The impact of defamilisation measures on gender and pensions: A comparison between the UK and seven other European countries

This article uses individual-based and state-led care-focused defamilisation indices to explore women's employment opportunities and experiences and their implications for pension contributions. These two types of defamilisation indices are applied to eight European countries (Belgium, Finland, France, the Netherlands, Denmark, Norway, Sweden and the UK) which shows that the UK has less generous defamilisation measures than its European counterparts. It indicates that the use of defamilisation measures along with pension policies which are not based on the male breadwinner ideology have the capacity to moderate economic inequalities between men and women in older age.


Introduction
Women are typically over represented in pensioner poverty. On average, in the EU, the percentage of men aged 65+ at risk of poverty was 14.6 per cent compared with 20.2 per cent of women in 2014 (Samek Ludovici et al, 2016). 1 The gendered nature of poverty in older age reflects women's constrained opportunities across the life-course, including their greater likelihood of undertaking caring responsibilities, its impact on employment and, subsequently, women's increased possibility of reaching retirement with inadequate pensions (Foster, 2014;Ginn, 2003;Grady, 2015;Vlachantoni, 2012). Levels of inequalities are also affected by the extent to which pension systems address these diverse experiences and compensate for relative disadvantages in the division of work and care.
Given that women are more likely to be under-saving for retirement it is particularly important to explore factors which reduce gendered pension inequalities. Focusing on European countries, and the UK in particular, this article outlines gender inequalities in employment and care provision and their implications for pension accumulation. It shows how pension income in older age is the result of the interplay between country-specific institutional frameworks and policies, and individual employment patterns over the life-course. These patterns are influenced by policies which assist individuals to undertake paid employment and contribute to pensions. These include the use of defamilisation measures.
We explore how these defamilisation measures have the potential to facilitate women's employment and improve opportunities to accumulate pension income. In doing so we seek insights from the Capacity Approach (CA). The CA is a broad normative framework for assessing individual well-being and social arrangements which focuses on the substantive freedom of people to choose and be what they value (Egdell and McQuaid, 2016;Sen, 1985). While this approach has been widely used in the study of gender inequality and family policies, few analysts have linked it to the examination of defamilisation measures (Kurowska, forthcoming;Lewis and Giullari, 2005). Furthermore, it is apparent that existing work in this field has not discussed the connection between defamilisation and pensions. Hence, this paper serves to contribute to knowledge advancement concerning how the core ideas of the CA link to the use of defamilisation measures to assist women to accumulate pensions.
Initially, the paper reviews the gendered nature of employment and care before presenting the challenges associated with the manner in which pension systems operate. It then discusses the concept of defamilisation and how it can be operationalised in the form of individual-based defamilisation and state-led carefocused defamilisation indices to explore women's employment experiences. These two types of defamilisation indices are then applied to eight European countries (Belgium, Finland, France, the Netherlands, Denmark, Norway, Sweden and the UK) to explore the potential barriers to women's employment and capacity to contribute to pensions. The characteristics of the UK are considered with reference to the seven other European countries. It shows that the increased role of neoliberalism in pensions, which emphasises deregulation, the curtailment of the welfare state and market solutions to social problems, is important in understanding women's pension experiences. The evidence provided by the indices indicates that the use of defamilisation measures along with pension policies which are not based on the male breadwinner ideology have the capacity to moderate economic inequalities in old age. However, not all the countries in the indices utilise defamilisation measures in the same manner. Finally, the paper discusses the importance of defamilisation measures and of developing pension schemes which do not discriminate against those who undertake caring responsibilities, in ensuring that CA values are adhered to, and reducing gendered pension inequalities in older age.

Women's employment
Recent decades have been characterised by an expansion in women's employment (Eurostat, 2016a). However, despite this, gender disparities in employment levels continue to exist. In 2015, the employment rate for men aged 60-64 stood at 75.9 per cent in the EU-28, compared with 64.3 per cent for women. The difference between employment rates by sex varies considerably among EU countries. For instance, it was as wide as 27.8 per cent in Malta and less than 2.5 per cent in Finland and Lithuania (Eurostat, 2016a). In the UK in 2013 around 69 per cent of women aged 16-64 were in work (compared with 79 per cent of men) representing an increase from 53 per cent in 1971 (ONS, 2013). There has been an expansion of service sector employment, particularly among women. This is characterised by low pay, flexible contracts and part-time work (Blackburn et al, 2015). In the UK in 2015, 43 per cent of employed women worked in part-time jobs compared with 12 per cent of men (PPI, 2016). In the EU, a third of employed women aged 15-64 worked part-time in 2014 (Eurostat, 2016a). In 2014, women's gross hourly earnings were on average 16.1 per cent below those of men in the European Union (Eurostat, 2016b) with gaps in annual earnings highest in the UK (31 per cent) (Samek Ludovici et al, 2016). Women's employment characteristics have repercussions for pension accumulation. For example, in the UK 54 per cent of men and 58 per cent of women working full-time were members of occupational pension schemes, but for women employed part-time, only 30 per cent were members (ONS, 2014). Furthermore, lower pay and time out of the labour market have implications for pension contributions (Ginn, 2003).
In many countries, including the UK, the traditional male breadwinner model has been modified, with woman supplementing the main family wage (earned by men) through intermittent and part-time jobs that fit with caring responsibilities (Loretto and Vickerstaff, 2015). Policy has often emphasised women's dual role as carers and employees (Foster, 2010). For instance, in the 'Europe 2020' strategy women aged 20-64 are one of the target groups for employment (European Commission, 2010a). However, the capacity to combine caring and employment depends on the generosity of parental leave and on state childcare services in addition to policies to support other forms of caring. Gendered aspects of such policies can serve to entrench women's responsibility, discouraging sharing of leave and care (Bennett and Daly, 2014;Lewis, 1997).

Gender and care provision
Gender inequalities in older age are affected by the provision of care, largely carried out by women. It is associated with lower earnings, long-term negative wage effects, and consequently, lower pension entitlements (Ginn, 2003;Grady, 2015;Möhring, 2015;Price, 2007). For instance, Sefton et al (2011), using longitudinal analysis from the British Household Panel Survey, reported a negative impact of motherhood on the income of older women in Britain given its links to employment. They found 42 per cent of married women aged 65+ without children had worked full-time for 30 years or more, whereas only 19 per cent of women who had one child, 12 per cent of women who had two children and 7 per cent of women who had three or more children, worked full-time for 30 years or more. In the UK, women report spending an average of 23 hours on caring for family members and 13 hours on housework each week with the equivalent figures for men being only ten hours and eight hours (Scott and Clery, 2013). Children still have a considerable impact on women's employment patterns. For instance, in the UK in 2010-12, 76 per cent of women aged 21-30 without children were working, compared to 44 per cent of women with children (PPI, 2016). Furthermore, in addition to influencing work patterns, new child-focused priorities for spending and financial pressures associated with childcare, can adversely affect mothers' propensity to save (Foster, 2017a).
It is often assumed that family care means childcare but in practice it includes caring for adults. For instance, caring for other relatives is increasingly common (in part as a result of an ageing population) and the responsibility disproportionately falls on women (Loretto and Vickerstaff, 2015). In the 50-64 age group in the UK, 12 per cent of women and 9 per cent of men were informal carers, mainly for older relatives (Stokes, 2013). Unpaid care for parents and in-laws is less visible than childcare, 2 including to employers who may view caring responsibilities for parents as less legitimate than obligations to small children (Loretto and Vickerstaff, 2015). In the years approaching State Pension Age (SPA), carers' weekly time spent on eldercare tends to increase, with implications for pension contributions (Ginn and Macintyre, 2013).
In the UK, free part-time nursery places for 3-4-year-olds and breakfast and after-school clubs have been introduced with a substantial increase in childcare places in the last 20 years. Yet provision still falls short of need (Daly and Scheiwe, 2010). Furthermore, women's low earnings can make such childcare unaffordable (Bennett and Daly, 2014). In the EU, approximately 30 per cent of women with caring responsibilities state that they are inactive or work on a part-time basis as a consequence of the lack of care services for children and other dependents (Leschke, 2011). State policies such as affordable, accessible high-quality childcare and wellpaid parental leave, as in the Nordic countries, can encourage women's labour force participation, while the absence of such incentives can disincentivise, and even penalise women's paid employment (Earles, 2013). While some feminist scholars interpret leave schemes as a way of supporting a traditional family model, in practice it can serve to support parental choice and gender equality (Frericks and Höppner, forthcoming). Ginn et al (2001) argued that many liberal welfare states in particular often employ a contradictory policy, whereby women's employment is encouraged but support, such as childcare or parental leave, is largely insufficient.

Gender and pension policy
Typical male working patterns are still predominantly the reference point for the calculation of pension entitlements, with gender differences in work and care overlooked (Grady, 2015). While many EU countries have implemented some form of credit for childcare in state pensions (Leitner, 2001) they are uncommon in private pensions. The extent to which pension systems compensate for gender inequalities is dependent on how it combines redistributive and non-redistributive components, including credits, in its entitlement structure (Vlachantoni, 2012). Pension systems that maintain a close link between lifetime earnings and pension income through a strong emphasis on private pension provision and a residual state scheme exacerbate the accumulated effects of gendered work histories on retirement income (Ginn, 2003).

The impact of defamilisation measures on gender and pensions 203
Over the last three decades the rise of a neoliberal political economy with an anticollectivist ideology has driven the promotion of private pensions through regressive tax relief subsidies and an emphasis on 'personal responsibility' and individualisation of risk. The retrenchment of state provision and increasing emphasis on non-state pension provision have been emphasised in an attempt to contain welfare costs and boost financial markets (Earles, 2013). At the same time changes in social norms and the decline of marriage as a lifelong contract have meant that reliance on a husband's income in later life is increasingly risky (Sefton et al, 2011). In Europe additional or new forms of private pension have been introduced (Germany, Sweden and the UK), the pension age has been raised (Austria, Germany, Portugal and the UK), pensions reduced for those who retire early (Sweden and Germany), the number of years of contributions raised for entitlement to full public pensions (Belgium, Sweden, Germany and the UK) and early retirement incentives reduced (Denmark, Finland, Germany, and the Netherlands) (Foster, 2014). Any move towards tighter links between contributions and benefits is likely to result in a widening of gender inequalities in pensions (Grech, 2015).
UK pensions have been characterised by numerous changes over recent years (see Foster, 2017b). For instance, the State Pension system has been substantially reformed since April 2016, with the Basic State Pension (BSP) and State Second Pension (S2P) replaced by a new Single-tier State Pension (STP) for those below the State Pension Age. While the STP should ultimately provide a stronger foundation than its predecessor for most in the short-term, there will be many 'losers' in the longerterm with the new system less generous than the current one for most people, given a lower accrual rate than the combined rate of the BSP and S2P.
The introduction of the STP does not represent an attempt to reverse the drive to promote private pensions. Rather it is deemed that a 'solid foundation' in the form of a 'more generous' state pension represents a suitable platform for auto-enrolment (Price, 2007). Auto-enrolment, introduced in a phased manner from July 2012, offers access to a portable occupational pension to millions of people without access to good-quality workplace provision, while enabling existing schemes with benefits or contributions above the National Employment Savings Trust (NEST) (the default option auto-enrolment scheme) minimum to continue. Eligible employees (some people are excluded as a result of age and earnings) are automatically opted into the scheme chosen by their employer but may withdraw. By October 2014 4.8 million people had been auto-enrolled with less than one-in-ten exercising their right to opt out (Webb, 2016). The stated logic behind auto-enrolment is that while structured advice and information may improve understanding, behavioural barriers, including myopia, cynicism and inertia, all inhibit pension saving (Foster, 2017a). Furthermore, a lack of understanding of risk and insufficient saving is offered as a justification for supporting the 'libertarian paternalism' inherent in auto-enrolment (Clark et al, 2012).
It is, however, apparent that even after automatic enrolment has been fully phased in, the distribution of tax relief will benefit higher rate taxpayers (mainly men) most. The NEST also incorporates features of investment choice and individualised risk. Contributing to such schemes may essentially be inadvisable due to potential interaction with means-testing in retirement (Price, 2007). For women, the risk of a wrong decision is especially high because unpredictable career development is combined with the uncertainties of future investment returns, charges and annuity rates (Ginn and Macintyre, 2013). As in other private pensions, no credits are provided in NESTs for periods of family care.
There has been a greater awareness of gender inequalities in pensions over recent years (European Commission, 2010b;Ginn and Macintyre, 2013;Pensions Commission, 2006). However, UK governments have largely failed to adequately address these gender inequalities through recent pension developments which, while encouraging pension saving, are largely gender blind (Grady, 2015). On the whole, when private pension schemes place high thresholds of eligibility in relation to years of service, earnings and/or level of contributions, women are more likely to be disadvantaged. Given the emphasis on individual responsibility for pension accumulation, the impact of employment characteristics on pensions is likely to continue to play a substantial role in women's access to pension provision. As such, it is important to explore the extent to which countries adopt strategies to enable women to participate in paid employment including defamilisation measures and/ or reward periods of care.

Defamilisation issues
This article uses two defamilisation indices (individual-based and state-led carefocused) to examine how government policies regarding defamilisation issues might affect women's pension accumulation. These two indices are developed on the basis of the work by Bambra (2004;2007) and Kroger (2011). Before providing details of the two indices, it is necessary to explain why we attach particular importance to these studies.
A number of welfare typologies based on gender which consider defamilisation have been developed (for example, Bambra, 2004;2007;Chau et al, 2016;Korpi, 2000;2010;Leitner, 2003;Saraceno and Keck, 2011). Most of these typologies are concerned with women's disadvantaged positions in the family in association with the dominance of the notion of men as the main wage earner. Women face two related defamilisation issues here (Saxonberg, 2013). First, women are expected to bear most of the caring responsibilities in the family. As a result, they may not have sufficient opportunities to develop their career and thus achieve financial autonomy, both in work and retirement (Kroger, 2011). Second, women are often financially dependent on men. This financial subordination to men leaves women little bargaining power over important family matters such as the gender division of caring responsibilities (Lister, 1994;Nyberg, 2002). We term these 'defamilisation issues' because the solutions to them are likely to involve attempts to reduce the significance of the family in the provision of care and/or financial support through the provision of measures such as care credits.
While these two defamilisation issues are highly related to each other, different studies place varying emphasis on them (Lohmann and Zagel, 2016). For instance, Bambra (2004;2007) focuses on ways in which women may achieve financial autonomy through taking part in the paid labour market. In her study of defamilisation patterns in 21 countries, she examined ways in which welfare states can enable women to function as independent workers (accumulating pensions in their own right) and reduce the economic importance of the family in women's lives (Bambra, 2007). She states that the conceptualisation of defamilisation in studies by Esping-Andersen (1999) and Korpi (2000) are problematic because they focus on reducing the family's caring The impact of defamilisation measures on gender and pensions 205 responsibilities instead of directly supporting women to achieve financial autonomy. In contrast, Bambra (2007) focuses on comparing the ways in which welfare states can enable women to function as independent workers and reduce the economic importance of the family in women's lives. Based on the data from the World Women Statistics Report, Bambra (2007) studies three factors -maternity leave compensation, compensated maternity leave and relative female labour participation.
This stress on economic dimensions of defamilisation is, however, not without criticism. It has been stated that defamilisation reforms often place too much emphasis on providing support for the commodification of women, but put too little emphasis on reducing the caring responsibilities of the family (Lohmann and Zagel, 2016). As a result, while women are expected to take an active role in paid employment, they do not necessarily see their caring responsibilities in the family significantly reduced. Hence, we argue here that defamilisation studies should focus not only on how to help women achieve financial autonomy but also on the reallocation of caring responsibilities within the family.
Some analysts are particularly concerned with the degree to which care policies make it possible for women (and men) to take part in society outside their homes and families (Esping-Andersen, 1999;Kroger, 2011;Leitner, 2003). It is believed that by providing non-family caring services (defamilisation measures) such as public childcare services, the caring responsibilities of women in particular can be reduced (Korpi, 2000;2010). As a result, women may have more time to develop their careers and to contribute to earnings-related pensions (Leitner, 2003).
In contrast to the defamilisation typology developed by Bambra (2007) concerning women's financial independence within the family, Kroger (2011) has developed another kind of defamilisation typology (the dedomestication typology) which stresses the importance of family's caring responsibilities being shared by non-family sectors (such as the government). Kroger (2011) asserts that the term defamilisation should be reserved for economic dimensions; and 'dedomestication' should be utilised to study the changes in the distribution of caring responsibilities in the family. Based on OECD data, Kroger (2011) focuses on five variables concerning childcare: the average hours of children's attendance in childcare per week; the right to a childcare place; the childcare cost for a dual-earner family (as percentage of net family income); the childcare cost for a sole parent family (as a percentage of net family income), and the child-staff ratio in formal childcare.
Given the related nature of the two typologies and the applicability of both to the study of women's opportunities to undertake paid employment and contribute to pensions, this paper utilises both sets of defamilisation indicators to study the defamilisation issues faced by women. These are termed individual-based and stateled care-focused defamilisation. The former is indebted to Bambra's (2004;2007) work on defamilisation. These measures are concerned with how welfare enables women to survive as independent workers and decrease the economic importance of the family in women's lives. The latter, expanded from Kroger's (2011) model of dedomestication, are concerned with the extent to which family caring responsibilities are relaxed via policy measures. Using the two types of defamilisation indicators will assist in identifying the level to which particular countries, including the UK, have embraced defamilisation measures, and consider the potential implications for pensions contributions and gender inequality in later life.

Methodology
The defamilisation indicators developed for this article cover eight countries (Denmark, Norway, Sweden, Finland, Belgium, France, the Netherlands and the UK). These eight countries have high levels of human development (United Nations, 2014). In 1990, Esping-Andersen presented the three worlds of welfare capitalism thesis based on the study of 18 OECD countries including these eight European countries. 3 According to Esping-Andersen (1990), these eight countries come from different worlds of welfare -Denmark, Norway, Sweden and Finland are members of a social democratic group; Belgium, France and the Netherlands members of a conservative group; and the UK is the only European country identified as a member of a liberal group. While Esping-Anderson's approach is not without criticism (Lewis, 1997) it is still commonly used to aid policy analysis (Ferragina and Seeleib-Kaiser, 2011).
In order to study these countries' individual-based and state-led care-focused defamilisation patterns, the World's Women statistic report was used (United Nations, 2010). 4 Z-scores were utilised to compare data from more than one distribution. A variable where the observations have been converted into Z-scores is known as a standardised variable. 5 A positive Z-score shows that the observation is greater than the mean (above average) whereas a negative Z-score shows that the observation is lower than the mean (below average). The Z-score will be zero if the observation equals the mean. Most Z-scores will lie in the range from Z = -2 to Z = 2. Values more than two standard deviations from the mean tend to be extreme values (outliers).
These measures are particularly useful in exploring the potential impact on pensions given that the links between caring (which extends beyond the period a child remains in the family home through its long-term impact on pay, promotion and levels of employment) and its impact on pay and pensions are well established (Ginn, 2003;Price, 2007). The individual-based and state-led care-focused defamilisation characteristics are now outlined.

Individual-based defamilisation
Previous research by Bambra (2007) used three variables to measure defamilisation -maternity leave compensation, compensated maternity leave and relative female labour participation. 6 According to Bambra, the first and the second variables, are indicators of the levels of support women are given when they have children. The relative female labour participation rate shows the extent to which the economy facilitates female employment. The formula for calculating this variable is the male participation rate minus the female participation rate. In accordance with Bambra (2007) we also used these three variables when considering the individual-based defamilisation patterns of the eight countries. In addition, two additional variables are used: the average female wage expressed as a percentage of male average wage and the relative time spent on unpaid work. Given the direct link between capacity to contribute to pensions and income, the average female wage is an important indicator for determining whether women can accumulate sufficient retirement income through participating in employment. The relative time spent on unpaid work is calculated as the difference between the average female and male time spent on unpaid work.

State-led care focused defamilisation
In carrying out his dedomestication study, Kroger (2011) focused on five variables which are found in OECD data sets: the average hours of children's attendance in childcare per week; the right to a childcare place; the childcare cost for a dual earner family (as a percentage of net family income); the childcare cost for a sole parent family (as a percentage of net family income); and the child-staff ratio in formal childcare. Kroger (2011) used the childcare costs for a dual-earner family and the childcare costs for a sole parent family to indicate the affordability of childcare. The child-staff ratio in formal childcare was also used to show the quality of the childcare services and the time replacement rate shows the importance of the childcare services in reducing the family's caring responsibilities. These five variables, using more up-to-date data, were also employed in this study as indicators of state-led care focused defamilisation patterns. Moreover, an additional indicator in the form of public expenditure on childcare was used. This indicator serves to show the financial commitment to the provision of childcare.

Findings
The information provided in Table 1 regarding the individual-based defamilisation indicators highlights that there is considerable variation in the eight countries' experience of these types of measures. It is apparent that the UK tends to fare unfavourably compared with the other countries in relation to these defamilisation indicators. The UK has higher than average Z-scores in the indicators associated with the relative female labour participation rate (1.07) and relative time spent on unpaid work (1.57), and has a low Z-score in the indicator related to the average female wage (expressed as a percentage of male average wage) (-2.05). While the UK compares slightly more favourably in relation to maternity provision it still tends to lag behind most of the other countries and, in particular, Denmark, Norway and Sweden. For instance, Table 1 shows that Denmark, Norway and Sweden have low Z-scores in the indicators concerned with the relative time spent on unpaid work and the relative female labour participation rate (for persons aged 15-64), and high Z-scores in the indicator concerned with the average female wage (expressed as a percentage of male average wage). It is evident that the Nordic countries in particular support female employment and have introduced generous parental leave schemes (Frericks and Höppner, forthcoming). This is also important given that wage penalties associated with time out of employment denies opportunities for significant wage growth and occupational mobility, and leads to a reduction in the retirement income of mothers (Dewilde, 2003;Kahn et al, 2014). Table 2 shows the Z-scores for the six state-led care-focused indicators in the eight different countries. Once again, the UK tends to lag behind the other countries in the study in relation to defamilisation measures. For instance, the UK has a low Z-score in relation to the average hours of children's attendance in childcare per week, and a high Z-score in relation to the childcare costs for a dual-earner family as a percentage of net family income (-1.78 and 2.12 respectively). There is a significant negative correlation between the average hours of attendance per week and the relative female labour participation (r=-0.776, p<0.05), a similar significant negative correlation between the average hours of attendance per week and the relative time spent on unpaid work (r=-0.718, p<0.05), and a significant positive correlation between the average hours of attendance per week and the average female wage (r=0.800, p<0.05). This indicates that greater provision and use of childcare services may assist women's participation in paid work (and their capacity to contribute to pensions). It is equally important to note that there is a significant negative correlation between the average hours of attendance per week and childcare costs for a dual-earner family (as a percentage of net family income) (r=-0.833, p<0.05). This concurs with Samek Ludovici et al's (2016) research, which shows that the negative effects on mothers' pension entitlements are greater in countries lacking affordable full-time childcare facilities.
Once again, the countries identified as belonging to the social democratic group by Esping-Andersen (1990), Denmark, Norway, Sweden and Finland, fared most favourably in relation to these indicators with the conservative countries France, Netherlands and Belgium tending to perform less well, but not to the extent of the UK, which represented the liberal regime. These findings are consistent with the literature on childcare provision which has identified that Nordic countries provide high levels of public childcare with liberal regimes offering more limited provision (Frericks and Höppner, forthcoming). The impact of defamilisation measures on gender and pensions 209

Discussion
It is evident from the data that there are considerable differences in both of the defamilisation indices between the countries. These differences have consequences for women's capacities to undertake paid work and contribute to pensions. For instance, the UK's performance in relation to the relative female labour participation rate is likely to have negative consequences on pensions given that Möhring (2015), using a combination of life-history data from the Survey of Health, Ageing and Retirement in Europe (SHARELIFE), found that on average each additional year in employment increases pension income by 6.7 per cent.
The findings show a correlation between childcare attendance and average weekly wages. This indicates that defamilisation measures such as access to and use of childcare facilities have important consequences for labour market experience. It is widely known that increased wages are also associated with further opportunities to contribute to pensions (Ginn and Macintyre, 2013). As such the extent to which defamilisation measures are employed has implications for the gendered nature of retirement.
The results indicate that cost is an important factor considered by parents in relation to the use of childcare services. In fact, the associations identified in the findings provide support for the existence of a chain of events caused by expensive 1 Those countries that offered a legal right to childcare for all under 3-year-old children were given value 3, countries where this right was limited to same age or user groups or to certain regions were given value 2, countries that offered no right for the service were given value 1.

Source: OECD, 2014
childcare services, which affect women's pension opportunities (Figure 1). At the same time, those women in highly paid employment are more likely to be in a position to afford expensive childcare services. Following the same logic, more women are likely to be assisted to accumulate pension incomes if governments highly subsidise childcare service.
Not all governments give sufficient recognition to the importance of defamilisation measures in assisting women to take part in the paid labour market (and thus accumulate pension incomes through individualised income-based contribution pension schemes). For instance, the findings indicate that the UK's defamilisation measures are limited compared with other countries in the study. Therefore, it is not surprising that the UK (43 per cent) has a wider than average gender pensions gap (Tinios et al, 2015). 7 There is also evidence that there are also considerable disparities between working age men's and women's levels of pension contributions in the UK (PPI, 2016). The UK pension system is also characterised by a relatively minimal redistributive element (Grech, 2015). The gender pension gap (individuals aged 65+) in the Netherlands (40 per cent) and France (39 per cent) is around the EU average, whereas the social democratic countries in the study are all considerably below average (Tinios et al, 2015). Therefore, it is evident that those countries with less generous defamilisation measures are also those countries with the largest gender gaps in pensioner poverty (Möhring, 2015). The impact of defamilisation measures (and increased access to employment) is also influenced by the extent to which pension schemes are earnings-related and reward caring responsibilities through generous care credits. Care credits in relation to childcare are integrated into the state provision of most earnings-related national pension systems in Europe in order to compensate for care-related career interruptions. This is less common in relation to private pensions, with only France of the countries in the study having an occupational scheme that takes into consideration periods of care leave (Samek Ludovici et al, 2016). Other forms of care credits (for older people, the disabled or severely ill family members) are less widespread, only being available in 12 member states including the UK. Generous pension care credits do not tend to sufficiently compensate for care-related cutbacks in working life (Ginn, 2003). They take no account of wage penalties associated with time out of employment, of difficulties when re-entering the labour market, of lost opportunities for skills development, or promotion and career advancement (Leschke, 2011). Möhring (2015) found that the broader level of redistribution in the pension system and gender inequality in work histories, are of much greater significance for income in Parents are discouraged from using the childcare services.
Women spend more time on unpaid work including taking care of their child.
Women face more pressure to quit their jobs for a short period or permanently in order to take care of their child after the end of the maternity leave.
Women have less chance of accumulating pension saving through the individualised income-based pension scheme.
The impact of defamilisation measures on gender and pensions 211 later life among mothers. The existence and generosity of pension care credits does not necessarily indicate how redistributive a pension system is. For instance, care entitlements are increasingly important in earnings-related systems, while they are not required in state pension systems that comprise universal residence-based benefits such as in Denmark and the Netherlands (Samek Ludovici et al, 2016).
A CA approach is useful in considering this relationship between defamilisation measures, care credits and pensions. The core idea of this approach is concerned with an individual's freedom to be and to do what he/she has reasons to value Sen, 1985). According to Deneulin (2011), the CA assumes that individuals have the agency to achieve based on their own understanding of what constitutes a good life rather than one imposed upon them. Following this logic, social policies should aim to increase people's real freedom to choose the ways of life they value instead of imposing a particular way of life upon them. Analysts have discussed various ways of strengthening people's freedom which are particularly relevant to assisting women to accumulate pension income (Clark, 2006;Giullari and Lewis, 2005). First, it is important to note that women should have access to a sufficient amount of negative freedom and positive freedom to organise their lives. Examples of measures that enhance negative freedom are protective policies against sex discrimination in the work place and the provision of basic rights (Walker and Maltby, 2012). Positive freedom is concerned with the resources that enable women to choose the option they prefer. These include factors such as subsidised childcare services. Both of these measures have the potential function of reducing women's caring responsibilities and their time spent on unpaid work, and thus, may give them more time to develop their own careers and pension rights.
Second, Giullari and Lewis (2005) argue that if women are to exercise real freedom to choose there must be equal freedom on the part of men and women to choose between alternatives (for example between work and care). They further argue that the real freedom to choose between paid work and care requires that a comparable value is attached to care. On the basis of this argument, it is important to stress that those women who choose to be a full-time family carer should be afforded the same opportunities to accumulate sufficient pension income for retirement as those who choose to take part in paid employment. This implies that at the same time as supporting women to develop their career and thus accumulate pension income through income-based contributory schemes, governments should provide extensive credits that could protect a family carer's state and private pension record during periods out of paid employment. However, in reality these policies are often limited. For example, in the UK there are concessions or credits provided in relation to state pensions for taking time off work to raise children up to the age of 12 and to act as carers for older relatives, but this does not adequately cover the time women often spend on caring duties. Furthermore, carers do not always claim state pension credits as a result of poor information and care credits are not provided in relation to private pensions, including NEST. Hence, women do not currently have the opportunity to exercise the freedom to choose between work and care without implications for their future pension income.
Therefore, in addition to providing enhanced defamilisation measures, there is a need for pension systems which do not operate in a gendered manner. In practice, pension systems, such as in the UK, institutionalise disadvantages experienced in the labour market by not fully recognising experiences and contributions that are not hetero-patriarchal (Grady, 2015). As a result, policy effort towards increasing provision for retirement through private pensions is likely to lead to greater income inequality between older women and men, and between those who have had an intermittent or low-paid employment history and those with an advantaged position in the labour market. Pension penalties arising from earlier caring roles will be magnified, with increasing income disparity among women in older age (Foster, 2010).
There needs to be a revaluation of productive and socially reproductive activities, valuing the ethic of care as much as the ethic of employment, and rebalancing the gender division of labour. It would be possible to redevelop pension systems in a manner which de-couples income in retirement from paid employment (Strauss, 2014) avoiding the penalty for caring years incurred in private pensions (Ginn and Macintyre, 2013). In effect, we need to move beyond policy that solely reinforces the liberal celebration of better-paid employment in the labour market and explore ways to reward all forms of work, including unpaid labour (Fraser, 2009). One option is to introduce an unconditional Citizens Pension set at an adequate level. This could be part funded through progressive taxation such as the removal of the tax-favoured status for private pension contributions, which exacerbate pension inequalities between men and women (Strauss, 2014). This would reduce gendered inequalities in pensions in retirement. In accordance with a CA approach improvements in defamilisation measures alongside a Citizen's Pension would enable women to choose between work and care, and between different ways of accumulating pension income. It represents a means of enabling women to choose the way of life they value.

Conclusion
Using evidence from the individual and state-based defamilisation indices this article shows that greater moves towards individual responsibility for pension provision may be particularly problematic for women in the UK given the limitations of many of the defamilisation measures identified. There is a danger that pension penalties arising from earlier caring roles and insufficient defamilisation measures will be magnified, with the effect of increasing income disparity between men and women in older age (Blackburn et al, 2015). As such there is a need for policymakers to recognise the very different trajectories often faced by women and not penalise women who diverge from typical male patterns of employment (Grady, 2015). This entails the need to develop strategies to enhance women's real freedom to choose between work and care and different ways of accumulating pension incomes with reference to the core idea of CA. Without such developments, the UK will continue to lag behind many other EU countries in relation to gender inequalities and pension accumulation in the future.
Finally, it is worth identifying what additional research in this area is required. First, despite limitations in available data, it would be beneficial to explore the importance of additional defamilisation measures to assist women to take part in the paid labour market. These include care services for older people and fathers' only leave. Second, it would be beneficial to explore trends in additional countries including those outside Europe. Third, it is important to acknowledge that while it is possible to statistically analyse the impact of prior events, including having children, on current pensions (see Möhring, 2015;Sefton et al, 2011), exploring the implications of current caring responsibilities and employment on pension receipt in this manner is challenging given the uncertainties in future life-course events and pension policy measures. It points to the need for the use of appropriately linked statistical and administrative databases in order to develop further comparative pension micro simulation data. Finally, as pensions can play an important role in preventing inequalities in working-life from persisting in older age by moderating the relationship between an individual's working life and status in retirement, there is a need for further in-depth discussion regarding the implementation of pension schemes which limit the impact of caring responsibilities on income in retirement.

Notes
1 While 65 years and older is commonly used in the literature when discussing pensioners, it is important to recognise that the age at which a pension is received varies in different countries. Furthermore, in some countries women are entitled to a state pension at an earlier age than their male counterparts. 2 There is also less reliable data on such forms of care, which often goes unrecognised. For example, it is not always formally perceived as a form of care by those who carry out such duties (Loretto and Vickerstaff, 2015). For these reasons, it is not explored in this analysis. 3 Esping-Andersen's project (1990) also covers four other European countries: Ireland, Austria, Switzerland and Italy. However, there is a lack of sufficient comparative data for us to study their individual-based defamilisation patterns or their state-led care-focused defamilisation patterns or both. 4 The 1995 United Nations Fourth World Conference on Women, governments adopted the Beijing Declaration and Platform for Action, which seeks to promote and protect the full enjoyment of all human rights and fundamental freedoms of all women throughout their life cycle. Guided by these principles the World's Women report presents statistics based on a comprehensive and careful assessment of a large set of available data from international and national statistical agencies (United Nations, 2014). 5 This statistical technique provides a simple way of summarising comparative data giving considerations to both rank order and the degree of dispersion. A Z-score is calculated using the equation: where Xi is the individual observation, X is the mean observation for all of the countries and SD is the standard deviation for all of the countries. A Z-score measures the number of standard deviations an observation is away from the mean. 6 The maternity leave compensation for duration is weighted in terms of the percentage of the normal wage a female worker receives during the compensated maternity leave period. The compensated maternity leave duration is measured by the number of weeks of paid maternity leave. 7 It is important to recognise that current pensioner's retirement income is related to their specific employment and pension contributions as well as the impact of pension policies throughout their life-course. As such it is not possible to accurately link current pensioner's retirement income to current worker's employment opportunities.