Page 1 of 20

Archives of Business Research – Vol. 12, No. 4

Publication Date: April 25, 2024

DOI:10.14738/abr.124.16774.

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement

Age of 62. Archives of Business Research, 12(4). 18-37.

Services for Science and Education – United Kingdom

The Effects of Increasing the Full Retirement Age from 66 to 67

on the Retirement Decision at the Early Retirement Age of 62

Chong Hwan Son

Department of Education and Social Sciences, Hudson Valley Community College

80 Vandenburgh Ave. Troy, New York 12180, U.S.A

ABSTRACT

In the United States, the full retirement age increases in two-month increments each

year until reaching 67 years of age in 2027 for individuals born from 1955 to 1960.

The increase in the full retirement age is equivalent to a decline in the Social

Security benefits for all new retirees. If the full retirement age is increased from 66

to 67 while the early retirement age remains at 62, then the monthly benefit

reduction at age 62 would increase from 25% to 30%. Despite the reduction in

Social Security benefits, a significant portion of newly retired workers claim their

Social Security benefits as soon as they turn 62, which is the earliest age to claim the

benefit. The Social Security Administration predicts that the reduction in Social

Security benefits can encourage workers to delay their retirement and that the

reduction can restore some financial balance to the Social Security system.

Therefore, studying the effects of increasing the full retirement age from 66 to 67

on the retirement decision at 62 is essential for planning future social security

reforms. This study conducts a Two-Stage Least Squares (2SLS) regression, using

cross-sectional and time series data from the Behavioral Risk Factor Surveillance

System (BRFSS) from 2016 to 2021. The empirical results show that the probability

of retirement for individuals at age 62 decreased from 2016 to 2021, indicating a

trend of delaying retirement as their full retirement age rises. During the same

period, the likelihood of men reporting being retired decreases by 35.56%, while

for women, it drops by 16.79%. It implies that if male and female workers choose to

exit the workforce solely based on the Social Security benefits, the increase in full

retirement age appears to have a more significant impact on delaying retirement

decisions for male workers than for female workers.

Keywords: Full Retirement Age, Early Retirement Age, Social Security Benefit, BRFSS, Life

Expectancy.

INTRODUCTION

The global average life expectancy has increased due to advancements in medicine, technology,

and hygiene even though the actual life expectancy for individuals may vary depending on age,

sex, race, ethnicity, geographic location, etc. As a result, people worldwide likely live much

longer and enjoy better health than their earlier generations. According to the study (Cheng et

al., 2020), an aging population is the most dominant global demographic trend, and most

countries are experiencing growth in both the size and the share of older people in their

population. In the United States, the average life expectancy at birth for males (females) had

increased from 74.1 (79.3) years in 2000 to 76.3 (81.4) years in 2019 (National Center for Health

Statistics, 2023). Also, the average life expectancy at age 65 has grown. The aging population

Page 2 of 20

19

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

corresponds with the increasing proportion of older people living in retirement (Horner and

Cullen, 2016; Romig, 2023). These increases in longevity and living in retirement contribute to

increases in lifetime Social Security benefits for retirees. Thus, this demographic shift has led

to concerns about the financial stability of Social Security systems. In the United States, Social

Security is facing a long-term financing problem. It is projected that the Social Security program

will only be able to pay the benefits on time until 2037 (Social Security Administration, 2010).

Several developed countries have raised the age at which individuals are eligible to receive

pensions, resulting in workers delaying their retirement to relieve the financial challenges

faced by social security systems. In the United States, individuals born from 1943 to 1954 have

a full retirement age of 66, but the Social Security Amendments of 1983 required a gradual

increase in the full retirement age. Hence, the full retirement age for individuals born between

1955 and 1960 is set to increase by two months each year until it reaches 67 in 2027

(Springstead, 2011). In 2017, the Social Security Administration (SSA) began to increase the full

retirement age from 66 to 67 (Table 1). Consequently, individuals born in 1960 or later receive

the full retirement benefits at age 67 and reach the early retirement age of 62 in 2022. As the

full retirement age increases, the Social Security benefits are reduced for those who claim early,

and the benefits increase relatively small for those who delay (Tables 2 and 3). The SSA predicts

that increasing the full retirement age can encourage workers to postpone their retirement

decisions and remain longer in the labor force even if the early retirement age remains 62

(Bonsang et al., 2012).

Elderly individuals rely on Social Security benefits as a form of financial security in their late

years. Hence, it is beneficial for individuals to delay claiming Social Security benefits to ensure

a higher level of consumption during their later years after they have retired (Guo et al., 2020).

However, according to their study, a substantial portion of individual workers tended to claim

their retirement benefits shortly after becoming eligible for it. They explained that the official

designations of early and full retirement ages may serve as a provision for elderly workers to

retire at these ages. Workers may also interpret these designations as advice from the

government on when to retire from work and claim their benefits (Coile et al., 2002; Behaghel

and Blau, 2012; Eibich, 2015; Hessel, 2016). In the United States, despite a substantial reduction

in Social Security benefits (Table 2), a significant portion of newly retired workers (29.14% in

2021, as per Table 4) claim their Social Security benefits as soon as they turn 62, which is the

earliest age to claim these benefits. Additionally, the second largest share of newly retired

workers (25.47% in 2021, as per Table 4) claims their Social Security benefits at age 66, which

is the full retirement age. These large spikes in retirement at the early and full retirement ages

are well-documented in the literature.

Individuals are eligible to receive Social Security benefits as early as 62 but can wait as late as

70. However, as mentioned earlier, if an individual claims these benefits before the full

retirement age, the monthly benefits are permanently reduced based on the number of months

claimed earlier (Table 2). Before reaching the full retirement age, there is a monthly reduction

rate of five-ninths of 1% for each of the 36 months. It equals a 62⁄3% reduction each year. For

each month earlier than 36 months (3 years) before the full retirement age, the monthly

reduction rate is five-twelfths of 1%, equivalent to a 5% reduction each year (Li, 2022). For

instance, if the full retirement age is increased from 66 to 67 while the early retirement age

Page 3 of 20

20

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

remains at 62, then the monthly benefit reduction at age 62 would increase from 25% to 30%.

Also, the delayed benefit credit at age 70 would decrease from 32% to 24%

Table 1: Age to Receive Full Social Security Benefits

Year of Birth Full Retirement Age

(FRA)

Year the individual

Attains Age 62

Year the Individual Attains Full

Retirement Age

1943-1954 66 2005-2016 2009-2020

1955 66 and 2 months 2017 2021 or 2022

1956 66 and 4 months 2018 2022 or 2023

1957 66 and 6 months 2019 2023 or 2024

1958 66 and 8 months 2020 2024 or 2025

1959 66 and 10 months 2021 2025 or 2026

1960 or later 67 2022 or later 2027 or later

Source: Social Security Administration, Annual Statistical Supplement to the Social Security Bulletin, 2022.

https://www.ssa.gov/policy/docs /statcomps/supplement/. Note: Persons born on January 1 of any year should

refer to the previous year of birth.

Table 2: Full Retirement Age (FRA) and Maximum Reduction of Retired-Worker

Benefits, by Year of Birth

Year of

Birth

Year the

individual

Attains Age 62

Full Retirement

Age

Year the Individual

Attains FRA

Maximum

Reduction

Months

Maximum

Reduction at age

62

1943-1954 2005-2016 66 2009-2020 48 25.00%

1955 2017 66 and 2 months 2021 or 2022 50 25.83%

1956 2018 66 and 4 months 2022 or 2023 52 26.67%

1957 2019 66 and 6 months 2023 or 2024 54 27.50%

1958 2020 66 and 8 months 2024 or 2025 56 28.33%

1959 2021 66 and 10 months 2025 or 2026 58 29.17%

1960 or

later

2022 or later 67 2027 or later 60 30.00%

Source: Social Security Administration, Annual Statistical Supplement to the Social Security Bulletin, 2022.

https://www.ssa.gov/policy/docs /statcomps/supplement/. Table 2.A17.1. Note: Persons born on January 1 of

any year should refer to the previous year of birth. For each of the 36 months immediately preceding the FRA,

the monthly rate of reduction from the full retirement benefit is five-ninths of 1%. This equals a 6 2/3%

reduction each year. For each month earlier than three years (36 months) before the FRA, the monthly rate of

reduction is five-twelfths of of 1%. This equals a 5% reduction each year. The earliest a worker can claim

retirement benefits is age 62. For a worker with an FRA of 67, claiming benefits at 62 results in a 30% reduction

in their monthly benefit.

Table 3: Benefit Increase for Delayed Retirement by Birth Year

Year of

Birth

Full Retirement

Age

Year the

Individual

Attains FRA

Monthly

Credit

Annual

Credit

Maximum

Credit

Months

Total Credit

at Age 70+

1943-1954 66 2009-2020 2/3 of 1% 8% 48 32%

1955 66 and 2 months 2021 or 2022 2/3 of 1% 8% 46 30 2/3%

1956 66 and 4 months 2022 or 2023 2/3 of 1% 8% 44 29 1/3%

1957 66 and 6 months 2023 or 2024 2/3 of 1% 8% 42 28%

1958 66 and 8 months 2024 or 2025 2/3 of 1% 8% 40 26 2/3%

1959 66 and 10

months

2025 or 2026 2/3 of 1% 8% 38 25 1/3%

Page 4 of 20

21

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

1960 or

later

67 2027 or later 2/3 of 1% 8% 36 24%

Source: Social Security Administration, Annual Statistical Supplement to the Social Security Bulletin, 2022.

https://www.ssa.gov/policy/docs /statcomps/supplement/. Table 2A.17.3. Note: Persons born on January 1 of

any year should refer to the previous year of birth.

Table 4: Distribution of Claiming Ages for Newly Retired Workers in 2021

Total Men Women

Age Number Proportion Number Proportion Number Proportion

Total 2,645,830 100.00% 1,318,462 100.00% 1,327,368 100.00%

62 770,971 29.14% 371,142 28.15% 399,829 30.12%

63 179,341 6.78% 87,359 6.63% 91,982 6.93%

64 203,634 7.70% 95,878 7.27% 107,756 8.12%

65 342,176 12.93% 169,540 12.86% 172,636 13.01%

66 673,808 25.47% 360,536 27.35% 313,272 23.60%

67 107,484 4.06% 59,051 4.48% 48,433 3.65%

68 63,232 2.39% 34,531 2.62% 28,701 2.16%

69 58,151 2.20% 30,619 2.32% 27,532 2.07%

70+ 261,433 9.88% 123,149 9.34% 138,284 10.42%

Source: Social Security Administration, Annual Statistical Supplement to the Social Security Bulletin, 2022.

https://www.ssa.gov/policy/docs /statcomps/supplement/. Table 6.A4. Note: New retired workers in 2021, in

which the full retirement age was 66 years and 10 months. This table excluded the 540,353 beneficiaries who

automatically converted from disability benefits to retirement benefits at the full retirement age. They made up

16.96% of total beneficiaries in 2021.

(Tables 2 and 3). Consequently, the increase in the full retirement age from age 66 to 67 is

equivalent to reducing the Social Security benefits for new retirees, regardless of the age at

which a worker claims the benefits.

The Social Security Administration (SSA) expects the reduction in Social Security benefits might

encourage workers to delay their retirement and stay longer in the labor force. Also, the SSA

predicts that reducing and delaying benefits due to the increase in the full retirement age can

restore some financial balance to the Social Security system. Therefore, studying the effects of

increasing the full retirement age from 66 to 67 on the likelihood of individuals delaying their

retirement decision at the early retirement age of 62 is essential for planning future social

security reforms. It is important because a significant proportion of newly retired workers in

the United States have chosen the retirement decision before or at the early retirement age of

62.

This study employs Two-Stage Least Squares (2SLS) regression analysis, utilizing cross- sectional and time-series data from the Behavioral Risk Factor Surveillance System (BRFSS)

from 2016 to 2021 for adults aged 55-74. The BRFSS is a health survey conducted

collaboratively by the Centers for Disease Control and Prevention (CDC) and all states in the

United States. The BRFSS is a monthly telephone survey designed to collect uniform data on

health-related risk behaviors, chronic diseases, health conditions, access to health care, and use

of preventive services among non-institutionalized adults in the United States. The data is state- specific to each state and relates to the leading causes of death and disability among individuals

aged 18 years or older. Section 3 presents a detailed analysis of the BRFSS data.

Page 5 of 20

22

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

ANALYTICAL FRAMEWORK

Individuals can receive Social Security retirement benefits as early as age 62. However, if they

choose to start receiving the benefits before, they reach their full retirement age, their benefits

are reduced by a small percentage each month. Currently, the full retirement age is gradually

increasing from 66 to 67 by adding two months every year, resulting in a reduction of 25% to

30% in monthly benefits if an individual claims the benefits at age 62 (Table 2). As described in

section 1, a significant portion of newly retired workers choose to retire at or before this age.

Therefore, it is crucial to investigate the likelihood of workers delaying their retirement

decisions at age 62 due to the increase in the full retirement age. The impacts of increasing the

full retirement age on older workers’ retirement decisions at age 62 are developed as follows.

R = f(d, z) (1)

The retirement status of respondents at the time of the survey interview is denoted by ‘R’. ‘d’

represents health status, socio-demographic variables (income level, marital status,

race/ethnicity, and educational achievement), and age of respondents who are 62 years old. ‘z’

is a vector of all other relevant factors that affect retirement decisions. As described earlier, if

the full retirement age increases, then workers face financial disincentives to retire at the early

retirement age of 62. If a worker expects to receive a certain amount of the monthly Social

Security benefits, the worker likely delays his/her retirement. Therefore, the probability of

retiring at the early retirement age of 62 would decrease as the full retirement age increases by

two months each year. The following is a linear probability model to estimate retirement

decisions.

Rijt = β0 + hijtβh + xijtkβk + age62tβa + sjδj + ttδt + uijt (2)

The subscript i denotes each observation. The binary variable Rijt indicates the retirement

status of an individual respondent i in state j during the interview year t. Individual workers’

health status is one of the important determinants to make a retirement decision (Horner and

Cullen, 2016; Zhu, 2016; Son, 2020). hijt represents the poor physical or mental health days of

individual i in state j at interview year t. The vector xijtk comprises personal characteristics,

including demographic variables, where k represents the number of explanatory variables, and

age62t is a binary variable and equals 1 if the respondent’s age is 62 years at interview year t.

sj is a state fixed-effect, and tt represents the interview year. β0 is a fixed non-random constant,

and βh is the coefficient of the poor health status. βk is a coefficient matrix of the personal

characteristics with a dimension k x 1, and βa is a coefficient matrix of the respondents aged 62

years with a dimension t x 1. δj is the coefficient of the individual state j, and δt represents the

coefficient of the interview year t. uijt is an error term of the ith observation in state j at year t.

It is an independent random variable with mean zero and constant standard error σu, and that

is generally assumed that Cov (rijt, uijt) = 0 and Cov (xijtk, uijt) = 0 (Mesbah, 2004; Jia and Lubetkin,

2009; Son and Lee, 2014) with expected value of uijt, E(uijt) = 0.

In the above Ordinary Least Squares (OLS) estimation, equation (2), if the poor health status

(hijt) is correlated with the error term (uijt) due to endogeneity, the OLS estimator of βh is

inconsistent. Theoretically, the poor health status (hijt) could be endogenous in equation (2)

because of omitted variables, such as unobserved individual heterogeneities (i.e., genetics, life

Page 6 of 20

23

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

history, time preference, financial stability, or unobservable health status) that could have an

impact on both retirement and health outcomes. In addition, the econometric concern, of

whether retirement is associated with health status, could result in reverse causality in the

relationship between retirement and health status in equation (2). It means that poor health

could influence people to retire early, whereas retirement could affect the retirees’ health,

leading to endogeneity (Motegi et al., 2016; Zhu, 2016; Son, 2020). As a result, the estimates of

the effects of poor health status on retirement are likely to be biased upward, overestimating

the negative effects of poor health on retirement. Therefore, the OLS regression coefficient (βh)

from equation (2) is likely inconsistent.

To account for the issues of endogeneity and potential reverse causality in equation (2), this

study utilizes the Instrumental Variable (IV) estimation method. This method is a well- established and proven approach in the literature and is relevant to the data analyzed in this

paper. Previous studies (Eibich, 2015; Godard, 2016; Hessel, 2016; Motegi et al., 2016; Zhu,

2016) have suggested that health status is a core factor affecting retirement decisions and that

instrumental variables could be used to estimate its impact. A good instrument should be

strongly correlated with poor health status, but it should not directly influence the retirement

decision. To this end, this paper uses the health behaviors of individual respondents, such as

exercise, smoking, and drinking, as the identifying instruments for poor health status in IV

estimation.

In the first stage, the health outcomes are estimated using the following equation.

hijt = β0 + hbijtβhb + xijtkβk + sjδj + ttδt + uijt (3)

The instrument used for measuring health outcomes (hijt) is hbijt, which is equal to 1 if an

individual respondent i reported participating in physical exercise, smoking, or drinking at

interview year t. The coefficient of individual state j is δj, and the coefficient of interview year t

is δt. This study replaces the poor health status (hijt) in equation (2) with the predicted value of

poor health outcomes (hijt) from equation (3). As a result, the empirical model becomes the

following.

Rijt = β0 + (predicted hijt)βh + xijtkβk + age62tβa + sjδj + ttδt + uijt (4)

In equation (4), the predicted hijt is exogenous to all observed and unobserved individual-level

characteristics. The state-fixed model controls for all time-invariant differences that remain

constant over time in the states, eliminates the impact of variations associated with other

relevant independent variables among states on the dependent variables, and accounts for the

effects of unobserved state-specific variables. For instance, according to the study (Knoll, 2011),

individuals in good health with higher socioeconomic status tend to work longer than those

who are less healthy and wealthy because financial concerns play a significant role in

retirement decisions. Hence, residents in states with different income levels exhibit distinct

behavioral aspects of retirement decisions. The year dummy variables control for the factors

that vary uniformly across the states a year.

Page 7 of 20

24

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

Table 5: Definition, Mean, and Standard Deviations of Variables

Variable Definition

Retirees Dichotomous variable that equals 1 if respondent is retired.

Poorhealth Number of days of poor physical health or mental health keep respondents from

doing their usual activities, such as self-care, work, or recreation during the past

30 days.

Income

Income 35 Respondent’s annual household income from all sources < $35,000.

Income 35-75 $35,000 ≤ respondent’s annual household income from all sources < $75,000.

Income 75+ Respondent’s annual household income from all sources ≥ $75,000.

Marital Status

Married Respondent is married or a member of an unmarried couple.

Divorced Respondent was divorced or separated.

Widowed Respondent was widowed.

Never married Respondent is never married.

Races

White, non-Hispanic Respondent is white only, non-Hispanic.

Black, non-Hispanic Respondent is black only, non-Hispanic.

Hispanic Respondent is Hispanic.

Others, non- Hispanic

Respondent is other race, non-white, non-black, and non-Hispanic.

Education

Less high school Respondent did not graduate high school

High school Respondent completed grade 12 or GED (high school graduate).

Some college Respondent completed college 1 year to 3 years (some college or technical

school).

College graduation Respondent completed college 4 years or more (college graduate).

Health behaviors

Exercise Dichotomous variable that equals 1 if respondent participates in any physical

activities or exercises such as running, calisthenics, golf, gardening, or walking for

exercise during the past month.

Current smoker Dichotomous variable that equals 1 if respondent now smokes cigarettes every

day or some days.

Current drinker Dichotomous variable that equals 1 if respondent had at least one drink of

alcohol in the past 30 days.

Age variables

Age62y2016 Respondent aged 62 years in 2016.

Age62y2017 Respondent aged 62 years in 2017.

Age62y2018 Respondent aged 62 years in 2018.

Age62y2019 Respondent aged 62 years in 2019.

Age62y2020 Respondent aged 62 years in 2020.

Age62y2021 Respondent aged 62 years in 2021.

Ageunder62y2016 Respondent aged 62 years or under in 2016.

Ageunder62y2017 Respondent aged 62 years or under in 2017.

Ageunder62y2018 Respondent aged 62 years or under in 2018.

Ageunder62y2019 Respondent aged 62 years or under in 2019.

Ageunder62y2020 Respondent aged 62 years or under in 2020.

Page 8 of 20

25

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

Ageunder62y2021 Respondent aged 62 years or under in 2021.

Source: 2016-2021 Behavioral Risk Factor Surveillance System (BRFSS) at https://www.cdc.gov/brfss/. Notes:

One drink is equivalent to a 12-ounce beer, a 5-ounce glass of wine, or a drink with one shot of liquor.

EMPIRICAL IMPLEMENTATION

The data used in the analysis was obtained from the Behavioral Risk Factor Surveillance System

(BRFSS) from 2016-2021, a nationwide health survey. The BRFSS is a uniform, state-specific

system of ongoing health-related telephone surveys. It collects data on health-related risk

behaviors, chronic diseases, health conditions, access to healthcare, and the use of preventive

services for non-institutionalized adults aged 18 or older residing in the United States. The

BRFSS is a collaborated project by the Center for Disease and Prevention (CDC) and all states in

the United States. The state health department manages the BRFSS operations and conducts the

interviews themselves or uses contractors. During the survey development, state health

departments collaborate with the CDC and follow guidelines provided by the states, with

technical assistance from the CDC. The BRFSS data is used by state health departments for

various purposes, including identifying demographic variations in health-related behaviors,

designing, implementing, and evaluating public health programs, addressing emergent and

critical health issues, proposing legislation for health initiatives, and measuring progress

toward state health objectives. This study uses the BRFSS data for adults aged 55-74 across all

50 states and Washington D.C. However, observations with missing values, unanswered

questions, don’t know/not sure responses, questions not asked, or refusals were excluded. Out

of the 2,145,061 observations, only 405,031 survey participants (18.88%) completed all the

questions about retirement status, physical or mental health status, socio-demographic

variables, and health behavior variables. The observations were analyzed separately in male

and female subgroups to study retirement decisions. Tables 5 and 6 contain descriptions and

summary statistics.

Table 6: Descriptive Statistics

Variable Overall (n = 405,031) Male (n = 163,685) Female (n = 241,346)

Retirees 0.359 (0.480) 0.364 (0.481) 0.356 (0.479)

Poor health 6.450 (10.222) 6.690 (10.516) 6.270 (9.983)

Income

Income 35 0.416 (0.493) 0.402 (0.490) 0.427 (0.495)

Income 35-75 0.290 (0.454) 0.2857 (0.452) 0.293 (0.455)

Income 75+ 0.294 (0.456) 0.312 (0.463) 0.280 (0.449)

Marital Status

Married 0.589 (0.492) 0.632 (0.482) 0.555 (0.497)

Divorced 0.219 (0.413) 0.209 (0.407) 0.226 (0.418)

Widowed 0.107 (0.308) 0.060 (0.238) 0.143 (0.350)

Never married 0.086 (0.280) 0.099 (0.298) 0.076 (0.265)

Races

White, non-Hispanic 0.736 (0.441) 0.728 (0.445) 0.743 (0.437)

Black, non-Hispanic 0.111 (0.314) 0.111 (0.314) 0.111 (0.315)

Hispanic 0.095 (0.293) 0.101 (0.301) 0.091 (0.287)

Others, non-Hispanic 0.058 (0.233) 0.061 (0.238) 0.055 (0.228)

Education

Page 9 of 20

26

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

Less high school 0.129 (0.335) 0.146 (0.353) 0.115 (0.319)

High school 0.272 (0.445) 0.280 (0.449) 0.267 (0.442)

Some college 0.333 (0.471) 0.311 (0.463) 0.351 (0.477)

College graduation 0.266 (0.442) 0.263 (0.441) 0.268 (0.443)

Health behaviors

Exercise 0.666 (0.472) 0.681 (0.466) 0.654 (0.476)

Current smoker 0.179 (0.384) 0.197 (0.398) 0.166 (0.372)

Current drinker 0.473 (0.499) 0.523 (0.450) 0.434 (0.496)

Age variables

Age62y2016 0.011 (0.105) 0.012 (0.108) 0.011 (0.102)

Age62y2017 0.011 (0.102) 0.010 (0.101) 0.011 (0.103)

Age62y2018 0.011 (0.103) 0.011 (0.104) 0.011 (0.102)

Age62y2019 0.012 (0.108) 0.012 (0.110) 0.012 (0.107)

Age62y2020 0.010 (0.099) 0.010 (0.099) 0.010 (0.100)

Age62y2021 0.010 (0.100) 0.010 (0.101) 0.010 (0.100)

Ageunder62y2016 0.082 (0.275) 0.086 (0.281) 0.079 (0.270

Ageunder62y2017 0.085 (0.279) 0.086 (0.280) 0.085 (0.278)

Ageunder62y2018 0.084 (0.278) 0.088 (0.284) 0.081 (0.273)

Ageunder62y2019 0.085 (0.279) 0.083 (0.277) 0.086 (0.281)

Ageunder62y2020 0.077 (0.266) 0.075 (0.263) 0.078 (0.268)

Ageunder62y2021 0.073 (0.261) 0.072 (0.259) 0.074 (0.262)

Notes: Percentages are calculated with the BRFSS sample weight factor. All percentages are statistically

significant using two tailed t-test with 95% confidence interval (CI).

Retirement Status

The retirement decision is one of the most crucial decisions an individual makes in their lifetime

because it can significantly impact their well-being for years to come, especially given the

increase in life expectancy. Retirement decisions are highly personal and consider health and

financial factors due to the significant loss of income (Knoll, 2011). Thus, most individuals

consider the eligible age for Social Security benefits when deciding whether to retire or not

since it has a significant impact on their financial situation. In this study, the retirement status

of the survey respondents was self-reported in the employment status section of the BRFSS

demographic data. The survey asked respondents about eight categories of employment status:

employed for wages, self-employed, out of work for one year or more, out of work less than one

year, a homemaker, a student, retired, and unable to work. Respondents were classified as

retired if they reported being retired at the interview. 36.40% of male respondents and 35.60%

of female respondents reported being retired (Table 6).

Explanatory Variables

Poor health status, socio-demographic variables (income level, marital status, race/ethnicity,

and education level), health behavioral variables (exercise, current smoker, and current

drinker), and year-age-related variables are independent variables.

Health Status:

A study conducted in 2016 by Leinonen et al. found that health was the primary concern for

individuals when deciding to retire, particularly in the case of early retirement. The present

Page 10 of 20

27

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

study drew information about physical and mental health from the health-days section of the

BRFSS. The survey included questions about the number of days in the last 30 days when

respondents experienced poor physical health, including physical illness and injury, and poor

mental health, which encompassed stress, depression, and emotional problems. The survey

then inquired about the number of days during the previous month when poor physical or

mental health prevented individuals from performing their regular activities, such as self-care,

work, or recreation. Participants reported the number of poor healthy days ranging from 0 to

30 days, with 30 days being the logical maximum per month. The average poor health days for

overalls, males, and females were 6.45, 6.69, and 6.27 days, respectively, in the last 30 days.

Socio-demographic Variables:

The study conducted by Gough et al. in 2008 analyzed the relationship between income and

retirement decisions in the UK and Italy from 1992 to 2002. They revealed that income was a

strong predictor of retirement decisions. Their results showed that people from high-income

groups in both countries tended to retire earlier, while those from low-income groups retired

later. The BRFSS asked participants about their annual household income from all sources,

ranging from less than $10,000 to $75,000 or more (2016 BRFSS – 2020 BRFSS) and from less

than $10,000 to $200,000 or more (2021 BRFSS). This study re-categorized the income levels

into three groups: less than $35,000, $35,000 to less than $75,000, and $75,000 or over. In the

income category, 41.60% of respondents reported an annual household income of less than

$35,000, 29.00% reported an income of $35,000 to less than $75,000, and 29.40% reported an

annual household income of $75,000 or more. The proportion of male respondents reporting a

household income of $75,000 or over was higher than the proportion of female respondents

reporting the same income level, as shown in Table 6.

The female labor force participation rate has significantly increased in recent decades, leading

to changes in retirement planning for couples (Boehnke and Gay, 2022). In the past, couples

typically focused on the husband’s retirement benefits, and his retirement decision would

impact the couple’s future benefits as a single decision unit. However, with women at present

accumulating substantial benefits in their names, retirement decisions have become more

challenging to coordinate. More than two-thirds of individuals of retirement age live as couples,

and the majority of the couples are dual-earners. As a result, many couples now need to

consider the impact of one spouse stopping work on both income and retirement benefits.

According to a study (Stancanelli, 2017), couples tend to retire at similar time, within a year of

each other. Since husbands are generally older than their wives, the increased female labor

force participation may lead to later retirement for men. Therefore, marital status would be an

important factor in an individual’s retirement decision. In the BRFSS, survey participants are

asked about six categories of marital status: married, divorced, widowed, separated, never

married, and a member of an unmarried couple. They are then classified into four marital

subgroups: married (married or a member of an unmarried couple), divorced (divorced or

separated), widowed, and never married. After regrouping, 58.9%, 21.9%, 10.7%, and 8.6% of

survey participants are married, divorced, widowed, and never married, respectively. The

proportion of married respondents is higher among males than among females, whereas the

proportion of widowed respondents is significantly lower among males than among females

(Table 6).

Page 11 of 20

28

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

According to a study conducted by Francis and Weller in 2021, there are disparities in

retirement security among households of different races and ethnicities. These disparities arise

due to factors, such as market discrimination, occupational segregation, and lower wages,

which lead to limited access to retirement savings and contributions. Furthermore, households

belonging to ethnic minorities have fewer chances of receiving intergenerational wealth

transfers, which could help alleviate debt burdens and enable access to education and

homeownership. These pathways are crucial for building wealth and retirement savings. This

study uses two questions from the BRFSS to determine race/ethnicity: “Which one of these

groups would you say best represents your race? (With seven response choices)” and “Are you

Hispanic or Latino? (yes/no)”. The BRFSS’s ‘calculated race variables’ section listed five

categories of race/ethnicity, which this study reduces to four: White, non-Hispanic; Black, non- Hispanic; Hispanic; and others (multi-racial or other race, non-Hispanic).

A study conducted by Venti and Wise in 2015 found that both men and women with less than a

high school degree were more likely to participate in Social Security Disability Insurance and

to claim Social Security benefits at an early age. In contrast, individuals with a college degree or

higher were less likely to rely on these benefits. In this study, the respondents are classified into

four levels of educational attainment: less than high school graduation, high school graduation,

attended some college or technical school, and four-year college graduation or more. 12.9%

have completed less than high school. In contrast, 27.2% have completed high school, 33.3%

attended some college, and 26.6% completed a four-year college degree or higher (Table 6).

Health Behavioral Variables:

It is well known that health behaviors are strongly associated with health outcomes. In the first

stage of the 2SLS regression model, three health behavioral factors (exercise, smoking, and

drinking) are independent variables. Exercise was defined as participating in any physical

activities or exercises other than job-related regular activities during the past month, including

running, calisthenics, golf, gardening, or walking for exercise. 66.6% of survey respondents

reported participating in such activities. Smoking was defined as those who smoked cigarettes

every day or some days. 17.9% of the respondents reported being current smokers. Drinking

was defined as a respondent who had at least one drink of alcohol in the past 30 days. One drink

was equivalent to a 12-ounce beer, a 5-ounce glass of wine, or a drink with one shot of liquor.

47.3% of the respondents reported being current drinkers.

Year-Age Group Variables:

During the survey, the participants were asked about their age, and the response was recorded

as a continuous variable. Two age-year variables were used in the empirical models: one for the

respondents who were 62 years old and the other for those who were 62 years old or younger.

ANALYSIS AND RESULTS

Descriptive Analysis

Table 6 presents descriptive statistics for the overall observations as well as for men and

women separately. The survey participants reported an average of 6.45 poor health days over

the past 30 days, taking into account the BRFSS sampling weighting factor. Interestingly, men

reported slightly more poor health days than women on average. Regarding annual household

income, 31.2% of male respondents reported an income of $75,000 or more, while 28.0% of

Page 12 of 20

29

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

female respondents reported the same income level. The table shows that 63.2% of male

respondents were married, compared with 55.5% of female respondents. Of those not married,

6% of male respondents were widowed, whereas 14.3% of female respondents were widowed.

Educational attainment did not show any significant gender differences. Finally, a high

proportion of male respondents engaged in all three health behaviors compared with female

respondents: exercise (68.1% vs. 65.4%), smoking (19.7% vs. 16.6%), and drinking (52.3% vs.

43.4%).

Table 7 displays the percentage of retirees among survey respondents who are 62 years old, 62

years old or under, 66 years old, and 66 years old or under. The table shows that the

proportions of retirees, who are 62 years old, for males and females are similar. However, for

all other age groups, the proportions of retirees among males are higher than the fractions of

retirees among females. In other words, males are more likely than females to retire at ages 62

or under, 66 years old, and 66 years old or under. It implies that females tend to stay in the

labor force longer than males. Regarding the overall observations, the descriptive analysis

demonstrates that the direction of the proportions of retirees for each age group from 2016 to

2021 is unclear. It means that the impact of the increases in the full retirement age on delaying

retirement decisions is ambiguous based on descriptive analysis.

Table 7: Proportion of Retirees in Age Group Each Year

Overall Male Female

Aged 62 years

2016 0.250 (0.433) 0.267 (0.442) 0.234 (0.424)

2017 0.270 (0.444) 0.261 (0.439) 0.276 (0.447)

2018 0.242 (0.429) 0.266 (0.442) 0.223 (0.416)

2019 0.231 (0.421) 0.251 (0.434) 0.213 (0.410)

2020 0.215 (0.411) 0.211 (0.408) 0.219 (0.413)

2021 0.233 (0.423) 0.209 (0.407) 0.252 (0.434)

Aged 62 years or under

2016 0.122 (0.328) 0.134 (0.340) 0.113 (0.317)

2017 0.122 (0.327) 0.126 (0.332) 0.119 (0.324)

2018 0.124 (0.330) 0.132 (0.338) 0.118 (0.323)

2019 0.120 (0.325) 0.130 (0.336) 0.112 (0.315)

2020 0.122 (0.327) 0.140 (0.347) 0.108 (0.310)

2021 0.116 (0.320) 0.112 (0.316) 0.118 (0.323)

Aged 66 years

2016 0.577 (0.494) 0.558 (0.497) 0.593 (0.491)

2017 0.537 (0.499) 0.588 (0.492) 0.499 (0.500)

2018 0.563 (0.496) 0.606 (0.489) 0.528 (0.499)

2019 0.579 (0.494) 0.584 (0.493) 0.574 (0.495)

2020 0.599 (0.490) 0.603 (0.489) 0.596 (0.491)

2021 0.527 (0.499) 0.519 (0.500) 0.533 (0.499)

Aged 66 years or under

2016 0.214 (0.410) 0.220 (0.414) 0.209 (0.406)

2017 0.206 (0.404) 0.212 (0.409) 0.200 (0.400)

2018 0.213 (0.409) 0.223 (0.416) 0.205 (0.404)

Page 13 of 20

30

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

2019 0.206 (0.404) 0.216 (0.411) 0.199 (0.399)

2020 0.215 (0.411) 0.225 (0.417) 0.208 (0.406)

2021 0.200 (0.400) 0.194 (0.396) 0.204 (0.403)

Notes: Percentages are calculated with the BRFSS sample weight factor. All percentages are statistically

significant using two tailed t-test with 95% confidence interval (CI).

Second Stage Regression Results

Table 8 presents the empirical results obtained from the 2SLS regression analysis with respect

to the probability of retirement for adults aged between 55 and 74. The empirical models

include poor health status, socio-demographic variables, age variables, state-fixed effects, and

year-fixed effects. The t-statistics, shown in the parenthesis for each coefficient, indicate that

all estimated coefficients are statistically significant at a 95 percent confidence interval for the

two-tailed test. The last row of the table displays the R-squared values, which indicate the

overall model fit, i.e., how well the estimated equation explains the observations in the sample.

Since the core determinants of retirement decisions, such as job characteristics, retirement

preferences, social participation, functional limitation, etc. (Venti and Wise, 2015; Boehnke and

Gay, 2022; Oduro et al., 2023) are absent in the models, lower R-squared values are expected.

The R-squared values, ranging from 0.047 to 0.261, are relatively low, suggesting that the

regression models are not a good fit for the data based on conventional regression analysis.

However, this study primarily focused on examining the retirement decisions for retirement- aged adults at the early retirement age of 62 due to the increase in full retirement age rather

than investigating the impact of the key factors on retirement decisions.

Poor Health:

Individual workers who have experienced poor health throughout their lifetime may retire

earlier or later than their plan. If the workers anticipate declines in expected annual earnings,

they might pursue early retirement. Conversely, if they expect an increase in expected income,

they might decide to stay longer in the labor market and retire later to compensate for their

past loss of earnings.

A systematic review conducted by Homaie Rad et al. in 2017 examined the effects of poor health

on early retirement. Their results showed a strong association between poor health and early

retirement, indicating that poor health increased the likelihood of early retirement. However,

from the results of the 2SLS regression analysis in Table 8, the signs of the estimated coefficients

for poor health are negative. It implies that poor health lowers the likelihood of reporting being

retired at the time of the interview. Although the magnitude of the effects is small, with the

estimated coefficients ranging from -0.009 to -0.01, the adverse effects remain consistent

across six empirical models.

Table 8: Second Stage Regression Results for Retirement Decision

Overall Male Female

Poorhealth -0.010

(-30.195)

-0.009

(-31.618)

-0.009

(-17.475)

-0.009

(-19.029)

-0.010

(-23.276)

-0.009

(-24.256)

Income

Income 35 - - - - - -

Income 35-75 -0.019

(-8.436)

-0.014

(-6.901)

-0.014

(-3.751)

-0.014

(-4.377)

-0.022

(-7.687)

-0.015

(-5.874)

Page 14 of 20

31

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

Income 75+ -0.187

(-68.159)

-0.115

(-47.443)

-0.213

(-48.363)

-0.145

(-37.051)

-0.167

(-47.558)

-0.093

(-30.047)

Marital Status

Married - - - - - -

Divorced -0.071

(-35.036)

-0.049

(-27.323)

-0.072

(-22.637)

-0.038

(-13.556)

-0.069

(-25.805)

-0.057

(-24.226)

Widowed 0.102

(41.118)

0.030

(13.619)

0.115

(23.896)

0.056

(13.050)

0.110

(37.284)

0.028

(10.743)

Never married -0.101

(-35.819)

-0.043

(-17.303)

-0.129

(-30.988)

-0.055

(-14.825)

-0.081

(-20.944)

-0.039

(-11.355)

Races

White, non-Hispanic - - - - - -

Black, non-Hispanic -0.038

(-12.476)

-0.015

(-5.556)

-0.038

(-7.708)

-0.016

(-3.699)

-0.037

(-9.714)

-0.013

(-3.898)

Hispanic -0.110

(-27.507)

-0.055

(-15.576)

-0.107

(-17.295)

-0.054

(-9.754)

-0.113

(-21.406)

-0.056

(-12.063)

Others, non- Hispanic

-0.060

(-17.398)

-0.027

(-8.893)

-0.054

(-10.314)

-0.022

(-4.804)

-0.069

(-15.005)

-0.034

(-8.347)

Education

Less high school - - - - - -

High school 0.077

(22.568)

0.067

(22.393)

0.061

(11.978)

0.053

(11.750)

0.096

(20.949)

0.085

(21.126)

Some college 0.117

(34.394)

0.088

(29.404)

0.115

(22.311)

0.074

(16.102)

0.130

(28.598)

0.110

(27.369))

College graduation 0.159

(44.827)

0.107

(34.187)

0.151

(28.062)

0.079

(16.538)

0.177

(37.209)

0.138

(33.101)

Age variables

Age62y2016 -0.132

(-18.305)

-0.135

(-11.777)

-0.131

(-14.041)

Age62y2017 -0.140

(-18.657)

-0.152

(-12.717)

-0.133

(-13.712)

Age62y2018 -0.152

(-19.493)

-0.138

(-11.525)

-0.162

(-15.810)

Age62y2019 -0.159

(-20.191)

-0.158

(-13.080

-0.160

(-15.453)

Age62y2020 -0.159

(-18.304)

-0.145

(-10.745)

-0.169

(-14.882)

Age62y2021 -0.165

(-19.603)

-0.183

(-13.957)

-0.153

(-13.906)

Ageunder62y2016 -0.462

(-150.292)

-0.468

(-95.132)

-0.461

(-116.983)

Ageunder62y2017 -0.462

(-145.760)

-0.457

(-90.750)

-0.466

(-114.655)

Ageunder62y2018 -0.468

(-143.983)

-0.464

(-91.552)

-0.473

(-111.619)

Ageunder62y2019 -0.470

(-141.235)

-0.459

(-88.253)

-0.479

(-110.675)

Ageunder62y2020 -0.473

(-131.476)

-0.458

(-80.268)

-0.483

(-104.584)

Ageunder62y2021 -0.481

(-137.826)

-0.469

(-85.635)

-0.491

(-108.383)

Year dummy yes yes yes yes yes Yes

Page 15 of 20

32

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

State dummy yes yes yes yes yes Yes

R-square 0.048 0.257 0.052 0.254 0.047 0.261

F-statistics 278.089 1916.878 122.716 764.389 164.009 1168.035

Observation 405,031 405,031 163,685 163,685 241,346 241,346

Notes: The t-statistics are in parentheses. All estimated coefficients are statistically significant at 95 percent level

of confidence except.

Demographic Variables:

Three levels of annual household income are in the empirical models, with an annual household

income of less than $35,000 serving as the reference for the models. The estimated coefficients

of the two other income levels, higher than $35,000, have negative signs and larger magnitudes

as the income level increases. It suggests that high-earning individuals likely stay in the labor

force longer and delay retirement. Especially respondents with an annual household income of

$75,000 or more are 18.7% less likely to report being retired than those with an annual

household income of less than $35,000. For men, the probability gap is even wider. Based on

the results, male respondents with an annual household income of $75,000 or more are 21.3%

less likely to report being retired than those with an annual household income of less than

$35,000.

The study (Whitaker and Bokemeier, 2018) examined the expected retirement age in the

United States due to demographic changes, such as an increase in life expectancy, a decline in

fertility rates, and decreased marriage rates. They explained that these changes have resulted

in the elderly population being more likely to spend a significant portion of their retirement

alone. The research found that the expected retirement age was significantly associated with

personal and family characteristics, especially marital status. In this study, the regression

results in Table 8 demonstrate that divorced or never married respondents, both men and

women, are less likely to be retired at the interview, whereas widowed respondents are more

likely to be retired than married respondents. However, the average age for the marital

category may bring in these results if age is positively associated with retirement. Based on the

estimated coefficients, respondents in higher average age categories of marital status (married:

62.23 years; divorced: 62.86 years; widowed: 65.91 years; and never married: 61.72 years) are

more likely to be retired at the interview. Especially the average age for widowed respondents

is 2.5 years higher than the average age for married respondents.

In a recent study conducted by Kim et al. in 2021, racial and ethnic differences in retirement

resources could be attributed to holding a retirement savings account. The study identified

homeownership, objective financial knowledge, planning horizon, and age as the main factors

that explained these differences. Their results showed that White respondents were more likely

to hold a retirement savings account compared with other racial and ethnic groups. It suggests

that racial disparities in retirement savings likely exist in the United States, affecting retirement

decisions. The estimated probabilities from the 2SLS regression analysis in Table 8 indicate that

White respondents are more likely to report being retired compared with other races, while

Hispanic respondents are the least likely to report being retired.

According to a study conducted by Hardy in 1984, the impact of education on occupational

status, career-entry position, and subsequent locations within the occupational structure was

significant. The study reported that workers with higher levels of education were more likely

Page 16 of 20

33

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

to delay retirement compared with those with lower levels of educational achievement.

Similarly, the empirical results of this study explain a negative correlation between education

achievement and retirement, meaning that an increase in education achievement leads to a

decrease in the probability of being retired. The average ages of respondents across different

education levels are similar (not shown in tables - less high school: 63.09 years; high school:

63.20 years; some college: 63.42 years; college graduation: 63.39 years).

Age-Year Variables:

This study investigates the retirement decision based on two age groups: respondents aged 62

years and respondents aged 62 years or under, as the full retirement age increases. The results

of the empirical analysis for respondents aged 62 years are presented in the first column of

Table 8. The probability of reporting being retired at age 62 decreased from 2016 to 2021,

indicating that individuals were choosing to delay retirement at age 62 as the full retirement

age increases. For overall observations, the probability of reporting themselves as retired for

respondents aged 62 years in 2016 is 0.132 less than the probability for other aged respondents

and 0.165 less than in 2021. For men aged 62 years, the estimated coefficient of the likelihood

of reporting being retired in 2016 is -0.135, and in 2021 it is -0.183. It implies that the likelihood

of reporting being retired decreased by 35.56% during the period. For women, the estimated

coefficient of the probability of reporting being retired in 2016 is -0.131, and in 2021 it is -

0.153. It explains the probability dropped by 16.79%. If both male and female workers exit from

the labor force only based on the Social Security benefits, the effects of the increase in full

retirement age on delaying retirement decisions for males appear to be greater than the effects

for females. These results suggest that the decline in the monetary value of Social Security

benefits due to the increase in the full retirement age can discourage individuals from retiring

at 62.

Table 8 also presents the estimated probabilities of being retired at the interview for

respondents aged 62 years or under in the second column. Signs of all estimated coefficients

are negatives, and the magnitude of these coefficients increases each year from 2016 to 2021.

It implies a gradual decline in the probability of reporting being retired at 62 or under since the

full retirement age increased by two months each year. For men, the estimated coefficient of

the likelihood of reporting being retired for this age category in 2016 is -0.462, and in 2021 it

is -0.481. The results also show no clear trend of changing the probability for men from 2016

to 2021, whereas the estimated coefficients for women consistently increase as the year passes.

DISCUSSION

In the United States, individuals born from 1943 to 1954 have a full retirement age of 66.

However, the Social Security Amendments of 1983 mandates a gradual increase in the full

retirement age. Currently, the full retirement age for individuals born from 1955 to 1960

increases by two-month increments each year until it reaches 67 years of age in 2027

(Springstead, 2011). Individuals are eligible for Social Security benefits anytime as early as 62

or as late as 70, with unreduced benefits available at the full retirement age. If an individual

files for Social Security benefits before reaching the full retirement age, the individual will

receive a permanently reduced monthly benefit based on the number of months claimed before

the full retirement age (Table 2). As a result, the increase in the full retirement age is equivalent

to a decline in the Social Security benefits for all new retirees. The Social Security

Page 17 of 20

34

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

Administration predicts that reducing Social Security benefits can encourage workers to

postpone their retirement and stay longer in the labor force. These could help to restore some

financial balance to the Social Security system. Therefore, studying the effects of increasing the

full retirement age from 66 to 67 on retirement decisions at 62 becomes essential for planning

future social security reforms. That is important because a significant proportion of workers

tend to exit the labor force at the age or even earlier.

This study conducts a Two-Stage Least Squares (2SLS) regression analysis to investigate the

likelihood of retirement for adults aged 62 years and adults aged 62 years or under, using cross- sectional and time-series data from 2016 to 2021. The full retirement age increases by two

months each year, beginning in 2017. Hence, Social Security benefits would be reduced by a

certain amount for individuals who claim the benefits at 62 after 2016. The empirical results

show that the probability of retirement for individuals at age 62 decreased from 2016 to 2021,

suggesting a trend of delaying retirement. During the same period, the likelihood of men

reporting being retired decreases by 35.56%, while for women, it drops by 16.79%. It indicates

that if male and female workers choose to exit the labor force solely based on the Social Security

benefits, the increase in full retirement age appears to have a more significant impact on

delaying retirement decisions for male workers than for female workers. The additional

findings regarding retirement decisions for adults aged 62 or under yield similar. In summary,

the empirical outcomes demonstrate that the decrease in the monetary value of Social Security

benefits due to the increase in the full retirement age can discourage individuals from retiring

at age 62 as well as age 62 or under.

It has been well documented that there are large spikes in retirement rates at both the early

and full retirement ages. Studies show that social security financial incentives have a significant

impact on these spikes, particularly at the age of 62 (Behaghel and Blau, 2012; Bonsang et al.,

2012; Eibich, 2015; Hessel, 2016). Similarly, in the United States, despite a large reduction in

Social Security benefits (Table 2), a significant proportion of newly retired workers (29.14% in

2021, as per Table 4) claim their Social Security benefits as soon as they turn 62, which is the

earliest age to claim these benefits. Remarkably, the empirical findings reveal that workers tend

to delay their retirement at 62 since the full retirement age started to increase. It appears that

the current social security reform has effectively achieved the goal of delaying individuals’

retirement decisions. Reducing benefits and delaying retirement-aged workers from claiming

benefits can provide financial respite to the Social Security program. Thus, given the current

trend of increasing life expectancy and Social Security imbalance, there may be proposals for

Social Security reforms to make further raise the retirement age.

The following are the reasonable supports for increasing the retirement age. First, recent

decades have seen an increase in the average life expectancy, implying that people can work up

until an older age. Second, older workers are now in better health conditions, which mean they

can continue working even at an older age. Third, the job characteristics in many industries

have become more suitable for older workers. However, there are concerns regarding the

increase in the retirement age due to differences in life expectancy among different

socioeconomic groups (Singh and Lee, 2021). Additionally, improvements in health status, job

characteristics, and an increase in life expectancy may not be equally distributed among

individuals with different attributes, such as sex, race, educational attainment, or income level.

Page 18 of 20

35

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

Furthermore, workers with health problems may be encouraged to apply for Social Security

disability benefits, leading to higher enrollment and costs for the program. Lastly, older

workers may become more vulnerable to the risk of unemployment as they would no longer be

eligible for Social Security retirement benefits.

References

Behaghel, Luc and Blau, David M. (2012). Framing social security reform: behavioral responses to changes in the

full retirement age. American Economic Journal: Economic Policy, Vol. 4 (4), pp. 41-67.

http://dx.doi.org/10.1257/pol.4.4.41.

Boehnke, Jörn and Gay, Victor. (2022). The missing men. World war I and female labor force participation.

Journal of Human Resources, Vol. 57 (4), pp. 1209-1241. doi: https://doi.org/10.3368/jhr.57.4.0419-10151R1.

Bonsang, Eric; Adam, Stéphane; and Perelman, Sergio. (2012). Does retirement affect cognitive function? Journal

of Health Economics, Vol. 31, No. 3, pp. 490-501. https://doi.org/10.1016/j.jhealeco.2012.03.005.

Cheng, Xunjie; Yang, Yang; Schwebel, David C.; Liu, Zuyun; Li, Li; Cheng, Peixia; Ning, Peishan; and Hu, Guoqing.

(2020). Population ageing and mortality during 1990-2017: a global decomposition analysis. PLOS Medicine, Vol.

17 (6): e1003138. https://doi.org/10.1371/journal.pmed.1003138.

Coile, Courtney; Diamonf, Peter; Gruber, Jonathan; and Jousten, Alain. (2002). Delays in claiming social security

benefits. Journal of Public Economics, Vol. 84 (3), pp. 357-385. https://doi.org/10.1016/S0047-2727(01)00129-

3.

Eibich, Peter. (2015). Understanding the effect of retirement on health: Mechanisms and heterogeneity. Journal

of Health Economics, Vol. 43, (Sept.), pp.1-12. doi: 10.1016/j.jhealeco.2015.05.001.

Francis, Dania V. and Weller, Christian E. (2021). Retirement inequality by race and ethnicity. Public Policy &

Aging Report, Vol. 31 (3), pp.83-88. https://doi.org/10.1093/ppar/prab009.

Godard, Mathilde. (2016). Gaining weight through retirement? Results from the SHARE survey. Journal of Health

Economics, Vol. 45, pp. 27-46. https://doi.org/10.1016/j.jhealeco.2015.11.002.

Gough, Orla; Adami, Roberta; and Waters, James. (2008). The effects of age and income on retirement deciaions:

A comparative analysis between Italy and the UK. Pensions, Vol. 13, pp. 167-175. doi: 10.1057/pm.2008.12.

Guo, Rui; Sun, Wei; Wang, Jianqiu; and Xiao, Gang. (2020). Why do retired workers claim social security benefits

so early? A potential explanation based on the cumulative prospect theory. Applied Economics, Vol. 52 (5), pp.

490-505. https://doi.org/10.1080/00036846.2019.1646874.

Hardy, Melissa A. (1884). Effects of education on retirement among White male wage- and salary workers.

Sociology of Education, Vol. 57 (2), pp. 84-98. https://doi.org/10.2307/2112631.

Hessel, Philipp. (2016). Does retirement (really) lead to worse health among European men and women across

all educational levels? Social Science & Medicine, Vol. 151, Issue C, pp. 19-26.

https://doi.org/10.1016/j.socscimed.2015.12.018.

Homaie Rad, Enayatollah; Rashidian, Arash; Arab, Mohamad; and Souri, Ali. (2017). Comparison the effect of

poor health and low income on early retirement: a systematic review and meta-analysis. Industrial Health,

Vol.55, pp. 306-313. doi: 10.2486/indhealth.2017-0010.

Horner, Elizabeth Mokyr and Cullen, Mark R. (2016). The impact of retirement on health: quasi-experimental

methods using administrative data. BMC Health Services Research, Vol. 16:68, pp. 1-9.

https://doi.org/10.1186/s12913-016-1318-5.

Page 19 of 20

36

Archives of Business Research (ABR) Vol. 12, Issue 4, April-2024

Services for Science and Education – United Kingdom

Jia, H. and Lubetkin, E. I. (2009). Time trends and seasonal patterns of health-related quality of life among U.S.

adults. Public Health Report, Vol. 124 (5), pp. 692-701. doi: 10.1177/003335490912400511.

Kim, Kyoung Tae; Cho, Soo Hyun; and DeVaney, Sharon A. (2021). Racial/ethnic difference in holding a

retirement saving motive. A decomposition analysis. The Journal of Consumer Affairs, Vol. 55, pp. 464-482. doi:

10.1111/joca.12338.

Knoll, Melissa A. Z. (2011). Behavioral and psychological aspects of the retirement decision. Social Security

Bulletin, Vol. 71 (4), pp. 15-32. https://www.ssa.gov/policy/docs/ssb/v71n4/v71n4p15.html.

Leinonen, Taina; Laaksonen, Mikko; Chandola, Tarani; and Martikainen, Pekka. (2016). Health as a predictor of

early retirement before and after introduction of a flexible statutory pension age in Finland. Social Science &

Medicine, Vol. 158, pp. 149-157. https://doi.org/10.1016/j.socscimed.2016.04.029.

Li, Zhe. (2022). The Social Security Retirement Age. CRS Report, Congressional Research Service.

https://crsreports.congress.gov. R44670.

Mesbah, Mounir. (2004). Measurement and analysis of health-related quality of life and environment data.

Environmetrics, Vol. 15, pp. 471-481. https://doi.org/10.1002/env.680.

Motegi, Hiroyuki; Nishimura, Yoshinori; and Terada, Kazuyuki. (2016). Does retirement change lifestyle habits?

The Japanese Economic Review, Vol. 67 (2), pp. 169-191. doi:10.1111/jere.12104.

National Center for Health Statistics. (2023). Health, United States, 2020-2021: Table LExpMort. Hyattsville, MD.

Available from: https://www.cdc.gov/nchs/hus/data-finder.htm.

Oduro, Joseph Kojo; Okyere, Joshua; and Nyador, Jonas Kwame Mawuli. (2023). Risky health behaviours and chronic

conditions among aged persons: analysis of SAGE selected countries. BMC Geriatrics, Vol. 23 (1): 145.

https://doi.org/10.1186/s12877-023-03836-y.

Romig, Kathleen. (2023). Raising social security’s retirement age would cut benefits for all new retirees: harm would

fall disproportionately on seniors with low income. Center on Budget and Policy Priorities, JSTOR.

http://www.jstor.org/stable/resrep49189.

Singh, Gopal K. and Lee, Hyunjung. (2021). Marked disparities in life expectancy by education, poverty level,

occupation, and housing tenure in the United States, 1997-2014. International Journal of Maternal and Child

Health and AIDS, Vol. 10 (1), pp. 7-18. doi: 10.21106/ijma.402.

Social Security Administration. (2010). Celebrating the 75th Anniversary of the Social Security Act. Social Security

Bulletin, Vol. 70 (3). 500 E. Street, SW, 8th Floor, Washington, DC 20254-0001.

Social Security Administration. (2022) The Annual Statistical Supplement to the Social Security Bulletin, 2022. 250 E.

Street SW, 8th Floor, Washington, DC 20254. ttps://www.ssa.gov/policy/docs /statcomps/supplement/.

Son, Chong-Hwan. (2020). The effects of retirement on health-related quality of life of retirement-aged adults for

four marital status subgroups. International Journal of Economic Sciences, Vol. IX (1), pp. 179-201. doi:

10.20472/ES.2020.9.1.010.

Son, C.H. and Lee, D.H. (2014). The effects of alcohol consumption patterns on cardiovascular disease: An

empirical study with Behavioral Risk Factor Surveillance System data. The Empirical Economics Letters, Vol. 13

(7), pp. 723-730.

Springstead, Glenn R. (2011). Distribution effects of accelerating and extending the increase in the full retirement

age. Policy Brief, No. 2011-01.

Page 20 of 20

37

Son, C. H. (2024). The Effects of Increasing the Full Retirement Age from 66 to 67 on the Retirement Decision at the Early Retirement Age of 62.

Archives of Business Research, 12(4). 18-37.

URL: http://doi.org/10.14738/abr.124.16774

Stancanelli, Elena. (2017). Couples’ retirement under individual pension design: A regression discontinuity study

for France. Labour Economics, Vol. 49, pp. 14-16. https://doi.org/10.1016/j.labeco.2017.08.009.

Venti, Steven and Wise, David A. (2015). The long reach of education: early retirement. The Journal of the

Economics of Aging, Vol. 6, pp. 133-148. doi: 10.1016/j.jeoa.2015.08.001.

Whitaker, Elizabeth Ann and Bokemeier, Janet L. (2018). Spousal, family and gender effects on expected

retirement age for married pre-retirees. Journal of Family and Economic Issues, Vol. 39, pp. 371-385.

https://doi.org/10.1007/s10834-018-9564-9.

Zhu, Rong. (2016). Retirement and its consequences for women’s health in Australia. Social Science & Medicine,

Vol. 163, (Aug.), pp. 117-125. doi: 10.1016/j.socscimed.2016.04.003.