A New Approach to Raising Social Security’s Earliest
Kelly Haverstick, Margarita Sapozhnikov, Robert K. Triest, and
While Social Security’s Normal Retirement Age (NRA) is increasing to 67, the Earliest
Eligibility Age (EEA) remains at 62. Similar plans to increase the EEA raise concerns that
they would create excessive hardship on workers who are worn‐out or in bad health. One
simple rule to increase the EEA is to tie an increase to the number of quarters of covered
earnings. Such a provision would allow those with long work lives—presumably the less
educated and lower paid—to quit earlier. We provide evidence that this simple rule would
not satisfy the goal of preventing undue hardship on certain workers. Therefore, this paper
considers an alternative policy that ties an increase in the EEA to individuals’ Average
Indexed Monthly Earnings (AIME). We show that allowing workers with low AIME to
continue to be eligible to receive benefits at age 62 has promise as a policy to protect
workers who have low earnings and are in poor health from hardship associated with an
increase in the EEA.
JEL Classifications: H55, J26, I12
Kelly Haverstick is a research economist at the Center for Retirement Research at Boston College (CRR).
Margarita Sapozhnikov is a senior associate at CRA International. Robert Triest is a senior economist and
policy advisor at the Federal Reserve Bank of Boston. At the time this paper was written, he was a visiting
scholar at the CRR. Triest’s email address is email@example.com. Natalia Zhivan is a graduate research
assistant at the CRR.
This paper, which may be revised, is available on the web site of the Federal Reserve Bank of Boston at
The views expressed in this paper are solely those of the author and do not necessarily represent those of the
Federal Reserve Bank of Boston, the Federal Reserve System, or the opinions or policy of SSA, any agency of
the federal government, or Boston College.
The research reported herein was performed pursuant to a grant from the U.S. Social Security Administration
(SSA) funded as part of the Retirement Research Consortium.
This version: October 2007
Social Security’s Normal Retirement Age (NRA) is currently rising from age 65 to
age 67. An individual must wait until the NRA to claim full benefits; benefits claimed at
an earlier age are actuarially reduced. Despite the rise in the NRA, the Earliest Eligibility
Age (EEA)—the earliest age individuals can claim reduced benefits—remains at 62. A
question facing policymakers is whether the EEA should be increased to 64 to match the
increase in the NRA. Increasing the EEA would not alter Social Security’s financial state
since benefit reductions for a person with average life expectancy are actuarially fair—
benefits are lowered to offset the longer claiming period. However, a rise in the EEA
would protect workers from facing an increased risk of impoverishment in old age due
to their myopically claiming benefits too early.
Raising the EEA involves making a tradeoff between ensuring retirement income
adequacy and insuring workers against the prospect that they will find it difficult to
work or find jobs as they age. As the NRA increases, there is a reduction in the benefits
payable to a worker claiming at any age below the NRA. This reduction raises the
concern that workers may myopically claim benefits too early, making it difficult for
them to maintain their living standard during retirement. The prospect of future
increases in Medicare premiums and out‐of‐pocket medical expenses imply that this
problem will likely grow in importance, especially in the case of widows or widowers
relying on survivors’ benefits as their primary source of income.
Increasing the EEA prevents workers from putting their future standard of living
at risk, but it does so at the cost of potentially causing hardship for those workers who
find it difficult to work at ages younger than the NRA. Individuals who have physically
demanding jobs may find work to be increasingly onerous as they age, yet they may not
have health problems serious enough to qualify for disability benefits. If such
individuals lack adequate private pension benefits or other financial resources, then an
increase in the EEA would require them to either rely on inadequate income or prolong
their working lives when they would instead prefer to retire.
To avoid hurting those unable to work, some have proposed tying the EEA to the
length of a worker’s labor force participation. The idea of differentiation based on labor
force participation rates has an intuitive appeal and is easy to implement. Those who
went to college and thus delayed entry into the workforce usually have a safer working
environment, higher incomes, better health, and longer life expectancies, and thus can
remain in the labor force more easily. While health status, ability to work, or even level
of education are not directly observed by the Social Security Administration,
information on earnings and number of quarters of covered earnings are routinely used
to determine eligibility and the level of monthly benefits.
One simple proposed scheme would give workers an option of retiring at 62 and
claiming unreduced benefits if they have spent at least 35 years in the labor force.
Everybody else’s EEA would be moved to 64. Credit years would be assigned to women
to reflect the number of years spent caring for young children.
Our analysis demonstrates that policy rules that tie eligibility for unreduced
benefits at age 62 to labor force participation fail to help those who are in poor health.
Unhealthy workers are simply unable to obtain 35 years of labor force participation,
while workers in good health satisfy the proposed test for eligibility. Healthy workers,
however, tend to postpone claiming and retiring until a later age regardless of the EEA.
Thus, tying the EEA to the length of a worker’s labor force participation would not help
many individuals who are unable to work because of poor health or an inability to find
jobs in their 60s.
While the number of covered quarters is a poor measure of health status,
earnings are a good predictor of health, wealth, and job prospects later in life. Our
analysis shows that tying the EEA to the Average Indexed Monthly Earnings (AIME)
may have some potential. Admittedly, this method is more complex (with decisions to
be made concerning cutoff values for eligibility), so further research is necessary to
determine the feasibility of such a rule.
This paper is organized as follows. Section II discusses the data used and
provides descriptive statistics for the sample. Section III summarizes the results of
bivariate probit regressions of the decisions to claim early benefits and to exit the labor
force. Section IV puts forth a simple rule for raising the EEA and considers its effects,
including why it does not meet our goal of protecting the most vulnerable workers. This
section then describes an alternative rule to raise the EEA based on a worker’s earnings;
this approach would better satisfy our main objective. Section V concludes.
II. Data and Sample
To consider potential rules for increasing the EEA, this analysis uses the RAND
public version of the Health and Retirement Study (HRS) and the SSA Administrative
Data (SSAA), which contain seven waves of data (1992–2004).1 The sample selection
process follows that of Panis et al. (2002) and is shown in Figure 1. We begin with 10,560
individuals. We require that respondents reach the age of 63 by the 2004 wave, have the
Social Security benefit type available, have quarters of covered earnings information, are
not widow(er)s, do not receive care for a dependent child under the age of 16, and do
not clearly misrepresent their information.2 This leaves a final sample of 3,277
Each individual in our sample is classified into one of six categories based on the
claim of Social Security benefits.3 These categories are determined using the “Type of
benefits received from Social Security” and the “Type of supplemental (other) benefits
received from Social Security” variables in the SSAA data:
“Takers”: Individuals who receive their own Old‐Age and Survivors
Insurance (OASI) benefits at 62 years of age;
“Postponers”: Individuals who are eligible for their own early OASI
benefits but claim after age 62;
1 Specifically, we use the HRS Cohort: Respondent Earnings and Benefits data file.
2Widow(er)s become eligible for benefits at the age of 60 and are not directly affected by a higher EEA.
Thus, 478 observations identified as widow(er)s were excluded.
3 These categories were defined using the method of Panis et al. (2002).
“Spousal Takers”: Individuals who receive OASI spousal benefits starting
at age 62;
“Spousal Postponers”: Individuals who are eligible for early OASI
spousal benefits but claim after age 62;
“Ineligibles”: Individuals who do not qualify for OASI benefits; and
“DI claimants”: Individuals who receive Disability Insurance (DI)
Table 1 shows the distribution of claimant types by gender. With the inclusion of
the 2000, 2002, and 2004 waves of data, these percentages are not statistically different
from those reported by Panis et al. (2002) with two exceptions due to female takers who
have dual entitlement.4 Our method classifies all individuals with dual entitlement into
the spousal benefits category.5 Table 2 shows the breakdown in the spousal benefit
category between “dual benefit” claimants and “spousal benefit only” claimants for
females and for the entire sample. Most individuals classified as spousal benefit takers
have dual entitlement rather than only spousal benefits. This variation in classification,
rather than a difference in the composition of the two samples, appears to be the reason
for the discrepancy in the proportions in these two groups.
Characteristics of EEA Claimants
Once individuals are classified into claimant categories, we can consider the
descriptive characteristics of the takers compared with the postponers. In determining a
rule for increasing the EEA, we begin by observing the characteristics of the individuals
4 See Table 3.2 in Panis et al. (2002) for comparison proportions.
5 “Under the dual entitlement provision, if a person is entitled to a larger Social Security benefit as a worker
than as a spouse, no spouse’s benefit is payable because the person is not considered dependent on the other
spouse (in this case, we classify the person as receiving worker benefits). If the benefit payable as a spouse
exceeds the worker’s benefit for a person, then the spouse’s benefit is offset by the amount of the worker’s
benefit (in this case, we classify the person as receiving spousal benefits). As a result of the dual entitlement
provision, nearly 6 million beneficiaries receive reduced benefits as a spouse—meaning they receive the
equivalent of their worker’s benefit or their spouse’s benefit, whichever is higher.” See U.S. Social Security
who claim benefits at the current EEA, 62, to gain insight on the people who would be
affected by the increase.
Table 3 contains summary statistics on the number of covered earnings quarters,
health, physical job, wealth, pension coverage, and education for OASI worker claimants
(based on their own work history). These summary statistics are reported by gender and
claimant type (taker or postponer). The first variable of interest, number of quarters of
covered earnings by age 62, shows no statistical differences between takers and
postponers within each gender. As expected, there are statistical differences in the
means and medians between the genders, with higher values for males than for females.
The next two rows of the table show that there are substantial differences in the
health of takers and postponers. Of those who report poor health at age 63, 40 percent
were male takers (compared with 34 percent of the entire sample), 25 percent were male
postponers (compared with 33 percent), 24 percent were female takers (compared with
19 percent), and 10 percent were female postponers (compared with 14 percent overall).
Individuals who report poor health or job‐limiting health are more likely to claim early
benefits. The opposite is true for those reporting having a physical job, as they were less
likely to be takers and more likely to be postponers. However, this finding may be
misleading since having a physical job is measured at age 63. Thus, if an individual has a
job (whether physical or not) at age 63, he is probably less likely to have already claimed
Social Security benefits.6
Male takers have lower mean and median non‐housing wealth than do male
postponers, while the mean and median for females indicate that wealthier women
claim benefits early. For both males and females, there is more variance in the takers’
wealth than in the postponers’ wealth, indicating that the group retiring later is more
homogeneous with respect to wealth. The heterogeneity in wealth among those taking
benefits early is suggestive that early takers are a mix of those who are financially secure
enough to retire early and those who take benefits early due to poor health or economic
6 See the next section for the relationship between age of claiming benefits and age of exiting the labor force.
As expected, claimants with defined benefit pension coverage are more likely to
retire early. This finding is consistent with the type of jobs generally covered by defined
benefit pensions, such as union jobs where employers frequently offer early retirement
incentives or physically demanding or dangerous public jobs, such as firefighters or
police officers. Defined benefit pensions also may provide the financial resources needed
to finance a comfortable early retirement.
Finally, educational attainment differs between the takers and postponers. One
might expect takers to be less educated in that they would have worked full‐time longer,
since presumably those with less education start their work lives earlier and are more
likely to be in physically demanding jobs. This expectation does seem to be the case.
Generally, within each gender, less‐educated individuals are more likely to be takers
and highly‐educated individuals are more likely to be postponers. This pattern suggests
a negative correlation between claiming benefits early and education level.
Table 4 contains the same summary statistics for females who claim spousal
benefits. As expected, women with dual entitlement have more covered quarters of
work than do women with spousal only benefits.
For claimants with dual entitlement, takers have a lower mean but higher
median non‐housing wealth than postponers, while the opposite is true within the
spousal only claimants. As is the case with those claiming worker benefits, the standard
deviation is higher for takers than for postponers in both categories. Again this pattern
indicates that the postponers are a more homogeneous group than the takers with
respect to wealth.
For education level, nothing noticeable appears in comparing takers and
postponers within the dual entitlement or spousal only categories. However, spousal
only claimants seem more likely to have less education than dual entitlement qualifiers.7
These descriptive statistics provide some insight into the characteristics of
individuals who would be affected by an increase in the EEA. There are differences in
7 One exception is that a substantial number of those with at least a college degree are spousal only
some characteristics between takers and postponers for women who are spousal
claimants, but clearer differences appear in the groups who claim worker benefits. Since
a goal of increasing the EEA is to increase labor force participation, this analysis
concentrates on differences between workers who claim early benefits and those who
Relationship between Claiming OASI Benefits and Labor Force Participation
It is important to distinguish between claiming Social Security benefits and
retiring, since these may not occur at the same time. Some workers may choose to retire
before reaching the EEA, and others may continue to work while receiving benefits.
From a policy standpoint, to the extent that early claimants continue to participate in the
labor market, it suggests that there would be fewer individuals who would need to be
protected against an increase in the EEA. The analysis in this section separately
measures the exit from the labor force for each individual in the sample and compares
the timing of this occurrence with the timing of claiming Social Security benefits.
About 55 percent of the sample of OASI claimants choose to claim benefits at age
62. Sixty‐one percent of women claim benefits at this age compared with only 50 percent
of men. A significant number of those receiving benefits at 62 do not exit the labor force
at the same time—about 40 percent of men and 34 percent of women continue working
(see Table 5).8 Not surprisingly, fewer spousal takers, about 26 percent, remain in the
labor force.9 These individuals likely have weaker ties to the labor force.
For the entire sample, a better overview of these two decisions appears in Figures
2 and 3. Figure 2 shows the distribution of the ages of claiming OASI benefits and the
age when exiting the labor force. Figure 3 contains these distributions by gender. Both
include worker and spousal claimants.
8 We refer to an individual as “staying in the labor force” if we observe him working at least one quarter
after receiving Social Security benefits. Note that this designation does not measure the length of time an
individual remains in the labor force while claiming benefits. We refer to any claimant who did not stay in
the labor force as “exiting the labor force” even though he or she may have never been in the labor force.
9 This result combines the two subgroups of dual entitlement takers and spousal only takers ((74+6)/(263+43)
Figure 2 shows that the decision to claim benefits is much more discrete than the
decision to exit the labor force. A substantial fraction of individuals stop working before
they are eligible to receive benefits, but many others continue working while receiving
benefits. While 55 percent of our sample claim benefits at 62, only 44 percent exit the
labor force by age 62. Although the age of exiting the labor force is more dispersed, there
is also a spike at age 62, presumably influenced by the ability to claim Social Security
benefits. Also note that the distribution has a normal shape, centered on the most
common ages for claiming Social Security benefits, 62 to 65. This pattern indicates that a
shift in the ages at which benefits can be claimed might shift not only the claiming age
but also the age of exiting the labor force.
Considering these distributions separately for males and females, Figure 3
provides some insight on how an increase in the EEA may affect men and women
differently. The distribution of ages of claiming benefits is tighter for women than for
men. For the ages when exiting the labor force, the distribution is more dispersed for
women than for men. This relative difference may imply that the rules based on the ages
of claiming Social Security benefits have a larger impact on men’s decisions to retire
than on women’s decisions to retire.
While the decisions to claim Social Security benefits and to exit the labor force
are not perfectly correlated, there seems to be a meaningful relationship. Assuming this
relationship holds, a universal increase in the EEA would also tend to increase the
average age of exiting the labor force. Thus, it seems a valid concern that raising the EEA
uniformly would cause hardship for those who would find it difficult to continue
working. To further identify these workers, the following section uses multivariate
regression analysis to estimate the effects of observable characteristics on the decisions
to claim benefits and exit the labor force.
III. Multivariate Analysis
Using the information from the previous section on distinguishing characteristics
of those who claim Social Security benefits early versus those who postpone claiming
benefits, we estimate a reduced form model of the choice to claim benefits and to exit the
labor force. Controlling for other factors that potentially affect these decisions, these
regressions test for the effect of the length of labor force participation on these choices.
We use a bivariate probit model to estimate these decisions, as we have evidence that
these decisions are made jointly. The model is:
0 , 01
0 , 01
where is the propensity to claim Social Security benefits and =1 means that the
individual claims benefits at the age of 62 or earlier; is the propensity to exit the labor
force, and =1 means that an individual exits the labor force at the age of 62 or earlier.
The set of explanatory variables and is the same
, and includes variables
discussed from Table 3.
We decompose the total non‐housing wealth distribution into five percentile
groups and create dummy variables for each quintile. We exclude the lowest quintile as
the reference group. We create dummy variables for each of the four education groups,
with the “Less than high school” group excluded. A dummy variable indicating being in
poor/fair health at age 63 is also included. Finally, we include the number of quarters of
covered earnings at age 63 and indicators of household private pension coverage:
having both a defined benefit and defined contribution plan, a defined benefit plan only,
or a defined contribution plan only.
The regression results are presented in Table 6. The coefficients on the
explanatory variables are proportional to their effect on the probability of the dependent
variable being equal to 1; our analysis here focuses on the direction of the effects and
their statistical significance. Although the coefficient estimates are somewhat suggestive
of there being a U‐shaped relationship between wealth and the probability of claiming
benefits at the EEA, the estimates are not statistically significant. In contrast, there is a
discernable effect of wealth on the decision to exit the labor force. Individuals in the
middle quintile of wealth are significantly less likely to exit the labor force by age 62
than individuals either at the bottom or the top of the wealth distribution.
Consistent with trends revealed in the descriptive statistics, more highly
educated males tend to stay in the labor force past the age of 62 and postpone claiming
benefits. Those with at least a college degree are less likely to claim benefits at the EEA
and to leave the labor force by age 62 than those with less education.
Another significant factor in these decisions is health. As expected, being in bad
health at age 63 indicates a higher likelihood of claiming early benefits and of exiting the
labor force at or before age 62.
We also expect the number of quarters of covered earnings to have a positive
impact on claiming benefits early. However, the opposite effect appears, in that the
number of covered quarters has a significant negative impact on retiring early. Those
with more time in the labor force (as measured by quarters of covered earnings) are less
likely to claim benefits at the EEA and to exit the labor force before age 63. This finding
may indicate that if an individual retires at 62 because of bad health, it is likely that he
has frequently been in bad health and thus unable to work as consistently as a healthier
Finally, these coefficients indicate that households with defined benefit pension
coverage are more likely to both leave the labor force early and claim early Social
Security benefits. As mentioned earlier, this finding is consistent with our hypothesis,
given the prevalence of early retirement incentives for workers with defined benefit
These regressions confirm the findings from the descriptive statistics. Many
explanatory variables have the expected effects on the decisions to claim benefits and to
exit the labor force by age 62. However, once we control for these other factors, more
years of earnings still lower the probability of claiming benefits and exiting the labor
IV. Proposed Policies to Raise the EEA
Using the information on factors affecting the decisions to claim and to exit the
labor force and the characteristics of early claimants, this section explores possible rules
to increase the EEA. In order to devise a plan to increase the EEA non‐uniformly so that
unhealthy, worn‐out workers could still retire at 62, we first consider a simple intuitive
rule that is relatively easy to implement: conditioning the EEA on workers’ covered
quarters of labor force participation. Unfortunately, evidence from the previous section’s
regression results plus additional support show why this simple rule would not meet
this goal. We then provide an alternative method that conditions the EEA on past
earnings. Although this approach is less intuitive, it is more effective in targeting less
healthy workers for eligibility to claim benefits early.
Tying the EEA to Workers’ Past Labor Force Participation
Allowing individuals with longer work histories to collect benefits at a younger
age is intuitively reasonable. Individuals engaged in physically demanding work tend to
start full‐time work at younger ages than the college‐educated workers with comfortable
jobs in office environments. Setting the EEA to a younger age for those with longer work
histories might permit workers who are most in need of retiring to do so, while
encouraging others to continue working.
To be more specific, assume that individuals with 35 years of labor force
participation may claim at 62 while all others would need to wait until 64. The
effectiveness of this rule in allowing those in poor health to retire earlier than those in
good health depends upon whether completed quarters of work at age 62 is a good
proxy for poor health. However, the results from the regression analysis described
above suggest that such a relationship is very unlikely. Our analysis in the previous
section reveals that longer work lives are associated with a tendency to postpone
retirement. Because poor health discourages labor force participation prior to age 62, the
individuals with the most covered quarters at age 62 are likely to be in relatively good
By relating the number of quarters of covered earnings to an individual’s health,
we provide additional evidence that the proposed policy would not protect unhealthy
workers. Figure 4 shows the mean number of quarters of covered earnings by age and
health status at age 63. The mean number of quarters worked is generally higher for
those in good health than for those in bad health. However, the difference is not obvious
until about age 50. This pattern suggests that whether individuals’ health problems
recorded at age 63 are recent or chronic, they have a larger impact on labor force
participation at older ages. Figure 5 shows the proportion of individuals reporting poor
or fair health by their number of covered quarters. The category including those with
fewer than 40 covered quarters (10 years) by the age of 62 has twice the proportion of
unhealthy individuals as the group with at least 160 covered quarters (40 years).
Table 7 also shows why linking the EEA to labor force participation does not
protect unhealthy, worn‐out workers who cannot afford to exit the labor market. It
provides information on certain characteristics of two groups: workers who satisfy the
condition of having 35 years (140 quarters) of earnings and those who do not. Again,
these data show that, of those reporting poor health, only 60 percent would be eligible to
claim benefits at age 62, compared with 70 percent of the full sample. Workers with at
least 35 years in the labor force by age 62 also tend to not be in the bottom of the wealth
distribution. While 70 percent of the sample have at least 35 years of earnings, only
about 46 percent of those in the lowest quintile of the wealth distribution would qualify
to claim benefits at age 62 using this rule.
While long working lives may have a negative impact on health, workers who Download full-text
have been in poor health are unable to stay in the labor force and thus do not satisfy the
rule exempting them from an increase in the EEA. Setting the EEA as a function of the
number of years in the labor force would seem to adversely affect the most vulnerable
group of workers: those who are in poor health and have Social Security benefits as their
primary source of retirement income.10
Tying the Earliest Eligibility Age to AIME
Instead of using the length of an individual’s work history to try to identify
vulnerable workers, perhaps a better approach would be to use their earnings history.
Figure 6 shows that the earnings of individuals in poor or fair health at age 63 are lower
than the earnings of healthy workers over the entire course of their lives. This finding is
consistent with the hypothesis that unhealthy workers have poor job opportunities.
However, it is also consistent with low earnings causing poor health. Sullivan and von
Wachter (2006) show that job loss reduces life expectancy. One possible explanation is
that displaced workers experience a lasting decrease in earnings and a rise in earnings
instability. Lack of access to health insurance, unhealthy lifestyles due to low earnings,
and stress associated with job loss and earnings instability may have a negative impact
The association between poor health at age 63 and low earnings throughout
much of one’s working life suggests that a reasonable alternative to tying the EEA to
covered quarters of labor force participation would be to instead base individuals’ EEA
on a measure of lifetime earnings. A natural measure of lifetime earnings is Average
10 Other rules to non‐uniformly increase the EEA were also considered. These rules were based on allowing
low‐educated workers the opportunity to claim benefits at 62 while raising the EEA to 64 for highly
educated individuals. The main issue with this method is that the Social Security Administration (SSA) does
not observe education level. Thus, we tested and found patterns of earnings histories that were correlated
with education levels which would allow the SSA to estimate the target group of low‐educated workers. The
standard deviation, growth, peaks, and timing of wages are indicators of an individual’s education.
However, there are two major problems with this method. First, it is rather complicated. Second, and more
importantly, it does not meet the goal of permitting unhealthy, worn‐out workers to claim benefits at 62,
allowing them to exit the labor force.