The effects of retirement on physical and mental health outcomes.
Dave, Dhaval ; Rashad, Inas ; Spasojevic, Jasmina 等
1. Introduction
Despite rising life expectancy, the average age at retirement has
been declining over the past four decades. Social Security data indicate
that the retirement age for men declined from 68.5 to 62.6 years, and
that for women declined from 67.9 to 62.5 years (Gendell 2001). (1) In a
recent study, Gruber and Wise (2005) note that many countries have
benefit structures that discourage work by lowering lifetime benefits to
people who work longer. There are strong incentives to retire built into
the U.S. Social Security system as well as many private pensions
(Quadagno and Quinn 1997). With an aging population retiring earlier and
an unfunded liability facing both Social Security and Medicare, policy
makers have pressed for several reforms including an increase in the
retirement age. (2)
Whether early retirement is individually or socially optimal
depends on how retirement affects subsequent health status, among other
things. While numerous studies have examined the effects of changes in
health on retirement behavior, research on how retirement impacts health
status has been sparse. Using seven longitudinal waves of the Health and
Retirement Study (HRS), spanning 1992 through 2005, the objective of
this study is to analyze the effects of full retirement on outcomes
related to physical and mental health. We are careful in noting that the
effect we are analyzing is not that of retirement per se, but rather the
change in environment that encompasses retirement, leading an individual
to invest more or less in their health. While we distinguish voluntary
versus involuntary retirement, the behavioral framework suggests that
even if retirement is voluntary, individual investments in health may
respond to changes in incentives post-retirement. If retirement improves
health outcomes, then evaluation of policies that prolong retirement
should account for the effect on health.
Results indicate that retirement has adverse health effects for the
average individual. Specifically, complete retirement leads to a 5-14%
increase in difficulties associated with mobility and daily activities,
4-6% increase in illness conditions, and 6-9% decline in mental health,
over an average post-retirement period of six years. (3) Further tests
suggest that the effects tend to operate through lifestyle changes
including declines in physical activity and social interactions. The
adverse health effects are mitigated if the individual is married and
has social support, continues to engage in physical activity
post-retirement, or continues to work part-time upon retirement. Some
evidence also suggests that the adverse effects of retirement on health
may be larger in the event of involuntary retirement.
2. Relevant Studies
The decision to retire is affected by numerous factors, including
availability of health insurance, Social Security eligibility, financial
resources, and spousal interdependence. Several studies also point to
health status as a significant determinant. Workers in poor
health--those who suffer from activity limitations and chronic health
conditions--are found to retire earlier than those who are healthy
(Belgrave, Haug, and Gomez-Bellenge 1987). Dwyer and Mitchell (1999),
using data from the HRS, find that health problems influence retirement
behavior more strongly than economic factors. Correcting for the
potential endogeneity of self-rated health due to "justification
bias," men in poor overall health expect to retire one to two years
earlier. Similarly, McGarry (2004) finds that those in poor health are
less likely to continue working than those in good health. Using data
from the HRS, she notes that changes in retirement expectations are
driven to a much greater degree by changes in health than by changes in
income or wealth. Ettner, Frank, and Kessler (1997) also indicate that
psychiatric disorders significantly reduce employment among both
genders. Several other studies similarly show that poor health motivates
early retirement, though the relative impact of health versus other
factors is debated. (4)
In contrast, very few studies have examined the impact in the other
direction--that is, how retirement affects subsequent health. This
question takes on added relevance given the shifting trends in labor
force attachment, aging of the population, and growth in health care
expenditures. Szinovacz and Davey (2004) find that depressive symptoms
increase for women post-retirement and are reinforced by the presence of
a spouse with functional limitations. A recent Whitehall II longitudinal
study of civil servants by Mein et al. (2003) compared 392 retired
individuals with 618 working participants at follow-up to determine if
retirement at age 60 is associated with changes in mental and physical
health. Their results indicate that mental health deteriorated among
those continuing to work; whereas, physical functioning deteriorated for
both workers and retirees.
A Kaiser Permanente study of members of a health maintenance
organization (ages 60-66) compared mental health and other health
behaviors of those who retired with those who did not (Midanik et al.
1995). Controlling for age, gender, marital status, and education,
retired members were more likely to have lower stress levels and engage
in regular exercise. No differences were found between the groups on
self-reported mental health status, coping, depression, smoking, and
alcohol consumption.
A follow-up study on 6257 active municipal employees in Finland
found an increase in musculoskeletal and cardiovascular diseases among
retired men (Tuomi et al. 1991). Ostberg and Samuelsson (1994), on the
other hand, find positive effects of retirement on health, as measured
by blood pressure, musculoskeletal diseases, psychiatric symptoms, and
physician visits. Salokangas and Joukamaa (1991) find mental health
improvements but no clear effect on physical health in a study of
Finnish individuals between the ages of 62 and 66 years. Bosse et al.
(1987) examine psychological symptoms in a sample of 1513 older men.
Controlling for physical health status, models indicate that retirees
reported more psychological symptoms than workers. The role of family
income (a correlate of retirement) as a determinant of good physical and
mental health is underscored in Ettner (1996). Using data from the
National Survey of Families and Households, the Survey of Income and
Program Participation, and the National Health Interview Survey,
instrumental variables estimates indicate that income is significantly
related to several measures of physical health in addition to measures
of depressive symptoms.
While these studies highlight important aspects of the interaction
between retirement and health, there is no consensus, and the studies
are also limited in several respects. Many use self-reported subjective
evaluation of health and are based on small selected samples, which
limits their external validity. Most of the studies are also based on
individuals in other countries, which have substantially different
norms, labor markets, and economic incentives embedded in their pension
systems relative to the United States. Several studies employ a simple
cross-sectional comparison between workers and retirees and ignore the
heterogeneity between the treatment and control. Data limitations also
preclude an extensive set of controls, and many do not account for
changes in income or assets post-retirement. Most importantly, none of
these studies account for biases due to endogeneity.
The present study exploits seven longitudinal waves of a
large-scale population survey of older adults in the United States.
Diverse health measures, including self-rated health and objective
functional and illness indicators, are used as the dependent outcomes.
The HRS data also allow for a rich set of controls, the exclusion of
which may have biased other studies. Panel data methodologies and
various specification checks are used to overcome unobserved
heterogeneity and endogeneity and disentangle the causal effect of
retirement on subsequent health.
3. Analytical Framework
The objective of this study is to assess the extent to which
complete retirement impacts health outcomes. This question can be framed
within the human capital model for the demand for health (Grossman
1972). Grossman combines the household production model of consumer
behavior with the theory of human capital investment to analyze an
individual's demand for health capital. In this paradigm,
individuals demand health for its consumptive and investment aspects.
That is, health capital directly increases utility and also reduces work
loss due to illness, consequently increasing healthy time and raising
earnings. This implies that upon retirement, the investment motive for
investing in health in order to raise productivity and earnings is no
longer present. We may therefore expect health to decline after
retirement. However, since healthy time enters into the utility function
as a consumption good, retirees may invest more in their health
post-retirement. In this case, we could expect health to increase after
retirement. Specifically, the effect of retirement on health depends on
how the retirement transition affects the marginal benefits and costs of
health capital, which in turn depends on the life cycle behavior of the
marginal value of time and the relative intensity of time versus market
inputs in the production of health capital. As the standard theoretical
framework does not deliver an unambiguous prediction, the effect of
retirement on health status remains an empirical question. (5)
Other specific mechanisms may further explain how investments in
health may be affected subsequent to retirement. Prior studies (Cohen
2004) suggest that social interactions are strongly associated with
physical and mental health. With social interactions in the form of
external memberships and religious attendance on the decline, social
networks formed at work take on added importance and may buffer
individuals from shocks that may otherwise impact health. The transition
from work to full retirement, by reducing the degree of social
interactions, may have a negative effect on mental and physical health.
Sugisawa et al. (1997) find that retirement reduced social contacts for
males over the age of 60 and induced social isolation. If social
isolation induces depression, this may also reinforce deterioration in
physical health, since they have been found to go hand in hand. (6) On
the other hand, to the extent that work is stress-enhancing and
utility-reducing, retirement may lead to better physical and mental
health.
Work and related actions may also be the primary form of physical
activity and exercise for many individuals. Grundy et al. (1999) report
that 27% of males and 31% of females get no regular physical activity
outside of work. The positive benefits of physical activity on health
indicators, including coronary heart disease, weight, diabetes,
hypertension, cholesterol, heart attack and stroke, cerebral blood flow,
overall mortality, and depression have been well-documented. (7) To the
extent that the shift from work to retirement leads to a decline in the
frequency or intensity of physical activity, retirement may lead to
worse health outcomes, ceteris paribus. On the other hand, physical
activity from the working years may be habit forming and may not decline
upon retirement, conditional on age effects.
The Grossman paradigm is a convenient abstraction in that it
assumes the individual has full control over their health. Thus, a
standard critique concerns the lack of uncertainty in the production of
health capital. However, these mechanisms suggest that the individual
does have some degree of control over their health in support of a
behavioral framework--for instance, through social interactions,
physical activity and exercise, risky behaviors such as smoking and
drinking, diet, and preventive health care utilization. While all health
outcomes have varying degrees of uncertainty, the indicators used in
this study are found to be responsive to health behaviors and lifestyle
factors and therefore, have a strong deterministic component. (8)
Lifestyle behaviors have been shown to be strong indicators of a variety
of health outcomes, including heart disease, depression, diabetes,
functional limitations, and other chronic disease (for example, see
Brach et al. 2004). Injury is more likely in certain populations given
the roles of job demands, living conditions, and lifestyle (Chau et al.
2007). Self-management is key in diseases such as diabetes (Tessier and
Lassmann-Vague 2007) and lifestyle changes that affect the metabolic
syndrome help to prevent illnesses such as heart disease and stroke
(Wong 2007). An abundance of literature also points to lifestyle as a
large determinant of obesity, which is associated with a host of
morbidities (National Institute of Diabetes and Digestive and Kidney
Diseases 1996; Rashad 2006).
Empirically identifying the causal effect of retirement on health
is complicated by two issues. First, an individual's retirement
behavior and health status may depend on a common set of unobserved
factors (for example, life history and time preference). Second,
retirement may be endogenous to health. In addition to retirement
affecting health outcomes, the literature has also identified causality
in the other direction.
Consider linear specifications of the structural demand function
for negative health outcomes ([H.sub.it]) and the labor supply function
representing retirement ([R.sub.it]): (9)
[H.sub.it] = [[alpha].sub.1] [R.sub.it] + [[alpha].sub.2][I.sub.it]
+ [[alpha].sub.3][X.sub.it] + [[alpha].sub.4] [[mu].sub.i] +
[[epsilon].sub.it], (1)
[R.sub.it] = [[beta].sub.1] [H.sub.it] + [[beta].sub.2][E.sub.it] +
[[beta].sub.3][X.sub.it] + [[beta].sub.4][[mu].sub.i] + [[eta].sub.it].
(2)
Equation 1 is a demand function for health ([H.sub.it]), which is a
function of retirement ([R.sub.it]), determinants of health such as
health insurance ([[I.sub.it]), observable characteristics such as age,
gender, race, and education ([X.sub.it]), and unobservable
characteristics pertaining to the individual, such as family background,
tolerance towards risk, and the rate of time preference ([[mu].sub.i]).
Equation 2 postulates labor supply in the form of full retirement
([R.sub.it]). The vector [E.sub.it] represents variables specific to the
retirement decision, such as employer-provided health insurance and
retiree access to health insurance. The vector [[mu].sub.i] denotes
unobserved determinants of retirement that may also influence health.
The subscripts refer to the ith individual in time period t.
The parameter of interest is [[alpha].sub.1], the structural effect
of retirement on negative health outcomes. Ordinary least squares (OLS)
estimation of Equation 1 may be biased. This is reflected in Equation 3,
the quasi-reduced form labor supply function, obtained by substitution
of Equation 1 into Equation 2:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)
If common unmeasured factors ([[mu].sub.i]) determine both health
and retirement ([[alpha].sub.4] [not equal to] 0 and [[beta].sub.4] [not
equal to] 0), then such unmeasured factors are likely to be correlated
with retirement ([[pi].sub.4] [not equal to] 0). The possibility that
health influences the decision to retire also leads to correlated errors
([[beta].sub.1] [not equal to] 0, [[pi].sub.5] [not equal to] 0).
The estimation strategy exploits the longitudinal panels of the
data to control for these biases. The HRS contains a rich set of
information on parental history, health insurance, and indicators for
tolerance towards risk and the rate of time preference. An extensive set
of controls, while informative, cannot fully bypass bias from unobserved
selection. The longitudinal aspect of the data, however, does allow for
the estimation of individual fixed effects (FE) models that control for
all unobserved time-invariant heterogeneity across individuals
([[mu].sub.i]).
Even after identifying off the within-person differences through
the fixed effects, reverse causality remains a concern ([[beta].sub.1]
[not equal to] 0). The sign of [[beta].sub.1] (the reverse effect of
health on retirement) is theoretically ambiguous, especially since poor
health may force some individuals to withdraw from the labor force and
others to work longer to pay medical bills (Anderson and Burkhauser
1985; Dwyer and Mitchell 1999; McGarry 2004). However, with respect to
the measures of health employed in this study, conditional on income or
wealth, it is generally found (as discussed in section 2) that poor
health drives early retirement. Thus, [[beta].sub.1] is likely to be
positive (negative health outcomes may motivate retirement), which
implies that the parameter [[pi].sub.3] is also positive. This would
impart a positive correlation between retirement ([R.sub.it]) and the
structural error term ([[epsilon].sub.it]) in the health demand
function. The effect of retirement on adverse health outcomes in the FE
models may therefore be overstated. (10)
To account for this bias, the sample is stratified across
individuals who had no major illnesses or health problems in the waves
prior to retirement and no reported worsening of health between adjacent
waves prior to retirement. For these individuals, retirement is likely
to be exogenous to health. Since they are physically and mentally
healthy in the waves prior to retirement, their subsequent retirement
cannot have been driven by poor health status. Individual FE
specifications estimated for the pre-retirement healthy sample will
therefore provide the cleanest post-retirement health effects for the
average healthy individual. (11) The identifying assumption is that for
individuals who are mentally and physically healthy at baseline prior to
retirement, the change in health status among those who retire later
serves as a good counterfactual for those who choose to retire earlier,
conditional on the individual fixed effects and the observed
characteristics. The comparison of the full-sample and the
stratified-sample marginal effects will also provide an additional check
for whether the endogeneity bias is being alleviated in the hypothesized
direction. Further specifications build on these and exploit the
longitudinal aspect of the data set to disentangle some of the driving
mechanisms by which retirement may impact health outcomes. Information
on the reported reasons for retirement also allows an alternative method
of identifying individuals whose retirement decisions may be exogenous
to their health.
4. Data
The analysis relies on the Health and Retirement Study (HRS), which
is conducted by the Institute for Social Research at the University of
Michigan. The HRS is an ongoing longitudinal study that began in 1992
and is repeated biennially. (12) Prior to 1998, the HRS cohort included
individuals born between 1931 and 1941, and a separate Study of Assets
and Health Dynamics among the Oldest Old (AHEAD) included individuals
born before 1924. Since 1998, AHEAD respondents have been contacted as
part of a joint data collection effort with the HRS, and the sample
frame was also expanded by including cohorts born between 1924 and 1930
and those born between 1942 and 1947. The present analysis utilizes the
seven waves, spanning 1992 through 2005, and restricts the sample to
older adults between the ages of 50 and 75. This yields a maximum sample
size of about 77,194 person-wave observations.
The HRS is administered for the specific purpose of studying
life-cycle changes in health and economic resources, and includes
detailed information on various health outcomes. A series of 12 measures
of physical and mental health are constructed from the data. A
dichotomous indicator is defined for whether the respondent self-reports
that his or her health is poor. Additional indicators are defined
separately for whether the respondent reports that he or she has been
diagnosed with the following illnesses: diabetes, heart disease, stroke,
high blood pressure, arthritis, and psychological problems. A composite
index measuring the number of these illnesses is also defined and ranges
from zero to six. Two additional composite indices are defined to
measure the number of difficulties associated with mobility and
activities of daily living (ADL), both ranging from zero to five
(detailed in Appendix 1). The HRS contains a depression scale, as
defined by the Center for Epidemiologic Studies (CES), which ranges from
zero to eight. This CESD score measures the sum of adverse mental health
symptoms for the past week, and studies have confirmed the validity and
reliability of the CESD scale as a screening instrument for the
identification of major depression in older adults (Irwin, Artin, and
Oxman 1999). These measures are chosen since they summarize a broad
range of physical and mental health outcomes and have some deterministic
component that can be affected in a behavioral framework. Specifically,
these measures are correlated with lifestyle factors such as diet,
exercise, smoking, and drinking and therefore, would be most likely to
reflect any causal effect of retirement through behavioral channels.
Dichotomous indicators are defined for complete retirement, if the
respondent reports that he is retired and not working, and for partial
retirement, if the respondent reports that he is retired but continues
to work part-time. Individuals otherwise not in the labor force,
including homemakers and the disabled, are excluded from the analysis.
Individuals who are partially retired are excluded when estimating the
effects of complete retirement on health. Similarly, individuals who are
fully retired are excluded from specifications estimating the effects of
partial retirement. Thus, in both analyses the reference category is
comprised of working individuals in the labor force, and this
facilitates the comparison of marginal effects across models.
Health outcomes differ across several observable socioeconomic and
demographic dimensions. Indicators for gender, race, ethnicity, marital
status, and no religious preference are defined and included in the
models. Age fixed effects control for any nonparametric declines in
health over the life cycle, allowing the retirement indicator to pick up
shocks beyond general age-related health deterioration. Real income is
calculated for each individual from all available sources including
earnings, pension, supplemental security, Social Security retirement,
and other government transfers deflated by the consumer price index.
(13)
An individual's health status may also depend on access to
care, which in turn is a function of health insurance coverage. The
respondent's health insurance status is determined from various
questions. A coverage indicator is defined for whether the individual
reports being covered by health insurance under any governmental program
including Medicare or Medicaid, under his own current or previous
employer, under his spouse's current or previous employer, or under
any other supplemental insurance.
The HRS further contains rich information on other variables that
may confound the relationship between retirement and health. (14) All
models include dichotomous indicators for year of the interview to
capture unobserved time-varying factors, and indicators for eight census
divisions to capture unobserved differentials in health care and
outcomes across the regions. Weighted means for all variables for the
full sample and samples stratified across retirement status are
presented in Table 1.
Table 1 indicates that about 38% of the sample is fully retired,
with an additional 12% partially retired. The means also indicate that
fully retired individuals are in poorer health. For instance, retirees
have 1.7 illnesses compared to one illness for those still working.
Similar statistically significant differences are observed for all other
indicators of physical and mental health. The figures further show that
retirement is correlated with other observed and sometimes unobserved
characteristics. For example, retired individuals have completed fewer
years of schooling and have less educated parents. Fewer retirees are
married, have a high income, or have no insurance coverage. They are
also more likely to be risk averse and differ somewhat in their
financial outlook. (15) Thus there may be "positive selection"
on observed characteristics--individuals who are retired are not a
random sample. They are also more likely to differ along characteristics
that generally are associated with worse health (less human capital,
less parental human capital, less income, non-married, Hispanic or other
race, generally more present-oriented, to name a few). Most notably,
retirees are older, with a mean age in the sample of 66, versus the mean
age of 58 for individuals who have not yet retired. The multivariate
models account for these differences.
5. Results
Table 2 presents estimation of the baseline specifications (Eqn. 1)
for various indicators of health. (16) In addition to basic demographic
measures, the extended specification (reported in column 1) includes
health insurance status, parental characteristics, proxies for risk and
time preference along with age, year, and census division indicators.
Appendix 2 details the effects of all independent variables on poor
health and mobility difficulties. The effects of the socio-demographic
measures are consistent with prior studies. (17) Several additional
effects are also noteworthy. Individuals with better health endowment,
as proxied by the life span of the parents, are healthier. Growing up
with more educated parents also improves adult health outcomes.
Risk-averse individuals are healthier since they may be less likely to
engage in risky activities, such as smoking or drinking, or work in
riskier occupations, which may adversely affect health (Barsky et al.
1997; Saffer and Dave 2005; Dave and Saffer 2007). Conditional on age,
individuals who are more future-oriented, as proxied by their planning
horizon, are also healthier. These individuals may also be less likely
to engage in risky health behaviors and may make greater investments in
their own health capital (Fuchs 1982). Health insurance has a negative
impact on health, likely reflecting adverse selection.
Conditional on these covariates, complete retirement has a
significant negative impact on health. For instance, it raises the
probability of poor health by 0.12 percentage points and increases the
number of mobility difficulties by 0.66 and the number of ADL
difficulties by 0.26. The magnitudes of the marginal effects on all
health measures are quite large in the extended models, relative to the
overall sample means. Potential selection on unobservable
characteristics and reverse causality may be driving the link between
health and retirement. Since the retirement decision and adult health
status are generally the result of an accumulation of life-cycle
decisions to invest in health and human capital, most of the effects of
retirement on health may reflect heterogeneity across individuals. The
longitudinal panels of the HRS allow for the estimation of individual FE
models (reported in specification 2) that account for this unobserved
heterogeneity. The marginal effects of retirement on health remain
significant, but decline substantially in magnitude by about 60%. This
indicates positive selection on unobservables. For instance, these
individuals may have made inadequate investments in their own human
capital or have dysfunctional family upbringing that may lead to labor
force withdrawal and worse adult health. This is consistent with the
unadjusted differences between retirees and workers (Table 1), which
also showed positive selection on observable factors.
While controlling for individual fixed effects diminishes the
magnitudes, retirement continues to have a significant adverse effect on
all proxies of physical and mental health. Results from the second
column of Table 2 show, for instance, that complete retirement worsens
mobility by 34%, leads to a 61.6% increase in difficulties associated
with activities of daily living (ADL), leads to a 7.9% increase in
illnesses, and worsens mental health by 1114.5%, relative to the sample
means.
Identifying off the within-individual variation, conditional on age
and income, the results are analogous to a pre- and post-retirement
difference in health status for each individual relative to others
retiring at different ages. However, the possibility remains that
retirement itself may be motivated by deteriorating health. This
endogeneity would inflate the negative effects of retirement on health.
The last row of Table 2 serves as a check and suggests that this is
indeed what may be occurring. Restricting the sample to never-smokers
and moderate drinkers, retirement is found to raise the probability of
cancer (excluding skin cancer) by 27%. It is implausible that
post-retirement lifestyle changes could cause such a large increase in
cancer; although, it needs to be noted that lifestyle factors have the
potential to affect certain types of cancer to some degree. (18) If
anything, retirement should have minimal or no impact on the probability
of contracting cancer for individuals who do not engage in risky
activities.
To aid in bypassing endogeneity, the last two columns of Table 2
present estimation of the individual FE models for samples restricted to
individuals who were physically and mentally healthy in the waves prior
to retirement. Specifically, the sample is limited to those with no
mobility difficulties, no illness conditions (diabetes, heart disease,
stroke, high blood pressure, arthritis, cancer, or lung disease), and no
reported psychological problems pre-retirement. Retirement for these
workers should not be motivated by poor health status and represents
labor force decisions orthogonal to current or past health. The effect
sizes in these models are expected to be smaller, given the positive
bias due to endogeneity (see footnote 10).
The third column of Table 2 shows that the negative effects of
complete retirement on health are indeed generally much smaller in
magnitude, though they remain statistically significant. Retirement
causes a 17-23% increase in difficulties associated with mobility and
daily activities and a 7% increase in illnesses. (19) It also leads to
about a 9% decline in mental health, as proxied by the CES Depression
Scale. Since the typical individual in the HRS is observed for three
post-retirement waves, these effects are being realized over six years
subsequent to retirement, on average. In addition, these specifications
show that while retirement negatively impacts health measures, which are
most likely to be correlated with lifestyle changes, it has no effect on
cancer, where we do not expect to find any large effect.
Prior studies have highlighted important, though not always
consistent, differences across gender. To maximize sample size,
differential effects by gender were estimated through an interaction
term for the specifications in Table 2 (results not reported). For
males, retirement generally leads to a larger decline in physical health
outcomes as proxied by self-reported health, difficulties in mobility
and daily activities, illness conditions, diabetes, heart disease, and
stroke. However, with respect to the CES Depression Scale, retirement is
found to have a larger negative effect for females. This differential
effect may be related to the reasons proposed for the overall larger
prevalence of depression and anxiety disorders among women at all stages
of life (Nolan-Hoeksema, Larson, and Grayson 1999).
Health Insurance
Withdrawal from the labor force before the age of 65 may be
accompanied by a change in health insurance status, which may also be
endogenous to health outcomes. The adverse health effects
post-retirement may reflect a decline in access to health care if
retired individuals lose their employer-sponsored coverage, are
ineligible for Medicare if younger than 65 years of age, and opt not to
purchase private insurance. Furthermore, those who retire may be more
likely to have retirement coverage, and health insurance may also be
picking up the propensity to be in poorer health. (20) This adverse
selection is apparent in the extended and FE specifications. Simple
means also show that retirees are more likely to be insured. To
ascertain that the retirement effects are not driven by selective
changes into and out of coverage or retiree access to coverage, the
sample is constrained to individuals who are consistently insured in all
waves. The marginal effects, presented in the last column of Table 2,
are not materially affected and remain statistically significant. (21)
Conditional on individual fixed effects, shifts in and out of health
insurance related to retirement do not play a major role in the
post-retirement decline in health.
Unobserved Health Shocks
While focusing on individuals who were healthy pre-retirement
bypasses endogeneity from observed health measures, one concern is that
these individuals may nevertheless have experienced a health shock
between waves that may not be reflected in the diagnosed or reported
health outcomes at each wave. Utilizing information on reported changes
in health status between waves and reported reasons for retirement
allows specification checks for this possibility. The first two columns
of Table 3 show individual FE results where the sample is restricted to
those who did not report any worsening of health in the wave of
retirement (relative to the prior wave) and also did not report any
worsening of health in the wave prior to retirement. Thus, for an
individual retiring in Wave 4 to make it into the sample, he must not
report any health deterioration between Waves 3 and 4, as well as
between Waves 2 and 3. Plausibly, for this individual, the retirement
decision is orthogonal to any reported health deterioration or shocks
between adjacent waves prior to their retirement. Specification 2
employs a more restrictive sample--that is, individuals who did not
report any worsening of health between adjacent waves and with no
observed ill-health measures prior to their retirement. Although the
effect sizes decline slightly in magnitude, the results remain generally
robust across all samples and health outcomes. The standard errors also
remain relatively stable across samples so as not to significantly alter
inferences, despite smaller sample sizes in specification 2.
In the HRS, reasons for retirement are probed at the time that the
individual first reports retirement, though there are various gaps and
inconsistencies across waves. Four indicators are found to be consistent
across waves with minimal missing observations. These include the
following reasons for retirement: (i) poor health; (ii) wanted to do
other things; (iii) wanted to spend more time with family; and (iv) did
not like work. Columns 3-7 of Table 4 present results where this
information is exploited. (22) While purging potential justification
bias, exploiting these measures also informs about the individuals'
preferences for retirement. They allow sample stratification of
relatively homogeneous groups of retirees, at least with respect to
their reported reason for retirement.
Specification 3 is restricted to the sample that excludes all
individuals who reported that poor health was an important reason in
their retirement decision. Across the four health indicators, complete
retirement is found to have a significant and adverse impact.
Specification 4 excludes all individuals who cite poor health as a
retirement reason, and further restricts the sample to individuals who
were healthy (with respect to the observed indicators) in the waves
prior to retirement. Thus, this sample also addresses the concern of
unobserved health shocks between waves. To the extent that the
individuals are healthy prior to retirement, and also do not attribute
their retirement to health reasons, retirement would be exogenous to
health status for this group. The results are not materially affected,
though there is an increase in the standard errors due to reduced sample
size. The effect sizes in models 3 and 4 are slightly smaller in
magnitude, yet this may be consistent with potential "justification
bias" that has been suggested in the literature. The concern is
that subjective reports of health are biased by individuals using poor
health as a justification for early retirement (Bound 1991; McGarry
2004). In this case, these restricted samples would be excluding
individuals who truly retired due to health reasons and also those who
may have retired for other reasons but are using their health as a
justification.
Models 5-7 look at the group of individuals who are healthy in the
waves prior to retirement, alternately stratifying by other
(non-health-related) reasons. Model 5 focuses on those who retired
"to do other things." Models 6 and 7 focus on those who
retired "to spend more time with family" and those who retired
because they "did not like work," respectively. The
coefficient magnitudes are robust across most of these specifications,
and also similar to the earlier models. Reduced sample sizes inflate the
standard errors; although, the inferences are generally not affected. It
is perhaps not surprising that for those who did not like work, the
results for the CESD scale are no longer significant.
Specification Checks
If these estimated effects are due to causal behavioral changes
prompted by retirement, the effects should be spread out over time and
not concentrated in the first wave post-retirement. If a substantial
health effect is observed in the first post-retirement observation, then
this would suggest that unobserved health shocks are motivating
retirement, or there are anticipation effects. (23) Specification 1 in
Table 4 estimates the individual FE models for the full sample that does
not adjust for endogeneity. As expected, there are large significantly
negative effects of retirement on health even in the first
post-retirement wave. This suggests that the effect cannot plausibly be
causal, reflecting endogenous selection and possible anticipation
effects. (24) Specification 2 re-estimates the FE model for the
preferred conservative sample of individuals who are healthy in the
waves prior to retirement. None of the first post-retirement wave
effects are significant, and most of the effect of retirement on health
is being realized in the latter periods. (The effect sizes for
depression are imprecisely measured due to inflated standard errors.)
Specification 3 restricts the analysis to only the first wave after
retirement. The effect sizes remain insignificant and small in
magnitude. These results are validating, in that they do not show any
large immediate effects that might cast doubt on a causal
interpretation.
As an additional falsification check, a pseudo-retirement indicator
is constructed to gauge whether the preferred models are bypassing the
endogeneity bias. It is defined such that an individual who retired in
Wave 5 is falsely assigned retirement in a prior wave (Wave 3 in this
case), and so on. Specifications 4 and 5 of Table 4 present the marginal
effects of pseudo-retirement on poor health outcomes for the extended
and preferred specifications. (25) Pseudo-retirement should have no
causal adverse effect on health outcomes, since it is not inherently
reflecting any real change in status. In the extended models, however,
the indicator has a strong, significantly negative effect on all
measures of health. This suggests that the effects are biased upwards
(in magnitude) due to endogeneity. Pseudo-retirement is picking up
systematic variations across individuals and other concurrent shifts
related to aging, health, and labor force behavior. If the preferred
specifications are successful in removing the endogeneity, then the
marginal effect of pseudo-retirement should decline to zero and become
insignificant. The fifth column reassuringly confirms this to be the
case.
Models based on instrumental variables (IV) are also estimated,
though they should be interpreted with caution due to the inherent
difficulties of identifying valid instruments. The sample was limited to
individuals who reported that they expect to retire at the same time as
their spouse, and further limited to those who report that they are not
concerned about inadequate retirement income. For this group, the
spouse's retirement status (complete, partial, or non-retired) is a
significant predictor for own-retirement status. The instruments are
orthogonal to own health, conditional on own retirement and wealth, and
they "pass" the overidentification test. Marginal effects from
the IV models fall somewhere between those from the full sample and the
pre-retirement healthy sample individual FE models. The standard errors
are larger, making the estimate imprecise for illness conditions. Models
are also estimated separately for those who retired at age 62.
Retirement at 62 is likely to have a larger exogenous component
(relative to other ages), since the spike is related to Social Security
and other pension eligibility. The marginal effects tend to be similar
to those based on the individual FE models for the pre-retirement
healthy sample. (26) These results are presented in the last two columns
of Table 4.
While selective attrition is a concern for all longitudinal
datasets, it can be especially relevant in our analysis of health
outcomes due to death-related attrition. In the HRS, the average
mortality rate between waves is 2.3%, consistent with the Social
Security Administration life table mortality rates (Kapteyn et al.
2006). Appendix 3 reports results from three different approaches to
inform on potential bias due to this attrition. The results are highly
similar across the three specifications, and estimates remain robust.
Stratifications
Table 5 estimates the preferred individual FE models for the
pre-retirement healthy sample, stratified across additional dimensions.
(27) These stratifications shed light on some of the possible mechanisms
for the post-retirement decline in health.
One hypothesis concerns the post-retirement reduction in social
interactions and support that were formed through and at work. Since
studies have linked social interactions to better health, the transition
from work to full retirement may lead to deteriorating mental and
physical health through this channel. In this case, the negative effects
of retirement should be larger for individuals without a spouse or a
partner. Social support from a spouse may help to buffer shocks and
offset some of the diminished external social interactions. The first
two rows of Table 5 confirm this direction of effect. Complete
retirement generally leads to worse health for single relative to
married individuals. The difference is especially large for mental
health, which is consistent with prior studies that show social
interactions and spousal support to have a significant effect on
depressive symptoms (Sugisawa et al. 2002; Cohen 2004).
For many individuals, work-related activities may constitute the
primary form of exercise and physical activity. If retirement leads to a
decline in the frequency or intensity of physical activity, then health
may deteriorate. The prevalence of engagement in physical activity
postretirement is similar for those individuals with physically
demanding work relative to others. The decline in physical activity
post-retirement is therefore steeper for individuals who had physically
demanding jobs prior to retirement. Ceterisparibus, retirement would be
expected to have a larger adverse health effect for these workers. The
next two rows stratify the sample across individuals who report that
their job required a great deal of physical effort almost all of the
time. Retirement is found to deteriorate physical health more for these
individuals relative to those in nonlaborious work. (28) Similarly,
retirement should cause the largest declines in health among those who
do not participate in vigorous physical activity post-retirement, to
substitute for the drop in work-related physical activity. The next two
samples, stratified across individuals who participate and do not
participate in physical exercise after retirement, show that the
marginal effects are indeed substantially larger for those who do not
remain physically active. Summary measures show that for individuals who
do not engage in physical activity, there is a slight increase in weight
and the probability of being overweight. This is consistent with the
transition from work to full retirement leading to negative lifestyle
factors that worsen health.
In standard models of labor supply, it is assumed that leisure is
utility-enhancing, and thus work is utility-diminishing. In this case,
retirement would be expected to yield benefits due to the increase in
leisure time, ceteris paribus. To the extent that this effect offsets
some of the negative health effects, retirement would be expected to
have a smaller adverse effect on health for those individuals who found
work especially distasteful or stressful. The next two models are
stratified across individuals who report that their work involved a
great deal of stress almost all of the time. For these individuals,
retirement is presumably stress-reducing, and consequently their decline
in physical and mental health is also expectedly smaller.
An additional stratification based on whether the retirement
decision was voluntary or involuntary is shown next. (29) Among those
who report that their retirement was "forced," we further
exclude from the analysis individuals reporting health as a retirement
reason. Conditions leading to forced retirement include job
displacements, employer policy towards older workers, care obligations,
and other personal reasons. Standard errors are inflated due to smaller
sample sizes; however, there is some evidence that the adverse effects
of retirement on health are larger in the event of forced retirement and
smaller in the event of voluntary retirement.
Where the negative health effects of full retirement are mediated
by other positive factors, the magnitudes are found to be smaller. An
additional robustness check is permitted by individuals who are
partially retired--that is, those who continue to do some part-time work
after retiring from their jobs. Complete retirement has adverse health
effects, consistent with an increase in the relative net price of health
investment, a decline in social interactions and a decline in
work-related physical activity. If this is a causal relationship, then
partial retirement would be expected to have little or no adverse health
effects since the incentive to avoid work loss from illness still
exists, which raises the marginal benefit of investing in health.
Part-time work may also impart positive effects through social support
and physical activity. The final two rows of Table 5 confirm this
pattern. Partial retirement generally has a much smaller negative effect
on health outcomes, relative to full retirement. It is found to
significantly increase the number of illness conditions by 0.055 (4.2%)
and difficulties in daily activities by 0.016 (10%), compared with 0.083
(6.4%) and 0.027 (17%) for complete retirement. For other measures of
physical and mental health, partial retirement has no significant
adverse effects.
6. Conclusion
While unadjusted differences document a strong negative effect of
complete retirement on health, the aim of this study was to examine how
much of this association is consistent with a causal mechanism, and how
much of it is being driven by non-random selection and endogeneity.
Estimates suggest that indeed most of the observed difference (80-90%)
is due to such confounding. However, a sizable residual effect remains
that is consistent with a behavioral framework. The absolute effects of
complete retirement include an increase in the number of mobility
difficulties by 0.08-0.10, an increase in the number of difficulties in
daily activities by 0.01-0.02, an increase in the number of illness
conditions by 0.05-0.08, and an increase in depression symptoms by
0.08-0.11. Evaluated relative to the sample mean, these translate into a
5-16% increase in difficulties associated with mobility and daily
activities, a 5-6% increase in illness conditions, and a 6-9% decline in
mental health. (30) These are average cumulative effects realized over a
period of about six years post-retirement.
Additional checks indicate that the effects tend to operate through
lifestyle changes including declines in physical activity and social
interactions. Future research should focus on these lifestyle shifts and
other channels by which retirement impacts health. The adverse health
effects are mitigated if the individual is married and has social
support, continues to engage in physical activity post-retirement, or
continues to work part-time upon retirement. There is also some evidence
that the adverse health effects are larger in the event of involuntary
retirement. In this case, programs that help older workers forced into
retirement find alternative employment opportunities may be
health-promoting. On the other hand, voluntary withdrawal from the labor
force also has some negative health impact that is consistent with
changes in health behaviors and lifestyle post-retirement. This does not
necessarily suggest that individuals who retire early or voluntarily are
irrational or that they have not considered the full implications of
retirement, including the change in environment or incentives. Indeed,
the behavioral framework presupposes some rationality. (31) However, if
retirement decisions are "forced" or if voluntary retirement
is rationally based on market constraints (delayed retirement credit in
Social Security or private pensions, incentives in defined-benefit
plans, labor market inflexibility regarding hours or work
opportunities), then there may be room for altering these market
constraints so as to improve the health of older adults, ceteris
paribus.
The adverse effects of retirement on subsequent health status found
in this study have held up to various specification and robustness
checks and yet, should nevertheless be interpreted with caution due to
the striking nature of the results. The estimates from this study have
policy implications and should be considered in any policy evaluation
that aims at shifting the retirement age. With the financial
difficulties facing Social Security and Medicare compounded by an aging
population retiring earlier, policy makers have pressed for higher
retirement ages. (32) For employer and private pension plans, 60 remains
a popular age for benefits eligibility. Furthermore, the Social Security
system, as well as many other private pension plans, contain incentives
that may discourage work for certain individuals. (33) In the presence
of negative health effects, policies that aim to increase the retirement
age may be desirable. Ceteris paribus, a higher retirement age, by
postponing or reducing poor health outcomes, will also consequently
reduce the utilization of health services by older adults conditional on
life expectancy, which may have implications for the projected increases
in Medicare expenditures. Thus, policies that raise the retirement age,
while improving the financial liability of Social Security, may also
curb the long-term growth in Medicare expenditures, even if the Medicare
eligibility age remains unchanged.
Appendix 1
Variable Definitions
Variable Definition
Retirement
Complete retirement Dichotomous indicator for whether
respondent is fully retired
Partial retirement Dichotomous indicator for whether
respondent is partially retired
Health outcomes
Good health Dichotomous indicator for whether
respondent reported health as being
excellent or very good
Poor health Dichotomous indicator for whether
respondent reported health as poor
Mobility difficulties Index for mobility problems ranging from 0
to 5, indicating the respondent reporting
any difficulty in walking one block,
walking several blocks, walking across a
room, climbing one flight of stairs, and
climbing several flights of stairs
Activities of daily Index for problems in activities of daily
living (ADL) living (ADL) ranging from 0 to 5,
difficulties indicating the respondent reporting any
difficulty in bathing, eating, getting
dressed, getting in/out of bed, and
walking across a room
Illness conditions Index of respondent's diagnosed conditions,
ranging from 0 to 6, indicating high
blood pressure, diabetes, heart problems,
stroke, psychiatric problems, and
arthritis
Diabetes Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she has diabetes
Heart disease Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she had a heart attack,
coronary heart disease, angina,
congestive heart failure, or other heart
problems
Stroke Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she had a stroke
High blood pressure Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she has high blood pressure
Arthritis Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she has arthritis or
rheumatism
Psychological problems Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she had emotional, nervous,
or psychiatric problems
Center for Index of mental health for respondent,
Epidemiologic ranging from 0 to 8, indicating the
Studies Depression negative mental health symptoms for last
(CESD) Scale week (depressed, everything an effort,
restless sleep, not happy, lonely, sad,
could not get going, and did not enjoy
life)
Cancer Dichotomous indicator for whether
respondent has ever been told by doctor
that he or she has had cancer or a
malignant tumor of any kind, except skin
cancer
Socio-demographic
Age Age of respondent
Male Dichotomous indicator for whether
respondent is male
Black Dichotomous indicator for whether
respondent is black but not Hispanic
Other race Dichotomous indicator for whether
respondent's race is other than white,
black, or Hispanic
Hispanic Dichotomous indicator for whether
respondent is Hispanic
Education Years of education completed
Married Dichotomous indicator for whether
respondent is married
No religious Dichotomous indicator for whether
preference respondent has no religious preference
Income Total individual income from all sources,
measured in thousands of 1982-1984
dollars
Health insurance Dichotomous indicator for whether
respondent has any type of health
insurance coverage
Mother's age Age of mother, or age at death
Father's age Age of father, or age at death
Mother's education Dichotomous indicator for whether
respondent's mother has attended eight or
more years of school
Father's education Dichotomous indicator for whether
respondent's father has attended eight
or more years of school
Native born Dichotomous indicator for whether
respondent was born in the United States
Risk averse Dichotomous indicator for whether
respondent is very risk averse
Planning horizon, 5-10 Dichotomous indicator for whether
years respondent's relevant financial planning
horizon is 5-10 years
Planning horizon, more Dichotomous indicator for whether
than 10 years respondent's relevant financial planning
horizon is greater than 10 years
Physical activity/stress
Vigorous physical Dichotomous variable that equals 1 if
activity respondent is physically active three or
more days a week
Physical work Dichotomous indicator for whether the
respondent's job required a lot of
physical effort most or all of the time
Stressful work Dichotomous indicator for whether the
respondent's job involved much stress
most or all of the time
Appendix 2
The Role of Complete Variables in Models of Health Outcomes on the
Full Sample (a)
Poor Health
Dependent Variable Individual Fixed
Specification Extended Effects
Complete retirement 0.1163 *** 0.0496 ***
(0.0041) (0.0035)
[1.971] [0.841]
Male 0.0152 *** --
(0.0029)
Black 0.0052 --
(0.0049)
Other race 0.0236 *** --
(0.0088)
Hispanic -0.0145 ** --
(0.0068)
Education -0.0102 *** --
(0.0006)
Married -0.0306 *** 0.0019
(0.0034) (0.0049)
No religious preference 0.0193 *** --
(0.0071)
Income -0.0001 *** -0.00001
(0.00003) (0.00002)
Health insurance 0.0137 *** 0.0101 **
(0.0045) (0.0047)
Mother's age -0.0003 *** --
(0.0001)
Father's age -0.0001 --
(0.0001)
Mother's education -0.0068 --
(0.0044)
Father's education -0.0125 *** --
(0.0040)
Native born 0.0016 --
(0.0054)
Risk averse -0.0051 * --
(0.0029)
Planning horizon, 5-10 years -0.0168 ***
(0.0030)
Planning horizon, more than -0.0149 *** --
10 years (0.0045)
Age fixed effects Yes Yes
Year fixed effects Yes Yes
Census division fixed effects Yes Yes
Individual fixed effects No Yes
Observations 53,551 75,727
Mobility Difficulties
Dependent Variable Individual Fixed
Specification Extended Effects
Complete retirement 0.6593 *** 0.2389 ***
(0.0202) (0.0152)
[0.942] [0.341]
Male -0.1981 *** --
(0.0161)
Black 0.0431 * --
(0.0249)
Other race 0.0593 --
(0.0402)
Hispanic -0.0956 *** --
(0.0340)
Education -0.0494 *** --
(0.0031)
Married -0.1790 *** 0.0152
(0.0183) (0.0221)
No religious preference 0.0321 --
(0.0345)
Income -0.0012 *** 0.00001
(0.0003) (0.0001)
Health insurance 0.1136 *** 0.0668 **
(0.0211) (0.0187)
Mother's age -0.0031 *** --
(0.0006)
Father's age -0.0024 *** --
(0.0006)
Mother's education -0.0011 --
(0.0231)
Father's education -0.1076 *** --
(0.0214)
Native born 0.1347 *** --
(0.0278)
Risk averse -0.0132 --
(0.0161)
Planning horizon, 5-10 years -0.1133 *** --
(0.0171)
Planning horizon, more than -0.1474 *** --
10 years (0.0248)
Age fixed effects Yes Yes
Year fixed effects Yes Yes
Census division fixed effects Yes Yes
Individual fixed effects No Yes
Observations 53,400 72,905
(a) Standard errors are robust clustered at the individual level and
reported in parentheses. Semi-elasticity of health outcome with
respect to retirement, evaluated at the sample mean, is reported in
brackets. Sample is limited to individuals ages 50 to 75. Significance
is defined as follows: *** significant at the 1% level; ** significant
at the 5% level; * significant at the 10% level.
Appendix 3
Sample Attrition (a)
Specification
1 2
Individual Individual
Fixed Effects Fixed Effects
Healthy Healthy
Pre-Retirement Pre-Retirement
Sample attrition: Sample attrition:
balanced panel excluding all
Dependent Variable waves 1-7 passive attritors
Mobility difficulties 0.1539 *** 0.1292 **
(0.0309) (0.0251)
[0.220] [0.185]
Activities of daily 0.0333 *** 0.0216 **
living (ADL) (0.0106) (0.0088)
difficulties [0.208] [0.135]
Illness conditions 0.0809 *** 0.0728 ***
(0.0271) (0.0238)
[0.062] [0.056]
Depression (CESD) 0.1141 ** 0.0930 *
scale (0.0587) (0.0548)
[0.092] [0.075]
Specification
3
Individual
Fixed Effects
Healthy
Pre-Retirement
Sample attrition:
inverse probability
Dependent Variable weighting
Mobility difficulties 0.1192 ***
(0.0290)
[0.170]
Activities of daily 0.0168
living (ADL) (0.0115)
difficulties [0.106]
Illness conditions 0.0510 *
(0.0268)
[0.039]
Depression (CESD) 0.1434 **
scale (0.0634)
[0.116]
(a) See notes to Table 2. Specification 1 limits the sample to
individuals observed in all seven waves. Specification 2 utilizes
a sample that excludes individuals who are known to exit the HRS
due to death at some future point in time. Specification 3 employs
inverse probability weights (IPW) to adjust for selection bias, where
individuals whose observable characteristics predict higher attrition
rates are given a larger weight in the regression. IPWs are predicted
using baseline characteristics (gender, race, ethnicity, education,
parental education, religion, and native born) along with other
time-varying factors (age indicators, wave indicators, and census
division indicators), lagged covariates (income, marital status, and
health insurance), and health status in the prior wave. Significance
is defined as follows: *** significant at the 1% level;
** significant at the 5% level; * significant at the 10% level.
We are grateful to Angela Dills, Michael Grossman, Julie Hotchkiss,
Richard Kaplan, Donald Kenkel, Sean Nicholson, Henry Saffer, and two
anonymous referees for helpful comments. In addition, we wish to thank
seminar participants at the 2007 International Health Economics
Association World Congress, the 2007 Allied Social Sciences Association
Conference, Andrew Young School of Policy Studies, and the 2006 Southern
Economics Association Conference for helpful comments on earlier
versions of the paper. The authors would also like to thank their
respective schools for research support.
Received June 2006; accepted December 2007.
References
Anderson, Kathryn H., and Richard V. Burkhauser. 1985. The
retirement-health nexus: A new measure of an old puzzle. Journal of
Human Resources 20:315-30.
Barsky, Robert B., F. Thomas Juster, Miles S. Kimball, and Matthew
D. Shapiro. 1997. Preference parameters and behavioral heterogeneity: An
experimental approach in the Health and Retirement Study. Quarterly
Journal of Economics 112:537-79.
Bazzoli, Gloria J. 1985. The early retirement decision: new
empirical evidence on the influence of health. Journal of Human
Resources 20:214-34.
Belgrave, Linda L., Marie R. Haug, and Francisco-Xavier
Gomez-Bellenge. 1987. Gender and race differences in effects of health
and pension on retirement before 65. Comprehensive Gerontology 1:109-17.
Borjas, George J. 2004. Labor economics. New York: McGraw-Hill.
Bosse, Raymond, Carolyn M. Aldwin, Richard Levenson, and David J.
Ekerdt. 1987. Mental health differences among retirees and workers:
findings from the Normative Aging Study. Psychology and Aging 2:383-9.
Bound, John. 1991. Self-reported versus objective measures of
health in retirement models. Journal of Human Resources 26:106-38.
Brach, Jennifer S., Eleanor M. Simonsick, Stephen Kritchevsky,
Kristine Yaffe, and Anne B. Newman. 2004. The association between
physical function and lifestyle activity and exercise in the health,
aging and body composition study. Journal of the American Geriatrics
Society 52:502-9.
Calle, Eugenia E., Carmen Rodriguez, Kimberly Walker-Thurmond, and
Michael J. Thun. 2003. Overweight, obesity, and mortality from cancer in
a prospectively studied cohort of U.S. adults. New England Journal of
Medicine 348:1625-38.
Chau, Nearkasen, Eve Bourgkard, Ashis Bhattacherjee, Jean-Francois
Ravaud, Marie Choquet, and Jean-Marie Mur. 2007. Associations of job,
living conditions and lifestyle with occupational injury in working
population: a population-based study. International Archives of
Occupational and Environmental Health 81:379-89.
Cohen, Sheldon. 2004. Social relationships and health. American
Psychologist 59:676-84.
Dave, Dhaval, and Henry Saffer. 2007. Risk tolerance and alcohol
demand among adults and older adults. NBER Working Paper No. 13482.
Dave, Dhaval, Inas Rashad, and Jasmina Spasojevic. 2006. The
effects of retirement on physical and mental health outcomes. NBER
Working Paper No. 12123.
Dwyer, Debra S., and Olivia S. Mitchell. 1999. Health problems as
determinants of retirement: Are self-rated measures endogenous? Journal
of Health Economies 182:173-93.
Ettner, Susan L. 1996. New evidence on the relationship between
income and health. Journal of Health Economies 15:67-85.
Ettner, Susan L., Richard G. Frank, and Ronald C. Kessler. 1997.
The impact of psychiatric disorders on labor market outcomes. Industrial
and Labor Relations Review 51:64-81.
Franco, Oscar H., Chris de Laet, Anna Peeters, Jacqueline Jonker,
Johan Mackenbach, and Wilma Nusselder. 2005.
Effects of physical activity on life expectancy with cardiovascular
disease. Archives of Internal Medicine 165:2355-60.
Fuchs, Victor R. 1982. Time preference and health: an exploratory
study. In Economic aspects of health, edited by Victor R. Fuchs.
Chicago: University of Chicago Press, pp. 93-120.
Gendell, Murray. 2001. Retirement age declines again in 1990s.
Monthly Labor Review 124:12-21.
Grossman, Michael. 1972. On the concept of health capital and the
demand for health. Journal of Political Economy 80:223-55.
Grossman, Michael, and Robert Kaestner. 1997. Effects of education
on health. In The social benefits of education, edited by Jere R.
Behrman and Nevzer Stacey. Ann Arbor, MI: University of Michigan Press,
pp. 69-123.
Gruber, Jonathan, and David Wise. 2005. Social security programs
and retirement around the world: Fiscal implications, introduction and
summary. NBER Working Paper No. 11290.
Grundy, Scott M., George Blackburn, Millicent Higgins, Ronald
Lauer, Michael G. Perri, and Donna Ryan. 1999. Physical activity in the
prevention and treatment of obesity and its comorbidities. Medicine and
Science in Sports and Exercise 31:S502-8.
Irwin, Michael, Kamal Haydari Artin, and Michael N. Oxman. 1999.
Screening for depression in the older adult: Criterion validity of the
10-item Center for Epidemiological Studies Depression Scale (CES-D).
Archives of Internal Medicine 159:1701-4.
Kapteyn, Arie, Pierre-Carl Michaud, James Smith, and Arthur van
Soest. 2006. Effects of attrition and non-response in the Health and
Retirement Study. RAND Working Paper WR-407.
Lee, I-Min, and Patrick J. Skerrett. 2001. Physical activity and
all-cause mortality: What is the dose-response relation? Medicine and
Science in Sports and Exercise 33:S459-71.
McGarry, Kathleen. 2004. Health and retirement: Do changes in
health affect retirement expectations? Journal of Human Resources
39:624-48.
Mein, Gill, Pekka Martikainen, Harry Hemingway, Stephen Stansfeld,
and Michael Marmot. 2003. Is retirement good or bad for mental and
physical health functioning? Whitehall II longitudinal study of civil
servants. Journal of Epidemiology and Community, Health 57:46. 9.
Midanik, Lorraine T., Krikor Soghikian, Laura J. Ransom, and Irene
S. Tekawa. 1995. The effect of retirement on mental health and health
behaviors: The Kaiser Permanente Retirement Study. Journals of
Gerontology Series B. Psychological Sciences and Social Sciences
50:$59-61.
Mitchell, Olivia S. 1992. Trends in pension benefit formulas and
retirement provisions. In Trends in Pensions: 1992, edited by John A.
Turner and D. J. Beller. Washington, DC: U.S. Government Printing
Office, pp. 177-216.
National Institute of Diabetes and Digestive and Kidney Diseases.
1996. Statistics related to overneight and obesity. Washington, DC: U.S.
Government Printing Office.
Nolan-Hoeksema, Susan, Judith Larson, and Carla Grayson. 1999.
Explaining the gender difference in depressive symptoms. Journal of
Personality, and Social Psychology 77:1061-72.
Ostberg, Henrik, and Sven-Marten Samuelsson. 1994. Occupational
retirement in women due to age. Scandinavian Journal of Social Medicine
2:90-6.
Quadagno, Jill, and Joseph Quinn. 1997. Does social security
discourage work? In Social Security in the 21st century, edited by Eric
R. Kingson and James H. Schulz. New York: Oxford University Press, pp.
127-46.
Rashad, Inas. 2006. Structural estimation of caloric intake,
exercise, smoking, and obesity. Quarterly Review of Economics and
Finance 46:268-83.
Rice, Nigel, Jennifer Roberts, and Andrew M. Jones. 2006. Sick of
work or too sick to work? Evidence on health shocks and early retirement
from the BHPS. Health, Econometrics and Data Group Working Paper 06113,
University of York.
Saffer, Henry, and Dhaval Dave. 2005. The effect of alcohol
consumption on the earnings of older workers. In Advances in health
economics and health services research, volume 16, edited by Bjorn
Lindgren and Michael Grossman. Greenwich, CT: JAI Press, pp. 61-90.
Salokangas, Raimo K., and Matti Joukamaa. 1991. Physical and mental
health changes in retirement age. Psychotherapy and Psychosomatics
55:100-7.
Sternberg, Esther M. 2001. The balance within: The science
connecting health and emotions. New York: W.H. Freeman.
Sugisawa, Atsudo, Hidehiro Sugisawa, Yomei Nakatani, and Hiroshi
Shibata. 1997. Effect of retirement on mental health and social
well-being among elderly Japanese. Nippon Koshu Eisei Zasshi 44:123-30.
Sugisawa, Hidehiro, Hiroshi Shibata, Gavin W. Hougham, Yoko
Sugihara, and Jersey Liang. 2002. The impact of social ties on
depressive symptoms in U.S. and Japanese elderly. Journal of Social
Issues 58:785-804.
Szinovacz, Maximiliane E., and Adam Davey. 2004. Retirement
transitions and spouse disability: Effects on depressive symptoms.
Journals of Gerontology Series B, Psychological Sciences and Social
Sciences 59:S333-42.
Tessier, Daniel M., and Veronique Lassmann-Vague, V. J. 2007.
Diabetes and education in the elderly. Diabetes & Metabolism
33:S75-8.
Tuomi, Kaija, Erkki Jarvinen, Leena Eskelinen, Juhani Ilmarinen,
and Matti Klockars. 1991. Effect of retirement on health and work
ability among municipal employees. Scandinavian Journal of Work,
Environment and Health 17:75-81.
Wong, Nathan D. 2007. Metabolic syndrome: Cardiovascular risk
assessment and management. American Journal of Cardiovascular Drugs
7:259-72.
(1) Recent data suggest a slight upturn in the trend towards early
retirement. However, it is not clear whether this reflects a structural
reversal or cyclical factors.
(2) As of 2002, the retirement age for full Social Security
eligibility was raised to 67 for those born in 1960 or later. (There is
a gradual increase in the retirement age from 65 to 67 for those born
between 1937 and 1960. Those born in 1938 fully retire at 65 and 2
months; those born in 1955 retire at 66 and 2 months, and so on.)
(3) These are relative effects evaluated at the overall sample
mean. The absolute effects of complete retirement are as follows: an
increase in the number of mobility difficulties by 0.08-0.10, an
increase in the number of difficulties in daily activities by 0.01-0.03,
an increase in the number of illness conditions by 0.05-0.08, and an
increase in depression symptoms by 0.08-0.11. Outcomes are defined in
Appendix 1.
(4) See, for example, Anderson and Burkhauser (1985), Bazzoli
(1985), and Rice, Roberts, and Jones (2006).
(5) More discussion on this can be found in Dave, Rashad, and
Spasojevic (2006).
(6) Sternberg (2001) documents how physical and psychological
stresses can lead to illness by adversely affecting immune and hormonal
responses. The direction of causality is not well established and may
run in both directions.
(7) See for example Lee and Skerrett (2001) and Franco et al.
(2005).
(8) In fact, it would be implausible (and we exploit this as a
specification check) to find that retirement has significant effects on
health shocks that are independent of individual behaviors.
(9) The health outcomes function is based on the demand for health
model in Grossman (1972). The retirement function is based on the
standard labor supply model (for example, see Borjas 2004). Intercepts
are suppressed for convenience.
(10) It can be shown that the bias due to structural endogeneity is
equal to , which is positive if [R.sub.it] and [[epsilon].sub.it] are
positively correlated.
(11) This is equivalent to a differenced specification with
individual fixed effects. Thus, the pre post difference in health status
is compared across individuals retiring at different ages, conditional
on the sample being healthy in all waves prior to retirement.
(12) Blacks, Hispanics, and Florida residents are oversampled.
Sampling weights are provided to adjust for unequal probabilities of
sample selection.
(13) Models were also estimated with alternate measures, including
net household assets and net household income. The results are not
materially affected. Since these measures are missing for a larger
proportion of the sample, reported specifications control for income
from all sources instead.
(14) Details on these variables are provided in Appendix 1.
(15) Questions on tolerance towards risk are asked only once to
each individual, and thus these variables do not vary over time in the
data set. See Barsky et al. (1997) for a detailed analysis of the risk
preference module in the HRS.
(16) Standard errors in all models are corrected for
autocorrelation at the individual level using STATA's cluster
option. Due to large sample sizes, we estimate OLS and linear fixed
effects models for all outcome variables. The within transformation in
the linear models also bypasses the incidental parameters problem in
estimating individual fixed effects that exist for some nonlinear
models.
(17) Blacks and other races are of significantly poor health
relative to whites. Prior studies document that education makes
individuals more efficient in producing health, and hence educated
individuals have better health outcomes (Grossman and Kaestner 1997).
Married individuals are also healthier, as are nonreligious individuals.
The marginal effect of income indicates that health is a normal good.
One of the channels by which retirement may affect health is through
income (Ettner 1996). Models which exclude income (not reported) yield
marginal effects of retirement on poor health outcomes that are only
slightly larger in magnitude. This indicates that the decline in income
upon retirement is not the main driver of the decline in health.
(18) This falsification test is not a perfect one. Evidence has
been put forth suggesting that some types of cancer are affected by
lifestyle, stressing good nutrition and physical activity in cancer
prevention (Calle et al. 2003). However, if large negative effects of
retirement on cancer are found for non-risk-engaging individuals, then
the specifications may still be reflecting endogeneity bias.
(19) The semi-elasticities represent the effect for the average
individual in the HRS sample, for transition from work to full
retirement. Due to small sample means for these adverse health measures,
small absolute effects can translate to large relative effects.
Assessing the effects for a one-standard deviation change in the
probability of retirement yields magnitudes that are about one-half
those reported in the text. It should be noted that these effects are
strictly applicable only to the pre-retirement healthy group of
individuals due to nonrandom sorting of pre-retirement healthy and
unhealthy individuals. As expected, the pre-retirement healthy group
differs along observable characteristics from those excluded in this
analysis. The average individual in this sample is more likely to be a
married, non-black, male who is more future-oriented and has about a
half-year more schooling, 16% more income, and more educated parents,
relative to the excluded individuals. To the extent that retirement may
magnify some of the channels for those who are unhealthy prior to
retirement, the decline in health post-retirement may be larger. In this
respect, these effects may be interpreted as lower-bound estimates.
(20) We thank an anonymous referee for highlighting this point.
(21) Models are also estimated, explicitly controlling for health
insurance status, history of coverage (number of prior waves respondent
was insured), and whether the respondent has access to retiree coverage
through their employer or their spouse's employer. There are no
significant differences in the results.
(22) We thank two anonymous referees for this suggestion.
(23) We thank an anonymous reviewer for this suggestion.
(24) Negative effects in the first post-retirement wave also may
potentially reflect direct transition or adjustment effects of
retirement, though large magnitudes may still cast doubt on a causal
interpretation.
(25) For this analysis, the sample is limited to non-retired
individuals. If retired individuals are included in the sample, the
pseudo-indicator may still pick up subsequent negative health effects of
actual retirement.
(26) In the HRS, the primary mode in retirement age is 62, followed
by 60 and 65. Examining only those individuals who retired at age 65
yields similar effects though they are imprecisely estimated due to
reduced sample sizes. In order to gauge the timing of moving into a sick
state, hazard models of poor health against retirement were also
implemented. For both the full and preferred samples, there is positive
duration dependence and retirement is found to increase the hazard of
subsequent poor health. Results are available upon request.
(27) Results are presented for the composite measures of physical
and mental health. Estimates for the separate illness conditions (such
as diabetes, high blood pressure, and heart disease) follow the same
pattern.
(28) Since the specification is limited to individuals who were
physically and mentally healthy pre-retirement, controlling for age and
individual fixed effects, it is unlikely that the post-retirement
worsening in health is significantly related to their work.
(29) While mandatory retirement was widespread in the United States
in the 1960s and 1970s, it was abolished in 1986 and is no longer in
practice due to anti-age-discrimination laws (some exceptions remain for
state and local police, firefighters, federal law enforcement and
corrections officers, air traffic controllers, and commercial airline
pilots). Since the HRS begins in 1992, this does not permit the use of
compulsory retirement rules for broad segments of the population as
exogenous shocks to retirement.
(30) If evaluated at the counterfactual prediction for each retired
observation and averaged over individuals, the semi-elasticities are
necessarily larger due to better health pre-retirement: about a 30%
increase in mobility and ADL difficulties, 22% increase in illness
conditions, and 20% increase in depressive symptoms.
(31) An alternative explanation involves hyperbolic discounting and
time-inconsistent preferences. When the individual is working, the
investment return from staying healthy in the form of higher income and
productivity provides a commitment device to continue investing in
health. Indeed, about 30% of Americans report no regular physical
activity outside of work. The individual may retire, knowing that he
will have more free time and thinking that he will continue to stay
active and invest in a healthy lifestyle. However, upon retirement, a
retiree with hyperbolic discounting may keep postponing such investments
(for example, joining a gym, eating healthy, staying active, or quitting
smoking) which in turn may adversely impact health.
(32) Alan Greenspan headed up the 1983 bipartisan commission that
raised the Social Security payroll tax and enacted an increase in the
retirement eligibility age. He continued thereafter to press for further
increases in the retirement age, given the improving feasibility of work
at older ages.
(33) See Mitchell (1992) and Quadagno and Quinn (1997), for
instance. It should be noted that Social Security's delayed
retirement credit has been increasing on a phased basis for individuals
born after 1928. While the initial credit rate of 4% may have been less
than actuarially fair, the applicable credit rate for prospective
retirees born after 1942 is 8%. This would seem, in fact, to be
actuarially fair.
Dhaval Dave, * Inas Rashad, ([dagger]) and Jasmina Spasojevic
([double dagger])
* Department of Economics, Bentley College and National Bureau of
Economic Research, 175 Forest Street, AAC 195, Waltham, MA 02452-4705,
USA; E-mail ddave@bentley.edu; corresponding author.
([dagger]) Department of Economics, Georgia State University &
National Bureau of Economic Research, Andrew Young School of Policy
Studies, P.O. Box 3992, Atlanta, GA 30302-3992, USA; E-mail
irashad@gsu.edu.
([double dagger]) Department of Public Affairs, Metropolitan
College, School for Public Affairs and Administration, 75 Varick Street,
New York, NY 10013, USA; E-mail jspasojevic@metropolitan.edu.
Table 1. Weighted Sample Means (a)
Variable All Retired
Retirement
Complete retirement 0.379 (0.485) 1.000 (0.000)
Partial retirement 0.115 *** (0.319) --
Health outcomes
Good health 0.489 *** (0.500) 0.369 (0.483)
Poor health 0.059 *** (0.235) 0.116 (0.320)
Mobility difficulties 0.700 *** (1.195) 1.144 (1.484)
Activities of daily living
(ADL) difficulties 0.160 *** (0.603) 0.314 (0.849)
Illness conditions 1.308 *** (1.168) 1.721 (1.269)
Diabetes 0.126 *** (0.332) 0.175 (0.380)
Heart disease 0.166 *** (0.372) 0.253 (0.435)
Stroke 0.043 *** (0.203) 0.078 (0.269)
High blood pressure 0.420 *** (0.494) 0.511 (0.500)
Arthritis 0.447 *** (0.497) 0.564 (0.496)
Psychological problems 0.108 *** (0.310) 0.143 (0.350)
Center for Epidemiologic
Studies Depression (CESD)
Scale 1.244 *** (1.799) 1.541 (1.984)
Cancer 0.091 *** (0.288) 0.129 (0.336)
Socio-demographic
Age 61.437 *** (7.083) 66.169 (6.378)
Male 0.510 *** (0.500) 0.476 (0.499)
Black 0.095 *** (0.293) 0.104 (0.305)
Other race 0.036 *** (0.187) 0.031 (0.175)
Hispanic 0.058 *** (0.234) 0.050 (0.218)
Education 12.779 *** (2.972) 12.186 (3.042)
Married 0.695 *** (0.460) 0.659 (0.474)
No religious preference 0.069 *** (0.253) 0.060 (0.238)
Income 16.884 *** (24.867) 9.931 (13.665)
Health insurance 0.935 *** (0.246) 0.968 (0.176)
Mother's age 75.349 *** (13.661) 75.678 (14.730)
Father's age 71.310 *** (14.013) 71.106 (14.381)
Mother's education 0.727 *** (0.445) 0.668 (0.471)
Father's education 0.644 *** (0.479) 0.584 (0.493)
Native born 0.918 *** (0.275) 0.929 (0.256)
Risk averse 0.634 *** (0.482) 0.666 (0.472)
Planning horizon, 5-10 years 0.295 *** (0.456) 0.272 (0.445)
Planning horizon, more than
10 years 0.104 (0.305) 0.101 (0.302)
Physical activity/stress
Vigorous physical activity 0.399 *** (0.490) 0.360 (0.480)
Physical work 0.353 ** (0.478) 0.362 (0.481)
Stressful work 0.543 *** (0.498) 0.559 (0.496)
Observations 77,194 31,411
Variable Non-Retired
Retirement
Complete retirement 0.000 (0.000)
Partial retirement --
Health outcomes
Good health 0.566 (0.496)
Poor health 0.022 (0.146)
Mobility difficulties 0.443 (0.894)
Activities of daily living
(ADL) difficulties 0.063 (0.335)
Illness conditions 1.057 (1.024)
Diabetes 0.096 (0.294)
Heart disease 0.111 (0.314)
Stroke 0.021 (0.143)
High blood pressure 0.364 (0.481)
Arthritis 0.380 (0.485)
Psychological problems 0.086 (0.281)
Center for Epidemiologic
Studies Depression (CESD)
Scale 1.056 (1.643)
Cancer 0.067 (0.250)
Socio-demographic
Age 58.365 (5.621)
Male 0.535 (0.499)
Black 0.089 (0.285)
Other race 0.039 (0.194)
Hispanic 0.063 (0.243)
Education 13.181 (2.833)
Married 0.718 (0.450)
No religious preference 0.075 (0.263)
Income 21.303 (28.996)
Health insurance 0.914 (0.280)
Mother's age 75.165 (12.892)
Father's age 71.428 (13.759)
Mother's education 0.769 (0.422)
Father's education 0.684 (0.465)
Native born 0.911 (0.285)
Risk averse 0.618 (0.486)
Planning horizon, 5-10 years 0.309 (0.462)
Planning horizon, more than
10 years 0.105 (0.307)
Physical activity/stress
Vigorous physical activity 0.421 (0.494)
Physical work 0.346 (0.476)
Stressful work 0.531 (0.499)
Observations 44,799
(a) Data are for individuals ages 50 to 75 from Waves 1 to 7 of
the Health and Retirement Study (HRS). Standard deviations are in
parentheses. Number of observations listed represents the maximum
number. For some variables, the actual sample size is slightly less
due to missing information. Retired and non-retired samples exclude
individuals who are partially retired. Asterisks denote that
the difference between the retired and non-retired samples is
statistically significant as follows: *** significant at the 1%
level; ** significant at the 5% level; * significant at the 10%
level.
Table 2. Retirement Effect on Aspects of Health Using Alternative
Samples (a)
Specification
2
1 Individual
Fixed
Dependent Variable Extended Effects
Poor health 0.1163 *** 0.0496 ***
(0.0041) (0.0035)
[1.971] [0.841]
Mobility difficulties 0.6593 *** 0.2389 ***
(0.0202) (0.0152)
[0.942] [0.341]
Activities of daily 0.2643 *** 0.0991 ***
living (ADL) (0.0105) (0.0088)
difficulties [1.652] [0.619]
Illness conditions 0.4972 *** 0.1038 ***
(0.0198) (0.0101)
[0.380] [0.079]
Diabetes 0.0657 *** 0.0128 ***
(0.0058) (0.0035)
[0.521] [0.102]
Heart disease 0.0987 *** 0.0270 ***
(0.0064) (0.0039)
[0.595] [0.163]
Stroke 0.0428 *** 0.0174 ***
(0.0036) (0.0026)
[0.995] [0.405]
High blood pressure 0.0910* ** 0.0114 **
(0.0084) (0.0045)
[0.217] [0.027]
Arthritis 0.1170 *** 0.0235 ***
(0.0084) (0.0050)
[0.262] [0.053]
Psychological 0.0824 *** 0.0124 ***
problems (0.0054) (0.0030)
[0.763] [0.115]
Depression (CESD) 0.4832 *** 0.1812 ***
scale (0.0267) (0.0236)
[0.388] [0.146]
Cancer (b) 0.0145 * 0.0187 ***
(0.0088) (0.0060)
[0.209] [0.268]
Specification
4
Individual
Fixed Effects
3
Healthy
Individual Pre-Retirement
Fixed Effects
Consistently
Healthy Insured in
Dependent Variable Pre-Retirement All Waves
Poor health 0.0268 *** 0.0252 ***
(0.0059) (0.0059)
[0.454] [0.427]
Mobility difficulties 0.1579 *** 0.1617 ***
(0.0264) (0.0273)
[0.226] [0.231]
Activities of daily 0.0267 *** 0.0237 **
living (ADL) (0.0100) (0.0116)
difficulties [0.167] [0.148]
Illness conditions 0.0847 *** 0.0699 ***
(0.0235) (0.0267)
[0.065] [0.053]
Diabetes 0.0131 ** 0.0142 **
(0.0064) (0.0058)
[0.104] [0.113]
Heart disease 0.0146 ** 0.0084
(0.0072) (0.0084)
[0.088] [0.051]
Stroke 0.0076 ** 0.0052
(0.0034) (0.0035)
[0.176] [0.121]
High blood pressure 0.0088 0.0089
(0.0112) (0.0125)
[0.021] [0.021]
Arthritis 0.0392 *** 0.0289 *
(0.0132) (0.0157)
[0.088] [0.065]
Psychological 0.0006 0.0031
problems (0.0050) (0.0055)
[0.006] [0.023]
Depression (CESD) 0.1157 ** 0.1367 **
scale (0.0546) (0.0626)
[0.093] [0.110]
Cancer (b) -0.00065 0.0081
(0.0083) (0.0094)
[0.009] [0.089]
(a) Standard errors are robust clustered at the individual level
and reported in parentheses. Semi-elasticity of health outcome with
respect to retirement, evaluated at the sample mean, is reported in
brackets. Sample is limited to individuals ages 50 to 75. Significance
is defined as follows: *** significant at the 1% level; ** significant
at the 5% level; * significant at the 10% level. Each cell represents
the marginal effect of Complete Retirement on the given health outcome
from a separate regression. Sample sizes range from 53,551 to 75,752
(specifications 1 & 2) and from 4951 to 5289 (specifications 3 & 4).
Each specification includes the same covariates listed in Appendix 2.
(b) Sample is limited to never-smokers and moderate drinkers.
Table 3. Exploring Unobserved Heterogeneity in a Model of Health:
Health Shocks and Preferences (a)
Specification
1 2 3
Individual Individual Individual
Fixed Fixed Fixed
Effects Effects Effects
Healthy
-- Pre-Retirement --
Did not report Did not report Health not a
worsening of worsening of reason for
Dependent Variable health health retirement
Mobility 0.1196 *** 0.0970 *** 0.0683 ***
difficulties (0.0178) (0.0224 (0.0143)
[0.171] [0.139] [0.098]
Activities of 0.0523 *** 0.0170 * 0.0279 ***
daily living (0.0094) (0.0098) (0.0069)
(ADL) [0.332] [0.106] [0.174]
difficulties
Illness 0.0650 *** 0.0692 *** 0.0446 ***
conditions (0.0113) (0.0231) (0.0105)
[0.050] [0.057] [0.034]
Depression 0.0693 ** 0.0759 0.0487 **
(CESD) scale (0.0318) (0.0546) (0.0242)
[0.056] [0.064] [0.039]
Specification
4 5
Individual Individual
Fixed Fixed
Effects Effects
Healthy Healthy
Pre-Retirement Pre-Retirement
Health not a
reason for Retired: To do
Dependent Variable retirement other things
Mobility 0.0804 *** 0.1143 ***
difficulties (0.0212) (0.0316)
[0.115] [0.163]
Activities of 0.0074 0.0332 ***
daily living (0.0070) (0.0117
(ADL) [0.046] [0.208]
difficulties
Illness 0.0488 ** 0.0835 ***
conditions (0.0230) (0.0286)
[0.037] [0.064]
Depression 0.0432 0.0477
(CESD) scale (0.0550) (0.0629)
[0.035] [0.038]
Specification
6 7
Individual Individual
Fixed Fixed
Effects Effects
Healthy Healthy
Pre-Retirement Pre-Retirement
Retired: To
spend more
time with Retired: Did
Dependent Variable family not like work
Mobility 0.0924 *** 0.1159 **
difficulties (0.0280) (0.0481)
[0.132] [0.166]
Activities of 0.0156 0.0102
daily living (0.0137) (0.0133)
(ADL) [0.098] [0.064]
difficulties
Illness 0.0745 *** 0.1881 ***
conditions (0.0274) (0.0437)
[0.057] [0.1441
Depression 0.0334 -0.0216
(CESD) scale (0.0685) (0.1213)
[0.028] [-0.017]
(a) See notes to Table 2. Sample sizes range from 4519 to 5289
(specifications 1-5) and 970 to 3193 (specifications 6-7).
Significance is defined as follows: *** significant at the 1% level;
** significant at the 5% level; * significant at the 10% level.
Table 4. Specification Checks (a)
Specification
Timing (b)
1 2
Individual Individual
Fixed Fixed
Effects Effects
Healthy
Full Sample Pre-Retirement
Decomposition: Decomposition:
timing of timing of
Dependent retirement retirement
Variable effect effect
Mobility Post-retirement 0.1878 *** 0.0312
difficulties Wave 1 (0.0225) (0.0397)
Post-retirement 0.2168 *** 0.0938 **
Waves 2+ (0.0275) (0.0406)
Activities of Post-retirement 0.0987 *** 0.0084
daily living Wave 1 (0.0130) (0.0172)
(ADL) Post-retirement 0.1331 *** 0.0407 **
difficulties Waves 2+ (0.0159) (0.0208)
Illness Post-retirement 0.1124 *** -0.0321
conditions Wave 1 (0.0149) (0.0344)
Post-retirement 0.1256 *** 0.1801 ***
Waves 2+ (0.0198) (0.0477)
Depression Post-retirement 0.1475 *** 0.0566
(CESD) Wave 1 (0.0365) (0.0807)
scale Post-retirement 0.1416 *** 0.0620
Waves 2+ (0.0440) (0.0822)
Specification
Timing (b)
3
Individual
Fixed
Effects
Healthy
Pre-Retirement
Restricting
effect to
first
Dependent post-retirement
Variable wave
Mobility Post-retirement 0.0326
difficulties Wave 1 (0.0398)
Post-retirement --
Waves 2+
Activities of Post-retirement 0.0029
daily living Wave 1 (0.0130)
(ADL) Post-retirement --
difficulties Waves 2+
Illness Post-retirement 0.0300
conditions Wave 1 (0.0472)
Post-retirement --
Waves 2+
Depression Post-retirement 0.0323
(CESD) Wave 1 (0.1228)
scale Post-retirement --
Waves 2+
Specification
Pseudo-Retirement (c)
4 5
Individual
Extended Fixed Effects
Dependent Healthy
Variable -- Pre-Retirement
Mobility Post-retirement
difficulties Wave 1 0.0858 *** -0.0061
Post-retirement (0.0213) (0.0239)
Waves 2+
Activities of Post-retirement
daily living Wave 1 0.0257 *** 0.0076
(ADL) Post-retirement (0.0084) (0.0051)
difficulties Waves 2+
Illness Post-retirement
conditions Wave 1 0.1251 *** -0.0043
Post-retirement (0.0236) (0.0119)
Waves 2+
Depression Post-retirement 0.1841 *** -0.0028
(CESD) Wave 1
scale Post-retirement (0.0368) (0.1056)
Waves 2+
Specification
Alternate Identification Methods
6 7
Individual
Fixed Effects,
Dependent Instrumental Retired at
Variable Variables (d) Age 62 (c)
Mobility Post-retirement 0.2299 *
difficulties Wave 1 (0.1357) 0.2005 ***
Post-retirement F = 12.15 *** (0.0447)
Waves 2+ Hansen J = 2.62
Activities of Post-retirement 0.1055 *
daily living Wave 1 (0.0616) 0.0885 ***
(ADL) Post-retirement F = 112.15 *** (0.0239)
difficulties Waves 2+ Hansen J = 0.55
Illness Post-retirement 0.2472
conditions Wave 1 (0.1545) 0.1030 ***
Post-retirement F = 112.15 *** (0.0290)
Waves 2+ Hansen J = 0.38
Depression Post-retirement 0.3589 *
(CESD) Wave 1 (0.2163) 0.1120
scale Post-retirement F = 122.14 *** (0.0724)
Waves 2+ Hansen J = 0.38
(a) The extended specification includes covariates listed in Appendix 2.
The individual fixed effects specification also includes married,
income, and fixed effects for age, year, and census division and is
limited to individuals who had no mobility difficulties, no illness
conditions, and no psychological problems in the wave prior to
retirement. Standard errors are robust clustered at the individual level
and reported in parentheses. Significance is defined as follows: ***
significant at the 1% level; ** significant at the 5% level; *
significant at the 10% level.
(b) See text and notes to Table 2.
(c) Each cell represents the marginal effect of pseudo-retired indicator
on the given health outcome from a separate regression. The sample is
further limited to non-retired individuals.
(d) Each cell represents the marginal effect of retired on the given
health outcome from a separate IV regression. The excluded instruments
are indicators for whether the spouse is completely or partially
retired. The sample is limited to married individuals who reported
that they plan on retiring at the same time as their spouse, and
they are not concerned about inadequate retirement income. Standard
errors are reported in parentheses. The joint F-statistic on the
excluded instruments is reported. Hansen J is the Chi-squared
statistic on the test of overidentifying restrictions.
(e) Each cell represents the marginal effect of retired on the given
health outcome from a separate regression. The sample is limited
to individuals who retired at age 62. Standard errors are reported
in parentheses.
Table 5. Stratified Samples (a)
Dependent Variable
Mobility
Specification Difficulties ADL Difficulties
Unmarried 0.1737 *** (0.0728) 0.0349 ** (0.0137)
Married 0.1487 ** (0.0262) 0.0160 (0.0115)
Job required physical
effort 0.2121 *** (0.0493) 0.0509 *** (0.0187)
Job did not require
physical effort 0.1303 *** (0.0321) 0.0192 (0.0134)
Non-participation in
vigorous physical
activity
post-retirement 0.2627 *** (0.0441) 0.0522 *** (0.0191)
Participation in vigorous
physical activity
post-retirement 0.0530 * (0.0292) 0.0048 (0.0079)
Job was non-stressful 0.1706 *** (0.0404) 0.0392 *** (0.0133)
Job was stressful 0.1503 *** (0.0369) 0.0267 (0.0165)
Retirement was involuntary
(Excluding health as a
reason) 0.1845 *** (0.0599) 0.0169 (0.0233)
Retirement was voluntary 0.0504 ** (0.0244) 0.0025 (0.0094)
Complete retirement
(reproduced from
Table 3) 0.1563 *** (0.0295) 0.0268 ** (0.0112)
Partial retirement (b) 0.0022 (0.0288) 0.0159 * (0.0085)
Dependent Variable
Specification Illness Conditions Depression Scale
Unmarried 0.0865 * (0.0450) 0.2215 * (0.1281)
Married 0.0865 *** (0.0277) 0.0903 (0.0621)
Job required physical
effort 0.1571 *** (0.0413) 0.1225 (0.1039)
Job did not require
physical effort 0.0553 * (0.0316) 0.1621 ** (0.0695)
Non-participation in
vigorous physical
activity
post-retirement 0.0862 ** (0.0368) 0.2349 *** (0.0831)
Participation in vigorous
physical activity
post-retirement 0.0847 ** (0.0301) -0.0362 (0.0718)
Job was non-stressful 0.1054 *** (0.0349) 0.1726 ** (0.0863)
Job was stressful 0.0843 ** (0.0347) 0.1477 * (0.0833)
Retirement was involuntary
(Excluding health as a
reason) 0.1079 * (0.0618) 0.1440 (0.1380)
Retirement was voluntary 0.0330 (0.0267) 0.0454 (0.0631)
Complete retirement
(reproduced from
Table 3) 0.0834 *** (0.0263) 0.1145 * (0.0616)
Partial retirement (b) 0.0549 * (0.0310) -0.0803 (0.0718)
(a) Each cell represents the marginal effect of retired on the given
health outcome from a separate regression. All specifications include
married (except in samples stratified by married), income, and fixed
effects for age, year, census division, and the individual. Standard
errors are robust clustered at the individual level and reported in
parentheses. Sample is limited to individuals ages 50 to 75, who had
no mobility difficulties, no illness conditions, and no psychological
problems in the wave prior to retirement. Significance is defined as
follows: *** significant at the 1% level; ** significant at the 5%
level; * significant at the 10% level.
(b) Sample excludes individuals who are completely retired.
