2. Understanding the gender pension gap beyond labour market drivers through a literature review

This chapter reviews the existing literature to identify possible findings beyond those directly linked to labour market differences, to explain the gap in retirement income that men and women receive in retirement.1

Chapter 1 shows that labour market conditions, such as lower employment rates and the gender gap in income, coming from lower wages and career breaks are probably the main driver for the gap in retirement income between men and women. However, there might also be behavioural and cultural elements, which could explain part of the gender pension gap without being directly connected to labour market differences. In defined contribution funded pension schemes, participants often need to make several decisions about their pension savings, contrary to most defined benefit pay-as-you-go pension arrangements where no active decisions are required from individuals. For instance, participants may need to choose their contribution rate, their contribution period, and the investment strategy for their portfolio of assets. Hence, differences between men and women in attitudes and personal preferences may play a role in the gender gap in retirement savings in funded pensions.

Behavioural and cultural elements may influence the actions of men and women on an individual basis. For instance, women can be found to often be more risk averse than men. There is abundant literature on the gender difference in risk aversion, and this element may play a role in the gap in retirement income received by men and women. Part of the gender gap might also be explained by a difference in financial literacy, either actual or perceived. Men and women may furthermore demonstrate a difference in their attitude towards saving, i.e. their time preference, with women more likely than men to spend their income, either for themselves or for others, rather than to save it.

Attitudes and societal features may also play a role in the interactions and relationships between individuals, which might have an impact on the gender gap in retirement income beyond the direct effect cultural differences may have on labour market differences. Marital choices may be a factor in the gender pension gap, either through a possible shift in the decision-making process among couples, or through the different consequences of divorce or widowhood over retirement savings for men and women. Gender stereotyping may be at play, which could encourage women to opt for solutions that can be more conservative than what their actual risk preferences should imply. Communication campaigns may also influence the gender pension gap by omitting to take into account certain needs such as how to compensate for the decrease in salary and contributions during parental leave.

However, the situation is evolving, with women increasingly saving for retirement, and the concern about the gender pension gap getting higher up on the agenda of policy makers and pension arrangements stakeholders.

This chapter explores the potential behavioural and cultural drivers for the gap in retirement income between men and women. The first section discusses gender differences in risk tolerance described in the literature. The second section examines differences in the financial literacy of men and women. The third section looks at the impact of the marital status on funded pensions, while the fourth section explores the potential role of gender stereotyping on the gender pension gap. The fifth section calls attention to the influence of communication and framing on retirement outcomes. The last section concludes by highlighting some of the more recent trends in the gender gap in retirement income.

Labour market differences explain a large part of the gender gap in retirement income, however behavioural and cultural elements might also be at play. This section examines how differences in risk aversion may potentially affect the gender gap in retirement income. It first presents studies demonstrating a lower financial risk tolerance among women than men. It then looks at differences in attitudes towards gambling and competition. Finally, it discusses how factors such as marital status and financial literacy may affect risk aversion.

Women tend to be more risk averse than men in the allocation of their financial assets. Garnick (2016[1]) shows that women in the United States tend to hold more low risk assets and less high risk assets than men, as part of their overall asset allocation. In 2014, women held, as a proportion of their overall assets, more low risk assets such as cash - 20.6% versus 15.6% for men, annuities - 11.4% versus 7.5% for men, and certificates of deposits - 7.1% versus 5.8% for men and had a lower allocation to higher risk assets such as stocks - 16.6% versus 21.1% for men, and mutual funds - 27.5% versus 33% for men.2 The study also claims that default schemes in private pension arrangements, such as qualified default investment alternatives in private plans in the United States, have decreased the investment return differences between men and women.3

The gender difference in financial risk aversion may be linked to factors such as overall wealth and income stability. Fisher and Yao (2017[2]) estimate the level of financial risk tolerance of single men and women in the United States using a decomposition technique.4,5 They find that while women do tend to exhibit lower levels of financial risk tolerance, with 39.6% of women (respectively 56.3% of men) reporting some risk tolerance, 11.4% (respectively 20.3%) high risk tolerance, and 2.6% (respectively 4.2%) substantial risk tolerance, this is determined not solely by their gender, but significantly by variables affecting their risk tolerance, in particular their level of income uncertainty and net worth. Indeed, income uncertainty exhibits a positive effect on financial risk tolerance for men, with men with uncertain incomes 66.4% more likely than men with certain income to declare having some risk tolerance, and 95.6% more likely to declare a high risk tolerance. For women, income uncertainty has a negative effect, decreasing the likelihood of having some risk tolerance by 14.1% compared to women with no income uncertainty, and by 6.8% for the high risk tolerance. For both men and women, an increase in net worth increases the odds of reporting high risk tolerance. Similarly, using a constant absolute risk aversion model to explain households’ investment behaviour, Makarov and Schornick (2010[3]) demonstrate that poorer households are less likely to participate in the stock market, and that among those participating, the portion of wealth invested in risky assets increases with wealth.

The lower risk tolerance of women cannot be explained only by lower financial knowledge, lower confidence, or by lower wealth levels. Baeckstrom, Marsh and Silvester (2019[4]) use data from British individuals holding between GBP 50 000 and GBP 2.5 million to invest, in order to verify whether wealthier investors would exhibit dissimilar attitudes towards investment risk and confidence, compared to average investors.6 Participants were asked to compare their own financial knowledge, risk appetite and confidence in their investment decisions to that of the general population. The study results show that women considered wealthy have similar confidence and financial knowledge levels compared to their male counterparts, and that risk tolerance is negatively correlated with investor age and positively with wealth. Regression analysis shows that investor risk tolerance, measured on a scale of 1 to 5, decreases by 0.2 for each additional 10 years of age, and is higher by 0.5 for investors with more than GBP 500 000 in wealth compared to investors with less than GBP 500 000. However, on average women in the panel still hold five percentage points more cash as a proportion of their investments than men, and consider themselves more risk averse in their investment decisions relative to men.

When provided with different investment strategy choices for their retirement savings, women are less likely to choose the high risk and high return option. Watson and McNaughton (2007[5]) examine the impact of gender on the superannuation fund risk preferences, and ultimately on the retirement benefits of staff in the Australian university sector, using data from members of the UniSuper defined contribution superannuation fund from 1997 to 2003.7 Plan members were offered seven different investment strategies to choose from, corresponding to six risk categories – two of the strategies offering 100% allocation to shares differ in their investment focus rather than risk profile. For members unwilling or unable to make an investment choice, a default option in the balanced plan had been set up. In order to obtain meaningful data on individuals’ risk preferences, the research eliminated those members who were allocated to the default plan, as the records did not allow differentiating those members who would have actively chosen the balanced strategy from those who would have stayed with the default plan in any case. The analysis shows that women in the panel are more risk averse than men, with 25% of men choosing the riskiest investment strategy versus 18% of women. Control variables were included in the regression in order to eliminate the potential interference of income and age profiles in the gender statistics, and women still generally chose less risky investment strategies than men for a similar age and level of income.

However, other studies argue that gender is only a minimal factor when studying retirement savings, and that structural labour market conditions only, rather than behavioural elements, drive the gender gap in retirement savings. Allport et al. (2019[6]) look at the impact of gender on retirement savings behaviour, in order to untangle the behavioural element from any structural factor, based on data from the United Kingdom’s National Employment Savings Trust (NEST) covering over seven million occupational defined contribution plan members at the end of 2018, and broken down in several earning bands. The data show that women appear to have higher median contributions and pension account balances across all considered earning bands, except for the highest income group. However, the data allow distinguishing regular contributors to the pension plan from those only making occasional contributions, in order to control for any difference attributable to occasional contributors, which might distort the overall picture.8 When focusing only on regular contributors, who made 12 contributions during the year 2018, the gender difference in median contributions and account balances disappears, except for individuals in the highest earnings band. The study also looks at attitudes towards saving, and shows that when controlling for earnings, there is no gender difference in voluntary contributions and active enrolment to the scheme. While more women actively opt in for pension contributions, this is attributable to the fact that three times more women in the sample earn less than the threshold for automatic enrolment. Additional contributions come equally from men and women - 50% each - with contribution levels higher for men, due to higher earnings. Hence, the authors conclude that gender is a minimal factor when studying retirement savings and that labour market conditions such as those linked to pay, positions, employment sectors, and career breaks drive the gender pension gap, rather than behavioural elements.9

Additionally, the increasing use of default parameters such as default investment strategies may reduce the impact of personal risk preferences on retirement outcomes between men and women. Default investment strategies are generally similar for all members or based on objective and gender-neutral parameters such as member age or time to retirement. Behavioural biases such as procrastination and inertia significantly limit the tendency of members to change their investment strategy once they are assigned one by default (OECD, 2018[7]). For example, Cronqvist et al. (2018[8]) show that in the absence of communication campaigns encouraging individuals to actively choose their own investment portfolio, only about 1% of those joining the Swedish Premium Pension system in 2016 declined the default fund and made an active choice. Hence, the implementation of defaults may alleviate the natural tendency of women to choose more conservative investment strategies.

Finally, differences in risk aversion levels between men and women may have different impacts on retirement outcomes at different times or for different cohorts depending on the performance of investment markets. On average over the longer-term, saving in a conservative investment strategy is expected to yield lower investment returns for retirement savings than a balanced or dynamic strategy. At times when financial markets fall however, risk averse savers with a defensive or conservative investment portfolio may end up with larger pension pots than savers with a dynamic or higher risk portfolio.

Women tend to perceive risky situations more as a threat than a challenge. Arch (1993[9]) looks at explanations for the fact that while there are attempts to equalize opportunities for women to attain positions of power and status in society, change is not happening as rapidly as expected. The author uses experiments from different studies to demonstrate that women tend to see situations that present either a physical or a social risk, more like a threat and to choose avoiding options whereas men perceive them more as a challenge, which may explain a potential discrepancy in participation in opportunities. Larkin and Pines (2003[10]) also show that the perception of potential public humiliation rather than the likelihood of failure drives women more than men when considering participating in a public performance. In their experiment, they find that women are more concerned about doing poorly in public, and are less likely to risk entering a public competition by fear of disappointment.

Similarly, the attitude of men and women towards competition may be driven by societal factors. In their literature and research review of gender differences in preferences, Croson and Gneezy (2009[11]) look at how women and men react to competition in different studies. Overall, women tend to choose not to compete more often than men do. Yet research also demonstrates that women placed in a competitive environment perform just as well as men. Looking at experiments with children of different ages demonstrates that the gender difference seems to have appeared from a certain age rather than for all age groups, suggesting an environmental cause, which they characterise as backlash. Comparing a patriarchal society (the Maasai in Tanzania) and a matrilineal society (the Khasi in India), Gneezy, Leonard and List (2009[12]) also show that men choose to compete significantly more than women in the patriarchal societies, and that women choose to compete significantly more in matrilineal societies, reinforcing the assumption that the societal structure influences the willingness to compete.

Men may be driven more by speculation than women in their investment decision-process and portfolio choices. Based on a questionnaire addressed to clients of an Italian bank in 2013, and after controlling for sociological, demographic and economic variables, Marinelli, Mazzoli and Palmucci (2017[13]) find that gender still explains many differences in investment behaviours, namely in the investment decision process, in risk preferences and in portfolio characteristics.10 The authors find no evidence of a difference in the quality of portfolios as illustrated by an assessment of liquidity and diversification. The data also show that women are more likely to declare relying on professional advice - 53.6% versus 44.3% of men, whereas men count more on autonomous decisions - 33.9% versus 26.3% of women. Men declare a greater risk tolerance, and to be driven more by speculation - 5% versus 2% of women, whereas women invest more with the aim of increasing their income. Consistent with this analysis, the survey by Ho (2018[14]) over Taiwanese finance students taking part in a virtual trading competition over 6 months shows that men are more likely to trade derivatives for speculation purposes than women in the sample, with a 20-percentage point difference in allocation between men and women.11

Women are less likely to gamble on events with objective probabilities, however, there is no effect of gender for subjective probabilities. Since most experiments demonstrating women’s lower risk tolerance are based on gambling situations with an objective probability of winning and losing, Sarin and Wieland (2016[15]) look at risk aversion levels for situations such as cultural or sports events in the United States, where no objective probability of success can be derived. Based on several experiments using different types of elicitation methods, the authors find that there is no gender difference in risk aversion for events with a subjective probability of success.12 The study compares these results to those obtained for gambles with objective probabilities, and finds that for gambles with objective probabilities, male respondents are twice as willing as women to pay to play the gambling game.13 For subjective probability events however, gender does not seem to influence the bet valuation, which is only a function of the subjective probability of wining perceived by the participant.

The distribution of risk tolerance may be much more dispersed for men than for women. Based on individual portfolio choices in the mandatory Swedish Premium Pension plan (PPM), Säve-Söderbergh (2012[16]) finds that there is little gender difference in portfolio risk for risk averse individuals, however men who choose risky portfolios take on significantly more risk than women who choose risky portfolios. The author examines the individual investment choices of participants by computing a measure of risk tolerance based on the standard deviation of the rate of return of funds in the pension portfolio for the past 36 months.14 In this study, a similar proportion – around 62% - of both men and women choose to select funds to invest their contributions, as opposed to staying with the default investment strategy of the PPM, and women are more risk averse on average than men, with risk tolerance measures of 19.8 and 20.8 respectively.15 However, when looking at the distribution of risk tolerance rather than at the mean, results indicate that the gender difference is explained in large part by risk tolerant men, i.e. men with a risk tolerance above the median male risk tolerance, who choose much riskier portfolios than risk tolerant women.

The gender difference in risk tolerance may be driven mostly by optimistic men who make significantly riskier choices than average men and women. Felton, Gibson and Sanbonmatsu (2003[17]) study the role of gender and optimism on the riskiness of investment choices of undergraduate business students from the Central Michigan University who participated in a portfolio simulation game on an individual basis.16 The results of the game were both monetary, as participants could win up to USD 500, and academic since their results formed part of their grade. Optimism was evaluated through the Revised Life Orientation Test, and both men and women reported similar levels of optimism.17 Risk indicators were built based on several parameters including the number of futures and options traded, the overall number of transactions, the number of companies invested in that traded on the Nasdaq stock market, on the New York Stock Exchange (NYSE) or on the American Stock Exchange (ASE).18 The data show that male students are more likely than female students to invest in futures and options on average and that optimism is a significant predictor of investment in futures and options for men. The authors therefore argue that the difference between men and women’s approach to futures and options is mostly driven by optimistic men, while optimism does not seem to affect the portfolio choices made by women. Looking at other risk indicators, men make significantly more transactions, and trade more on the Nasdaq market than women, and men have a standard deviation of their portfolio value approximately twice as large as that of women. The authors conclude that while male students take on significantly more risk than their female counterparts on average, the gender difference may be driven by optimistic men.

Over-confidence in their own capabilities and knowledge may hurt men’s investment returns. Research on households from the United States finds that men are more likely to be over-confident in their investment decisions and experience than women, and that this leads them to negatively affect their investment returns by trading excessively (Barber and Odean, 2001[18]).19 More (62.5%) men than women (47.8%) believe they have good or extensive experience, and more (2.8%) men than women (2.1%) believe they can outperform the market.20 In addition, while both men and women tend to make poor portfolio decisions and to sell those stocks that have the best performances, women modify their portfolio less than men - 53% and 77% respectively, on an annual basis. Analysing the impact of over-confidence and excessive trading on net portfolio returns shows that men’s returns are lowered by 2.65 percentage points a year, as opposed to 1.72 percentage points for women.

Married individuals may be more risk averse, which can have long-lasting consequences on the gender pension gap (Hinz, McCarthy and Turner, 1997[19]). When controlling for income levels, using the share of stocks in pension investment portfolios as a risk tolerance measure shows that marriage reduces the risk tolerance of individuals in the United States.21 Men invest more in stocks than women on average, 45% and 28% of their investment portfolio respectively. However, when breaking down risk tolerance by marital status, unmarried men demonstrate the riskiest investment profile, while married men and unmarried women show similar risk tolerance levels, and married women appear to be the most conservative in their investment risk profile.

However, the impact of marital status on financial risk tolerance may be more complex and vary by gender. Looking at participants in the Swedish PPM, Palme, Sundén and Söderlind (2004[20]) find that married men are less likely to accept financial risk than single men, but that married women are less risk averse than single women.22 This analysis controls for the level of household income and expected pension benefits, and uses the standard deviation of the rate of return of the pension portfolio for the past 36 months as a measure of risk tolerance. The effect of the marital status on risk aversion for men and women can therefore not be considered as straightforward and might depend on additional parameters linked for instance to country specificities such as the overall pension system functioning.

Financial education may mitigate the gender gap in risk aversion. Comparing the general population in the United States to a sample of highly educated individuals, and to finance professors, Hibbert, Lawrence and Prakash (2013[21]) show that financial education lessens the gender difference in risk tolerance.23 Comparing the general population with the highly educated shows that income and education are the most important variables in explaining risk aversion, however highly educated women are still significantly more risk averse than their male counterparts. Nevertheless the survey of finance professors shows no gender difference in risk aversion, with both male and female finance professors significantly less risk averse than the panel of highly educated participants.

The impact of financial education on risk aversion is also confirmed by comparing a sample of the general population’s holding of investment in stocks to that of highly educated individuals and to economists in Denmark (Christiansen, Joensen and Rangvid, 2008[22]).24 Economists in the panel are more likely to participate in the stock market – between 37% and 47% depending on years – than the overall panel, which has a participation rate of 23%, and than highly educated investors. The stock market participation rate of investors who are economists also increases significantly (by 6 percentage points) at the time when they complete their education, as does that of individuals moving in with an economist (by 5 percentage points). The duration of economics education also influences the stock market participation, as medium or long durations both significantly increase the stock market participation probability compared to a short economics education. Compared to several other education fields, having an economics education has by far the largest marginal effect on individuals’ holding of stocks.

Perceived, rather than actual financial education, might also play a role in explaining risk aversion differences between genders.25 The gender gap in stock market participation is usually explained by women’s lower financial knowledge, their lower numeracy, their lack of familiarity with financial products or their lower risk tolerance. The study of German households by Bannier and Neubert (2016[23]) extends the analysis by looking at a larger universe of investment products and comparing standard risky products such as stocks and mutual funds to more sophisticated products such as hedge funds.26 It also looks at different dimensions of financial knowledge, i.e. the actual versus the perceived financial literacy in order to examine the combined role of financial literacy and risk tolerance for investment decisions. While for men, a decrease in either the actual or the perceived financial literacy decreases the investment in standard risky products, for women only actual financial literacy appears meaningful. For both men (-11%) and women (-8%), standard risky investments decrease with risk tolerance. When controlling for actual financial literacy and risk tolerance together, gender does not play a significant role in investing in either standard or sophisticated assets.

Financial education may help increase the risk-return combination of investments, for a given level of risk tolerance. Comparing the levels of risk aversion of Chinese and American students, Pyles et al. (2016[24]) find that Chinese individuals tend to perceive themselves as more risk tolerant, although this does not necessarily translate into significant differences in portfolio composition. The authors also find that both American and Chinese female students exhibit lower risk tolerance levels than their male counterparts, and that financial education – assessed by having taking a finance class - increases the Sharpe ratio, which is a measure of the return of an investment compared to its risk. The results also show that the Sharpe ratio is unaffected by gender, meaning that women choose portfolios with a lower risk and return combination, which are as efficient as men’s per value of risk.

Financial literacy is an important factor to ensure individuals have the knowledge and understanding to make choices that will affect when and in which material conditions they may retire. Men and women may exhibit differences in their understanding of financial concepts and in their attitudes towards saving (Atkinson and Messy, 2012[25]). This section first looks at the influence of gender on financial literacy and retirement readiness, and then examines the potential gender differences in priorities between spending and saving, before highlighting examples of retirement-related financial education campaigns targeted at women.

Women have lower levels of financial literacy overall. The OECD - International Network for Financial Education (INFE) pilot study on financial education examines the impact of several socio-economic factors such as age, gender and income on the level of financial knowledge in 14 countries (Atkinson and Messy, 2012[25]). Financial knowledge is evaluated by 8 questions covering key financial concepts, financial behaviours such as thinking before making a purchase, paying bills on time, budgeting, saving and borrowing, and attitudes towards long-term financial plans, as opposed to short-term. Results from the study show that women have much lower levels of financial knowledge than men in all of the countries studied except Hungary where results are similar for men and women. In several countries such as Norway, Poland and the United Kingdom for example, there is a 20-percentage point difference between men and women attaining a high knowledge score, in favour of men. Women are also less likely to gain high scores for financial behaviour, although this is not true for the Czech Republic, Estonia, Ireland and Norway where the opposite is found. In most of the countries surveyed, women are more likely to have a positive attitude towards long-term saving than men. Overall, none of the 14 countries surveyed has women score higher on the combined measure of financial literacy. The 2015 OECD/INFE survey of financial literacy, which expands the initial pilot study to include 30 countries, shows that the trend is still valid as women have less financial knowledge than men in 19 of the participating jurisdictions, with no significant gender differences in the remaining countries (OECD, 2017[26]).

Women’s lower financial literacy may deter them from planning for retirement and may be linked to lower coverage by private pensions in several countries.

Men are more financially resilient than women in many countries. Taking into account financial knowledge reduces this gender difference (OECD, 2017[26]). Financial resilience denotes the ability of individuals to cope with a shock such as having to face major unexpected expenses, covering living expenses in case of an income loss, and supporting themselves in retirement independently of their spouse or family. In many countries participating in the 2015 OECD/INFE survey of financial literacy, men are more likely than women to report counting on their private pensions, accumulated financial and other assets to finance their retirement. Women are less likely than men to support themselves in retirement independently of their spouse in all 30 countries of the 2015 OECD/INFE survey of financial literacy, except Brazil, Korea and Lithuania. In several countries such as Belgium, Canada, New Zealand and Portugal, the financial resilience of women becomes closer to that of men when controlling for the level of financial knowledge.

Surveying women aged over 50 in the United States shows a strong positive correlation between financial literacy and planning (Lusardi and Mitchell, 2008[27]).27 Overall, financial literacy results for the considered population are low, with 61.9% of respondents correctly answering a basic question on interest rates, as are the number of respondents (30.9%) who have ever attempted to plan for retirement. However, those respondents who can answer the financial literacy questions correctly, especially the question on risk diversification, plan significantly more than the overall sample.

Prior to the introduction of automatic enrolment to occupational pension schemes in 2012, many more women than men had a limited understanding of pensions and many more women than men also did not have appropriate private pension coverage in Great Britain (MacLeod et al., 2012[28]).28 Overall 46% of women had never had a private pension at all, compared to 35% of men, and 23% of women actually had no resources at all for later life compared to 15% of men. This number increased to 42% for women with no educational qualifications, compared to 19% of all respondents. The study also showed that 71% of women, compared to 56% of men, felt that pensions seemed so complicated that they could not understand what was best to do. The study additionally found that 28% of women, compared to 13% of men, were “scared” by having to deal with their pension; and that 36% of women, compared to 25% of men, altogether avoided thinking of retirement.

In Ireland, crossing data from The Irish Longitudinal Study on Ageing (TILDA), and Growing Up in Ireland (GUI) surveys shows that women have lower levels of financial literacy than men, and that higher levels of female educational attainment reduce the gender gap for private and occupational pensions, throughout the income distribution (Nolan et al., 2019[29]). Almost 30% of boys aged 17 could correctly answer the three financial literacy questions of the GUI survey in 2014, compared to only 14% of girls of the same age. Similarly, for individuals aged over 54, the TILDA 2014 study shows that 17.2% of men could answer all financial literacy questions correctly, compared to 7.5% of women, and that women are more likely to answer ‘don’t know’ to questions related to financial literacy than their male counterparts, by up to 9.9 percentage points. When decomposing the gender gap in occupational and private pensions according to educational levels attained, the authors find that higher education levels reduce the gender pension gap for all income groups, with higher effects for individuals in the lower deciles of the pension income distribution.

The gender gap in financial literacy is lower when women are involved in financial decision-making (Nolan et al., 2019[29]). While 54.5% of households in the Irish TILDA survey nominate the male partner as the key decision-maker for financial decisions, 74.4% nominate the female partner for family-related decisions. Financial literacy is higher for men on average, whatever the household decision-making combination, with a 0.54 point gap in favour of men in families where the man is in charge of financial decisions and the women in charge of family decisions, compared to 0.08 point when the woman is in charge of financial and the man in charge of family decisions, out of a maximum total score of 4.

Women may prioritise more than men current spending for themselves or others over saving for retirement. Long-term planning can be linked to the level of financial knowledge, nevertheless it may also be influenced by a difference in priorities. Saving for retirement requires prioritising future over short-term well-being. It also implies saving for one-self rather than spending, potentially for others. Several pieces of research have shown a difference in preferences between men and women over these choices.

Women may be or feel more vulnerable to short-term financial hardship and hence avoid or delay saving for retirement. Just under half of men and women in the United Kingdom may not save adequately for retirement, and around 17% of both men and women do not save at all (Scottish Widows, 2018[30]).29 Nevertheless a gender disparity exists among the younger cohorts, aged 22 to 29, in which 46% of men and 33% of women save adequately for retirement, while 17% of men and 25% of women do not save at all. The study argues that barriers to saving or to saving more for retirement are mainly linked to the absence of an access to emergency savings in case of financial hardship, with 42% of women aged 22 to 29 currently not saving for retirement confirming they would be likely to start, and an additional 19% currently contributing to a private pension likely to increase their savings, if they could have access to emergency funds in case of financial hardship. This is consistent with findings from Henry (2014[31]) in the United States, showing that the average share of consumption of American households devoted to personal insurance and pensions is highly dependent on their income level. Households in the lowest income quintile allocated 2.32% of their total consumption to pension and insurance on average over the period 1984-2012, compared to 13.92% for those in the highest income quintile.

Women may be more likely to spend their income on family members and gifts than men, thereby saving less, including for retirement. A study of couple households aged 55 and above in the United Kingdom shows that women are responsible for respectively 76% and 66% of household expenditures linked to the day-to-day needs of children under the age of 15 (including grand-children) and gifts (Age UK, 2018[32]). Conversely, men in the survey made 55% of expenditures related to savings.

Women take time off to care for a relative more often than men do, which is another way of prioritising current spending over future well-being.30

The gender employment gap has halved on average between the 1940s- and 1970s-born generations in OECD countries, nonetheless women are still more likely than men to work part-time and to take career breaks, which can create obstacles to adequate saving for retirement (OECD, 2017[33]). Women also bear most of the burden of unpaid household chores and are more likely to care for both children and older relatives.

Caregiving responsibilities can adversely affect the retirement savings and health of the individual providing this unpaid care, who is more often a woman than a man. American women are more likely to be caregivers for their family, with 56% of pre-retired women and 35% of pre-retired men, and 63% of retired women compared to 43% of retired men reporting having provided caregiving services (Rappaport et al., 2017[34]). Caregiving has emotional, physical and financial consequences for the caregiver. Among pre-retired respondents, 19% declare that caregiving has a major or catastrophic burden physically, 11% financially and 35% emotionally.31 Women in the United States have a higher focus on others and are more likely to see their personal needs as secondary according to several surveys, focus groups and interviews (Rappaport, 2018[35]).32

Women, including the highly educated, often take career breaks to raise their children. In the United States, women work fewer years on average than men – 29 versus 38 respectively – mainly because of responsibilities linked to child-bearing and caring for the elderly (Garnick, 2016[1]).33 Women increasingly take time off to raise their children, with almost 30% of stay-at-home mothers recorded in 2016, compared to 23% in 2000. Among the highly educated, this trend also holds with 25% of stay-at-home mothers having a university degree in 2016, compared to 7% in 1970.

Women taking career breaks to look after family may be the biggest factor in the gender pension gap. Looking at both public and private pension sources in the United Kingdom shows that the different working patterns between men and women are the largest factor explaining the difference in pension savings between men and women aged over 50, representing 47 percentage points of the 49% difference in pension wealth between men and women (Pensions Policy Institute, 2019[36]). This analysis interestingly does not note any behavioural or cultural effect in the gender pension gap linked to investment choices or risk aversion for instance. Policy options such as making sure employer contributions are based on the pre-maternity leave salary or are increased when taking time off or working part-time to take on caring responsibilities are proposed and quantitatively analysed to reduce the impact of caring duties on retirement savings.

Women also have a higher longevity, which implies a higher life expectancy in both health and disability, and a higher probability of living alone in retirement.34 Women face healthcare costs 11% higher than men, because of their longer life expectancy, and these represent a growing portion of their retirement budget as they age (Garnick, 2016[1]). For instance at age 65, American women have a life expectancy of 19.5 years, made of 13.7 years in good health, and 5.8 years in mild or severe disability, compared to a total life expectancy of 15.3 years for men aged 65, of which 12.3 years in health and 3 years in disability. The associated long-term care costs incurred by women are therefore higher than for men, who are also more likely to have a family caregiver. The unpaid caregiving responsibilities of women can therefore imply less savings during working years, and also more spending in old age.

However, gender differences in attitudes towards long-term saving may not be universal (OECD, 2017[37]). In Norway for instance, over 80% of women, compared with 65% of men, achieve the minimum target score of three out of five for financial attitudes in the OECD/INFE survey, which assesses the preference for the long-term versus the tendency to live for today and spend rather than save.35

Several jurisdictions have put in place policy initiatives to address the gender gap in financial education (OECD, 2013[38]).

Financial education programmes may aim at informing women about the challenges and needs linked to retirement, with a focus on gender-related specificities and circumstances. For example, New Zealand’s “Women in Super” network organises meetings and events since 2001, for women to learn more about the superannuation system and their specific needs linked to retirement. It was inspired by a similar initiative launched in Australia in 1994.

Other programmes are designed for particular groups of women that are considered as most vulnerable, such as young women, elderly women, low-income and marginalised women. Turkey’s capital market board for instance launched a financial literacy programme in 2010 targeted at unemployed, unbanked and low-income women in order to increase their understanding of saving, debt and investment. In Singapore, the Tsao Foundation “Financial Education Programme for Mature Women” is designed for women between the ages of 40 and 60, and aims at helping them be financially independent in their older years.

Societal perceptions, attitudes and interactions may play a role in driving the gender gap in retirement income, in addition to individual behavioural biases that may affect women and have implications for their pension savings. This section examines dynamics of the pension gap for couples versus single individuals. It starts by focusing on financial decision-making at the household level, and then examines the impact of the marital status on the retirement outcomes of men and women.

Financial decisions taken at the household level may not allow understanding the investment decision dynamics within couples. Bajtelsmit and Bernasek (1996[39]) propose several possible causes such as discrimination, choices, family responsibilities, as well as potential biological and social determinism for the wealth, income and employment differences explaining that American women on average have lower wealth levels than men, which is expected to increase their absolute risk aversion. However, the authors note that understanding the investment decision-making process within households is important to draw conclusions, especially related to gender differences.

There can be inconsistencies in the individual responses of household members related to financial decision-making, which can have implications for studying the gender pension gap. Members of Australian couples aged 21 to 80 years old asked to name the main household decision-maker for choices related to savings, investments and borrowings, report different answers (Johnston, Kassenboehmer and Shields, 2016[40]). Both men and women report similar frequencies of shared decision-making, around 70%. However, 22.6% of men report being the main decision-maker, which is confirmed by 15.7% of women, and 12.6% of women report being the main decision-maker, which is confirmed by 7.1% of men. These inconsistencies may have implications when analysing differences in financial education, risk aversion and ultimately the gender pension gap. Married couples are more likely to have a male decision-maker than cohabiting couples. Women with more education are more likely to be decision-makers. Couples with working females are more likely to have female decision-makers. The gender wage gap is significantly smaller when females are the decision-makers, from AUD 40 000 to AUD 8 000 in the studied sample. Finally, using detailed information on household wealth, the study finds that households in which the male is the main financial decision-maker are significantly more likely to hold financial assets and less likely to hold their wealth in real estate than households with either shared or female decision-making.

Single women may experience an increased pension gap compared to married women. Couples may be able to mutualise their pension savings in order to manage their consumption smoothing, effectively reducing the gender pension gap. However, single retirees cannot and may suffer a higher gender pension gap.

The effect of having retirement savings on the income of single individuals aged over 65 in Canada is higher for women than for men (Richard Shillington et al., 2016[41]).36 Although poverty in Canada has declined sharply between 1976 and 1995, it has been slowly increasing since 1996, especially for single individuals. Poverty rates for single individuals aged over 65 in Canada are higher for women than for men, at 28% and 24% respectively, compared with 11.1% for the overall population – including couples - of this age category in 2013. The authors focus on the difference of income for seniors depending on whether or not they have individual retirement savings, either occupational or personal, and excluding any payments from the public Canada or Quebec Pension Plan and show that the effect is greater for single individuals than for couples, and for women than for men. While the average family aged over 65 without a pension has an income of CAD 52 000, it reaches CAD 68 000 if there is a private pension plan. For single individuals, the difference is even greater since having a private pension increases income for single men from CAD 26 000 to CAD 46 000, as opposed to CAD 19 800 and CAD 39 000 for single women without and with a private pension respectively.

The proportion of women living in poverty increases with age, as more women are alone in old age. In the United States, women represent 57% of the 65 to 84 age group, and 71% of the 85 and above age group, and the proportion of women who are widowed increases with age, as does the proportion of women living in poverty (Bajtelsmit et al., 2005[42]).37 The combination of women’s higher life expectancy and of observed age differences within couples implies that women live increasingly longer alone. Women who outlive their male partner live on average 11.5 years alone, whereas only one third of males outlive their female partners, by an average of 8.8 years (Garnick, 2016[1]). Similarly, in the United Kingdom, 53% of men over the age of 80 live in a couple, compared to only 14% of women of the same age group (Portas, 2019[43]). And 75% of men with a defined contribution pension scheme believe it includes a financial benefit for their partner or dependents in event of their death whereas this is not an automatic feature of defined contribution schemes, implying that bereaved women may live on less income than they previously expected (Age UK, 2018[32]).38

Even within couples, a gender gap in retirement income may exist and increase with age and pension wealth. Focusing on individual-level occupational pension wealth for six European countries, Schneebaum et al. (2018[44]) demonstrate that there is an increase in the gender pension wealth gap as the absolute level of pension wealth and age - both strongly correlated - increase, at the disadvantage of women.39 While most other parameters to quantify the wealth gap are observed for single households only because of the difficulty of correctly allocating assets within households, this result based on individual-level assets seems to show that the gap exists for the entire population, i.e. also for married couples or partners belonging to the same household.

The impact of a divorce on pension wealth can be significant, and increase the gender pension gap as pension savings may often be overlooked in divorce orders. In the United States, the poverty rate of divorced women is significantly higher than that of married women, at 20.4% and 4.3% respectively (Bajtelsmit et al., 2005[42]). The authors suggest this could be linked to a lack of awareness from the public that pension assets can be divided in a divorce, as other financial assets. In the United Kingdom, the median pension wealth of divorced women is reduced by 50% to GBP 26 100 compared to the overall female population, as opposed to approximately one third for men to GBP 103 500 (Now Pensions, 2019[45]). Out of the 118 142 divorce orders pronounced by British family courts in 2018, only 4 632 (i.e. less than 4%) included a pension attachment order, which divides pension entitlements accumulated during marriage between former spouses once they are in payment. However, other types of pension split orders also exist, such as pension offsetting and pension sharing orders (see Chapter 4 for further detail). Studying randomly selected divorce files in England and Wales courts between 2011 and 2012, Woodward and Sefton (2014[46]) found that 66% of files disclosed one or more relevant pension accounts but no pension split orders were pronounced, compared to 14% of cases which included an order to split pension assets between former spouses.40

The negative effect of divorce on women’s retirement outcomes may be augmented as couples increasingly split at ages closer to retirement. New Zealand data from 1997 and 2019 shows that the median age for divorce has increased from 40.6 to 47 for men, and from 37.9 to 44.4 for women over the period. Dale and St John (2020[47]) argue that divorcing at later ages implies that there is less time to rebuild pension entitlements if there is an imbalance between the pension assets of former partners.

Gender stereotyping refers to a generalised and preconceived view about some of the roles, capabilities and attributes performed or possessed by men and women. This section focuses on how gender stereotyping may play a role in several decisions and choices having an impact on the gender pension gap. It looks at the role gender may play on educational and career choices, before focusing on the gender imbalance in caregiving responsibilities and the impact of childcare costs on women’s retirement savings. It then examines the influence of gender on the opening of a pension account and gifting by grandparents, and on recommendations provided by financial advisors.

Gender stereotyping and gendered expectations still drive educational and career choices (OECD, 2017[26]).

Gender stereotyping can have long-lasting effects for women’s work and income prospects. Gender gaps persist in mathematics and numeracy across the OECD, and girls are more likely than boys to hold negative perceptions of their abilities in mathematics. This can deter women’s work prospects and their ability to make important financial decisions. Several government-led initiatives targeted at women have recently been set up, such as in Poland to encourage women to participate in the stock market, or in Australia to prepare for important life events such as purchasing a home or having children. In most OECD countries, women still spend more time than men carrying out unpaid family work and there is a close link between the gender gap in paid and unpaid work hours, i.e. in countries where men and women share unpaid work hours more equally, they also spend a closer number of hours working in the labour market (OECD, 2017[26]).

The economic participation and progression of women may be affected by cultural and behavioural biases in the United Kingdom (Government Equalities Office, 2019[48]). Female students are under-represented in science subjects both at General Certificate of Secondary Education (GCSE) and A levels, and only 17% of small and medium-size enterprises (SMEs) are majority led by women, compared to 43% that are entirely led by men.41 However, the over-representation of women in the public sector also implies that women with a pension are more likely to have a defined benefit plan than men, which may assist in reducing the gender pension gap (Now Pensions, 2019[45]).

In several countries, women still often are the primary family caregiver, and the moment in which the gender pension gap starts increasing coincides with the arrival of a child. In the United Kingdom, the average hourly gender pay gap between mothers and fathers, which is one of the main drivers of the gender pension gap, increases from around 10% to around 30% in favour of men from before the birth of a first child to the time the child is aged 13 (Mcguinness and Pyper, 2018[49]). The impact on hourly salary is compounded by the fact that when the first child reaches age 20, mothers have on average been working full-time for 10 years less than fathers. By that time, mothers have indeed on average worked part-time for seven years more than fathers, and have been out of paid work for three years more. Another study by Which? (2019[50]) notes that women working part time for ten years to take on caregiving responsibilities for children have pension pots between 15% and 20% lower than those working full time, and a likelihood 15% lower of receiving an adequate income in retirement.

Childcare costs may have a double effect on women’s saving for retirement, both in absolute and relative terms. High childcare costs might imply leaving a full-time paid position, altogether or for a part-time one, hence decreasing pensionable salary and linked employer and employee contributions in absolute terms. They can also lead women to opt-out of non-mandatory schemes, or else reduce their voluntary contribution levels.

While automatic enrolment in the United Kingdom has enabled to increase the participation rate in the private occupational pension sector, childcare costs may play a role in women actively choosing to opt out (Prabhakar, 2017[51]). A focus group interview shows that the lack of affordability is the main reason for individuals to opt out after having been automatically enrolled by their employer, and that opt-out rates vary across different age groups.42 Women in all age groups are more likely to opt out because of a lack of affordability. Women and not men, explicitly mention childcare costs as a reason for opting out.43

Childcare costs can make it financially more efficient in the short-term for women to reduce their working hours or leave employment (The People’s Pension, 2019[52]; Imeson, 2019[53]). The labour market impact of childcare costs may be significant, with studies showing that up to half of the women in the United Kingdom have reduced their working hours after having children, and over a third have left employment after the birth of a child, primarily because it did not make financial sense, over the short term, to work compared to childcare costs. Long-term pension implications do not seem to form part of the parameters taken into account when deciding to reduce paid work because of childcare costs.

Women working part time following the birth of a child may have reduced opportunities and work prospects, in particular if childcare costs require them to reduce travelling to work time (Mcguinness and Pyper, 2018[49]). In the United Kingdom, gender differences in part-time working seem to explain more of the widening in the overall gender pay gap than the time women spend out of employment following the birth of a child. An increasing “gender commuting gap”, i.e. the growing difference between the time men and women spend to travel to work in the 10 years following the birth a child, may imply that women have lower prospects of employment due to a constraint on their available time to commute to work, as travelling times for women decrease sharply from the birth of a child while they remain stable for men.

Different gender perceptions and expectations from family members may influence the propensity of parents and grandparents to open a pension account and make contributions for their children and grand-children. In countries where pension accounts may be opened for individuals before they start working, a gender difference in gifting patterns may have long lasting effects on the gap in pension assets between men and women. Tax data from the United Kingdom has shown that about 20 000 boys under the age of 16 had received money in a pension account for the fiscal year 2016-2017, compared to only about 13 000 girls of the same age. In addition, the tax relief of 20% attracted by contributions made on behalf of British children, which is added to the pension pot, and the compounding effect, may increase the impact of the difference in gifting patterns by family members towards young boys and girls.

Women may be more risk averse than men on average, nevertheless they are in addition, also often subject to gender stereotyping from financial advisors who assume an even greater risk aversion. Financial advisors may advise them on overly conservative investment options because of an unconscious bias.

Financial advisors may overestimate men’s risk tolerance and underestimate women’s because of gender stereotyping. Roszkowski and Grable (2005[54]) use four methods of regression and analysis to demonstrate that gender is a significant predictor of advisors’ rating of risk tolerance, and that there is a tendency to overestimate men’s risk tolerance and to underestimate women’s.44 Controlling for additional economical and educational factors that could explain the apparent gender stereotyping, the authors find that while there is a tendency to overestimate the risk tolerance of high income clients and to underestimate that of low income clients, gender remains the main predicting variable. Overall, they find a 0.41 correlation between advisors’ estimates and actual risk tolerance indicators, and could not verify whether there is a difference in stereotyping between male and female advisors, due to the small sample of female advisors. This study shows that while women tend to exhibit lower risk tolerance levels, their investment choices may be further influenced or even limited by gender stereotypes demonstrated by investment advisors. Financial advisors should therefore rely on objective measures of risk tolerance in order to provide suitable advice to their clients and avoid being biased by stereotyping.

The gender difference in risk tolerance may be situational and amplified by the gender of financial advisors as wealthy women advised by male advisors report lower levels of risk tolerance, confidence and perceived financial literacy than wealthy women advised by female advisors (Baeckstrom, Marsh and Silvester, 2019[4]). The authors compare risk tolerance perception and the cash holdings of wealthy investors with and without an investment advisor, and find that having a financial advisor has an effect on risk tolerance and on cash holdings. Advised investors indeed report a higher risk tolerance perception and cash holdings 15% lower than self-advised investors. They also find that the gender of the financial advisor has a significant impact on the risk tolerance of advised women, as wealthy women advised by men feel less confident, less knowledgeable, more risk averse and hold 11% more cash than wealthy women advised by women. Wealthy women advised by women report the highest risk tolerance and make the lowest portfolio allocation to risk-free assets of the sample, including wealthy men. The authors therefore conclude that the gender difference in risk tolerance is situational rather than universally true for all women.

Financial advisors and fiduciaries should understand the specific financial needs of women and not assume women universally have a lower risk tolerance. In order to provide advice that is in their clients’ best interest, it is important that financial advisors understand the ways in which income uncertainty and net worth may influence the risk tolerance levels of women, as the effects may be different for women and for men (Fisher and Yao, 2017[2]).

This section looks at how communication and marketing may also be factors affecting the gap in retirement income between men and women. It starts by looking at how men and women may perceive information about retirement savings differently, then focuses on the need to target any communication campaign to its audience, including according to gender, before describing concrete examples of successful communication campaigns targeted at women.

The way information is presented may play a different role in financial decision-making for men and women, alongside risk tolerance and financial literacy. For issues related to investment and pensions, women may be influenced more by information going against their preconceived notions. By using negative framing, i.e. by highlighting the potential financial losses associated with one or the other choice in an experiment, Agnew et al. (2008[55]) show that women are significantly more influenced by an information bias going against their preconceived notions.45 When asked to choose between purchasing an annuity and investing their old age savings, women on average choose to annuitise more often than men, for a given level of risk aversion and financial literacy. Additionally, more risk averse individuals are more likely to choose an annuity rather than an investment option, regardless of gender. However, depending on whether the information is framed in favour of purchasing an annuity or in favour of investing, and controlling for the level of risk tolerance and financial literacy, men and women are influenced differently. While men are similarly influenced by information highlighting the disadvantages of investing or annuitising, women are significantly more influenced by an information bias towards investing that highlights the drawbacks of annuities. Taking into account the dissimilar impact of framing on men and women can be of particular importance when designing communication campaigns and when setting default investment strategies.

Although risk preferences may play a role, most of the gender difference in expected retirement benefits may be linked to contribution levels. Based on average returns expected for the different investment strategies available to Australian members of the UniSuper superannuation scheme, Watson and McNaughton (2007[5]) find that income, thus contributions, differences between men and women explain most of the difference in expected retirement benefits. Marketing and information campaigns targeted towards the specific needs of women can therefore be set up to better communicate towards women in order to increase their contributions and their financial education.

Women’s behaviour may be more variable than men’s and influenced by their environment. In addition to their work on risk preferences, Croson and Gneezy (2009[11]) look at social preferences and the impact of gender on altruism, envy, inequality aversion and reciprocity, comparing different studies using several methods such as ultimatum games, trust games or dictator games.46 The authors find that while results across studies may vary, overall women tend to be more responsive to the conditions of the experiment. For example, in an ultimatum game, women respond more positively if the bargaining party is sitting opposite them rather than behind a curtain. Similarly, in the trust game, women trust less than men when they only have written information about the person they are asked to trust, and more than men when they are provided with a picture of this person. Women’s behaviour can therefore be more variable than men’s according to the context of the experiment, which could validate the need for targeted communication campaigns towards women, or specific training for financial advisors.

Targeted communication campaigns may be effective in increasing women’s participation in private defined contribution plans. Anderson and Collins (2017[56]) examine the effect of a financial education campaign called EMPOWER (Embracing and Promoting Options for Women to Enhance Retirement) and designed to improve the participation of women to a supplementary occupational pension plan set up for non-school public agencies workers in Wisconsin. The campaign consisted of monthly emails to all workers in the agencies, posters in the workplace, invitations to webinars, and live events available to women workers only. The agencies which put EMPOWER in place were on average those with the highest initial gender gaps, and results were observed before and after the campaign, for men and women, for agencies participating and for those not participating in EMPOWER. The campaign is estimated to have had an overall effect of increasing participation by 2.6%, closing the gender gap by more than half. The effect was largest for younger workers, i.e. those who might not have been as aware of or preoccupied by their retirement options, and for lower-earning workers. However, the contribution rates of workers who were already participating did not significantly increase with the campaign.

Communication on the effects of specific life events, such as parental leave or divorce, on retirement income may be a way to increase women’s awareness on retirement planning. Providing explanations on the potential effect of parental leave on retirement savings, and options to overcome these effects may help to address part of the gender gap in retirement income. Adapting communication to behavioural elements, such as risk aversion and confidence, and taking gender and life events into account when designing communication campaigns could be a solution to address the potential gender differences in behaviours related to pensions. This has only started to be considered by policy makers, pension schemes and trustees very recently (Imeson, 2019[57]).

The literature shows that differences in risk tolerance, financial literacy and in attitudes towards saving can have important effects on the gender gap in retirement savings. Societal factors such as the marital status, gender stereotyping and communication may also amplify this gap between men and women.

Although there is abundant literature on potential labour, behavioural, and cultural elements potentially explaining the gender gap in retirement income, research seems to indicate that the issue is now higher up on governments, policy makers, pension stakeholders, and investors’ agenda. This indicates that solutions may be sought and that improvements may be expected in the future. Improvements in the imbalance between men and women’s retirement income have already started to appear in several countries. For example, the European Commission’s “Mind the gap in Pensions” project includes an analysis of some of the possible causes for the gender gap in pensions, including the impact of framing and career choices on the income received in retirement by men and women (Dekkers, Hoorens and Van Den Bosch, 2020[58]).

The Group of 20 (G20) endorsed a policy guidance on addressing women’s and girls’ needs for financial awareness and education in 2013 (OECD, 2013[59]). To overcome some of the challenges linked to the gender disparity in financial education and the gender differences in financial needs, policy guidance providing an international framework as well as delivery tools for policy makers was prepared by the OECD, and endorsed by leaders of the G20 in September 2013. It recommends considering national priorities and needs to identify and address the areas in which the development of financial education approaches targeted specifically at women and girls may be required.

The role of men and women in society has changed and gender stereotyping is decreasing in several jurisdictions. A change in mentalities with respect to the role of men and women in society has occurred in the United Kingdom since 1984, when 43% of respondents agreed that men should earn money while women should take care of the family, versus 8% in 2017. Additionally, there has been a shift in the representation of women in company boards and in prominent public roles (Government Equalities Office, 2019[48]).

There has been an increase in women’s participation and savings rate in several OECD countries, which will eventually lead to a decrease in the gender gap in retirement savings. Working women aged 30 to 39 currently have a higher pension participation rate than men in the United Kingdom (Now Pensions, 2019[45]).

Active ownership by institutional investors might help reduce the occupational pension gap. Institutional investors such as for instance state pension funds in the United States have started showing more interest in the way companies treat gender neutrality, by asking them to disclose information on the gender pay and promotion gap for instance (Temple-West, 2019[60]). These pension funds see issues related to gender inequality as potential risks for the future and hence part of their fiduciary duty towards their members. The disclosure of information can affect the voting behaviour of investors in investee companies’ annual general meetings. Active investors can decide to sanction management in their voting decisions if they think gender equality is not sufficiently targeted, or they can engage with management in order to initiate changes and improvements. Such active participation of investors in management decisions of companies can encourage companies to tackle the gender gap in their treatment of employees in several aspects, including in matters related to occupational pension and the gender pension gap.

Allowing defined contribution pension scheme members to choose their investment strategy may be positive, even for participants with a lower risk tolerance. Australia’s UniSuper scheme moving from a single investment strategy to a choice between seven options with different risk and return profiles has had a positive impact on expected retirement incomes of both men and women, despite women choosing more conservative investment strategies than men on average (Watson and McNaughton, 2007[5]). Consistent with the recommendation to allow individual participants in defined contribution pension arrangements to choose their investment strategy stated from the OECD Roadmap for the Good Design of Defined Contribution Pension Plans (OECD, 2012[61]), imposing a single investment strategy may therefore not be a suitable solution to counterbalance the gender gap in retirement income.


[32] Age UK (2018), For love and money: Women’s pensions, expenditure and decision-making in retirement.

[55] Agnew, J. et al. (2008), Who chooses annuities? An experimental investigation of the role of gender, framing, and defaults, https://doi.org/10.1257/aer.98.2.418.

[6] Allport, W. et al. (2019), How the UK Saves: Effects of gender on retirement savings behaviour Member experience from the National Employment Savings Trust (NEST), Vanguard and NEST Insight.

[56] Anderson, D. and J. Collins (2017), Can knowledge empower women to save more for retirement?, https://crr.bc.edu/wp-content/uploads/2017/09/wp_2017-12.pdf (accessed on 25 February 2019).

[9] Arch, E. (1993), “Risk-Taking: A Motivational Basis for Sex Differences”, Psychological Reports, Vol. 73/1, pp. 3-11, https://doi.org/10.2466/pr0.1993.73.1.3.

[25] Atkinson, A. and F. Messy (2012), “Measuring Financial Literacy: Results of the OECD / International Network on Financial Education (INFE) Pilot Study”, OECD Working Papers on Finance, Insurance and Private Pensions, No. 15, OECD Publishing, Paris, https://dx.doi.org/10.1787/5k9csfs90fr4-en.

[4] Baeckstrom, Y., I. Marsh and J. Silvester (2019), Financial Advice, Wealth and Gender: Risk Tolerance, Knowledge and Confidence, https://www.bloomberg.com/news/articles/2018-06-14/millionaires-now-control-.

[39] Bajtelsmit, V. and A. Bernasek (1996), Why do women invest differently than men?, Association for Financial Counseling and Planning Education, http://ssrn.com/abstract=2238.

[42] Bajtelsmit, V. et al. (2005), Women and Pensions, https://doi.org/10.1111/j.1540-6296.2005.00047.x.

[23] Bannier, C. and M. Neubert (2016), “Gender differences in financial risk taking: The role of financial literacy and risk tolerance”, Economics Letters, Vol. 145, pp. 130-135, https://doi.org/10.1016/j.econlet.2016.05.033.

[18] Barber, B. and T. Odean (2001), Boys will be boys:gender, overconfidence, and common stock investment.

[69] Barker, A. (2018), “Increasing inclusiveness for women, youth and seniors in Canada”, OECD Economics Department Working Papers No.1519, OECD Publishing, Paris, https://doi.org/10.1787/83cb8b8d-en.

[22] Christiansen, C., J. Joensen and J. Rangvid (2008), “Are economists more likely to hold stocks?”, Review of Finance, Vol. 12/3, pp. 465-496, https://doi.org/10.1093/rof/rfm026.

[8] Cronqvist, H. et al. (2018), When Nudges Are Forever: Inertia in the Swedish Premium Pension Plan, https://ssrn.com/abstract=3099886.

[11] Croson, R. and U. Gneezy (2009), “Gender differences in preferences”, Journal of Economic Literature, Vol. 47/2, pp. 448-474, https://doi.org/10.1257/jel.47.2.448.

[47] Dale, M. and S. St John (2020), Women and Retirement in a post COVID-19 world.

[58] Dekkers, G., V. Hoorens and K. Van Den Bosch (2020), Project text “Mind the Gap”.

[67] Even, W. and D. Macpherson (1990), “The Gender Gap in Pensions and Wages”, The Review of Economics and Statistics, Vol. 72, pp. 259-265.

[17] Felton, J., B. Gibson and D. Sanbonmatsu (2003), “Preference for Risk in Investing as a Function of Trait Optimism and Gender”, Journal of Behavioral Finance, Vol. 4/1, pp. 33-40, https://doi.org/10.1207/S15327760JPFM0401_05.

[2] Fisher, P. and R. Yao (2017), “Gender differences in financial risk tolerance”, Journal of Economic Psychology, Vol. 61, pp. 191-202, https://doi.org/10.1016/j.joep.2017.03.006.

[1] Garnick, D. (2016), Income Insights: Gender Retirement Gap.

[68] Gherzi, S. et al. (2014), “The meerkat effect: Personality and market returns affect investors’ portfolio monitoring behaviour”, Journal of Economic Behavior and Organization, Vol. 107/PB, pp. 512-526, https://doi.org/10.1016/j.jebo.2014.07.013.

[12] Gneezy, U., K. Leonard and J. List (2009), “Gender differences in competition: Evidence from a matrilineal and a patriarchal society”, Econometrica, Vol. 77/5, pp. 1637-1664, https://doi.org/10.3982/ECTA6690.

[63] Goverment Equalities Office (2019), Gender equality at every stage: a roadmap for change, HM Government.

[48] Government Equalities Office (2019), Gender Equality Monitor, HM Government.

[62] Government Equalities Office (2019), The case for change, HM Government.

[31] Henry, L. (2014), “Income inequality and income class consumption patterns”, Economic Commentary - the Federal Reserve Bank of Cleveland.

[21] Hibbert, A., E. Lawrence and A. Prakash (2013), “Does knowledge of finance mitigate the gender difference in financial risk-aversion?”, Global Finance Journal, Vol. 24/2, pp. 140-152, https://doi.org/10.1016/j.gfj.2013.07.002.

[19] Hinz, R., D. McCarthy and J. Turner (1997), “Are women conservative investors? Gender differences in participant-directed pension investments”, in Positioning pensions for the twenty-first century, The Pension Research Council (the Wharton School of the University of Pennsylvania) and University of Pennsylvania Press (Philadelphia).

[14] Ho, C. (2018), “The Effects of Social Interaction and Psychological Bias on Trading Behavior: Evidence from a Laboratory Experiment”, Journal of Mathematical Finance, Vol. 08/01, pp. 178-196, https://doi.org/10.4236/jmf.2018.81014.

[53] Imeson, S. (2019), Bridging the gender pensions gap - How the industry can help improve retirement adequacy for women, Pensions-expert.com, https://www.pensions-expert.com/Law-Regulation/How-schemes-can-play-their-part-in-bridging-the-gender-pensions-gap?ct=true (accessed on 23 August 2019).

[57] Imeson, S. (2019), How schemes can play their part in bridging the gender pensions gap, Pensions-Expert.com, https://www.pensions-expert.com/DC-Auto-enrolment/How-schemes-can-play-their-part-in-bridging-the-gender-pensions-gap?ct=true (accessed on 3 October 2019).

[40] Johnston, D., S. Kassenboehmer and M. Shields (2016), “Financial decision-making in the household: Exploring the importance of survey respondent, health, cognitive ability and personality”, Journal of Economic Behavior and Organization, Vol. 132, pp. 42-61, https://doi.org/10.1016/j.jebo.2016.09.014.

[66] Kesavayuth, D., R. Rosenman and V. Zikos (2016), “Retirement, personality, and well-being”, Economic Inquiry, Vol. 54/2, pp. 733-750, https://doi.org/10.1111/ecin.12307.

[10] Larkin, J. and H. Pines (2003), Gender and Risk in Public Performance.

[27] Lusardi, A. and O. Mitchell (2008), “Planning and financial literacy : how do women fare?”, American Economic Review: Papers & Proceedings, Vol. 98/2, pp. 413-417, https://www.dartmouth.edu/~alusardi/Papers/AER-FinalPublishedVersion.pdf.

[28] MacLeod, P. et al. (2012), Attitudes to pensions : the 2012 survey, Department for Work and Pensions.

[3] Makarov, D. and A. Schornick (2010), Explaining Households’ Investment Behavior, http://ssrn.com/abstract=1623785.

[13] Marinelli, N., C. Mazzoli and F. Palmucci (2017), “How does gender really affect investment behavior?”, Economics Letters, Vol. 151, pp. 58-61, https://doi.org/10.1016/j.econlet.2016.12.006.

[49] Mcguinness, F. and D. Pyper (2018), The Gender Pay Gap, http://www.parliament.uk/commons-library%7cintranet.parliament.uk/commons-library%[email protected]%7c@commonslibrary.

[29] Nolan, A. et al. (2019), Gender, pensions and income in retirement, Economic and Social Research Institute, Dublin, Ireland, https://doi.org/10.26504/rs87.

[45] Now Pensions (2019), Facing an unequal future Closing the gender pensions gap, https://www.nowpensions.com/wp-content/uploads/2019/07/NP-gender-pensions-gap-report.pdf.

[7] OECD (2018), “Improving retirement incomes considering behavioural biases and limited financial knowledge”, in OECD Pensions Outlook 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/pens_outlook-2018-8-en.

[37] OECD (2017), G20/OECD INFE report on adult financial literacy in G20 countries, http://www.oecd.org/daf/fin/financial-education/G20-OECD-INFE-report-adult-financial-literacy-in-G20-countries.pdf.

[33] OECD (2017), Preventing Ageing Unequally, OECD Publishing, Paris, https://doi.org/10.1787/9789264279087-en.

[26] OECD (2017), The Pursuit of Gender Equality: An Uphill Battle, OECD Publishing, Paris, https://doi.org/10.1787/9789264281318-en.

[59] OECD (2013), G20 OECD/INFE policy guidance on addressing women’s and girls’ needs for financial awareness and education, http://www.oecd.org/daf/fin/financial-education/G20-Women-Girls-Fin-Ed-Policy-Guidance-2013.pdf.

[38] OECD (2013), Women and Financial Education: Evidence, Policy Responses and Guidance, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264202733-en.

[61] OECD (2012), The OECD roadmap for the good design of defined contribution pension plans, http://www.oecd.org/pensions/designingfundedpensionplans.htm.

[20] Palme, M., A. Sundén and P. Söderlind (2004), INVESTMENT CHOICE IN THE SWEDISH PREMIUM PENSION PLAN.

[36] Pensions Policy Institute (2019), Understanding the Gender Pensions Gap, http://www.pensionspolicyinstitute.org.uk.

[43] Portas, J. (2019), Securing the financial future of the next generation, Insuring Women’s Futures.

[51] Prabhakar, R. (2017), “Why do people opt-out or not opt-out of automatic enrolment? A focus group study of automatic enrolment into a workplace pension in the United Kingdom”, Journal of European Social Policy, Vol. 27/5, pp. 447-457, https://doi.org/10.1177/0958928717717656.

[24] Pyles, M. et al. (2016), “Cultural influences on risk tolerance and portfolio creation”, Journal of Behavioral and Experimental Finance, Vol. 9, pp. 43-55, https://doi.org/10.1016/j.jbef.2016.01.001.

[35] Rappaport, A. (2018), Women and Post Retirement Risks, Society of Actuaries.

[34] Rappaport, A. et al. (2017), Aging and Retirement Caregiving for Older Individuals: Perspectives of the Caregiver and the Care Recipient 2017 Risks and Process of Retirement Survey Report Retirement Survey Report Key Findings and Issues, http://www.soa.org.

[41] Richard Shillington, B. et al. (2016), An Analysis of the Economic Circumstances of Canadian Seniors, Broadbent Institute.

[54] Roszkowski, M. and J. Grable (2005), “Gender Stereotypes in Advisors’ Clinical Judgments of Financial Risk Tolerance: Objects in the Mirror Are Closer than They Appear”, Journal of Behavioral Finance, Vol. 6/4, pp. 181-191, https://doi.org/10.1207/s15427579jpfm0604_2.

[15] Sarin, R. and A. Wieland (2016), “Risk aversion for decisions under uncertainty: Are there gender differences?”, Journal of Behavioral and Experimental Economics, Vol. 60, pp. 1-8, https://doi.org/10.1016/j.socec.2015.10.007.

[16] Säve-Söderbergh, J. (2012), “Self-Directed Pensions: Gender, Risk, and Portfolio Choices”, Scandinavian Journal of Economics, Vol. 114/3, pp. 705-728, https://doi.org/10.1111/j.1467-9442.2012.01710.x.

[44] Schneebaum, A. et al. (2018), “The Gender Wealth Gap Across European Countries”, Review of Income and Wealth, https://doi.org/10.1111/roiw.12281.

[30] Scottish Widows (2018), The 2018 annual women and retirement report.

[64] Sierminska, E., J. Frick and M. Grabka (2009), “Examining the Gender Wealth Gap in Germany”, SSRN, https://doi.org/10.2139/ssrn.1424899.

[60] Temple-West, P. (2019), Pension funds step up campaign to reduce gender pay gap, Financial Times, https://www.ft.com/content/10f9caca-c1e0-11e9-a8e9-296ca66511c9 (accessed on 20 August 2019).

[52] The People’s Pension (2019), The Gender Pensions Gap: Tackling the motherhood penalty.

[65] Thurley, D. et al. (2018), Women and Pensions, House of Commons Library, https://doi.org/10.1111/j.1540-6296.2005.00047.x.

[5] Watson, J. and M. McNaughton (2007), “Gender differences in risk aversion and expected retirement benefits”, Financial Analysts Journal, Vol. 63/4, pp. 52-62, https://doi.org/10.2469/faj.v63.n4.4749.

[50] Which? (2019), Top up the Pots - Achieving adequate retirement incomes with automatic enrolment.

[46] Woodward, H. and M. Sefton (2014), Pensions on divorce: an empirical study, Cardiff University.


← 1. This literature review attempts to cover research from as many countries as possible, although some countries such as the United Kingdom and the United States may appear more often than others, reflecting the abundant literature currently available in these two OECD jurisdictions.

← 2. 2014 data from the Cogent Reports and Investor Brandscape Custom Data Tables.

← 3. The study refers to 401(k) and 403(b) private pension plans.

← 4. 2013 United States Survey of Consumer Finances data.

← 5. Married and partnered respondents were sampled out as the survey data would not allow to differentiate financial decisions made or influenced by spouses or partners. Final sample size: 2 246 individuals.

← 6. Sample size: 500 individuals.

← 7. UniSuper is the sole provider of retirement funds for Australian universities staff, both academic and general. Sample size: 32 061.

← 8. This control was suggested by the observation that men in the lowest earnings brackets and who contribute occasionally have lower contributions and account balances than their gender counterparts, mostly attributable to more transient employment patterns and higher job turnover.

← 9. Given that NEST does not cover the higher earnings categories of the population however, a caveat on data representativeness should be mentioned as results might therefore not hold for higher socio-economic groups.

← 10. Sample size: 2 374 individuals.

← 11. Sample size: 88 individuals.

← 12. Elicitation methods used for the valuation of the bet included the price at which participants would be willing to sell a lottery ticket paying if their choice was correct, an ordinal comparison of outcomes, and a certainty equivalent, i.e. the amount that participants would accept for being indifferent between participating or not in the betting game.

← 13. The objective probability game was a lottery game with 10 white balls and 20 yellow balls, asking participants to state how much they would be willing to pay to play a game where they would win USD 100 if a white ball was drawn.

← 14. Data from the 1999 Swedish Household Survey on Income (HINK) and information on individuals’ investment choices in the PPM individual accounts. Sample size: 17 987 individuals.

← 15. This result may seem surprising given the low rate of people choosing their own investment funds previously presented in the study by Cronqvist et al. (2018[8]) for 2016. However, Cronqvist et al. (2018[8]) show that this rate is heavily dependent on communication campaigns, and went from 66.6% in 2000, to 18.4% in 2001 and down to 0.9% in 2019. Large communication campaigns were carried out in 2000 when the new Premium Pension system was introduced for all workers in Sweden, by both the Swedish government and individual funds, to encourage individuals to select their own investment funds instead of choosing the default option. Communication and advertising efforts were significantly reduced in subsequent years, when only new entrants in the labour market were joining the PPM.

← 16. Sample size: 66 individuals.

← 17. The Revised Life Orientation Test (LOT-R) is a 10-item scale that measures how optimistic or pessimistic people feel about the future. It was developed in 1994 by psychologists Scheier, Carver and Bridges.

← 18. In this analysis, the number of futures and options traded is considered as showing a greater risk tolerance through speculation, stocks traded on the Nasdaq are considered more volatile hence positively associated with risk tolerance, while stocks traded on the NYSE or the ASE are considered to represent more traditional blue-chip companies and hence negatively associated with risk tolerance.

← 19. Sample size: 37 664 households.

← 20. The gender difference being statistically significant for both indicators.

← 21. Data from the 1990 survey of participants in the Federal government’s Thrift Savings Plan (TSP).

← 22. Data from the 2010 Longitudinal INdividual DAta panel (LINDA) and information on individuals’ investment choices in the PPM individual accounts in 2010. Sample size 244 750 individuals.

← 23. The general population data is drawn from the 2004 Federal Reserve Board’s Survey of Consumer Finance – SCF, of which 4 216 adults with a college degree were considered highly educated. The finance faculty sample size was 1 147.

← 24. The general population sample was built from the Danish Institute of Governmental Research data, covering 10% of the Danish population from 1997 to 2001, i.e. more than 400 000 individual investors with annual data. Highly educated individuals were the 19 233 individuals in the panel with at least 18 years of schooling, corresponding to holding a master’s degree or above. Economists were the 5 148 individuals in the sample with economics education of 2, 4 or 6 years beyond high school.

← 25. Perceived financial education refers to an individual’s own assessment of their financial literacy compared to the average person, while actual financial education can be assessed by objective means such as through questionnaires or surveys.

← 26. Based on 2 047 responses from the German household survey SAVE for 2009.

← 27. Sample size: 758 individuals.

← 28. Sample size: 1 949 randomly selected adults.

← 29. Data based on 5 148 adults in the United Kingdom for the year 2018. Adequate saving for retirement defined for individuals as either saving at least 12% of income or expecting that their main retirement income will come from a defined benefit plan.

← 30. This may be due to personal preferences and/or to societal expectations.

← 31. 2017 Society of Actuaries’ Risks Survey.

← 32. Combining data from the 2014 American Community Survey, the 2017 Society of Actuaries’ Risks and the Process of Retirement Survey, focus groups and interviews.

← 33. 2014 data from Social Security Administration.

← 34. According to the 2017 Society of Actuaries Risk Survey, 87% of women aged 85 and over are single – either unmarried or widowed, compared to 49% of men aged 85 and over.

← 35. Sample size: 1 031 individuals.

← 36. Statistics Canada and OECD data.

← 37. United States Census Bureau 2000 data.

← 38. Survey of people aged between 55 and 70, and living with a partner. Sample size: 1 010.

← 39. Based on data from the Household Finance and Consumption Survey (HFCS) for Austria, Belgium, France, Germany, the Slovak Republic and Spain.

← 40. Sample size: 369. In 20% of cases, no relevant pensions were disclosed by members of the couple.

← 41. Majority-led meaning controlled by a single woman or having a management team of which a majority were women.

← 42. Sample size: 44 individuals. Opt-out rates were 7% for participants under 30, 9% for participants aged 30 to 49, and 23% for those aged 50 and above.

← 43. Reasons to opt out mentioned were lack of affordability, the presence of more financially attractive alternatives to a pension, and limited time for returns to be generated by a private pension for some participants aged 50 and above.

← 44. Sample size: 183 financial advisors in the insurance industry and 290 clients.

← 45. Using a sample of 845 United States participants aged 19 to 89.

← 46. In an ultimatum game, one player - the allocator - is endowed with a sum of money and asked to split it between herself and an anonymous player - the recipient. The recipient may either accept the allocator’s proposal or reject it, in which case neither of the players will receive anything. In a trust game, one player can send all, some, or none of her endowment to a second player. The amount sent is multiplied by 3, and received by the second player, who can then decide to return as much or as little of the money in her possession to the first player. A dictator game is similar to the second stage of a trust game, as only one player chooses whether and how much of her endowment to send to another anonymous player.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2021

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at http://www.oecd.org/termsandconditions.