Gender and proposed Auto-enrolment in the Republic of Ireland: Lessons from the UK

The introduction of “ soft ” compulsion in the form of Auto-enrolment into non-state pensions has been seen as a key policy response to the challenges presented by an ageing population and concerns about under-saving for retirement in the UK. Since its introduction in 2012, amongst eligible employees in the private sector, pension participation had risen by over 31 percentage points to 73% of eligible employees in 2016. Despite these trends, Auto-enrolment in the UK has not been without criticism, particularly in terms of its exclusion of certain groups, including carers, amongst whom females are over-represented. The Republic of Ireland (ROI) has recently announced its intention to implement an Auto-enrolment pension scheme. As such, this article examines the UK's experience of rolling out Auto-enrolment policy and considers lessons that could be learned by the ROI from the UK in its pursuit of Auto-enrolment, with a particular focus on women's pensions. Initially it outlines the current Irish pension system, the gendered nature of pensions, and the proposed Auto-enrolment system in ROI. Then it discusses the UK's experience of Auto-enrolment, with a particular focus on gender, before examining the lessons the ROI can learn from the UK's Auto-enrolment policy in relation to women and pensions. Finally, it concludes that Auto-enrolment alone will not resolve the gendered nature of pensions in the ROI and Auto-enrolment in the ROI.


| INTRODUCTION
Concerns about the future sustainability of pensions have received considerable attention over recent years in the context of population ageing. This has informed pension policy reform debates across the globe. Along with alleged macroeconomic disutilities of public pensions, the notion of a "demographic timebomb" hasbeenusedto legitimate retrenchment, in addition to promoting the role of the private sector in pensions (Ebbinghaus, 2015).
This pension crisis narrative has emphasised the need to encourage individual responsibility for pension saving, in order to limit "under-saving" for retirement (Foster & Heneghan, 2018;Ginn & MacIntyre, 2013). One of the key policy developments to encourage additional (private) pension saving in the UK has been the employment of "soft" compulsion through the introduction of Auto-enrolment into non-state pensions. Since October 2012 (depending on employer size), workers in the UK within a particular age and income bracket have been automatically enrolled into workplace pensions (with the option to opt-out). The UK is one of the four countries (New Zealand, Italy and Turkey are the others) to have introduced Auto-enrolment at a national level, although there are plans to introduce Auto-enrolment in Poland, Russia and Lithuania (Pension Policy Institute [PPI], 2017). In addition, the Government of the Republic of Ireland (ROI) has recently proposed the implementation of a state sponsored supplementary retirement savings system by 2022, whereby workers will be automatically enrolled (The Government of Ireland, 2018).
The ROI, in its "National Pensions Framework" (Department of Social and Family Affairs, 2010), targeted a figure of 70% of the working population aged between 30 and 65 making private pension provision. Against the backdrop of not making this target, Auto-enrolment has been proposed as a way of helping achieve this goal. A national system of Auto-enrolment in the ROI is seen as the middle ground between a system in which saving into a private pension is compulsory (such as in Australia and several Latin American countries), and a system where private pension saving is voluntary (such as in much of Western Europe). It is perceived as a way of reducing the burden on the state, and increasing pension coverage by moving away from a purely voluntary approach to retirement saving (The Government of Ireland, 2018). This is significant given that more than half (54.2%) of workers without occupational pension coverage from their employers, state that their employer does not offer any pension provision (CSO, 2019a). It is seen as a strategy to immediately increase pension saving, thereby reducing the proportion of the working age cohort with insufficient resources in retirement.
The ROI has taken a close interest in pension developments in the UK and, in particular, the coverage rate of Auto-enrolment. Auto-enrolment has been largely hailed as a success in terms of increasing the number of people saving for retirement in the UK (Prabhakar, 2017). Since its introduction in 2012, amongst eligible employees in the private sector, pension participation had risen by over 31 percentage points to 73% of eligible employees in 2016. It has been stated that it has reduced levels of under-saving for retirement by 2 million people (DWP, 2017) with annual saving in workplace pensions predicted to have increased by £17 billion by 2019/20 (Thurley, 2019). Furthermore, the largest increases in those saving into private pensions have been witnessed amongst those traditionally with least access to workplace pensions including: the lowest earners; younger people (especially aged 20-29) and women (DWP, 2017).
Despite these trends, Auto-enrolment in the UK has not been without criticism, particularly in terms of its exclusion of certain groups, in addition to uncertainties regarding the pensions people will receive in retirement (Ginn & MacIntyre, 2013;Thurley, 2019). Although it has increased the number of low earners contributing to private pensions, significantly more of which are women than men, it also operates in a manner which excludes many marginalised individuals, including those not in paid employment (Grady, 2015). This has implications for gendered inequalities in later life (Strauss, 2014). These gendered inequalities in older age relate to women's more diverse employment histories, lower earnings and more limited engagement with pension planning throughout the lifecourse (MacLeod et al., 2012). These gendered inequalities in pensions are not specific to the UK. In most countries, including the ROI, women are less likely than men to receive a pension, and when they do, levels are usually lower.
Across European Organisation for Economic Co-operation and Development (OECD) countries and the USA, pension payments to individuals aged 65 and over were 28% lower, on average, for women than for men (OECD, 2015). In ROI this gender pensions gap is 26.1%, which is actually lower than the UK figure of 34.8% (European Commission, 2018).
There is a danger that a greater reliance on private pensions that transmit market-income inequalities from working life to retirement will ultimately widen the gender gap in pensions. Therefore, it is important to consider the potential gendered nature of pension reforms, including the introduction of Auto-enrolment in the ROI. This is especially important given that there is a tendency for pension reforms to be gender blind (Duvvury, Ní Léime, Callan, Price, & Simpson, 2012;Grady, 2015). This article represents the first to examine the UK's experience of rolling out Auto-enrolment policy and consider lessons that could be learned by the ROI from the UK in its pursuit of Autoenrolment and in particular, the implications for women's pensions. In order to do this, initially the article outlines the current pension system in the ROI and the challenges it faces in relation to women's pensions, before exploring the proposed Auto-enrolment system in ROI. The article then discusses the UK's experience of Auto-enrolment since its introduction in 2012, with a particular focus on gender. It examines the lessons the ROI can learn from the UK's policy in relation to women and pensions, before concluding that Auto-enrolment alone will not resolve the gendered nature of pensions.

| THE PENSION SYSTEM IN THE ROI
The Irish pension system is a multi-pillar system (Hughes & Maher, 2016;Mulligan, Wijeratne, & Maher, 2019;Natali, 2008), in which private pensions play a long-standing role. Pensions were initially introduced in the ROI based on UK's Old Age Pensions Act of 1908, as a state (means-tested) pension (Pillar 1), payable from the age of 70. In 1960, the ROI's state pension was reformed and a mandatory contributory (Pillar 2) pension (flat-rate) was introduced, and paid to all employees who met the social insurance contribution conditions required for a contributory pension. The original means-tested pension is still retained as a form of security for people without sufficient social insurance contributions. In addition, if individuals preferred and were financially capable of supplementing their state pension, they could do so through occupational or personal pensions (Pillar 3). Pillars 2 and 3 are voluntary, and despite this, some employers make membership a condition of employment, such as in public services (Hughes & Maher, 2016).
Despite a number of attempts to change the Irish pensions system following the introduction of the contributory social insurance pension in 1960, it went through relatively little structural change (Bridgen, 2019). In the 1970s the system saw a movement towards strengthening the first pillar pensions, while the 1980s represented a critical juncture in policy terms, with individuals encouraged to take responsibility for their pensions using market-based solu- recommendations were made that an element of compulsion was required to widen coverage of occupational pensions and reduce reliance on the state. As a result to address this, in 2015 a Universal Retirement Savings Group was established. This group made recommendations for a roadmap and a timeline for the introduction of a universal retirement savings scheme (Department of Social Protection, 2015). This has significant potential implications for pensions in the ROI, including the circumstances of female pensioners.

| WOMEN AND PENSIONS IN ROI
Similar to many countries, the ROI pension system is facing considerable challenges. For instance, the ROI has an ageing population and high levels of insecure employment (Maroukis, Iglicka, & Gmaj, 2011), which have implications for pensions sustainability. Furthermore, there are challenges in terms of inequalities in the distribution of pensions in the ROI. Women, for example, tend to have a lower level of pension than their male counterparts, with only 35% of women currently receiving the full rate of state pension (Department of Social Protection, 2015). The gap is also significant in terms of second and third tier pensions (occupational and personal pension). Although amongst employees, pension coverage (including occupational pension coverage from current and previous employments, and personal pensions), for men and women is fairly similar (57.1% and 55.4%, respectively) (CSO, 2019a) overall, using data from the 2014 Survey on Income and Living Conditions (SILC), Collins (forthcoming) found that men contribute between 30 and 35% more than women to their private pension (with that gap persisting across earners in most age groups). Furthermore, amongst retired people, 55% of retired men receive a private or occupational pension, compared to only 28% of women (Nolan, Whelan, McGuinness, & Maître, 2019).
Women's employment characteristics have traditionally adversely affected these pension patterns in the ROI (Ní Léime, Duvvury, & Finn, 2017). For instance, 40 years ago women were paid approximately half the salary of men, and from the 1930s to 1973 many women were not allowed to work after they married as a result of a marriage bar.
These developments have a long-term effect on pension contributions. Even today, although the marriage bar no longer exists, and women are more involved in work, inequality is a persistent feature of women's position in Irish society (Duvvury, Ní Léime, & Callan, 2018). In particular, gendered norms of care still limit the extent to which many women can seek and maintain paid employment (Duvvury et al., 2012). Collins (forthcoming) found the gender contributions gap for women with children at home is 39% compared to 30% for women in households without. Pension coverage is also lower amongst part-time workers (who are over-represented amongst women workers-almost 70% of all part-time workers are women [NWCI, n.d.]), with just 36% of part-time workers having pension coverage compared to 61.4% of full-time workers (CSO, 2019a). The employment rate for women in the ROI in 2016 was 59.5%, compared with 69.9% for their male counterparts, creating gender divisions in access to work-based pension schemes. This was below the EU average of 61.4% for women (CSO, 2016). Women in ROI represent 60% of all those who are low paid (Collins, 2016). Of the 137,300 employees who reported earning the National Minimum Wage or less, 75,900 (55.3%) were female while 61,300 (44.7%) were male (CSO, 2019b).
These underlying divisions in men and women's employment and income are a key driver of gender differences in pension contributions in ROI. As such, moves to encourage individual responsibility in relation to pensions can have particularly negative implications for women, given gender inequality in employment patterns often transforms into gender differences in coverage of private pensions (Jethwa, 2019). There is a danger in policy adopting a onesize-fits-all approach which is based on the assumption that work is freely available and accessible, and provides a sufficient income to contribute to a pension. Unfortunately, these norms do not hold for many women (Ni Leime & Street, 2017). In the state system some recognition of caring commitments is provided through care credits. Individuals with previous social insurance contributions through work, who subsequently spend up to 20 years out of the workforce caring full-time for children (up to the age of 12), or an incapacitated adult, can have these periods disregarded when calculating the yearly average social insurance contribution for the purposes of qualifying for a state contributory pension. However, there is still a requirement to have paid 520 contributions, which is equivalent to 10 years of social insurance contributions to qualify for this contributory state pension, which can be difficult for women who are working part-time and in precarious employment (Wijeratne, Mulligan, & Maher, 2018).

| PROPOSED AUTOMATIC ENROLMENT IN THE ROI
In order to enhance levels of individual saving for retirement the Irish government has proposed the introduction of a system of Auto-enrolment pensions from 2022. Taoiseach Leo Varadkar, head of the Irish Government, stated that the Government wanted to "create a new and necessary culture of personal retirement saving in Ireland" (The Department of Taoiseach, 2019). The Department of Employment Affairs and Social Protection (DEASP, 2018) launched the process of developing the policy. The proposal, which is called the "StrawMan," involves the government providing a proposal which is then refined through public consultation and a review by the Interdepartmental Pensions Reform and Taxation Group (IDPRTG), in order to enhance its effectiveness (DEASP, 2018). Similar to the UK, the Auto-enrolment proposal entails anyone fulfilling a defined eligibility criteria without a private pension plan (people already paying into a private scheme will not be enrolled), being automatically entered into a retirement savings scheme, which they could subsequently choose to leave.
The proposed introduction of Auto-enrolment stems from a concern that other policy mechanisms have failed to sufficiently encourage further saving for retirement (The Government of Ireland, 2018). There is a concern that many people will not be able to sustain their desired standard of living when they retire (The Government of Ireland, 2018). Immediate needs and desires mean sacrifices are often made to people's pension saving (Quilgars, Jones, & Abbott, 2008). Competing demands at different points of individuals' lives, such as student debt and housing costs, or childcare costs, can mean pension saving is not always prioritised (Foster & Heneghan, 2018). As such, persuading people that saving for retirement is both affordable and desirable can be challenging. Behavioural barriers, including myopia, cynicism and inertia, may still stymie increases in saving even when saving is affordable (Wicks & Horack, 2009). Furthermore, the complexity of pensions can deter decision-making leading to inaction. This can result in financial needs in retirement not being addressed until it is too late to significantly change the outcome (Foster, 2017).
Jerry Moriarty, chief executive of the Irish Association of Pension Funds, stated that the pensions sector had been "crying out" for simplification in the ROI (Reeve, 2018). This was deemed to be especially important due to low levels of financial literacy and numeracy (Nolan et al., 2019). This can lead to financial decisions being made which are not consistent with financial needs (Lusardi & Mitchell, 2014). For instance, employees often do not take advantage of employer matching contributions; and are risk averse (especially women) when faced with asset allocation and portfolio choices, but alternatively are risk-seeking in the face of losses (Clark, Strauss, & Knox-Hayes, 2012;Mulligan et al., 2019).
Insufficient saving and limited pensions knowledge has been used to advocate the "libertarian paternalism" inherent in Auto-enrolment (Clark & Strauss, 2008). This process may not conflict with liberal values (underlying policy developments in the ROI and UK), but instead represents a form of "asymmetric paternalism" whereby individuals can still make their own choices by positively rejecting the defaults in Auto-enrolment (Mabbett, 2012). In a sense, within a libertarian paternalist framework, the state plays the role of enabler and, paradoxically, commands individual choice and responsibility (Strauss, 2014). Auto-enrolment is associated with the concept of "nudge," linked to the view that behaviour does not always represent an entirely rational or instrumental response to the situation in which someone finds him/herself. Thaler and Sunstein (2008) have advocated an approach whereby "choice architecture," a central component of the market economy, is amended in order to counter the effects of irrational consumer behaviour.
Concerns about the future burden on the state as a result of under saving for retirement in the ROI have fuelled the call for Auto-enrolment. Currently two-thirds of private sector employees in the ROI have no workplace pension and are not saving for their retirement, which has created concerns about poverty levels in older age (Duvvury et al., 2018). According to the proposed system, employees aged between 23 and 60, earning more than €20,000 a year and not already contributing to a workplace pension, would be automatically enrolled in the new system (The Government of Ireland, 2018). The self-employed and employees outside the age and earnings-band thresholds would not be automatically enrolled but have the option to "opt-in" if they wished to do so. Employees can opt-out of the system at the end of a minimum membership period (during months seven and eight of membership) and cannot opt-out after this period, but they may be able to suspend or temporarily stop their payment of contributions.
This will be decided at a later stage (The Government of Ireland, 2018). The opt-out period is notably a narrower window compared to options available in other countries such as the UK.
Employees will be free to choose from a range of retirement savings fund options from the registered providers.
The system will have four providers with three pension plan options with three risk profiles (low, medium and high), which will give 12 choices for each employee. If an employee does not select a registered provider or fund of their choice, the system will allocate their payments to one of the dedicated funds on their behalf. This default option may be common as people struggle to choose from the various choices, especially given evidence of limited pensions literacy (Mulligan et al., 2019;Nolan et al., 2019).
In the proposals employees who are members of the system will initially contribute a minimum of 1% of gross earnings-that is, earnings before tax and Pay Related Social Insurance (PRSI) deductions. The member's minimum contributions will be increased on a phased basis as the Auto-enrolment system is rolled out at 1.5% for workers and employers for 3 years, starting in 2022, to reach a maximum of 6% each at the beginning of 2032 (10 years later).
Those enrolled in the Auto-enrolment system in the years following 2022 would invest the same percentage as everyone else. The government plans to maintain administrative fees for all registered providers and funds as low as possible. The current plan is that the most that can be charged for the yearly management of the funds will be 0.5% a year of a member's savings fund. High administration fees can reduce the income in retirement substantially. Furthermore, economies-of-scale can be found by having the collection of fees administered by a single organisation (or a limited number of approved funders) rather than a large number of pension providers (Tapia & Yermo, 2008).
Like in the UK, the Government will also provide an incentive to save. The level of this incentive and how it might be paid will be finalised following the consultation process. At the moment, the government is suggesting that the state contributes €1 for every €3 saved by the member. Registered providers will deliver services on a Defined Contribution (DC) basis (where the risk is borne by the employee given that the value of investments can go up or down), rather than as Defined Benefits (DB) (where employers take the investment risk). One of the advantages of the new system, if it is implemented, will be that each member will be able to take their account or "retirement pot" when they move jobs, operating on a "pot-follows-member" approach. People will be able to begin to draw down benefits from the system when they reach state pension age (this age was increased to 66 in 2014 (Duvvury et al., 2018)). There will not be any flexible retirement options or allowances made for withdrawals from the fund prior to retirement age (The Government of Ireland, 2018).
Employers will be provided with support in delivering on their duties in the roll-out of Auto-enrolment. They will have to enroll employees and organise the transfer of contributions to the Central Processing Authority (CPA). However, they will not have the responsibility of selecting a registered provider or savings fund option for their employees, given that employees will select their own registered provider. Employers will match the member's contributions up to an eventual maximum of 6% (initially limited to 1% of the member's earnings to a ceiling of €75,000, increasing at 1% a year to a maximum of 6% from year six). There has been some concern regarding the employer contribution rate. For instance, the Irish Small Firms Association, which represents a diverse membership of businesses with less than 50 employees, view 6% as too high for smaller companies (Public consultation on Strawman Automatic Enrolment proposal in Ireland, held in Galway, October 15, 2018). The employer contributions will continue to be deductible for corporation tax purposes. If employers fail to enroll their employees, or deduct and transfer contributions as required, they will have to pay a penalty.

| THE UK'S AUTOMATIC ENROLMENT EXPERIENCE
Auto-enrolment was introduced gradually in the UK from October 2012, with workers in a particular age and income range automatically enrolled into workplace pensions (with the option to opt-out). Each employer was allocated a staging date to rollout Auto-enrolment for employees based on the PAYE (Pay-as-you-earn income tax payroll) size in 2012. It entails automatically enrolling eligible individuals (generally low-to-median earners) without access to good quality workplace pensions into a low cost portable occupational pension. At the same time, it allows existing schemes, with benefits or contributions above the National Employment Savings Trust (NEST) (the default option Auto-enrolment scheme) minimum, to continue. Minimum contributions have been gradually increased since its introduction and are set at 4% for the employee, 3% for the employer and 1% in tax relief in 2019. There is an assumption with Auto-enrolment that, generally, people understand it is necessary to save for retirement yet are not prepared to proactively enroll in a pension (MacLeod et al., 2012).
It was envisaged that financial incentives of investment growth, along with employer and government contributions would persuade most of the auto-enrolled who are tempted to opt-out to resist (Wicks & Horack, 2009). Given the low levels of opting out so far this appears to be largely the case (DWP, 2017).
Employees may opt-out through an opt-out notice. They have up to one calendar month to opt-out once they are enrolled in a scheme if they want to receive a full refund on their contributions. If people opt-out at a later stage, then their payments may stay in the pension until they retire. They will be re-enrolled after about 3 years and re-assessed if they change employment (The Pensions Regulator, 2017). A 0.75% charge cap is in place for accumulation in Auto-enrolment.
There are certain conditions which need to be fulfilled to be eligible for Auto-enrolment. First, people need to be aged 22 years or over, but below state retirement age. Second, people need to earn within particular income boundaries. The annual earnings threshold above which every employee should be automatically enrolled is £10,000 (2019/20) (within a single employment). If an individual is not auto-enrolled into a workplace pension because their earnings are below the trigger, they have the right to opt-in and receive an employer contribution if they earn above the lower limit of the qualifying earnings band (over £6,136). Contributions are payable on band earnings up to a limit of £50,000 (2019/20) (Thurley, 2019).
Overall, Auto-enrolment has been successful in encouraging greater numbers of people to contribute to a pension (PPI, 2018), with the largest increases in those contributing to a pension witnessed amongst the youngest age cohorts (Cribb & Emmerson, 2016). Thus far the average opt-out rate for auto-enrolled employees has been fairly consistent since its implementation at around 9% (DWP, 2017). Lack of affordability or alternative provisions for retirement, (especially amongst older cohorts) have been cited as the main reasons for opting-out (DWP, 2014).
Pension freedoms were introduced in April 2015. These are likely to have significant implications for the dissimulation phase of Auto-enrolment pensions. In many ways these freedoms could be seen as diametrically opposed to the libertarian paternalism inherent in Auto-enrolment. The pension freedom reforms mean that people aged 55 years or over now have far greater freedom over how to access their DC pension pots. Previously, themajorityofpeoplewererequiredtopurchaseanannuity (commonly purchased through their provider). While annuities can still be purchased, there is now the flexibility to take their whole pensions pot in cash, leave funds invested while withdrawing income through drawdown, or leave the whole pot untouched (House of Commons Work and Pensions Select Committee, 2018). Given that Auto-enrolment will see the expansion of DC pots, pension freedoms will take on an increasing significance. How informed people are about appropriate investment decisions will be vital. The Financial Conduct Authority (FCA) (2017: 15) was "particularly concerned that consumers end up with an investment strategy that may not be suitable given the consumers" risk tolerance and what they intend to do with their pots in the future'. It has been suggested that impartial guidance appointment immediately prior to accessing a pension pot would result in better consumer outcomes (House of Commons Work and Pensions Committee, 2017).

| GENDER DIMENSIONS OF AUTOMATIC ENROLMENT IN THE UK
It has been claimed that Auto-enrolment will "help more women to save for later life" (Webb, 2014). This is important given that women's greater provision of family care in the UK (and beyond) often leaves them with inadequate pensions (Foster, Heneghan, & Wijeratne, 2019). Thus far Auto-enrolment in the UK has helped participation rates for women start to catch up with those for men. In 2012, 40% of eligible women in the private sector saved into a workplace pension, rising to 73% in 2016 (DWP, 2017). Despite increasing the number of women saving into private pensions, research by Now: Pensions (2018) suggests pension inequalities will persist, and in 40 years' time pension pots will have grown to £50,514 for men and £40,332 for women (assuming a 3% investment growth). If a five-year career break is applied, this figure falls to £33,986 for women (this is based on Auto-enrolment minimum contributions and takes into account mean salaries for men and women over their working lives). Auto-enrolment still excludes a greater number of women than men, given that it provides no credits for times out of employment and excludes the lowest earners (PPI, 2018). 1.2 million women (in relationships) with dependent children are currently looking after their family and are missing out on Auto-enrolment contributions, with an additional 1.4 million employed mothers with dependent children not earning above the £10,000 threshold to qualify for Auto-enrolment contributions (Jethwa, 2019). The PPI (2018) found that over three quarters (77%) of employees earning less than the Auto-enrolment trigger are women. NOW: Pensions' Director of Policy, Adrian Boulding, has referred to this as a form of "covert sexual discrimination" (Now: Pensions, 2016). Furthermore, almost 106,000 working people are not being auto-enrolled into a pension because their earnings come from more than one job and 70% of these are women (Citizens Advice, 2017). Over 50% of part-time workers earn less than the Auto-enrolment trigger and 81% of part-time workers are women (DWP, 2017). The PPI (2016) stated that 6% of men and 22% of women could be made eligible for Auto-enrolment if the earnings trigger and earnings bands were to be removed. There is also a concern that with higher statutory minimum employee contributions phased in, this could ultimately drive opt-outs (Jethwa, 2019;WBG, 2020). Women are already more likely than men to state that they have opted out of a pension because of other financial commitments (37% compared to 21%) (Scottish Widows, 2019). Gender divisions may increase as women's more limited earnings are stretched further.
So the current structure of automatic enrolment means there are gaps in coverage, in particular for those in low paid jobs (disproportionately women) and younger workers (it excludes those under 22) (DWP, 2017). However, even when people have access to Auto-enrolment it is unclear whether it will pay for people to save into Autoenrolment schemes. The default Auto-enrolment option, the NEST, a DC scheme, operates in a manner where there is no guarantee the fund at retirement will exceed the value of contributions. For women, the risk of a wrong decision is especially high given that more unpredictable careers combined with the uncertainties of investment returns, charges and annuity (WBG, 2020). Decision-making is especially challenging where individuals, especially women, have unpredictable life-course trajectories (Price, 2015) and have limited knowledge and understanding of where to locate appropriate impartial advice (Foster, 2012). Furthermore, women have lower levels of self-reported pension knowledge than their male counterparts (Scottish Widows, 2019), with 41% of men stating they know "a lot" or "a fair amount" about workplace pensions compared to 33% of women (DWP, 2020).
Recommendations in a review of Auto-enrolment by the DWP (2017) proposed to change the framework for Auto-enrolment so that pension contributions would be calculated from the first pound earned, rather than from the lower earnings limit, currently set at £6,136 (2019-20), to benefit low earners in particular. Removing the lower earnings limit would incentivise those with low earnings (predominantly women) to opt-in to a workplace pension as employers would contribute from the first pound (Jethwa, 2019).
It has also identified the need to explore coverage of those in multiple jobs (predominantly women) to encourage them to opt-in to their workplace pension scheme. The PPI (2016) calculates that if the income from both first and second jobs was taken into account when assessing eligibility for automatic enrolment a further 80,000 people (60,000 women and 20,000 men) would be included. The position of self-employed workers has also been considered and the need to encourage additional saving. Finally, in a move which would encourage saving amongst younger people, the DWP (2017) proposed reducing the lower age limit for Auto-enrolment from 22 to 18 years. This may be particularly important for women who are more likely to have interrupted employment histories resulting from caring responsibilities.
These potential developments do not detract from the fact that there are doubts about whether people contributing to auto-enrolled pensions will save enough for an adequate income in retirement (Ginn & MacIntyre, 2013).
Within Auto-enrolment there are concerns that the contribution rate of 8% is too low. The PPI (2017) conducted an international comparison of Auto-enrolment schemes and found that UK employment contributions lag far behind those of its European partners. For instance, someone earning £28,900 would need to put away £3,250 a year (11.2% of salary), to even have a chance of a comfortable retirement (Morley, 2014). Given that Auto-enrolment only requires 8% contributions, this leaves a 3.2% gap for this salary group. The lowest paid, amongst whom women are in the majority, tend to save a smaller proportion of their income than men. Webb (2016) advocates a system of "automatic escalation", where individuals pre-commit part of any future pay-rise to increasing their rate of pension contributions. For those with lower incomes, this escalation amount could be capped at a level proportionate to their income band (Turner, Hughes, & Maher, 2015). Given that men are more likely to be in paid employment and have higher earnings than women, they would be likely to experience the greater impact from increased contribution levels (Jethwa, 2019).
In its current form there is a danger of burdening workers, especially women, with the risk and responsibility for assuring an adequate income in retirement without guaranteeing adequacy (Clark et al., 2012). Thaler and Benartzi (2004) state that the inertia that rationalises why Auto-enrolment has the effect of increasing participation rates can also lead to low saving rates for those who do participate. This is because once workers are enrolled into a scheme they tend to stick to the same savings rate and do not look to increase their contributions or undertake additional saving. Furthermore, default options are commonly interpreted as an endorsement by the government, or indeed their employer, as to where they should contribute into a pension scheme and to what extent (Lusardi & Mitchell, 2014). As such, there is a danger that people will end up with lower levels of pension saving than expected. This is particularly problematic for those on low earnings.

| AUTO-ENROLMENT AND GENDER IN ROI: LESSONS FROM THE UK
There are a number of lessons to be learnt by the ROI in relation to Auto-enrolment from the UK's experiences.
While the dominant account of Auto-enrolment suggests that a "level playing field has been created" as the ability to independently save has been extended to women, we have shown that this is only part of the narrative in relation to Auto-enrolment in the UK (Grady, 2015). The plans to introduce Auto-enrolment in the ROI by 2022, will undoubtedly increase participation in workplace-based pension schemes, by overcoming barriers such as listlessness and complexity, which discourages opt-in participation. However, similar to the UK, there is a danger that the manner in which it operates will lead to the exclusion of a greater number of women than men. Table 1 sets out some of the key features of Auto-enrolment in the UK and the proposed features in the ROI, and whether these are likely to have a gendered impact. For instance, it shows how the proposed system, as in the UK, will, unlike state pensions, offer no credits for periods of unpaid care. This will create gender disparities given that women are more likely to undertake caring responsibilities than men (Duvvury et al., 2012).
The DC nature of the proposed system in the ROI makes it difficult to determine whether it will definitely pay to save. Similar to the UK, the risk of inappropriate decision-making is particularly problematic for women, since they have unpredictable careers and relationship statuses, combined with uncertainty regarding future returns on the auto-enrolled pension funds. These trends make appropriate decision-making regarding pensions challenging. In particular, those individuals who are some distance from retirement struggle to identify how much pension saving will be required to ensure the standard of living they require in retirement (MacLeod et al., 2012). The proposed system in the ROI involves the use of 12 different options depending on how risk averse individuals are, with a default option also provided. Given limited pension knowledge, particularly amongst women, a lack of understanding and availability of impartial advice could lead to inappropriate decision-making. This indicates the need for further impartial information tailored to individual requirements explained in clear language.
The planned eligibility criteria for Auto-enrolment-€20,000 a year-will disproportionately exclude more women than men given that women are more likely to work in lower paid positions (Maroukis et al., 2011). It is also considerably higher than the minimum eligibility criteria in the UK. In the ROI, the largest pensions contribution gap between men and women of 40%, is actually found amongst earners with an annual gross income of between €10,000 and €18,000 (Collins, forthcoming), who would not be auto-enrolled with this threshold. There is a need to ensure that pension contributions would be calculated from the first pound earned, something that the UK is now exploring (DWP, 2017). This will benefit low earners the most, amongst whom women are over-represented. The overall levels of contributions proposed by the ROI are higher than the UK (14% of salary). In the UK concerns have Yes-given women's lower earnings on average than men the higher earnings threshold excludes more women than men from automatically being auto-enrolled. This is potentially particularly problematic in the ROI.
Does the auto-enrolment qualifying figure include multiple forms of employment?
No No Yes-given that women are more likely to undertake multiple (low paid) forms of employment in the UK and the ROI they are less likely to be automatically included in the auto-enrolment threshold.
Are self-employed workers automatically auto-enrolled?

No No
Yes-men are more highly represented amongst self-employed workers in the UK and the ROI and, as such it will impact on more men than women. However, both countries have witnessed increasing proportions of women become self-employed.
Are credits provided during times of caring?

No No
Yes-caring for children and families is much more common amongst women than men in the UK and the ROI.
Employee contributions 4% of earnings 6% of earnings Yes-The ROI's proposed contribution level is higher than in the UK. Higher contribution rates have the effect of increasing the prospect of pension adequacy, but they also increase inequalities between men and women's pensions given men's greater likelihood of being employed and contributing to a pension. They may also result in greater numbers of women (with lower incomes) opting out.
been expressed about whether this contribution rate is too low (Grady, 2015;Thurley, 2019). This is something the ROI will need to monitor.
In the UK the coverage of those in multiple jobs (predominantly women) has been identified as a challenge and is something the ROI is likely to find problematic, especially given the proposed eligibility criteria of €20,000 a year, which is higher than the UK. The ROI has also proposed that it will be implemented with employees aged between 23 and 60 years (The Government of Ireland, 2018). Pension coverage in ROI is currently lowest amongst younger workers, with only 16.3% of those working aged 20-24 years old contributing to a private pension (CSO, 2019a), so this would have limited impact on those least likely to contribute to a pension. Given that women are more likely to have interrupted employment histories, it would be beneficial if Auto-enrolment was introduced for younger individuals and up until the pension age. In the UK, the Government's review of Auto-enrolment in 2017 determined that the upper age limit for Auto-enrolment should still be aligned to the State Pension age (SPA) (with individuals able to opt-in to continue contributing) (Thurley, 2019). However, in a move which would encourage saving amongst younger people, the DWP (2017) has proposed reducing the lower age limit for Auto-enrolment from 22 to 18 years. The ROI should consider following this approach.
The other group that have been identified as being in a problematic position in the UK in relation to Autoenrolment are self-employed workers (DWP, 2017), where the proportion of self-employed people actively contributing to a pension has decreased steadily since the late 2000s, from 27% in 2008-09 to 15% in 2017-18 (DWP, 2020). In the UK the changing nature of self-employment is likely to exacerbate under-saving amongst many women in particular, as the number of female self-employed workers is growing at a faster rate than men (34% of the self-employed workforce are women, compared to 27% in 2005 (Scottish Widows, 2019)). Self-employed men and those with higher income are also more likely to save into a pension (DWP, 2020). In the ROI the self-employed would also not be auto-enrolled. Although the self-employed are a smaller group than employees in ROI, they are more concentrated on lower incomes, with 33.1% (approximately 80,000) earning less than €10,000 per annum and 32.3% (78,000) earning between €10,000 and €25,000 (Collins, 2016). In the ROI approximately half of selfemployed workers aged 20-69 years had pension coverage (including from previous employment or personal pensions) compared to 56.3% of all workers aged 20-69 years (CSO, 2019a). This is an area which requires further exploration given increasing levels of self-employment in the ROI, including amongst women.
There will be more limited opportunities to opt-out of the proposed Auto-enrolment scheme in the ROI than the UK. In the UK employees have up to one calendar month to opt-out once enrolled if they want a full refund. If people opt-out at a later stage, then their payments may stay in the pension until they retire (The Pensions Regulator, 2017). In the ROI it has been proposed that employees will be able to opt-out of the system at the end of a minimum membership period (during months seven and eight of membership) but employees cannot opt-out after this period, though they may be able to suspend or temporarily stop their payment of contributions (The Government of Ireland, 2018). Given that unexpected financial needs (which may be in relation to changing family circumstances) can arise, and are more likely to affect women, it is important that there is some flexibility or "choice" within the system. However, if people do wish to opt-out of the scheme it is important that they are made aware of the consequences.

| CONCLUSION
Whilst Auto-enrolment has led to an increasing number of women contributing to workplace pensions in the UK, it is too soon to hail it an overwhelming success. Auto-enrolment still excludes more women than men. The gender gap in private pension accumulation remains substantial (WBG, 2020). Women currently earn on average approximately 18% less than men in the UK, which could contribute to a reduction of pension wealth of 28% for those approaching retirement (Jethwa, 2019). In its current form, Auto-enrolment will not address these inequalities. If the ROI follows the path of the UK, in the manner proposed, it is likely that policy effort towards increasing provision for retirement through Auto-enrolment will ultimately lead to greater income inequality between older women and men, and between those with intermittent or low-paid employment histories and those with an advantaged employment position. While Auto-enrolment can assist in narrowing the participation gap between men and women (Scottish Widows, 2019), the structure of Auto-enrolment contributes to the gender pensions gap since it is based on an individual's pay and men generally earn more than women in the UK and Ireland (Jethwa, 2019;Nolan et al., 2019).
There are lessons to be learned from the UK's experience for the ROI. However, if the UK and ROI want to limit gender inequalities in retirement Auto-enrolment alone is not the answer. It needs to be placed within a context of collective policy solutions to pension challenges, rather than framing pensions in a purely individualised neo-liberal discourse and policy framework which adversely impacts on women's pension prospects. For instance, it would be possible to redevelop pension systems in a manner which de-couples income in retirement from labour market participation (Strauss, 2014) thus avoiding the penalty for caring years incurred in private pensions (Ginn & MacIntyre, 2013). In effect, there is a need to move beyond policies that solely reinforce the liberal celebration of better paid employment and explore ways to reward all forms of work, including unpaid labour (Strauss, 2014). This does not occur in Auto-enrolment in the UK and, unless significant changes take place with the ROI's proposals, the outcomes are likely to be similar. Therefore, it is necessary to undertake a gender impact assessment of Autoenrolment, and pension reforms more generally, to ensure women are not disadvantaged in the implementation of pension policy.