Workers’ Expectations About Their Future Social Security Benefits: How Realistic Are They?

Introduction

This study examines workers' expectations about their future Social Security retirement benefits. We compile and analyze data from surveys and associated economic studies that cover 50 years, from 1971 to 2020. The surveys show that some workers substantially underestimate their future Social Security benefits relative to projections from the Social Security actuaries. This finding suggests that efforts to inform workers about the value of their future Social Security benefits need to improve. This article catalogs many of the various past and present surveys that have asked Americans about their expectations of the future of the Social Security program and their own benefits, and reports the various surveys' findings. The article provides insight into ways the Social Security Administration (SSA) can improve its communication and outreach, particularly regarding the future of Social Security and the way benefits are calculated.

We first discuss different theories on the formation of expectations, which may explain some of the errors workers make in assessing their prospects for Social Security benefits. We then discuss the survey results on expectations about future benefits. We compare the results across several dimensions—different survey questions, different years, and different demographic and socioeconomic groups.

Theories About Expectation Formation: What Do We Know About Workers' Expectations?

Das, Kuhnen, and Nagel (2017) identified socioeconomic status (SES) categories based on income and educational attainment and found that people with higher SES are more optimistic about future macroeconomic developments, including business conditions, the national unemployment rate, and stock market returns. A comparison of the study participants' predictions with those of professional forecasters and with historical data reveals that the difference by SES in expectations reflects excessive pessimism among low-SES individuals. This finding suggests that forecasts made by people with high SES may be closer to objective forecasts than those of people with low SES.

Norr (2017) attributed pessimistic views on the future of Social Security to negativity bias—the tendency to exaggerate negative information, such as reports of the need to reform Social Security to ensure program solvency. Bordalo, Gennaioli, and Schleifer (2018) presented a “diagnostic expectations” model that accounts for a behavioral tendency to overweight a narrow range of possible future events in light of incoming information. One example of this tendency is an overreaction to news; in response to negative news, people may develop overly pessimistic expectations.

Lack of information about Social Security benefits may explain why some workers' expectations about their future benefits differ substantially from estimates based on the projections of the Social Security actuaries. Surveys have found low levels of Social Security program literacy among U.S. workers (Mitchell 1988; Lusardi and Mitchell 2007; Yoong, Rabinovich, and Wah 2015). Although many workers have some basic knowledge about program rules, they often lack the knowledge of program details that would allow them to make informed retirement-saving and other financial decisions (Smith and Couch 2014). Of key relevance to our study, Greenwald and others (2010) found that only 32 percent of survey respondents aged 25–65 felt that they were very knowledgeable about how much they will receive in future Social Security benefits. Only 22 percent of the youngest workers (those aged 25–34) felt very knowledgeable, but the percentages rose with age. This finding raises the question of whether low-knowledge workers will generally underestimate or overestimate their future benefits, or whether the error in their estimates will be random. Another factor that may affect workers' uncertainty about their future Social Security benefits may be how much labor market uncertainty they face (Mitchell and Turner 2010).

Surveys on Social Security Expectations

We examine data on expectations from 18 singular or recurring surveys, of which some of the latter have been conducted for many years. In all, we reviewed more than 60 individual surveys with more than 130,000 respondents. Appendix A profiles the surveys and provides some key details about them. It identifies the sponsor or name of each survey; whether the survey was conducted by phone, via the Internet, or in person; the periodicity and date or dates of the survey; and the size and type of the group or groups sampled for the survey. The appendix profiles provide a high-level overview rather than a precise, detailed delineation of all the surveys reviewed.

Surveys such as those conducted by the Employee Benefit Research Institute (EBRI), Gallup, the Transamerica Institute, the Nationwide Retirement Institute, and Aegon recur regularly. The questionnaires for recurring surveys often differ in content across the years; in some cases, the differences are substantial enough that they might be considered distinct surveys. Other surveys have occurred irregularly or were conducted one time only. Our analysis, though extensive, does not encompass all possible sources of survey information on Social Security benefit expectations.¹

Findings

In this section, we first examine expected likelihoods of receiving Social Security benefits. Then, we examine expectations about benefit levels. Finally, we examine uncertainty and differences in expectations among subgroups.²

Workers' Expectations About Future Receipt of Social Security Benefits

Although nearly all workers will eventually receive Social Security benefits, many survey respondents expect not to receive them. Whitman, Reznik, and Shoffner (2011) estimated that 86 percent of individuals aged 62–84 were current Social Security beneficiaries in 2010, and projected that another 10 percent were to be future beneficiaries, leaving only 4 percent who would not receive benefits. Many of the workers who never qualify to receive Social Security benefits have an insufficient work history in covered employment; these workers are likely to have low earnings or to be immigrants, and in particular to have immigrated at older ages. In addition, some workers will not qualify for benefits because they work for a state or local government that provides substantial pension coverage, which exempts them from Social Security coverage. Roughly one-quarter of state and local government employees, or 6.5 million workers, are not covered by Social Security (Quinby, Aubry, and Munnell 2020).

EBRI has conducted its Retirement Confidence Survey (RCS) since 1991. In the 1996 RCS, 23 percent of respondents expected that Social Security benefits would “not [be] a source of income in retirement” (EBRI 2016). Respondents were not asked why they thought they would not receive future benefits. However, in 1996, the trustees of the Social Security trust funds projected that the Old-Age and Survivors Insurance program would become insolvent 35 years later, in 2031 (Board of Trustees 1996).^3,4 RCS respondents expressed similarly high levels of pessimism in each year from 2010 through 2013, when 21 percent did not expect to receive future Social Security benefits (EBRI 2016).

Concerning results for survey questions that elicit a yes or no response about future expectations, Dominitz and Manski (2006) suggested that respondents answer “no” when they view the probability of the queried outcome to be less than 50 percent. Thus, in the case of the 1996 RCS, Dominitz and Manski would argue that 23 percent should be interpreted as the share of respondents who viewed the probability of receiving future benefits as less than 50 percent, not as the share of respondents who viewed the probability as necessarily zero. Nevertheless, 23 percent seems to represent a high degree of pessimism, given the reality that all but about 4 percent of U.S. workers eventually receive Social Security benefits and that even potential insolvency would not eliminate benefits.

Armour (2017), using various 2010–2017 iterations of the RAND Corporation's American Life Panel, found that only about half of the respondents who were not receiving Social Security benefits at the time of the survey expected to receive them in the future. The author examined whether SSA's use of Social Security Statements to inform workers of program provisions and their earnings histories, and to estimate their future benefits, affected their expectations. The Statement is available online to individuals who sign up for a my Social Security account. For workers aged 60 or older who are not yet receiving benefits and who do not have a my Social Security account, SSA automatically mails a Statement each year until the benefit is claimed. From fiscal year 2000 through February 2011, SSA had mailed the Statement annually to all nonbeneficiary workers aged 25 or older. In a regression analysis, Armour found that having received a Social Security Statement raised the percentage of respondents expecting to receive a future Social Security benefit from about 50 percent to 62 percent—meaning that 38 percent of Statement recipients still expected not to receive benefits.

In a later study, Armour (2020) used a specialized American Life Panel survey to examine the effect of receiving a Social Security Statement during the years 2014–2016 on workers' expectations. He limited his sample to people who had worked at least 10 years (and earned 40 credits) in covered employment, enough to qualify for Social Security benefits. Because SSA had mailed Statements only to workers reaching a multiple-of-5 age during the study period, the sample allowed comparisons between Statement recipients and nonrecipients. Armour found that Statement recipients were likelier than nonrecipients to expect to receive future benefits, but they were also more likely to expect Congress to reduce future benefits. Among respondents who in 2013 had not expected to receive Social Security benefits, receiving a Statement increased the likelihood of expecting benefits by nearly 49 percentage points as of 2017. The effects on expectations were larger for workers who had lower knowledge about Social Security before receiving the Statement than for those with greater prior knowledge. The likelihood of expecting to receive Social Security benefits increased from 64 percent for those who had not received a Statement since 2011 to 71 percent among those who received one in 2014 or 2015 and to 78 percent for those who most recently received one in 2016. However, all of the effects were short-lived. For respondents who had last received the Statement 2 or more years before the survey, none of the effects were significant. Among Statement nonrecipients who reported that they did not expect to receive Social Security benefits, 53 percent cited as their reason that they did not expect Social Security to “be around long enough.” Future research might investigate why that belief is prevalent. The second most common reason was the belief that the respondent would not accrue enough work credits to qualify for benefits, even though the sample consisted entirely of workers with sufficient work histories to qualify at the time of the survey.

These results provide insights into how SSA communications might better inform the public and minimize the erroneous expectation among many workers that they will not receive Social Security benefits. However, the Statement's effect on younger workers' expectations may be limited, because only 44 percent of respondents younger than 40 who were mailed a Statement recalled having received it (Armour 2020). Smith (2020) found that receipt of two or more Statements had a stronger effect on workers' benefit-claiming decisions (specifically, the decision to defer claiming) than receipt of one Statement. Her results show that an informational intervention (such as Statement mailings) can be an effective way to improve people's expectations and that a comprehensive financial literacy program is not required to achieve that goal.

Expectations About Level of Benefits

Bernheim (1987) used SSA's longitudinal Retirement History Survey, with respondents initially surveyed at ages 58 to 63 in 1969, to study the difference between the Social Security benefit amounts the respondents expected as workers and the amounts they later received as retirees. Because the survey was longitudinal, Bernheim was able to compare responses to the same question across various survey intervals. He found that the expected future benefits reported in the 1971 survey, when the respondents were aged 60 to 65, were about 10 percent lower than the amounts they later received, on average. He concluded that people near retirement tended to underestimate their future benefits. Widows and single women tended to underestimate by the largest amount—18.9 percent and 16.0 percent, respectively. However, not all workers underestimated their future Social Security benefits. One in six overestimated their future benefits by at least 25 percent, while one in twelve overestimated their future benefits by at least 50 percent. Men were more likely to overestimate their future benefits than women. For example, among singles, one-fifth of men overestimated their future benefits, compared with one-tenth of women.

The next set of surveys we analyze asked respondents how confident they were about the future of Social Security. In 1977, Congress established a National Commission on Social Security to examine potential fundamental changes to the entire system. A survey of workers conducted in 1979 for the Commission's report found that only 32 percent of respondents were confident that Social Security would have sufficient funds to provide full benefits for them (National Commission on Social Security 1981, Appendix A). In this survey, the respondents who expressed confidence presumably considered the likelihood of future benefit receipt to be greater than 50 percent, but they were not asked to report their degree of confidence between 50 percent and 100 percent.

Researchers have used a variety of approaches to measure workers' uncertainty about future Social Security benefits. Studies have measured the subjective probability that workers place on the chances that they will receive any benefits or that they will receive the full amount of promised benefits. Delavande and Rohwedder (2011) used a 2007 Internet supplement to the University of Michigan's 2006 Health and Retirement Study. They found that 25 percent of respondents aged 52 or older assign a 62 percent or lower probability of receiving any Social Security benefits. This result is congruent with the previously mentioned 2010–2013 RCS results (EBRI 2016), which can be interpreted as indicating that 21 percent of respondents assigned a 50 percent or lower probability to the eventual receipt of Social Security benefits.

Yoong, Rabinovich, and Wah (2015), using the University of Southern California's Understanding America Study (UAS), found that only 4 percent of nonretirees are “very confident” that Social Security retirement benefits will “be there for [them] when [they] retire,” while 41 percent are “not at all” confident. The wording “be there” is ambiguous, with different respondents probably assigning different meanings, making it difficult to interpret that particular result. For example, some workers may be answering not that they expect to receive no benefits but rather that they expect to receive benefits that will be inadequate for their needs.

Aegon Retirement Readiness Surveys found that 32 percent of respondents in 2015 were concerned that their Social Security benefits would be less than they expected, a proportion that rose to 37 percent in 2017 (Aegon 2015, 2017). In addition, according to AARP (2015), 19 percent of survey respondents incorrectly think that the potential depletion of the Social Security trust funds' reserves means that the system would not be able to pay any benefits.

As noted earlier, interpreting the results of many of the earlier surveys reviewed here is complicated by the uncertainty over the probability that workers assign to receiving no benefits from Social Security. The University of Wisconsin's Survey of Economic Expectations (SEE) was a 1999–2002 telephone poll that included several unambiguously worded questions about Social Security expectations (Dominitz and Manski 2006). The first such question was “Think ahead to when you are about to turn 70 years old and suppose that you are not working at that time. What is the per cent chance that you will be eligible to collect any Social Security retirement benefits at that time?” Respondents who reported a positive probability of eligibility were then asked a series of questions designed to elicit the range within which they expected their benefit amount to be (conditional on eligibility). However, the potential specificity of these responses was hindered by relatively lower response rates to those questions. The response rate for the full set of Social Security questions was only 66 percent, compared with an 80 percent response rate for questions on expectations of personal income 1 year ahead. Further, about 80 percent of the nonresponses for the Social Security questions occurred when people did not report the lowest and highest possible benefits they thought they might receive.

Among all respondents, 9.4 percent reported expecting zero chance of receiving Social Security benefits. This share of respondents is much lower than those found in surveys on Social Security benefit expectations wherein the percentage likelihood of receiving no Social Security benefits is not specified.

Differences in Benefit Expectations Across Demographic Categories and Over Time

Brown and others (2017) conducted a UAS survey module that innovatively addressed people's expectations about future Social Security benefits. The survey posed a hypothetical scenario in which potential beneficiaries could pay or receive a one-time lump sum in exchange for a lifetime $100 increase or decrease in monthly Social Security benefits, then asked respondents to advise a hypothetical 60-year-old about the lump sum amounts to demand for such trade-offs. The advice the survey respondents provided presumably reflected their expectations about future Social Security benefits being paid as promised, expectations shaped in part by news and rhetoric they had heard about future Social Security financing. If people believed that Social Security would cease providing benefits in the future, they presumably would pay nothing for increased benefits and would be willing to “sell” a decrease in monthly benefits for a relatively low lump sum. The median amount people were hypothetically willing to pay for an extra $100 a month in benefits for the remainder of their life was $4,750, while the median amount they were willing to accept to take a $100 a month cut in benefits was $16,250. Thus, people would require substantially more money to accept a cut in benefits than to purchase a benefit increase of the same amount.

About 12 percent of the UAS module respondents were not willing to pay anything for the extra benefits, while another 10 percent were willing to pay very little. About 10 percent were willing to purchase the increased annuity stream for about $1,200, an amount they would get back in the first year. According to the authors, these results indicate that respondents would not be willing to purchase an increase in their Social Security benefits equivalent to purchasing a price-indexed annuity on an extremely favorable basis. Thus, the results are consistent with a small percentage of the population having very negative views about the future of Social Security.

Because studies have found heterogeneity in worker expectations about other future circumstances, it is not surprising that heterogeneity is also found for expectations about Social Security benefits. The next subsections investigate survey data on expectations disaggregated by selected demographic and economic group.

Different Expectations by Age

The 1999–2002 SEE asked its respondents (aged 18–69) questions designed to gauge their combined expectations of personal eligibility for Social Security benefits and the program's ability to provide benefits to eligible participants. For the entire sample, the median reported probability of benefit receipt was 60 percent, meaning that 50 percent of the sample thought that their chance of receiving benefits was less than 60 percent. The study found that for people aged 30, the median probability of receiving benefits at age 70 was 40 percent, and that the median probability rose as the respondents' age increased, to 100 percent at age 65 (Table 1) (Dominitz and Manski 2006).

In addition, the percentage of people who reported a zero percent chance that they would receive Social Security benefits at age 70 rose from 10 percent for respondents aged 20 to 17 percent among those aged 30, then decreased to 2 percent for those aged 65. Although most surveys do not ask their participants why they chose their particular responses, the SEE had an open-ended question asking why people thought they would not receive Social Security benefits. About two-thirds of those who reported they had a zero percent chance of receiving future Social Security benefits believed that the program would either cease to exist or no longer provide benefits (not shown). Much smaller shares of the respondents reporting zero chance for benefits believed that they would not live long enough to receive them (1 percent) or that prospective rule changes such as raising the retirement age or means-testing benefits would prohibit them from receiving benefits (4 percent). In addition, some believed that they would not receive benefits because they would not have worked the required 40 credits (or at least 10 years) in covered employment (12 percent).

Dominitz and Manski found that at the individual level, subjective uncertainty about the range of likely benefits is large among young people but it decreases with age. The authors measured the subjective uncertainty as the difference between the minimum level of benefits that the respondents expected to receive with 25 percent likelihood and the minimum level of benefits that they expected to receive with 75 percent likelihood. Dominitz and Manski argued that decreasing uncertainty with age makes sense because uncertainty about both the person's earnings history and the future structure of Social Security decrease, at least for those nearing retirement. They noted, however, that even for middle-aged persons, uncertainty remains fairly high. They found that the cross-sectional median of the interquartile range of expectations (25^th percentile versus 75^th percentile of expected annual benefits) for respondents aged 55 was $6,100.

By separately measuring the probability of eligibility for benefits and expectations of benefit levels conditional on receipt, Dominitz and Manski concluded that variations by age in expected benefits stem mainly from different expectations about the survival of the Social Security system, rather than different expectations about its generosity should it continue to exist. The authors did not investigate why young people are more likely to think that Social Security will cease to exist, but that may in part reflect the greater uncertainty that naturally accompanies a longer projection period.

Luttmer and Samwick (2015) designed and fielded a module of the Ipsos KnowledgePanel Internet survey and found that expected benefits as a percentage of scheduled benefits rise with age, which fits the perception that people near retirement will both have realistic expectations and be less likely to face future benefit reductions. The authors reported that 91 percent of respondents were aware of projected future financial shortfalls for Social Security. They found that, on average, individuals aged 25–59 expected to receive only 60 percent of the Social Security benefits that they were scheduled to receive. On average, individuals aged 25–29, 30–34, and 35–39 expected to receive roughly 50 percent of promised benefits, with individuals in older age groups expecting higher percentages.

In a Gallup news release, McCarthy (2018) reported the response to this survey question: “Now I am going to read a list of problems facing the country. For each one, please tell me if you personally worry about this problem a great deal, a fair amount, only a little bit, or not at all. Do you worry about the Social Security system?” Since 2005, the percentage of working-age respondents who said they worry “a great deal” about the future of Social Security has increased with age. This result—increasing worry with age—is not necessarily inconsistent with the previously discussed finding that the percentage of promised benefits one expects to receive also increases with age. The Gallup survey may indicate that people think more seriously about retirement as they get older.

The 2010 EBRI RCS found that only 67 percent of workers aged 25–34 expected to receive Social Security benefits, compared with 92 percent of workers aged 55 or older. In the 2012 RCS, only 65 percent of workers aged 25–34 expected to receive future Social Security benefits (EBRI 2010, 2012).

The American Academy of Actuaries provided us with unpublished survey data for 2016 indicating that the percentage of Americans expecting to receive Social Security benefits increases with age, from 48.7 percent for those aged 18–34 to 69.7 percent for those aged 35–54 and to 90.9 percent for those aged 55–64. Thus, from this survey, roughly 50 percent of workers younger than 35 had pessimistic benefit expectations. However, this survey aligns with other studies in finding that worker age is inversely associated with pessimism about the receipt of future Social Security benefits.

The finding that 69.7 percent of workers aged 35–54 expect to receive any Social Security benefits is of particular interest. Of these workers, who are old enough to be saving and planning for their retirement, 30 percent did not expect to receive benefits. Although some retirees do not receive benefits for various reasons, those individuals account for about 4 percent of the retirement-age population. Thus, roughly 25 percent of respondents aged 35–54 erroneously expect not to receive any benefits from Social Security.

An online survey conducted in 2019 as part of Morning Consult's Longevity Project (Konish 2020) asked participants which financial resources they were counting on in retirement. Among baby boomers (born 1946–1964), 83 percent said they were counting on Social Security. That percentage declined for each successively younger generation—64 percent for generation X (born 1965–1980), 42 percent for millennials (born 1981–1994), and 38 percent for generation Z (born 1995–2015).

Differences by Race and Ethnicity

Cohen, Luttig, and Rogowski (2017), using the GenForward Survey, explored the views of millennials (persons aged 18–34 in 2017) by race and ethnicity and found that most were not confident in the future of Social Security. White respondents were the most likely to lack confidence (77 percent), and 37 percent did not expect to rely on Social Security at all. Latino respondents were the least pessimistic, yet 66 percent of them lacked confidence, and 32 percent expected not to receive any benefits. Black respondents were the least likely to expect not to receive benefits (29 percent). Among both Asian and Black respondents, 73 percent lacked confidence; and like the Latino participants, 32 percent of Asian respondents did not expect to receive benefits.

Differences by Income

The 2016 American Academy of Actuaries survey asked respondents with different earnings levels whether they expected to receive income from selected sources, including Social Security, in retirement. The percentage of people expecting to receive Social Security benefits increased with earnings: 55.8 percent of those with earnings under $40,000, 74.5 percent of those with earnings from $40,000 to $99,999, and 71.2 percent of those with earnings of $100,000 or more. This pattern is roughly consistent with higher-income people having greater financial literacy (Lusardi and Mitchell 2014) and thus being more likely to understand that they will receive Social Security benefits. It is also consistent with the finding, cited earlier, that people with higher income tend to report being more optimistic about the economy in the future. However, it may also reflect the fact that income tends to rise with age, and the percentage of people expecting to receive Social Security benefits also rises with age.

Since 2005, Gallup has tracked respondents' views on the future of Social Security. Consistently, lower-income Americans have been more likely to register a “great deal” of worry about the future of Social Security than those in other income groups, while the highest income group has been least likely to express a great deal of worry (McCarthy 2018). A possible explanation is that lower-income Americans depend the most on the program as a source of retirement income. However, because they tend to have relatively low financial literacy (Lusardi and Mitchell 2014), it could also be that they are the least knowledgeable about the future of Social Security.

Differences by Education and Cognitive Ability

Perez-Arce, Rabinovich, and Yoong (2019) conducted a randomized survey as part of the UAS in which they presented respondents with alternative Social Security reforms: raising the payroll tax rate, raising the payroll tax rate ceiling, or reducing benefits. The authors found that people with higher educational attainment, financial literacy, and cognitive ability were more likely to adjust their expectations about future Social Security benefits rationally and consistently in response to changes caused by the potential alternative reforms.

Differences Over Time

Changes over time in Americans' expectations about future Social Security benefits is a topic of particular interest because differing expectations may provide evidence of the effects of changes in the Social Security actuaries' predictions of future Social Security financing. Beginning in 1968, the American Council of Life Insurers (then known as the American Council of Life Insurance [ACLI]) surveyed Americans' confidence in Social Security's future in its annual Monitoring Attitudes of the Public survey; our review focuses on the 1975–1988 iterations. The 1975 survey found that 63 percent of Americans were very or somewhat confident about the future of Social Security, with only 10 percent reporting that they were not at all confident (Reno and Friedland 1997). This survey established a baseline of high positive expectations among respondents. Three years later, in 1978, only 39 percent reported confidence, while 21 percent felt “not at all confident.” Notably, the 1978 survey followed media reporting of negative information about the future of Social Security that had emerged from public policy discussions related to reform legislation passed in 1977. A brief and modest rebound in confidence followed; but it fell again around the time of the 1983 Social Security reforms, to percentages in the low 30s in 1982–1984. Following the 1983 reforms, confidence rose until 1990, but then declined once again until 1994, perhaps because of further negative portrayals of the Social Security system by interest groups, which were later amplified in the media (Myers 1997). ⁵

In their analysis of the ACLI data, Reno and Friedland (1997) found that people who have low confidence in Social Security nonetheless indicate that they expect to receive benefits, although they tend to underestimate their likely future benefit amounts. Using earlier ACLI surveys, Sherman (1989) presented data for selected years 1975–1988 on workers' level of confidence in the future of Social Security (Table 2). She found that a majority of workers did not express that they were “very or somewhat” confident in the future of Social Security after 1976.

Since 2001, Gallup has asked workers “When you retire, will Social Security be a major source of income, a minor source of income, or not a source at all?” (Gallup 2020). Since 2016, the share of respondents expecting Social Security not to be an income source has trended downward (Table 3). Broadly consistent with Gallup, EBRI has found a downward trend since 2013 in the percentage of people expecting Social Security not to be a retirement income source. The trend may be due in part to the effect of SSA Statement mailings in 2014–2016 (Smith 2020). As noted earlier, from 2010 through 2013, the EBRI RCS found that 21 percent of workers expected not to receive Social Security benefits in retirement. The shares of respondents expecting no benefits declined to 20 percent in 2014, 19 percent in 2015, and 15 percent in 2016 (EBRI 2016). Those figures dipped further in 2017 and 2018, to 12 percent and 13 percent of workers, respectively (EBRI 2017, 2018). The decline from 2016 to 2017 may be due in part to the switch from a phone survey to an online survey, given that some of the characteristics of phone respondents and online respondents may differ. Nonetheless, there is clear evidence of a decline in the percentage of the population expecting Social Security benefits not to be a retirement income source.

Not all the data are consistent across surveys. The 2016 American Academy of Actuaries survey found that 34 percent of Americans did not expect to receive any Social Security benefits, in contrast with 15 percent of workers in the EBRI RCS for that year (and 12 percent of those in the 2017 RCS). Some of the discrepancy may be the result of different samples: The RCS interviews persons aged 25 or older, while the Actuaries survey covers persons aged 18 or older. In the 2018 RCS, 20 percent of workers younger than 45 did not expect to receive Social Security benefits, so differences in the age distribution of the two samples could be a factor; but they are not large enough to fully explain the difference.

Differences by Sex

The longitudinal Transamerica Retirement Survey annually asked workers to identify their greatest fears about their eventual retirement, such as the fear that Social Security benefits would be reduced or cease to exist in the future. The 2015 survey restricted the sample to workers aged 50 or older but subsequent surveys queried workers aged 18 or older. The older workers surveyed in 2015 were less likely to express that fear (33 percent) than younger workers surveyed later (Table 4). In 2017, 48 percent of workers in for-profit companies expressed that fear, but the percentage fell to 39 percent in 2019. In 2007, the same question had been asked of a sample of single women, who constitute about half the adult female U.S. population, and 45 percent expressed concern about Social Security (not shown; Transamerica 2008). A similar question in the 2012 survey asked respondents to indicate a single greatest fear, rather than being able to indicate several fears about retirement. In that survey, 18 percent of women and 12 percent of men listed Social Security ceasing to exist or reducing benefits as their single greatest fear (Transamerica 2013).

Responding to another annual Transamerica survey question from 2014 through 2018, roughly three-quarters of workers expressed concern that Social Security would “not be there” for them (Table 5). Of particular interest, women were more pessimistic than men, with roughly 80 percent of women and 72 percent of men expressing that concern each year. Other studies have found evidence that women are more financially risk-averse than men are. For example, Hinz, McCarthy, and Turner (1997) presented evidence that women are more conservative in their pension investments.

Differences by Worker/Retiree Status

Although most of the surveys we reviewed were administered to workers, the two surveys we discuss here asked current retirees to compare their actual Social Security benefit levels with the levels they had expected before retirement.

Using a UAS survey module administered to retirees, Prados and Kapteyn (2019) found that 21.4 percent of respondents reported that the Social Security benefits they received differed substantially from the amount they had expected. Most retirees reported receiving less than they had expected. Respondents with lower educational attainment were more likely to have expected their benefits to be substantially higher than the amount they received.

A Nationwide Retirement Institute survey conducted in 2019 asked recent retirees, “Is your Social Security benefit what you expected?” Eight percent of respondents were receiving higher Social Security benefits than they had expected, while 22 percent were receiving lower Social Security benefits than they had expected (Harris Poll 2019).

These results indicate that workers are more likely to overestimate their future benefits than to underestimate them. One possible explanation is that some respondents received lower-than-expected benefits because they retired earlier than they had expected.

Conclusions

This article examines data from 18 different surveys—some of them recurring—that explore workers' expectations about their future Social Security benefits. The surveys span 50 years, from 1971 to 2020. We review results from more than 60 distinct survey instruments with a total sample size of more than 130,000.

We find considerable heterogeneity in workers' expectations of their future Social Security retirement benefits by sociodemographic characteristics. We also find changes in expectations over time. Many workers are pessimistic about their future Social Security benefits, while some are overly optimistic. Evidence suggests that people's expectations become more accurate and more optimistic as they age.

This article contributes six findings. First, workers' expectations about Social Security have changed considerably over time. Second, a substantial minority of workers have expressed pessimistic views on the future of Social Security. Third, the extent of pessimism about the future of Social Security varies among racial and ethnic groups. Fourth, pessimism about receiving Social Security benefits tends to decrease with age. Fifth, women tend to be more pessimistic about the future of Social Security than men. Sixth, a small minority of workers are overly optimistic about their future Social Security benefits. The article also investigates theories of why some workers have negative expectations about Social Security benefits. Our findings support the use of informational intervention to provide workers with more realistic expectations about their future Social Security benefits. We elaborate on each of these findings, on our investigation on negative expectations, and on the role of informational interventions below.

1. Workers' expectations about Social Security have changed over time. A 1975 survey conducted by the ACLI found that 63 percent of Americans were very confident or somewhat confident about the future of Social Security, with only 10 percent reporting that they were not at all confident. That survey provides an important baseline, marking a high point in optimism about Social Security in our study period. In the data we examine, a majority of Americans have not expressed confidence in the future of Social Security since 1976.
2. A substantial minority of workers are pessimistic about their future Social Security benefits. Some workers believe that their future benefits will be substantially lower than the projections of the Social Security actuaries (or even that the program will no longer exist). Recent research by Smith (2020) suggests that effective informational interventions by SSA could provide workers with more realistic expectations about future Social Security benefits.
3. Some groups of workers are more likely than others to have pessimistic expectations about Social Security. For example, a 2017 survey found that 37 percent of White workers aged 18–34 expected to receive nothing from Social Security, compared with 29 percent of Black workers in that age group. A better understanding of which racial/ethnic groups have pessimistic views about future Social Security benefits could inform targeting strategies for informational interventions.
4. Pessimism about expected future Social Security benefits decreases with age. This does not appear to be a function of age but instead, a logical result of being closer in time to benefit eligibility, and therefore having less predictive uncertainty. Thus, workers with overly pessimistic views at younger ages tend to have more realistic expectations as they approach retirement age.
5. Women tend to be more pessimistic about the level of their future Social Security benefits than men. Some degree of pessimism may be viewed as conservative planning. People who underestimate their future benefits and react by increasing their retirement savings will be in relatively better financial condition in retirement. This gender difference might be a fruitful topic for further research.
6. Although many workers are pessimistic about future Social Security benefits, some workers, particularly those with lower income or education, overestimate their future benefits as they near retirement. As noted in item 4, workers' expectations change as they age. Thus, some workers who are pessimistic at younger ages make the opposite error as they approach retirement. Surveys of workers' Social Security expectations have generally overlooked the possibility of workers expecting higher benefits than they will receive.
7. We investigate factors that might account for pessimistic expectations. In addition to the effects of age, financial literacy, and income on worker expectations, psychologists have theorized that pessimistic overreactions are a common response to negative news, such as Social Security's well-publicized future financing problems if Congress does not enact reforms. Despite the availability of accurate information about Social Security's future finances, some people believe misleading statements suggesting that the system is “going bankrupt” and assume that, in the event of trust fund insolvency, no benefits will be paid at all. Further, a general pessimism has been observed among lower socioeconomic status groups (Das, Kuhnen, and Nagel 2017). These findings suggest a role for informational interventions, such as providing Social Security Statements and other communications more frequently.⁶
8. Informational interventions improve workers' understanding of their future Social Security benefits and can enable them to form more realistic expectations. In light of the findings and investigation listed above, the content and effective targeting of information from SSA could reduce worker apprehension and misunderstanding about their future benefits. Informing people that Social Security will continue paying benefits even if trust fund reserves are depleted would likely alleviate the unrealistic expectations of some workers. Although a broad program to raise financial literacy could address the lack of knowledge about future Social Security benefit levels among some groups of workers, we argue that a more direct, targeted approach may be more effective in promoting realistic expectations. For example, SSA communications could be targeted to younger workers and those with relatively low earnings when they qualify for future benefits, to apprise them that they have attained that significant milestone.

Appendix A: Survey Profiles

The entries below highlight the key characteristics of the 18 surveys we reviewed for this analysis, including name, sponsor (if different from the name), and mode. We provide the dates of the survey iterations we reviewed for this article, which do not necessarily encompass a given survey's full history. We identify the type and size of the sampled populations. We also provide a link to the survey's home page or other source of information, if one is available.

The surveys listed below share many sampling and methodological characteristics. They generally seek nationally representative samples of respondents and the results are weighted by common demographic characteristics such as age, sex, race, ethnicity, education, region, and income (among others) to reflect the distribution of the general population, based on sources such as the Census Bureau's Current Population Survey or American Community Survey. Significant additions or exceptions are briefly noted.

AARP Retirement Survey

Aegon Retirement Readiness Survey

American Academy of Actuaries

American Life Panel (ALP)

Gallup

GenForward Survey

Greenwald & Associates

Health and Retirement Study (HRS)

Longevity Project

Luttmer and Samwick Module of the KnowledgePanel Internet Survey

Monitoring Attitudes of the Public

Nationwide Retirement Institute

Nationwide Survey of Attitudes Toward Social Security (“Hart Survey”)

Retirement Confidence Survey (RCS)

Retirement History Survey (RHS)

Survey of Economic Expectations (SEE)

Transamerica Institute

Understanding America Study (UAS)