Why Are Women More Pessimistic About Social Security's Future Than Men?

Introduction

In studies of public expectations about the future of the Social Security Old-Age and Survivors Insurance (OASI) program, researchers have found that women are more pessimistic than men. In this article, we review an extensive literature and apply an original analysis of survey data from 2020 to examine why men's and women's Social Security expectations differ.¹ We consider economic and behavioral explanations for those differences, such as the potential role of economic vulnerability. If women or men have unrealistically pessimistic or optimistic Social Security expectations, they may make poor decisions about how much to save for retirement. They may also view their Social Security benefits as less or more valuable than they are (Burkhauser and Turner 1985).

Statistically, men and women face different prospects for their future Social Security benefits. For example, women typically rely on Social Security benefits for longer periods than men. In 2020, women who reached age 65 were expected to live, on average, an additional 23.9 years, compared with 21.4 years for men.² Women represented 55.3 percent of OASI beneficiaries aged 62 or older and 63.9 percent of beneficiaries aged 85 or older. Yet the median earnings of women aged 15–64 who worked full-time for 50 weeks or more in 2019 were $45,000, compared with $54,000 for men. Correspondingly, the average annual Social Security benefit received by women aged 65 or older was $13,505 in 2019, compared with $17,374 for men. If Social Security did not also provide benefits to spouses, divorced ex-spouses, and widow(er)s of covered workers—most of which are paid to women—the retirement income gap would be wider still (Social Security Administration [SSA] 2021a).³

This introduction is followed by a literature review presented in three parts, moving from general issues of pessimism or optimism and economic expectations to expectations about Social Security benefits and then to expectations about Social Security system solvency. The first part of the literature review, divided into five subsections, covers psychological and socioeconomic factors underlying expectations about economic conditions in general. Then, a section explores the types of risks to Social Security benefits faced by all future beneficiaries, which may differ between women and men. Part two of the literature review focuses on expectations about one's Social Security benefits, divided into subsections examining each of four factors that potentially underlie gender differences. Part three concludes the literature review by turning to focus on issues related specifically to the solvency of the Social Security trust fund and on gender differences in those expectations.

The literature review is followed by a section that uses detailed information from AARP's 85^th Anniversary of Social Security Survey (Perron 2020) to analyze gender differences in Social Security expectations. In two subsections, we first present selected survey results in detail; then, we describe our multivariate analysis and present the results of our logistic regressions. A concluding section summarizes our findings.⁴

Literature Review Part I: Psychological and Economic Research

This section addresses research on how expectations are formed and acted on. It is arranged in five subsections. The first subsection briefly reviews literature on the psychology of optimism and pessimism. The second subsection reviews studies examining expectations about specific economic variables, which may provide insight into expectations about Social Security. The third subsection discusses the literature exploring gender differences in economic expectations. The fourth subsection discusses the effects of income and education on expectations. The fifth subsection discusses whether gender differences in trust in government might play a role in Social Security expectations.

The Psychology of Optimism and Pessimism

Psychologists define dispositional pessimism as a general tendency to expect negative outcomes (Carver, Scheier, and Segerstrom 2010) and many studies have examined whether the tendency is more prevalent on some topics or among some groups than others. Perozek (2008) found that women are more likely to underestimate their life expectancy than men but did not explore reasons for the difference. Comerford (2021) posited that differences in predicted life expectancy might be related to the fact that women have a higher probability of living to an older age than men, rather than reflecting an innate gender difference in pessimism.

Hinz and others (2017) used a standard psychological instrument⁵ to test for gender differences in dispositional optimism and pessimism in a sample of 10,000 German adults. The authors treated optimism and pessimism as two distinct factors rather than a locus along a single continuum. They found that men were slightly less optimistic than women, but there were no gender differences in dispositional pessimism. Extremera, Durán, and Rey (2007) used a different psychological test but likewise considered optimism and pessimism as distinct factors. Studying Spanish adolescents, they found that girls were both more pessimistic and less optimistic than boys. Heinonen and others (2006), using the same test that Hinz and others used in their 2017 study, found that dispositional optimism was linked to childhood socioeconomic status (SES).

Differences in optimism and pessimism might be innate, or the result of childhood experiences or adult socioeconomic circumstances, or a combination of these factors. Further, they can vary depending on the object of one's expectations. Based on their literature survey, Chopik and others (2020) wrote that optimism (or pessimism) is partially heritable but is also influenced by life events and circumstances.

Economics Literature Addressing Expectations

Bordalo, Gennaioli, and Shleifer (2016) presented a “diagnostic expectations” model that explores a behavioral tendency to overreact to incoming negative news and form excessively pessimistic expectations. However, the study did not examine gender differences in diagnostic expectations. Norr (2017) attributed the negative views of some U.S. workers on the future of Social Security to “negativity bias,” or a tendency to exaggerate negative information, such as widely available news or opinions about the need for Social Security reform to preserve program solvency. According to AARP, 19 percent of survey respondents incorrectly believed that the potential depletion of the Social Security trust fund reserve would leave the system unable to pay any benefits (Williams 2015).

Burke and Manz (2014) found that economic literacy among survey respondents was positively associated with an ability to forecast inflation accurately, irrespective of sex and other socioeconomic characteristics. In particular, the forecasts of respondents with lower economic literacy tended to overestimate price increases. This finding suggests that different levels of knowledge about the financing of the Social Security program may help to explain differences in expectations.

Gender Differences in Economic Expectations

Harris, Jenkins, and Glaser (2006) argued that in evaluating negative expectations, it is helpful to distinguish between the probability of a negative occurrence and the severity of its consequences. The authors studied risky behavior and found that women tend to assume that a negative outcome is more probable than men do, yet for the expected severity of negative outcomes, the results were mixed. Marin and others (2012) found that women are more likely than men to remember, and to experience stress from, negative news.

Jacobsen and others (2014) found that men are more optimistic than women on a broad range of economic topics, including the stock market, economic growth, interest rates, and inflation. Gender differences persist after statistically controlling for other characteristics such as income, employment status, wealth, education, and marital status. The authors found that women are less optimistic than men about being able to retire at an age that is acceptable to them.

Bryan and Venkatu (2001) found that women perceive higher rates of past inflation and predict higher future inflation than men, even after controlling for age, education, income, marital status, and race. The authors found that gender differences in inflation perceptions may be influenced by prices of specific goods, such as gasoline, as well as by general perceptions. Consumers who purchase different baskets of goods have different experiences of price increases and form inflation expectations that vary demographically based on those experiences (Jonung 1981). Bjuggren and Elert (2019) found that men in Sweden are more optimistic about the economy than women.

Wong and Hardy (2009) studied women's expectations about retirement and found substantial heterogeneity in expected retirement ages both for different individuals in a given time and for the same individuals over a 7-year period. Same-person fluctuations suggested considerable uncertainty and variability in expectations about retirement income and Social Security benefits over time.

Effects of Income and Education on Expectations

Das, Kuhnen, and Nagel (2017) found that people with higher SES are more optimistic about future macroeconomic developments such as business conditions, the unemployment rate, and stock market returns. Conversely, low-SES individuals report excessively pessimistic expectations compared with those of professional forecasters and historical data.

Extending Das, Kuhnen, and Nagel's findings to Social Security, our study suggests that people with low SES may also be more pessimistic about their future benefits than people with higher SES. To the extent that women have lower SES than men, their expectations would presumably be more pessimistic. In the United States, women, on average, have lower income, fewer hours and years worked, and lower-earning careers than men, in part because they are more likely to pause their careers to care for family members (Erosa and others 2022). We discuss our empirical findings on these themes later.

Providing evidence on the economic differences between men and women, Table 1 shows real median annual earnings (in 2019 dollars) for men and women aged 50–59 in 1989 and 2019. In both years, real median earnings were substantially lower for women than for men. However, women's real median earnings as a percentage of men's rose substantially, from 48.9 percent to 67.4 percent. Over the period, real median earnings increased from $30,210 to $39,940 for women but decreased for men, from $61,820 to $59,220.

Mercer CFA Institute (2021, Chapter 4), studying the retirement income systems of 43 countries and defining pension income to include Social Security benefits, identified a gender pension gap in every country's retirement system. The study found that women's average pension income in the United States is approximately 34 percent lower than men's because women have lower earnings and more career pauses for family caregiving.

The wage gap influences gender differences in poverty. Although the difference in U.S. men's and women's poverty rates narrows from ages 35 to 64, the gap widens at older ages. More than one in eight women (13.2 percent) aged 75 or older live in poverty, compared with 8.8 percent of men (Bleiweiss, Boesch, and Gaines 2020). Mortality rates are higher for people in poverty than for those who are not. When measuring the poverty risk of the aged, correcting for this mortality difference (thereby avoiding a survivorship sample selection bias) increases the estimated risk relative to an estimate based on traditional poverty measures, for both men and women (Muller and Turner 2022). In other words, traditional measures understate the risk of falling into poverty at older ages because disproportionate shares of people in poverty at younger ages leave the sample because of their relatively high mortality. This effectively reduces the gender gap in poverty because men's higher mortality rates cause a greater sample selection bias for men than for women.

To the extent that people with lower incomes are more likely to have negative economic expectations across a range of variables, gender differences in economic circumstances may help explain findings that women tend to have more negative expectations about Social Security than men. Thus, greater negativity about Social Security among women may reflect a general pattern in male-female differences in economic expectations.

Trust in Government

The literature we review identifies gender differences in financial and economic expectations. By contrast, McDermott and Jones (2020) did not find gender differences in their literature review on trust in government. Thus, issues of trust presumably do not affect the gender differences in pessimism about Social Security's future.

How Women's and Men's Social Security Risks Differ

To some extent, men and women face different potential risks to future Social Security benefits. These risks can be divided into three types. The first type, political risk, relates to any future changes in Social Security legislation. Such changes would be motivated primarily by the need to restore long-term solvency to the Social Security trust funds and could alter the distributions of program contributions and benefits. The second is future earnings risk, discussed in more detail in a following section. The third is temporal risk, or the length of time an individual is exposed to the other Social Security risks. The temporal risk is greater for women than for men because of their longer life expectancy.

We explore the extent to which these risks, or the perception of them, vary by sex. For example, political risks may vary in part because women generally have lower lifetime earnings and, for that reason, depend more than men on Social Security benefits in retirement. Thus, the risk related to any future legislated changes in Social Security benefits may be greater for women than for men.

Literature Review Part II: Factors Underlying Gender Differences in Social Security Expectations

Recent research on workers' expectations about their future Social Security benefits includes an international study (Turner and others 2019) and a U.S. study (Turner and Rajnes 2021). This article extends that research by focusing on gender differences in Social Security expectations. This section continues the literature review by exploring the specific factors that may underlie those gender differences.

Level of Information

Surveys that address specific Social Security topics indicate that U.S. workers generally have limited knowledge about the program, particularly regarding how benefits are calculated (Lusardi and Mitchell 2007; Greenwald and others 2010; Yoong, Rabinovich, and Wah 2015). A 2021 survey found that, in general, men correctly answered specific questions about Social Security, such as whether benefits are protected from inflation, more often than women (Nationwide Retirement Institute 2021). Surveys have also determined that women generally have lower financial literacy test scores than men (Lusardi and Mitchell 2007).

In their literature review, Turner and Rajnes (2021) found that people with lower incomes tend to be more pessimistic about the amount of their future Social Security benefit. However, women's expectations might partially reflect conservatism in planning rather than pessimism alone. People who underestimate their future Social Security benefits may increase their retirement savings, leading to a better-funded and more diversified retirement portfolio.

AARP's 2015 survey on Social Security and retirement expectations likewise found that women were more pessimistic than men about the future of Social Security (Williams 2015). When asked to respond to the statement, “Social Security will not be there for you when you need it,” 72 percent of women and 56 percent of men agreed. To a parallel statement, “Social Security won't be enough for you to get by on,” the response was nearly identical, with 72 percent of women and 57 percent of men agreeing. Yet despite those negative views, 84 percent of women and 75 percent of men responded that they plan to rely on Social Security as a substantial or somewhat substantial source of income in retirement, possibly for lack of other options.

Women likely to receive eventual Social Security benefits based on the earnings record of a spouse may be less knowledgeable about how benefits are calculated than are women who will receive benefits based on their earnings. In 2014, just over 50 percent of female Social Security beneficiaries received benefits based on their earnings alone (Iams 2016). In SSA (2021b), the agency projects that 57.5 percent of female beneficiaries older than 60 will receive benefits based solely on their earnings record in 2025 and that by 2095, more than 70 percent of women will receive such benefits. As a result, women's expectations may become more like men's over time as their labor market experiences and Social Security eligibility become increasingly comparable.

Marital Status

Bernheim (1987) used the longitudinal Retirement History Survey to compare workers' predictions of their future Social Security benefits to the actual benefits they later received. The initial panel of respondents comprised individuals aged 58–63 in 1969. Bernheim found that expected benefits were about 10 percent lower on average than the actual benefits respondents later received. He concluded that even people close to retirement tended to be pessimistic and to underestimate their future benefits. Widows and single women made the most accurate—and most conservative—estimates, perhaps because many of them knew they would depend on Social Security benefits as their primary income source. Married men were the least conservative and least accurate in their estimates. The Retirement History Survey did not include married women, a major data shortcoming. The survey's “single women” category combined the never married and the divorced.

However, not everyone underestimates their future Social Security benefits. Of those who overestimated their future benefits in Bernheim (1987), one in six did so by at least 25 percent, and one in 12 overestimated them by at least 50 percent. Men were more likely to overestimate their future benefits than women. Among singles, one-fifth of men overestimated their future benefits, compared with one-tenth of women.

Quinby and Wettstein (2021) found that marital status had an insignificant effect on the difference between expected and scheduled benefits. The authors did not explore gender differences.

Financial Literacy

Prados and Kapteyn (2019) surveyed individuals aged 30 or older who did not have a disability and were not retired. The authors calculated the respondents' likely future Social Security benefits and compared their calculations to the respondents' expectations. They found that men's predictions were more accurate than women's. Further, men overall (as well as men and women who reported less uncertainty about their future benefits) were less likely to overestimate them. However, the gender effect was insignificant when the authors controlled for financial literacy and attitudes toward planning. Thus, gender differences in financial literacy may be a factor in gender differences in expectations.

Some people may have overestimated their future Social Security benefits because they retired sooner than they planned, resulting in lower-than-expected benefits. In a survey of retirees, the Employee Benefits Research Institute (2021) found that 46 percent had retired earlier than planned. Among those, 34 percent retired because of unexpected adverse events such as health problems or disability, 25 percent retired because of workplace- or employer-related changes, and 41 percent retired because they could afford to.

Nationwide Retirement Institute (2021) surveyed adults aged 25 or older not yet receiving Social Security benefits and found that 62 percent of women and 38 percent of men responded that they did not know or were not sure what their Social Security benefit amounts would be. When asked the age at which they could receive full Social Security benefits, 46 percent of women and 31 percent of men reported they did not know. When asked if they knew how to maximize their future Social Security benefit, 47 percent of women and 61 percent of men responded that they did. The question did not define “maximize,” however, so it may have been interpreted as maximizing annual benefits, which would entail postponing benefit claiming to age 70.⁶

Earnings and Their Effect on Future Benefits

Accurately estimating future Social Security benefits may be more difficult for women than for men because women are more likely to experience substantial variability in year-to-year earnings. This is partly because they are more likely than men to move from full-time to part-time work or to leave the labor force (Congressional Budget Office 2008; Mitchell and Turner 2010).

The COVID-19 pandemic has demonstrated that women tend to face more labor market risks than men (Muir and Turner 2022). For example, during 2020–2021, married women with school-aged children suffered greater income loss than their husbands because they assumed more of the additional child-care responsibilities when schools transitioned to distance learning (Calarco and others 2021).

Hegewisch (2016), writing before the pandemic, noted that women are more likely than men to give up their jobs when a family member needs serious care. Women may be more pessimistic about their future Social Security benefits, given that they are subject to greater earnings risks than men.

Research suggests that women are more likely than men to feel underprepared for retirement, which may be due in part to the gender gaps in wages and pensions, with women being disadvantaged in both. Using an online survey of 6,372 workers at for-profit companies with five or more employees, Transamerica Center for Retirement Studies (2018, 195) found that only 12 percent of women were “very confident” that they would “be able to fully retire with a lifestyle they consider comfortable,” compared with 24 percent of men.

Literature Review Part III: Expectations Focused on Social Security Solvency

Expectations about Social Security's future hinge on one's perceptions of what lies ahead for the OASI Trust Fund. Those perceptions are shaped by information that can come from a wide variety of sources and lead to a wide variety of conclusions.

Rational Expectations and Likely Scenarios

In trying to predict the timing and nature of program changes, one might assume that Congress will follow the pattern of the 1983 Social Security reforms. Those reforms were not enacted until a financing shortfall was imminent, which would have adversely affected payments to current retirees. In that respect, the 1983 reforms continued a pattern of legislative procrastination that has historically characterized attempts to deal with Social Security financing (Turner 2017). Other than procrastination, the aspects of the 1983 reforms that might be duplicated in future reforms are difficult to predict. Because one cannot foresee how future reforms would affect different age cohorts or how substantial their effect on benefits would be, concerns about future benefits are understandable. The 2022 Social Security Trustees Report (Board of Trustees 2022) projects that the OASI Trust Fund will enable SSA to pay scheduled benefits fully through 2034 and to pay 77 percent of scheduled benefits after that.⁷

Gender Differences in Solvency Expectations

Quinby and Wettstein (2021) studied how media reports on prospective Social Security financing shortfalls affect worker expectations about future benefits. Holding marital status constant, they found that men reported a significantly higher ratio of expected benefits to scheduled benefits—in other words, a lower future reduction in benefits—than women.

Beyond the question of expected benefit amounts is the level of confidence that benefits will be paid at all. In an online survey of 3,109 workers, 80 percent of women and 72 percent of men expressed concern that Social Security would “not be there for them” (Collinson, Rowey, and Cho 2021). Other evidence suggests that women are more risk averse than men, as Hinz, McCarthy, and Turner (1997) found in their study focusing on pension investments.

Women are more likely than men to worry that Social Security will “run out of money” during their lifetime—74 percent versus 65 percent (Nationwide Retirement Institute 2022). By contrast, men are more likely than women to answer factual questions about Social Security provisions correctly, suggesting that pessimism about the financing of future benefits may be related to lower levels of program knowledge. Women are also more likely than men to agree with the statement that “COVID heightens worries about Social Security funding.”

The different economic risks typically faced by men and women may explain gender differences in expectations. As women tend to rely more on Social Security than men, they are more vulnerable to risks to future Social Security benefit levels. Men are also generally more familiar with Social Security program details than women. Evidence from other research on expectations suggests that people tend to acquire information when they perceive a benefit to doing so (Roth, Settele, and Wohlfart 2022).

To summarize this literature review, a variety of factors may explain why women are more pessimistic than men about the future of Social Security. For example, men tend to be more optimistic than women in their economic outlooks (Barber and Odean 2001; Niederle and Vesterlund 2007); men have higher earnings than women and, perhaps as a result, are less risk-averse than women (Hinz, McCarthy, and Turner 1997; Cortés and others 2020); men are generally more knowledgeable about Social Security than women (Nationwide Retirement Institute 2021, 2022); and women have greater earnings volatility, which poses greater risks to their future benefits, than men (Muir and Turner 2022).

Findings from the AARP 85^th Anniversary of Social Security Survey

In this section, we explore results of a 2020 online and telephone survey of 1,441 respondents aged 18 or older commissioned by AARP to celebrate the 85^th anniversary of the Social Security Act's passage (Perron 2020).⁸ First, we present AARP survey results detailing respondents' views on Social Security. Then we describe and present the results of a multivariate analysis of confidence in Social Security that controls for key socioeconomic variables such as income, education, and age.⁹

Respondents' Views

In the AARP survey, 63 percent of men and 73 percent of women said that Social Security is one of the most important U.S. social programs and that they “would suffer if it ceased to exist” (Table 2).¹⁰ Highlighting concerns about the sustainability of Social Security and gender differences in those concerns, 70 percent of men and 81 percent of women agreed with the statement, “Social Security will not be there for you when you need it.”

Among men who indicated that they are not confident in the future of Social Security, 31 percent also reported that they do not trust government programs. The corresponding figure among women was similar, at 28 percent. Thus, as noted in the literature review, differences in trust in government do not underlie gender differences in views about the future of Social Security.

However, there are gender differences in knowledge about Social Security financing. In particular, 23 percent of men who reported that they are not confident in Social Security's future believed that the program is running out of money, as did 29 percent of women. The program is facing a shortfall, but it will be able to pay most scheduled benefits and it is not running out of money. Thus, gender differences in knowledge about Social Security financing are also associated with lack of confidence in the program's future.

Women are more likely than men to express lack of confidence in the future of Social Security. When asked their level of confidence in “the future of the Social Security program,” 53 percent of men and 61 percent of women responded they were either “not too” or “not at all” confident (Table 3). Seventeen percent of each were “not at all confident.”

Table 4 examines confidence in Social Security not only by sex but also among three subgroups—current Social Security beneficiaries, nonbeneficiaries with a spouse who is receiving benefits, and nonbeneficiaries without a spouse who is receiving benefits.

Becoming a beneficiary or being married to one increases confidence in the future of Social Security. Current beneficiaries and the nonbeneficiary spouses of current beneficiaries are much more confident in the future of Social Security than nonbeneficiaries with no spousal beneficiaries.¹¹ Hou (2022) argues that the risk of cuts in future Social Security benefits is low for current beneficiaries. Although the AARP survey results for Social Security beneficiaries generally indicate that beneficiaries support that assertion, 29.4 percent of men and 33.4 percent of women who are current beneficiaries express little or no confidence in Social Security's future. Thus, a slight gender difference in pessimism emerges among beneficiaries. Among nonbeneficiaries whose spouse receives benefits, the gender difference is sharper: 36.4 percent of men lack confidence in Social Security's future versus a majority (51.7 percent) of women.

The 2020 AARP survey includes a follow-up question that asks respondents who express a lack of confidence why they lack confidence in the future of the Social Security program (Table 5). The response “I don't trust the government to keep its promises” was selected by 31 percent of men and 28 percent of women. With the statement “I think the money is running out,” 23 percent of men and 29 percent of women agreed, aligning with findings of greater fear about program sustainability among women from other studies (for example, Williams 2015).

Among respondents who believed that “the money is running out” and were then asked to explain the implications of the statement “the Social Security trust fund will be exhausted in 15 years,” 37 percent of men and 45 percent of women indicated that it means “Social Security will not be able to pay any benefits” (not shown). Thus, the gender difference may, in part, reflect gender gaps in knowledge about the meaning of the financing shortfall.

Respondents who expressed confidence in Social Security's future were asked the main reason for their confidence. For both men and women, the most common responses were, “It has been around for many years” (34 percent) and “It has always paid its benefits” (28 percent; Table 6). “I trust the government to keep its promises” was chosen by 21 percent of men and 20 percent of women. Thus, we find virtually no gender differences in reasons for confidence in Social Security.

Reflecting widespread lack of knowledge about Social Security financing, Table 7 shows that majorities of both men and women do not know how the depletion of the Social Security trust fund reserves would affect benefits. Only 30 percent of men and 22 percent of women know that Social Security would continue to pay reduced benefits even in the absence of reforms to restore solvency. Women (45 percent) are more likely than men (37 percent) to believe incorrectly that Social Security will not be able to pay any benefits. Thus, the gender difference in negative expectations about the future of Social Security is partially due to a gender difference in knowledge about Social Security financing.

Table 8 shows respondent beliefs about the results of a projected depletion of the Social Security trust fund reserves by current beneficiary status and presence or absence of confidence in Social Security.¹² Surprisingly, 23 percent of beneficiaries with confidence in Social Security believe that the insolvency of the trust fund would mean that Social Security could not pay any benefits. That mistaken view is also held by 33 percent of nonbeneficiaries who report confidence in Social Security.

Multivariate Analysis

As noted earlier, becoming an OASI beneficiary increases confidence in Social Security. Among AARP survey respondents 64 percent of beneficiaries reported confidence in Social Security's future compared with only 35 percent of nonbeneficiaries (Table 9). Other differences between beneficiaries and nonbeneficiaries are evident, as well. For example, 29 percent of beneficiaries believed they were “very informed” about the program, compared with 15 percent of nonbeneficiaries. Furthermore, 3 percent of beneficiaries were “very concerned” that Social Security would not “be there when they need it,” compared with 12 percent of nonbeneficiaries.

In view of significant differences between the OASI beneficiary and nonbeneficiary populations, we focus our multivariate analysis on nonbeneficiaries who do not have a spousal beneficiary, a sample of 1,028 persons. We find a gender gap in reported confidence among these nonbeneficiaries of 7 percentage points, with 37 percent of men and 30 percent of women being confident in Social Security's future (Table 4).

Table 10 reports descriptive statistics for our regression sample of nonbeneficiaries. Women, on average, are more educated and earn less than men: 40 percent of women and 34 percent of men have a college degree, and 56 percent of women and 66 percent of men earn at least $50,000 annually.

We conducted logistic regression analysis to assess whether women are more pessimistic about Social Security's future because of innate or dispositional differences or because of differences in socioeconomic status. Our findings are reported as odds ratios, defined as the probability of having greater confidence in Social Security divided by the probability of lacking confidence in the program. A dependent variable equal to 1 indicates confidence; a dependent variable equal to 0 indicates no confidence. Estimated odds ratios above 1 indicate above-average confidence (a positive effect), while odds ratios below 1 indicate below-average confidence (a negative effect).

We use a range of socioeconomic and demographic independent variables in the analysis, including self-reported knowledge of Social Security, education, income, age, race, and marital status. Table 11 reports the results for four models, each with a different mix of variables. Holding constant the effect of these other variables, we find a significant negative effect on confidence among women with an annual income of $50,000 or higher, but the gender effect by itself is not statistically significant. In other words, for women with an annual income of less than $50,000, the gender gap in confidence in the future of Social Security is explained by the economic and demographic variables included in the regressions.

Thus, the gender difference in pessimistic Social Security expectations found in the survey results is largely not innate or dispositional, as it is not found for most women but is the result of the differences in socioeconomic variables. We tested the robustness of this result by omitting self-stated knowledge about Social Security, and the result was unchanged. The result also did not differ from those of other variations in the regressions reported in Table 11 (namely, omitting marital status, race, and age squared).

Odds ratios for the person's age, program knowledge, education, and income variables are statistically significant in explaining differences in the level of confidence in the future of Social Security. With age having a significantly negative effect and age squared having a significantly positive effect, the effect of age plus age squared is U-shaped, first having a negative effect on confidence in Social Security at younger ages, then reversing and having a positive effect at older ages. Respondents who reported they were knowledgeable about Social Security had greater confidence in the future of the program, with women rating themselves as less knowledgeable than men did. College-educated persons had less confidence, with women being more likely to have a college education than men. As noted above, having higher income (greater than $50,000) among women had a negative effect on confidence, but without the gender interaction variable, the effect of higher income by itself is insignificant. Odds ratios for the race and marital-status variables were not statistically significant. Thus, our results suggest that gender differences in pessimism about Social Security's future can be explained mostly by socioeconomic variables rather than innate or dispositional factors.

Conclusions

Social Security expectations differ substantially between men and women. This article examines the extent, nature, and causes of these gender differences in expectations. It focuses on why women tend to be more pessimistic than men about the future of Social Security.

Several explanations for the gender differences are suggested in the literature and by the descriptive statistics in our data analysis. Men are generally more economically optimistic than women. Women also tend to be less informed on Social Security program provisions. It is impossible to know how Social Security reforms will affect future benefits, but these explanations could all play a role in women being more pessimistic than men for the future of Social Security.

We reviewed survey responses on two aspects of Social Security benefits. First, in our literature review, we looked at respondent expectations about their future benefits. Second, we looked at their expectations about the future of the Social Security program overall. Expectations about one's benefits and the program's future differed. Benefit expectations are influenced by uncertainty about one's future earnings and by one's level of knowledge of the Social Security provisions. For the program overall, expectations are presumably related to beliefs about future Social Security reform legislation.

Predicting one's Social Security benefits is likely more difficult for women than for men. The variability of future income may affect the value of benefits for women who will claim benefits based on their earnings records. In addition, the variability of future spousal earnings may affect the value of future benefits for women whose claim will be based on their spouse's earnings record. In predicting what lies ahead for the program, different expectations about Social Security reform may reflect attitudes towards risk explored in the behavioral and economic studies we reviewed, with women tending to be more risk averse than men.

Gender differences in Social Security expectations have been observed for years. Our literature review and empirical analysis suggest multiple behavioral and economic explanations for the well-established pattern of women having more pessimistic expectations about Social Security than men. Possible reasons include:

• Women have lower personal incomes than men, and lower-income individuals tend to have more pessimistic expectations.
• Women tend to be more risk averse than men.
• Women tend to be less knowledgeable about Social Security than men.
• Women are more likely to recognize Social Security's projected financing problems.
• Women may have more difficulty in accurately estimating future Social Security benefits because they are likely to have more variability in their future earnings.

The gender differences in expectations that have been found in numerous studies, as well as in our regression analysis of nonbeneficiaries, are considerably diminished when we restrict the study population to Social Security beneficiaries (Table 4). Becoming an OASI beneficiary reduces the uncertainty about the future that some people felt earlier in life. Furthermore, we note the difference between dispositional optimism, which is a psychological concept, and our measure of optimism, which is affected by knowledge and experience, and, hence, is more of an economic concept.

Our findings support the conclusion that gender differences in pessimism about Social Security's future are not innate or dispositional, and do not arise from gender differences in trust in government, but can be largely explained by socioeconomic differences.