There is a fallacy in circulation that Americans tend to stop working in their early 60s, shortly after becoming eligible for reduced Social Security benefits at age 62 and well before the designated “full” retirement age of 66 (Rivlin 2010). As this issue brief shows:

• Labor force participation among older workers is high by historical standards.
• The myth of early retirement persists partly because of misleading measures of the retirement age which purport to show that people retire in their early 60s.
• A new measure of retirement that better captures what we mean by retirement (a measure that excludes self-described “retirees” still working for pay, people with disabilities, and full-time caregivers) shows that people in the United States typically retire in their mid-60s (age 65.5), not in their early 60s.

Trends in labor force participation

The shift to 401(k)s over the past quarter century, the gradual phase-in of Social Security benefit cuts enacted in 1983, and other factors have reversed a post-World War II trend toward early retirement (Munnell 2011). In fact, the best available measure of older workers—those aged 55 to 64—shows historically high labor force participation: The share of 55- to 64-year-olds in the labor force is the highest since the Bureau of Labor Statistics started measuring labor force participation in 1948. Though the labor force participation rate for the group of Americans age 65 and older has fallen somewhat, this broad age group includes a growing share of people in their 80s and 90s who have very low participation rates. Overall, the share of Americans age 55 and older who are in the labor force is now the same as it was a half a century ago and is only slightly below the postwar peak, as shown in Figure A.

Misleading measures perpetuate myth of early retirement

The age at which workers are eligible for unreduced Social Security benefits, currently 66, is often referred to as the “normal retirement age,” though the Social Security Administration uses the term “full retirement age.” The term “normal retirement age” is potentially confusing because it does not necessarily correspond to the age when the typical worker retires. Nevertheless, many experts refer to retirement in one’s early 60s as “early retirement” and view it as a cause for concern based on the assumption that many workers are not well informed about how their standard of living will be affected. It is also popularly believed that early retirement affects the sustainability of Social Security even though monthly benefits are adjusted so that the value of lifetime benefits remains roughly the same whether a worker retires at age 62 or 70.

Despite evidence to the contrary, the myth that people are retiring in their early 60s persists in part due to potentially misleading measures of the “average retirement age” that include people who are still working, who were forced to leave the workforce due to disability, or who were never in the workforce to begin with.

Existing measures generally fall into three categories: those based on Social Security take-up; those based on surveys asking people if they consider themselves retired; and those based on the labor force participation rate of older workers. The main problem with measures based on Social Security take-up or surveys is that they typically include many “retirees” who are still working for pay. On the other hand, measures based on labor force participation lump retirees in with others who are not in the labor force, such as disabled people and full-time caregivers.

As shown in Figure B, on average, people start accessing their Social Security retirement benefits at age 63.8. This measure declined by more than five years between 1945 and 1970 but has remained relatively stable since then. It has not increased in recent years, despite the higher labor force participation of older workers, because the share of Social Security beneficiaries who are still working has grown (Social Security Administration 2002 and 2010a).

This measure includes only people receiving “retired worker” benefits (as opposed to disabled worker, spousal, or dependent benefits). A variation that includes older workers who become eligible for Social Security disability benefits puts the average “retirement” age at 62.6 for men and 62.5 for women in 2005–10 (Gendell 2008). The average retirement age according to this alternative measure has actually fallen slightly in recent decades because it conflates two trends—later retirement for healthy workers and early exit for workers with disabilities. Though the increase in disability rolls is a valid concern, the two trends have different policy implications and the measure does not reflect what most people think of when they hear the word “retirement.”

As mentioned earlier, the main problem with relying on Social Security take-up is that many Social Security recipients are still working for pay. Nearly half (45 percent) of recipient “units” (individuals and married couples) age 62–64 also had earnings from a job or jobs, and these earnings constituted nearly one-fourth (24.2 percent) of their total income. These are not necessarily part-time jobs: The typical (median) beneficiary unit with earned income had earnings of $28,800 (Social Security Administration 2010a). Though these statistics include retirees with working spouses and those who worked part of a year before retiring, they also include people who are still working full-time and full-year, especially since Social Security rules changed to encourage people to keep working.¹

This ambiguity also affects statistics based on self-assessments of retirement status. Numerous surveys confirm that many self-described “retirees” are still working for pay. For example, the 2011 Retirement Confidence Survey conducted by the Employee Benefit Research Institute found that 30 percent of retirees reported retiring before age 60, and 38 percent reported retiring between age 60 and 64. However, nearly one in four (23 percent) of these self-described retirees were still working for pay (Helman 2011). Unlike the Social Security take-up data, this survey measure presumably does not count people who worked part of a year before retiring.

A less subjective way to define retirement is based on labor force participation, excluding anyone who is working or who is unemployed and seeking work. Researchers at the Center for Retirement Research (CRR) at Boston College, for example, define the “average retirement age” as the age at which labor force participation drops below half. By this measure, CRR estimates that the average retirement age is around 64 for men and 62 for women (Munnell 2011).

The problem with this measure is that it pegs the retirement age too low by counting full-time homemakers, people with long-term disabilities, and others who are outside the workforce as retired. For example, for workers born in 1944 (the cohort that attained Social Security’s normal retirement age of 66 in 2010), the CRR methodology yields an average retirement age of 62.2 for men and women combined (Figure C). However, the labor force participation rate for this cohort peaked at 84.3 percent at age 46. That means that nearly 16 percent of this cohort was not in the labor force at the time when their labor force participation was highest. So when their labor force participation rate dropped below 50 percent, nearly six of 10 people who were working at the peak were still working.

A new measure

A more logical estimate of the average retirement age is the age at which half of the workforce (as opposed to half of the civilian population) has left the labor force. According to Current Population Survey data, the labor force participation rate falls to half of its peak at age 64.1.² But this measure still includes older workers who left the workforce due to disability. A better estimate of the average retirement age is the age at which half of the nondisabled workforce has exited the labor force: 65.5 (Figure C).³ Note that this still understates the average age of voluntary retirement since it includes workers who dropped out of the labor force after losing their jobs. However, it does count anyone who works at all (even very part-time) as not retired.

Policy implications

Americans’ supposed proclivity for early retirement is often cited in policy discussions, including arguments for cutting Social Security benefits.For example, Erskine Bowles and Alan Simpson, co-chairs of President Obama’s fiscal commission, cited an average retirement age of 62 in arguing for an increase in the age at which people are first eligible for reduced Social Security retirement benefits (62) and the age at which they can start receiving full retirement benefits (currently 66, but increasing to 67 for workers born in 1960 and later) (National Commission on Fiscal Responsibility and Reform 2010). Maya MacGuineas of the Committee for a Responsible Federal Budget and others have made similar arguments (MacGuineas 2009).

There are a couple of problems with the argument for raising the retirement age to offset alleged early retirement. First, it implies that the timing of retirement affects Social Security’s finances. It does not. Contrary to popular belief, the average age of take-up does not affect Social Security’s finances because benefits are adjusted for early retirement in a way that equalizes the value of lifetime benefits. In other words, early retirees receive lower monthly benefits over a longer period of time.

Second, the suggestion is that workers can offset benefit cuts by working longer.⁴ But as we have shown, people already are working longer—to an average age of 65.5. So when policymakers say that people should be working longer, they need to be clear about what they exactly they are saying, which is, “65- and 66-year-olds need to be working longer,” not the less radical-sounding “62- and 63-year-olds need to be working longer.” In short, focusing on Americans’ supposed proclivity for early retirement makes proposed cuts appear less harsh, since it is easier to suggest that workers in their early 60s postpone retirement than to ask people to work into their late 60s or beyond. And because many people are unaware that lifetime Social Security benefits are the same whether a worker retires at age 62 or 70,⁵ the idea that Americans tend to retire early reinforces the myth that Social Security is in trouble because Americans are living longer but not working longer.

Clearly, previous measures of the “average retirement age” do not describe retirement as it is commonly understood—older workers opting to leave the workforce—because the measures either include people who are still working or count full-time caregivers and disabled people. The fact that the average nondisabled worker is still working in his or her mid-60s calls into question the common assumption that workers will be able to offset proposed Social Security cuts by working longer into old age.

It could be argued that nonworking spouses and disabled workers are properly included in these measures because they are also eligible for Social Security benefits. While it is true that these groups are also Social Security beneficiaries, they affect Social Security’s finances in different ways. For example, while the average retirement age of women has fluctuated over the years, labor force participation of women in their prime working years has risen. Social Security’s finances have been more affected by the influx of married women into the workforce than by changes in the average retirement age of women, and this effect is unambiguously positive since benefits for nonworking spouses are not paid for by higher contributions from married workers.

The higher participation of women has more than offset a decline in the participation of men. Figure D shows an overall increase in labor force participation despite ups and downs in the participation rate for the oldest workers. This has had a positive impact on Social Security’s finances.

The decline in work among men in their 40s and 50s raises concerns about a lack of jobs for less-educated older male workers, but this is a separate issue from Social Security. A focus on the average retirement age and labor force participation at the end of people’s working lives often lends support to policies that would extend working lives into old age, when it might be wiser to focus on increasing labor force participation at younger ages. Even though work effort for the population overall has increased, there is still room for improvement. For example, disability rates have increased due to many factors, including the impact of a weak job market on workers in poor health and the decline in health insurance coverage. Raising the Social Security and Medicare eligibility ages would exacerbate this trend, but expansionary macroeconomic policies and other job-creation measures would alleviate it. Such policies would also help middle-aged men without college degrees, teenagers, and others who are more likely to drop out of the workforce when jobs are scarce.

The focus on the average retirement age is a distraction from the more important question of how much Americans work over the course of their entire working lives. The broader question of whether Social Security’s projected shortfall is due to a growing imbalance between work and leisure is addressed at greater length in a longer EPI briefing paper (Morrissey 2011). However, as long as people pay attention to these statistics and draw policy prescriptions from them, they should be as accurate as possible.

—This paper was prepared with research assistance from Hilary Wething

Endnotes

1. Social Security rules allow beneficiaries below the full retirement age to earn up to $14,160 without a reduction in benefits. Benefits are reduced by $.50 for every $1 of earnings above $14,160. Beneficiaries at or above the full retirement age can have unlimited earned income without a reduction in benefits. It is important to keep in mind that the decision to apply for Social Security benefits while still working does not negatively affect Social Security’s finances since benefits are actuarially adjusted to maintain the same lifetime benefits regardless of the age of take-up.

2. There are flows into and out of the labor force at all ages. Thus, the peak participation rate does not capture everyone who ever worked, such as women who worked before marriage. But data on peak participation is more readily available than data on lifetime participation, and it is not clear whether people who were only briefly in the labor force should be considered “retired,” especially if they never became eligible for retired worker benefits.

3. The labor force participation rate for the nondisabled population was calculated by dividing the labor force by the total population minus the estimated Social Security disabled population by age and birth cohort (Social Security publishes breakdowns of the disabled population by five-year age groups). The Social Security disabled population includes people who were active-duty military personnel before becoming disabled and thus were not included in the civilian population and labor force estimates from the Census Bureau’s Current Population Survey. However, this group is too small to make a significant difference in these calculations.

4. Raising the normal retirement age is the same as an across-the-board benefit cut. Though some workers may be able to offset such a cut by working longer, most will likely experience it as a reduction in retirement income.

5. Social Security benefits increase by roughly 7–8 percent per year between age 62 and 70 (Social Security Administration 2010b). These adjustments are designed to approximately equalize lifetime benefits since those who retire later receive benefits for fewer years.

References

Gendell, Murray. 2008. “Older workers: increasing their labor force participation and hours of work.” Monthly Labor Review, pp. 41–54.

Helman, Ruth et al. 2011. The 2011 Retirement Confidence Survey: Confidence Drops to Record Lows, Reflecting the ‘New Normal.’ Employee Benefit Research Institute Issue Brief no. 35. Washington, D.C.: EBRI. http://www.ebri.org/pdf/surveys/rcs/2011/EBRI_03-2011_No355_RCS-11.pdf

King, Mariam et al. 2010. Integrated Public Use Microdata Series (IPUMS), Current Population Survey: Version 3.0. [Machine-readable database]. Minneapolis: University of Minnesota.

MacGuineas, Maya. 2009. Five Steps to Cut the National Debt. Washington, D.C.: The Committee for a Responsible Federal Budget. http://crfb.org/document/five-steps-cut-national-debt

Morrissey, Monique. 2011. Beyond ‘Normal’: Raising the Retirement Age Is the Wrong Approach for Social Security. Economic Policy Institute Briefing Paper #287. Washington, D.C.: EPI. http://www.epi.org/publication/bp287/

Munnell, Alicia. 2011. What is the Average Retirement Age? Center for Retirement Research at Boston College Brief No. 11-11. Boston: CRR. http://crr.bc.edu/briefs/what_is_the_average_retirement_age.html

National Commission on Fiscal Responsibility and Reform. 2010. The Moment of Truth, Report of the National Commission on Fiscal Responsibility and Reform. Washington, D.C.: The White House.  http://momentoftruthproject.org/sites/default/files/TheMomentofTruth12_1_2010.pdf

Rivlin, Alice. 2010. The Right Reason for Saving Social Security. Washington, D.C.: The Brookings Institution. http://www.brookings.edu/opinions/2010/0722_saving_social_security_rivlin.aspx

Social Security Administration. 2002. Income of the Population 55 or Older, 2000, Washington, D.C.: SSA.  http://www.ssa.gov/policy/docs/statcomps/income_pop55/2000/index.html

Social Security Administration. 2010a. Income of the Population 55 or Older, 2008, Washington, D.C.: SSA. http://www.ssa.gov/policy/docs/statcomps/income_pop55/

Social Security Administration. 2010b. “Effect of Early or Delayed Retirement on Retirement Benefits,” Office of the Chief Actuary Web page. http://www.ssa.gov/OACT/ProgData/ar_drc.html

Social Security Administration. 2011. Annual Statistical Supplement to the Social Security Bulletin, 2011, Washington, D.C.: SSA.  http://www.ssa.gov/policy/docs/statcomps/supplement/2011/