & Smyth, 2000). Never-partnered women and lesbians were disadvantaged by these same assumptions, as they had to work within structures that assumed that, as women, they were economically dependent on a higher paid male breadwinner.
The importance of male breadwinner assumptions was prominent in the comments of many participants. For example, Eloise (61, metropolitan, admin worker) explained that her husband believed that the role of a wife was to stay home and look after the children, and this was a belief she internalized after marriage. She later divorced and faced many difficulties in raising two children and supporting herself. Another participant described similar assumptions in a past relationship: [He] was always . . . of the view that he was a more serious professional and he was the one who should be thinking about [financial matters] and I wasn't that relevant. (Peta, 60, metropolitan, migrant, self-employed community development worker) This deprioritization of women's paid work was not limited only to attitudes of participants and their partners: It had important effects on the participants' employment behavior, decreasing their ability to follow the path of the "normal taxpayer" upon which the superannuation structure is based. Ellen's description of her working history illustrates this phenomenon very clearly: My husband, from the time we were married to today has always worked full-time. But [me], I was full time, I had to quit my job to get married, then I was unemployed for four months, then I worked, then he was transferred, I had to quit my job, move to a new town, then got a job after another couple of months and then I got pregnant. I had to then quit my job, then I was out of work for seven years and then I had to get a part-time job and I think it was say, 16 hours and . . . it was on what we would have termed "pathetic money" type thing and he's still working fulltime. (Ellen,56,regional,community worker) She further explained that being married meant that she took a job "for the cash, not for the career" and actively prevented her from thinking about following a long-term employment path.
The deprioritization of the participants' careers often meant that they invested significant unpaid labor and/or financial resources in their male partners' careers, instead of in their own. For example, Georgie (57, metropolitan, dental nurse) supported her ex-husband financially through 5 years of university, which allowed him to gain skills leading to a relatively prosperous career. Georgie herself never had the opportunity to do tertiary study and received little of the long-term financial gains from her ex-husband's career.
The participants did not have support in unpaid housework and child rearing as their partners did. This prevented them from studying and fundamentally altered the kind of paid work they could do, as their work had to fit around their unpaid duties in the home. As one participant remarked, wryly, He was a hard worker but he could be a hard worker because he came home [and] everything was clean. (Giulia,68,metropolitan,migrant,bookkeeper) Migrant participants described ways in which their culture involved stronger or different norms relating to women's roles compared with mainstream Australian culture. These cultural values often had a significant effect of discouraging or preventing them from pursuing better paid, higher status work. Peta (60, metropolitan, migrant, self-employed community worker), for example, explained that norms around the type of work appropriate for women, and its value, were very strong in South Africa, suggesting that this had an effect on her sense of her own worth.
Huda's (regional, migrant, child care worker) explanation of the expectations placed on women in her culture was similarly illuminating. She explained that in Iraq, culture, tradition, and religion combine to prevent women from acting independently. Even living in Australia with her daughter and infant grandson is considered unacceptable by her community as they are living without an adult male. Huda's example suggests how older migrant women from cultures where women are expected to rely entirely on a male "breadwinner" are particularly disadvantaged when moving to Australia, where the retirement savings system expects women to have worked independently for their entire adult life.
Unpaid care work. The corollary to the deprioritization of women's paid work was the prioritization of their unpaid care work. High unpaid care workloads, by occupying women's time and attention, reduced their ability to do paid work. Every participant mentioned having spent a significant amount of time caring for others over the course of her life. Some cared for an extraordinary number of people: Lorna (59, metropolitan, community worker) cared not only for her daughter alone but also for her grandson, her stepson and stepson's son, a second stepson, her god-daughter and god-daughter's daughter, her sister, brother-in-law, and niece, her father, and the child of a close friend. Another participant, a lesbian feminist activist, described the large amount of time she has spent and continues to spend caring for her friends (now aging) in the lesbian community. These examples show that care commitments were not limited to blood relations.
Many of the participants had children and changed their pattern of paid work to care for them, either stopping all together or working part-time. This reduced their superannuation in every case. Furthermore, re-entering the workforce after having children was difficult for many participants due to the gap in their working history and the lack of affordable child care. Two participants even mentioned that it was more economical for them to stay at home because what they were capable of earning, as women who had just spent time out of the workforce, was less than the cost of child care. Having children therefore prevented women from following the path of the "normal taxpayer." The Aboriginal participants' accounts suggested a somewhat different relationship with care, as they all brought up more than one generation of children. As Rosalie (77, regional, Aboriginal, community activist) explained, " . . . this is a part of our community. We do that, we bring up grandchildren." Another participant, Iris (55, regional, Aboriginal, self-employed caterer) cared for six of her own children, then later four grandchildren, a nephew, and multiple unrelated children, mostly after separating from her husband. The other Aboriginal participants had similar stories: Mel (regional, Aboriginal, never in paid work) was even known as "Nanna Mel" "because it looks like everyone in [town] are my grandkids. I've looked after my four and then I looked after my daughter's four." This greater care load affected their ability to work and earn superannuation.
Unpaid care work was clearly gendered, with none of the participants' male partners ever reducing or ceasing paid work for caring responsibilities. Care of the elderly was also heavily gendered, with a number of participants commenting that caring for their aging parents was expected, even demanded of them, but not of their brothers.
Even though [my parents] have got six kids, I'm the only one who really does anything. Which gives me the shits as a feminist because most of them are boys. (Lesley,60,metropolitan,activist) Some participants gave a concrete explanation of how their unpaid care work reduced their ability to do paid work, thus reducing their superannuation. This was particularly stark when they spent a protracted length of time out of the workforce due to caring. Lavender (72, regional, medical secretary) cared for her husband for 3 years until his death from cancer and then had great difficulty re-entering the workforce. Iris (55, regional, Aboriginal, self-employed caterer) had to stop paid work when her mother had a stroke because she was driving more than a thousand kilometers every week to care for her, while also caring for four grandchildren. Joan (64, metropolitan, small business owner) had to give up the business she started with her husband when he became a quadriplegic following a car accident.
We had a pet shop at that time and a two-year-old. [My husband] spent two and a half years in hospital with bad head injuries. I cared for him for the next 19 years. (Joan, 64, metropolitan, small business owner) A number of the participants were heavily involved in activism and/or community work (caring for the community). The time and energy they dedicated to these activities reduced their ability to engage in paid work and accumulate superannuation. One participant explained, I've lived in poverty all my life, more or less voluntarily because my primary thing was my activism and not making money or establishing any security for myself. Which may have been a mistake but that was the judgment I made. (Lesley,60,metropolitan,activist) Lesley was clear that activism was significant work, saying "I work all the time, I do my voluntary activist work," and felt it unfair that activists are not supported in old age despite the contribution they have made to society. This theme that caring work was real work that should be recognized with support into old age was echoed by a number of participants: I feel really sad more than anything that with all I have contributed . . . I've contributed so much, family and community that it's coming down to this. (Helena,57,metropolitan,psychotherapist) Disability, ill health, and older age. Although only one participant was disabled in the common-usage sense of the term, almost every participant mentioned health problems that affected their ability to work; in many cases, these were conditions linked to growing older. These health problems limited the participants' ability to accumulate superannuation in the last years of their working lives, usually by forcing them to quit or reduce working hours.
A number of participants had recently been forced to stop work or reduce hours due to health problems. Adele (62, metropolitan, admin worker) could not work because of advanced osteoarthritis. Brenda (59, regional, migrant, nurse) stopped work after a workplace-related injury, for which she was denied compensation. Lorna (59, metropolitan, community worker), suffering from the long-term effects of being hit by a bus as a child, was unable to cope with the physical stress of working the day shift as a nurse. Fatigue prevented Giulia (68,metropolitan,migrant,bookkeeper) from returning to work since being diagnosed with breast cancer 9 years ago. Some participants' paid work was affected by mental health problems. Moira (73, metropolitan, migrant, nurse) stopped working due to nervous stress following an attack by a male psychiatric patient. Eleanor (metropolitan, public servant) left her high-pressure job at a police station following a "nervous breakdown," and ongoing mental health problems have since left her struggling to retain short-term casual jobs, which do not pay superannuation. Fiona (metropolitan, disability support worker), suffering from post-traumatic stress due to childhood sexual abuse by her grandfather, had to leave jobs because they were not flexible enough to accommodate her needs.
I've got to be realistic about what I can do because healing from past trauma is also work. You have to work at being mentally well, [it's] not an automatic. (Fiona, metropolitan, disability support worker) What is notable about these circumstances is that many of the women were not very old: At least 19 of the participants were younger than 65, the current age of age pension eligibility, increasing to 67 by 2023 (Department of Human Services [DHS], 2014).
Violence. Nine participants mentioned experiencing violence from their male partner or a family member, even though this question was not asked. This may be evidence of both its prevalence and its significance to women's retirement savings. In some cases, the effect of this violence on paid work and superannuation balances was clear. The clearest example was that of Mel (regional, Aboriginal, never in paid work), who described being abused by her father and bringing up her brothers after her father murdered her mother. Later, she went through a series of violent relationships with men, some of whom stalked her even as she moved interstate. Living in shelters for many years, it is hardly surprising that under these circumstances she never did any paid work, and has no superannuation. Another participant, Elizabeth (55, regional, community manager), pulled all her savings out of superannuation to move herself and her children interstate, to escape a relationship that threatened to become violent. Withdrawing her super at that time meant that, despite having had relatively well-paid work, she now has little superannuation.
One participant, a lesbian, described the physical and sexual violence experienced by lesbians in the 1970s and 1980s. This included rape "as punishment for being lesbians," indicating that high levels of homophobia in mainstream society posed additional obstacles to lesbians in obtaining paid work.
Although the participants did not usually make an explicit link between experiences of violence and paid work, these testimonies nonetheless suggest how violence can have a significant effect on women's ability to work and therefore to earn superannuation. Although these findings are consistent with what little research exists on this subject both internationally and in Australia (Moe & Bell, 2004;Zufferey et al., 2010), there is a need to explore this question further; particularly the effects of violence during childhood and early adult life on employment and retirement savings.
The results of this study are not well explained by theories relying on women's preferences for work (e.g., Hakim, 2000); they are better understood in terms of constraints operating on women's ability to engage in well-paid work that would allow them to accumulate sufficient retirement savings. These constraints take the form of gender norms (Duncan et al., 2003) and the institutionalization of such norms of gender inequality in society (Crompton, 2006;Hill, 2007). These results suggest that researchers and policymakers should focus their analyses of women and paid work around constraints rather than preferences.

Implications
Cohort effects are important in this study: Some of the results are most applicable only to women aged more than 55, whereas others are applicable to all Australian women. Particularly, a large part of the working lives of the participants in this study took place before the introduction of compulsory superannuation in 1992 and related measures such as the 2002 legislation allowing for superannuation splitting in divorce settlements. This contributed to their low superannuation balances, as they were not building up superannuation over the entire course of their working life; indeed, they were operating under the assumption that the pension would be available to them in older age (see Jefferson, 2005). Similarly, they did not benefit from higher levels of awareness of superannuation and retirement saving existing since 1992. The participants also experienced greater discrimination in the workforce than younger women, especially because some forms of discrimination, such as women's lack of access to superannuation, have been eliminated. The existence of these cohort effects suggests that government policy should provide additional support for older women, perhaps by lowering the pension qualification age or providing an increased pension allowance for older women.
There are positive aspects of Australia's employment regime, such as the strong minimum wage, and significant progress has been made on reducing gender inequality through legislative change. However, discrimination and disadvantage persist. As of February 2015, the gender pay gap-the difference between women's and men's average weekly full-time equivalent earnings-was 18.8% (Workplace Gender Equality Agency [WGEA], 2015), and in 2012, the difference in pay between new male and female graduates was Aus$5,000 per annum (WGEA, 2013); an ongoing investigation has found that pregnancy discrimination is experienced by 49% of mothers (AHRC, 2014). Furthermore, the abuse experienced by Australia's first female prime minister, Julia Gillard, suggests that women's right to high-status work in the public sphere is still contested (Summers, 2013). Consequently, superannuation imbalances between the sexes are projected to continue (AHRC, 2014).
As a result of this continuing inequity, acknowledging the fact that many women will have no choice but to continue to rely on the pension (Jefferson, 2005;Sadiq, 2010) by raising the payment level and lowering the age of eligibility for all women is an appropriate short-term policy change. However, a key aim of Australian government policy is to reduce reliance on the age pension and increase reliance on personal assets (Australian Government Treasury, 2013). In the long term, the implementation of policies addressing the barriers to well-paid work for women will help to achieve this aim.
Addressing the deprioritization of women's paid work and its effects should be a key policy aim. This can be achieved through new legislation over unpaid care. Policy should both challenge the assumption that women will perform the majority of unpaid care work and ensure that carers are compensated for the unpaid work they do in their retirement savings. An example of the first type of reform is the introduction of incentives for equal take-up of parental leave in couples: This could take the form of a government-funded time-bonus of additional paid leave when it is equally split. An example of the second type of reform is the introduction of "carer credits" paid directly to individuals' superannuation accounts while they are caring for children; such a scheme could be extended to all carers (and perhaps even activists) as well as parents (AHRC, 2013;Hodgson & Marriott, 2013). Carer credits would also provide recognition of unpaid care work as valid, important work.
Many women experience poverty following divorce due to the imbalance of paid and unpaid work between partners within heterosexual couples. To reduce this, incentives to increase superannuation contributions from one spouse to another could be strengthened (Hodgson & Marriott, 2013). Changes to the Family Law Act could also work to increase instances of superannuation splitting as part of a divorce settlement, which is currently rare. A key possible change is to move from the current arrangement in which a superannuation split is only considered if one or both parties ask for it, to one in which the court is required to consider whether or not a split be made, or be satisfied as to why it is not provided for (Parkinson et al., 2013).
The results of this study indicate that disability and health problems, often relating to aging, are a major barrier to continued paid work for older women (see also McVicar & Wilkins, 2013). Measures such as greater flexibility in working hours will help to retain older people with disabilities and health problems in the workforce (AHRC, 2012). Workplace policies must reach a balance between protection from risk of physical injury while preventing stigmatization of older workers: The most effective policies involve systematic initiatives preventing the accumulation of injuries over the course of working lives (Brooke, Taylor, McLoughlin, & Di Biase, 2013).
The results of this study suggest the significant impact of male violence against women on women's poverty in later life. The universal provision of paid domestic/family violence leave, which would allow women to take leave to, for example, attend legal appointments and find alternative accommodation, would help women to keep their jobs while experiencing violence (

Limitations
The small sample size, although typical for a qualitative study, is a limitation. Not all characteristics of poorer older women were represented: For example, there were no homeless women or refugee women, and no metropolitan Aboriginal or Torres Strait Islander women. Although these findings may not be generalizable, however, they are potentially transferable to similar groups.

Conclusion
The failure of public and academic debate to place the issue of superannuation in the wider context of women's lives, including the barriers to their securing well-paid, superannuation-providing work, contributes to the neglect of disadvantaged older women by public policy and wider society. This increasing neglect is particularly concerning given the continuing reliance of older women on the pension (Jefferson, 2005) and the fact that the first generation of retirees whose entire working life has taken place since 1992 will not retire for another 30 years (de Silva & Harnath, 2011). Meanwhile, the challenges in saving for retirement faced by women have by no means been solved, meaning that women growing old in the future will face similar deprivation and inequity in superannuation balances.