Does Working Longer Make People Healthier and Happier?

(1)

Does Working Longer Make People Healthier and Happier?

Calvo, Esteban

CENTER FOR RETIREMENT RESEARCH AT BOSTON COLLEGE

2006

Online at https://mpra.ub.uni-muenchen.de/5606/

MPRA Paper No. 5606, posted 05 Nov 2007 UTC

(2)

By Esteban Calvo*

D OES W ORKING L ONGER M AKE P EOPLE

H EALTHIER AND H APPIER ?

In t r o d u ct io n

Financing retirem ent is one of the m ajor challenges facing an aging U.S. population. If individuals continue to retire in their early 6 0 s, m any will be hard pressed to m aintain an adequate standard of living throughout retirem ent due to the declining role of Social Security, the shift to 40 1(k) plans, and low per- son al savin g rates. Com bin e the retirem en t in com e crunch with the dram atic increase in life expectancy, and continued em ploym ent in later life appears to be an attractive option.

While it is clear that working longer would benefit older Am ericans financially, less attention has focused on the n on -m on etary effects of work at older ages. Thisbriefaddresses the im pact of late-life paid work on physical and psychological well-being. The first section reviews the literature on work at older ages and elderly well-being. The second section describes the an alysis. The third an d fourth section s presen t the results. The fifth section iden tifies vulnerable groups. A final section offers concluding thoughts.

Fin d in g s f r o m Pr evio u s St u d ies

Much of the literature relevant to older workers and well-being is actually focused on other issues, such as healthy agin g an d retirem en t an d n ot on the effects of work per se. Studies of healthy aging suggest that, for m any individuals, working to an older age could have a n um ber of positive physical an d psychological effects. The prim ary reason is that opportunities for active engagem ent tend to decline in later life, and this trend tends to adversely im pact overall well- being.¹ Sin ce m ost jobs require workers to en gage in a num ber of productive and social activities, working longer m ight bring a num ber of benefits associated with these activities. Prolonged em ploym ent during later life allows m any older workers to continue doing activities sim ilar to those perform ed in m iddle-age.

In this sense, working longer appears to be helpful in m aintaining m eaning and a sense of purpose in life, as well as adaptin g to agin g.²

Research on retirem ent, as opposed to work, also provides im portant clues regarding the potential consequences of working longer. Work is fundam ental to one's identity; therefore, leaving one's job im plies a partial loss of identity, which m ay produce lower lev- els of psychological well-being. Conversely, retire-

* Esteban Calvo is a graduate student in Sociology at Boston College and a graduate research assistant at the Center for Retirement Research at Boston College. The author thanks the following people for their comments on and other forms of help in connection with this brief: Paula Erráz uriz , Andy Eschtruth, Margaret Ford, Jessica Johnson, Alicia H . Munnell, Steven Sass, Mauricio Soto, and John W illiamson. This brief was adapted from a longer article by Esteban Calvo (forthcoming).

Search for other publications on this topic at:

www.bc.edu/ crr/ pub_wor.shtm l LEARN MORE

(3)

m ent can be viewed as a gain in freedom to pursue other activities or interests, thus enhancing older individuals' well-being. To the extent that retirem ent im plies the loss of valued rewards, it will be viewed negatively and perceived as a difficult adjustm ent.

But retirem ent is also an experience of release, though its benefits vary between long-term and recent retirees.³ An other im portan t factor in the in di- vidual’s response to retirem ent is the subjective sense of control. Voluntarily leaving a job could provide a sense of control in the work/ retirem ent decision, thus m oderating any negative aspects of retirem ent.⁴ Although retirem en t m ay n ot be a choice for som e particular groups (e.g. people constrained by financial or health problem s), the vast m ajority of retirees cur- rently leave their jobs voluntarily.⁵

Although research on healthy aging and retire- m ent provides valuable insights, only a few studies have directly addressed the effects of late-life work on the physical and psychological well-being of older Am erican s. Most of these studies have foun d positive effects of work on well-being. The workplace appears as an im portant source of friends, confidants, and social support, even after long-term retirem ent as retirees tend to retain the network of friends they developed when they were workin g.⁶ Job-related rewards seem to have a positive im pact on m ental health.⁷ Perform in g m ore than 10 0 an n ual hours of volun teer or paid work beyon d the retirem en t age has an independent beneficial effect on older adults' self- rated health an d survival.⁸ It has also been suggested that the conditions of work are m ore influential on physical an d m en tal health than the pure am oun t of work. Nam ely, older em ployees who work in low- stress jobs with the hours they desire, experience better health. And surprisingly, physically dem anding jobs have a positive effect on the physical health of those aged 6 5 and over.⁹

Most of the studies m en tion ed above iden tified relation ships between work an d well-bein g usin g data from a single point in tim e. These cross-sectional studies cannot establish causal relationships, because people with higher well-being m ay be m ore likely to rem ain at work. For exam ple, individuals with no physical disability are m ore likely to con tin ue working than those with a physical lim itation. Only longitudinal studies — which follow the sam e individuals over tim e — can differentiate whether health problem s cause cessation of work or con tin ued work causes better health. The an alysis below builds on the previous studies, but uses a longitudinal design, includes m ore control variables, and expands the n otion of well-bein g to in clude m ore in dicators.

Met h o d o l o g y

The following analysis assesses the im pact of individuals' work status in 20 0 0 on their physical an d psychological well-being in 20 0 2. It uses longitudinal data drawn from the H ealth an d Retirem en t Study (H RS) an d the RAND-H RS data base.¹⁰ The sam ple analyzed is com posed of individuals aged 59 to 6 9 who were workin g or n ot-workin g in the year 20 0 0 an d were alive in 20 0 2. People dead in 20 0 2 were included in the regressions analyzing m ortality.¹¹

The variable "paid work status" m easures if people are workin g for pay or n ot. It is related to the variable labor force participation, but is not the sam e.

As an illustration, for the sam ple analyzed, 74 percen t of those n ot workin g for pay are retired, 18 percent are not in the labor force (e.g. wom en who never worked outside of the hom e), 6 percen t are disabled, and 2 percent are unem ployed or partially retired.

Well-being was assessed by using a com prehen- sive set of six m easures:¹²

.

self-rated health;

.

self-rated m em ory;

.

activities of daily livin g (ADL) such as bathin g, eatin g, an d dressin g;

.

instrum ental activities of daily living (IADL) such as handling m oney, shopping, and cookin g;

.

m ood indicators such as depression, loneli- ness, and happiness; and

.

m ortality.¹³

The study controls for previous well-being by including status in 19 9 8.¹⁴ It also includes variables for healthy behaviors, such as whether the respon- den t sm okes, the n um ber of drin ks the respon den t has per day, and whether the respondent exercises vigorously. Finally, this study also controls for dem o- graphic and socioeconom ic factors.¹⁵

In sum , the m ethodology of this study enables us to determ ine whether health problem s cause cessation of work or continued work causes better health (see Figure 1).

Ho w Wo r k at Ol d er Ag es Af f ect s Heal t h

The results of the analysis of the im pact of later-life work on well-being are presented in Appendix Table 1 (Model 1 through Model 3). They show that those who worked in 20 0 0 ten ded to report greater well- being in 20 0 2 than those who did not work in 20 0 0 .

(4)

FIGURE 2. EFFECTS OFPAIDWORK ATOLDERAGES ON THE PROBABILITY OFREPORTINGFAIR/ POORSELF-RELATED

HEALTH(20 0 2)

Source: Author’s calculations (probit estim ation reporting discrete and m arginal coefficients) using data from the 19 9 8, 20 0 0 , and 20 0 2 waves of the H ealth and Retirement Study.

FIGURE 1. TIMEPERIODSUSED INDATAANALYSIS

Source: Author’s m ethodology.

For exam ple, Figure 2 illustrates that paid work at older ages reduced the probability of reporting fair/ poor health in 20 0 2 by 6 percent. Other variables that had a statistically significant im pact on the probability of reporting fair/ poor health are high- lighted using dark bars. While working, exercising, being white, m arried, and having m ore education an d in com e had a favorable im pact, poor physical and psychological health in 19 9 8, sm oking, being m ale an d H ispan ic had an un favorable im pact on self-rated health.

As illustrated in Appendix Figures 1 through 4, work also had a m odest, but statistically significant, beneficial im pact in the cases of ADL, IADL, m ood indicators, and m ortality. Paid work at older ages reduced the probability of having any ADL lim itation by 2 percen t, of havin g an y IADL lim itation by 2 percent, of reporting negative m ood indicators by 2 percent, and of dying by 1 percent.¹⁶ Paid work did not

significantly affect self-rated m em ory, illustrating that other factors are better predictors than work (e.g.

being m ale and initial self-rated m em ory).

Not surprisingly, well-being baselines and healthy behaviors are related to follow-up physical and psychological well-being. But after controlling for baseline effects and other variables, these findings suggest that there is a causal relationship between work and well-being.¹⁷

Ho w Un d esir abl e Jo bs Af f ect Psych o l o g ical Heal t h

The results presented above suggest that longer working lives have beneficial effects on individuals' physical and psychological well-being. H owever, this analysis did not distinguish between different types of jobs. Previous literature suggests that the characteristics of work m atter, and we all know that jobs are not all the sam e. Work can be backbreaking, stress- ful, and tedious. Therefore, this next exercise explores the hypothesis that undesirable jobs could have detrim en tal effects on in dividuals' well-bein g.

"Undesirable jobs" are those that have excessive dem ands or otherwise cause dissatisfaction.¹⁸ They are the type of jobs that people would probably not choose voluntarily.¹⁹

The an alysis of a subgroup of older workers suggests that having an undesirable job does not change the favorable effects of paid work on self-rated health, ADL and IADL (see results for Model 4 in Appendix Table 1). H owever, it does have an im pact on follow- up m ood indicators and m ortality. H igher job satis-

Work (20 0 0 )

H ispan ic (20 0 0 )

Percent

-25 -20 -15 -10 -5 0 5 10 15 20 25 Married (20 0 0 )

IADL lim it (19 9 8) Neg. m ood (19 9 8)

Male (20 0 0 ) Sm oking (20 0 0 )

Incom e (20 0 0 ) Education (20 0 0 ) Drin ks (20 0 0 ) Exercise (20 0 0 ) Age (20 0 0 )

Diseases (19 9 8) Bad m em ory (19 9 8)

White (20 0 0 )

ADL lim it (19 9 8)

Bad health (19 9 8) 38%

Statistically significant Statistically insignificant 19 9 8

Baselin e

H ealth Work Status

Dem ographic Factors Socioecon om ic

Factors

H ealth

20 0 0 20 0 2

(5)

Source: Author’s calculations using data from the 20 0 0 wave of the H ealth and Retirement Study.

*Note: The sam ple size is 6 ,511.

faction is associated with an im proved m ood. Those who do n ot en joy goin g to work are 5 percen t m ore likely to report negative m ood indicators. Lower job dem an ds are statistically associated with survival but have only a trivial effect.^{2 0}

Vu l n er abl e Gr o u ps

While working longer seem s beneficial for m ost, policym akers should con sider the poten tial n egative con - sequences of forcing all people to work, for exam ple, by raising the early eligibility age for Social Security ben efits.²¹ In rough num bers, people who retire involuntarily constitute about 20 percent of the population analyzed (see Figure 3). For this group, work is not an alternative.^{2 2} Also, 15 percen t of the population analyzed was working in undesirable jobs, and 40 percent was working in less dem anding and m ore satisfying jobs. H owever, if the early retirem ent age were raised, an im portant group of people would be forced to work in late-life, regardless of whether or not they enjoy working or have a highly dem anding job. For this group, continuing work would have opposin g effects: ben eficial in term s of self-rated health, ADL and IADL, but harm ful for their m ood, and to a lesser extent, their survival. For both vulnerable groups, workin g fewer an n ual hours, reducin g work-strain, or participating in other social or productive activities — like volun teerin g — seem m ore ben - eficial than working longer.

Co n cl u sio n

This analysis provides several interesting insights into the relationship between work at older ages and well-being. It suggests that longer working lives will help m ost people m aintain their overall well-being.

While workin g lon ger seem s ben eficial for m ost people, it will likely have negative consequences for som e. The type of job seem s to be a critical factor.

Undesirable jobs can wash out the potential favorable effects of work. Another critical factor is the opportu- nity to continue working. Older workers m ay be will- ingto prolong paid work, but, in order to find a job, they n eed to be ableto work and have a real dem and for their labor. Policym akers need to consider these factors when evaluating proposals to keep people in the labor force.

These findings suggest interesting areas for future research. For exam ple, an increase in the early retirem en t age reduces the ability of people to voluntarily decide their labor force participation. Less control in the work/ retirem ent decision could have an adverse im pact on the well-being of older

in dividuals.^{2 3} Also variations in the benefits of work as people m ove through their 6 0 s could be exam ined.

Even though work at older ages seem s beneficial for m any, the benefits m ay decrease, or stop increasing, after a certain age or am ount of tim e worked per year.

15%

20%

25% 40%

FIGURE3. VULNERABLEGROUPS BYPAIDWORKSTATUS

Working in desirable job Not working (other than forced)

Not working and forced into retirem ent

Working in undesirable job Vulnerable

Groups

(6)

En d n o t es

1 Much evidence suggests that m ore active elderly people are less likely to die or have various physical and m ental illnesses (Brum m ett et. al., 20 0 1; Everard et. al., 20 0 0 ; Rowe and Kahn, 19 9 8; and Siegrist, et al., 20 0 4). Rem aining active and socially engaged seem s to have a positive im pact on iden tity (Erickson et al., 19 86 ). Social integration, social support, and positive interactions tend to be associated with positive well-being, even after controlling for other variables related to social com petence (Cohen, 20 0 4).

Social and productive activities appear as effective as fitness activities in lowering the risk of death (Glass, et. al, 19 9 9 ). And the risk of social isolation is com - parable to that associated with cigarette sm oking (H ouse, 20 0 1).

2 Different authors suggest this interpretation, but none of them identify a com pulsory reason for m al- adjustm ent in retirem ent (Atchley, 19 9 9 ; and Rowe and Kahn, 19 9 8). Num erous studies suggest that non-paid activities, such as volunteering, care-giving, housework, and inform al help to friends, can be as beneficial as paid work for older individuals' well- being. H owever, they do not provide the benefits of additional incom e that paid work does. For research regarding non-paid activities, see: H erzog et al.

(19 9 1); H erzog et al. (19 9 8); Glass et al. (19 9 9 ); Luoh and H erzog (20 0 2); and Morrow-H owell et al. (20 0 3).

3 Bossé et al. (19 9 0 ); Elder (19 9 5); Kim an d Moen (20 0 2); and Kosloski et al. (20 0 1).

4 Volunteerism is an expression of self-efficacy, m as- tery, self-directedness, self-agency, autonom y, self- determ ination, and other form s of sense of control that are paralleled by biological responses that buffer stress reactions. Sim ilarly, the degree of voluntariness in the work/ retirem ent decision partially m edi- ates the well-bein g outcom es of either work or retire- m en t (Ben der an d Jivan , 20 0 5; H erzog et al., 19 9 8;

Gallo et al., 20 0 0 ; and Kim and Moen, 20 0 2).

5 According to the H RS (20 0 2), about 30 percent of the retirees were forced to retire. Prudential (20 0 5) finds that 38 percent fall into this group.

6 Bossé et al. (19 9 0 ).

7 Erickson et al. (19 86 ).

8 Addition al hours do n ot in crease the ben efit. See Luoh an d H erzog (20 0 2).

9 H erzog et al. (19 9 1).

10 The H RS is a n ation ally represen tative pan el survey of U.S. households with in dividuals aged 51-6 1 in 19 9 2 and their spouses. The RAND Center for the Study of Agin g created the RAND-H RS data based on H RS data.

11 The sam ple size varies for each dependent variable accordin g to available data. The data used in all of the regressions were adjusted to the weights of the gener- al population .

12 The effect of work on these six m easures of well- being (given the set of control variables) was estim at- ed using probit regression m odels for each well-being outcom e. Probit regressions fit m axim um -likelihood m odels, allowing the estim ation of the probability of each well-bein g outcom e, con dition al on a set of in dependent variables. Rather than reporting the regular probit coefficients, this briefreports the discrete and m arginal effects, holding all other variables at their m ean, a technique described in Long and Freese, 20 0 4.

13 The six m easures of well-bein g were coded as dum m y variables. For m ore inform ation about related m easurem en ts in the H RS, see Sain t Clair et al.

(20 0 4); Steffick (20 0 0 ); Fonda and H erzog (20 0 4);

an d Ofstedal et al. (20 0 5).

14 Baselines also consider chronic diseases, a binary variable that m easures the presen ce or absen ce of an y of eight chronic diseases: high blood pressure, dia- betes, cancer, lung disease, heart disease, stroke, psy- chiatric problem s, and arthritis. More in form ation about the m easures of chronic diseases can be found in Saint Clair et al. (20 0 4); and Fisher et al. (20 0 5).

15 Gender, race, ethnicity, and m arital status were coded as dum m y variables. H ousehold incom e was m easured in in crem en ts of $10 0 ,0 0 0 dollars for a better graphical presen tation of the data.

16 Control variables generally behaved as expected.

Generally, the corresponding baseline of the dependent variable had the higher coefficient. H owever, self- rated health an d sm okin g were two of the m ost con- stant predictors of well-being outcom es.

(7)

17 Since groups of workers and non-workers were not random ly assigned and data were collected every two years, these results should be interpreted with caution no m atter how rigorous the controls included in the m ode appear.

18 Job satisfaction m easures how m uch people en joy going to work. "Work dem ands" is an index of job characteristics including: physical effort, lifting heavy loads, stooping/ kneeling/ crouching, good eyesight, an d stress. The in dex run s from 0 to 5.

19 Additional data regarding voluntariness and sense of control in relation to work status and work pat- terns are insufficient for the sam ple analyzed.

20 These effects are both sim ilar and contrasting to cross-sectional effects reported in H erzog et al.

(19 9 8). Job satisfaction — un derstood as the m ost intrinsic non-econom ic job reward — positively affects psychological well-being. H owever, this study found detrim ental effects of physical dem ands on older adults' m ood indicators. These results m ust be interpreted with special caution, because people will ten d to rem ain at work if they en joy workin g an d if they do not have a highly dem anding job.

21 Raising the early eligibility age could also reduce the flexibility to work part tim e as a transitional path from full-tim e work to retirem ent.

22 Previous studies also suggest that overall well- being is likely to be im paired for those forced into retirem en t (Ben der an d Jivan , 20 0 5; H erzog et al., 19 9 8; Gallo et al., 20 0 0 ; and Kim and Moen, 20 0 2).

People are forced into retirem ent m ainly due to health reasons and job displacem ent (Prudential, 20 0 5).

23 Fortunately, increasing labor force participation and m aintaining choice are not incom patible (H erzog et al., 19 9 1).

(8)

Ref er en ces

Atchley, Robert C. 19 9 9 . Continuity and Adaptation in Aging: Creating Positive Experiences. Baltim ore, MD:

Johns H opkins University Press.

Bender, Keith A. and Natalia A. Jivan. 20 0 5. W hat Makes Retirees H appy? Issue in Brief28. Chestnut H ill, MA: Center for Retirem ent Research at Boston College.

Bossé, Raym ond, Carolyn M. Aldwin, Michael R.

Levenson, Kathryn Workm an-Daniels, and David J.

Ekerdt. 19 9 0 . "Differences in Social Support am ong Retirees and Workers: Findings from the Norm ative Aging Study." Psychology and Aging5(1): 41-47.

Brum m ett, Beverly H ., John C. Barefoot, Ilene C.

Siegler, Nancy E. Clapp-Channing, Barbara L. Lytle, H ayden B Bosworth, Redford B. William s, and Daniel B. Marx. 20 0 1. "Characteristics of Socially Isolated Patients with Coronary Artery Disease Who Are at Elevated Risk of Mortality." Psychosomatic Medicine 6 3:26 7-272.

Cohen, Sheldon. 20 0 4. "Social Relationship and H ealth." American Psychologist59 :6 76 -6 84.

Elder, G. H ., Jr. 19 9 5. “The Life Course Paradigm : Social Change and Individual Developm ent.” In Moen, P.; Elder, G. Jr. & Lusher, K. (Eds.), Examining Lives in Context: Perspectives on the Ecology of H uman Development.Washington, DC: Am erican

Psychological Association .

Erickson, Erik H ., Joan M. Erikson, H elen Q.

Kivnick. 19 86 . Vital Involvement in Old Age.New York: Norton .

Everard, Kelly M., H elen W. Lach, Edwin B. Fisher, and Carolyn M. Baum . 20 0 0 . "Relationship of Activity and Social Supports to the Functional H ealth of Older Adults." Journal of Gerontology: Social Sciences 55(4): 20 8-S212.

Fisher, Gwenith G., Jessica D. Faul, David R. Weir, and Robert B. Wallace. 20 0 5. “Docum entation of Chronic Diseases Measures in the H ealth and Retirem ent Study (H RS/ AH EAD).” H RS/ AH EAD Documentation Report, Ann Arbor: Survey Research Center University of Michigan.

Fonda, Stephanie and Regula H erzog. 20 0 4.

“Docum entation of Physical Functioning Measured in the H ealth and Retirem ent Study and H ealth Dyn am ics am on g the Oldest Old.” H RS/ AH EAD Documentation Report, Ann Arbor: Survey Research Center University of Michigan.

Gallo, William T., Elizabeth H . Bradley, Michele Siegel, and Stanislav V. Kasl. 20 0 0 . "H ealth Effects of Involuntary Job Loss am ong Older Workers: Findings from the H ealth and Retirem ent Survey." Journal of Gerontology: Social Sciences55B(3): S131-S140 .

Glass, Thom as A., Carlos Mendes de Leon, Richard A Marottoli, and Lisa F Berkm an. 19 9 9 . "Population Based Study of Social and Productive Activities as Predictors of Survival am ong Elderly Am ericans."

British Medical Journal(319 ): 478-483.

H erzog, Regula A., Jam es S. H ouse, and Jam es N.

Morgan. 19 9 1. "Relation of Work and Retirem ent to H ealth and Well-being in Older Age." Psychology and Aging6 (2): 20 1-211.

H erzog, Regula A., H azel R. Markus, Melissa M.

Franks, and Diane H olm berg. 19 9 8. "Activities and Well-being in Older Age: Effects of Self-concept and Educational Attainm ent." Psychology and Aging13(2):

179 -185.

H ouse, Jam es S. 20 0 1. "Social Isolation Kills, but H ow and Why?" Psychosomatic Medicine6 3:273-274.

Kim , Jungm een E. and Phyllis Moen. 20 0 2.

“Retirem ent Transitions, Gender, and Psychological Well-being: a Life-course, Ecological Model." Journal of Gerontology: Psychological Sciences57B(3): P212- P222.

Kosloski, Karl, David Ekerdt, and Stanley De Viney.

20 0 1. "The Role of Job-related Rewards in Retirem ent Planning." Journal of Gerontology: Psychological Sciences56 B(3): 16 0 -16 9 .

Long, J. Scott and Jerem y Freese. 20 0 4. Regression Models for Categorical Dependent Variables Using STATA. Texas: STATA Press.

Luoh, Ming-Ching and Regula A H erzog. 20 0 2.

"Individual Consequences of Volunteer and Paid Work in Old Age: H ealth and Mortality." Journal of H ealth and Social Behavior43(4): 459 -50 9 .

(9)

Morrow-H owell, Nancy, Jim H interlong, Philip Rozario, and Fengyan Tang. 20 0 3. "Effects of Volunteering on the Well-being of Older Adults."

Journal of Gerontology: Social Sciences58B(3): S137- S145.

Ofstedal, Mary B., Gwenith G. Fisher., and Regula A.

H erzog. 20 0 5. Documentation of Cognitive Functioning Measures in the H ealth and Retirement Study. An n Arbor, MI: Survey Research, University of Michigan.

Prudential Financial. 20 0 5. Roadblocks to Retirement.

A Report on W hat H appens W hen Living Today Gets in the W ay of Financial Security Tomorrow. Pruden tial's Four Pillars of Retirem ent Series.

Rowe, John W. and Robert L. Kahn. 19 9 8. Successful Aging. New York: Pantheon.

Saint Clair, Patricia, Delia Bugliari, Sandi Chien, Steven H aider, Orla H ayden, Michael H urd, Serhii Ilchuk, Gabriela Lopez, David Loughran, Constantijn Panis, Philip Pantoja, Monika Reti, Julie

Zissim opolous. 20 0 4. RAND H RS Data Documentation, Version D. Labor and Population Program , RAND Cen ter for the Study of Agin g.

Siegrist, Johannes, Olaf Von Dem Knesebeck, and Craig Evan Pollack. 20 0 4. "Social Productivity and Well-being of Older People: a Sociological

Exploration." Social Theory and H ealth2:1-17.

Steffick, Dian e E. 20 0 0 . Documentation of Affective Functioning Measures in the H ealth and Retirement Study.H RS/ AH EAD Docum entation Report, Ann Arbor: Survey Research Center University of Michigan.

University of Michigan. 19 9 8, 20 0 0 and 20 0 2.

H ealth and Retirement Study.

http:/ / hrsonline.isr.um ich.edu/

(10)

Appen d ix

(11)

Variables Self-Rated H ealth (20 0 2) (Fair-Poor vs. Excellent-Good)

ADL (2002)

(Any Lim itation vs. No Lim itation)

IADL (20 0 2)

(Any Limitation vs. No Limitation) Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4 Working for Pay (2000) -.158*** -.0 6 2*** -.058*** - -.094*** -.0 26 *** -.0 24** - -.0 80 *** -.0 26 *** -.0 21** -

Enjoy Working (20 0 0 ) - - - -.0 32 - - - -.004 - - - -.0 20

Dem anding Job (20 0 0 ) - - - .004 - - - -.002 - - - .0 0 0

Fair/ Poor Self-Rated H ealth (19 9 8) .40 3*** .376*** .322*** .074*** .0 70 *** .06 9 *** .0 59 *** .057*** .0 16

Any ADL Lim itation (19 9 8) .104*** .096*** .064* .255*** .252*** .19 1*** .082*** .0 80 *** .085***

Any IADL Lim itation (19 9 8) .0 45* .0 37 -.0 45 .0 6 1*** .06 2*** .0 21 .135*** .135*** .118***

Negative Mood Indicators (19 9 8) .09 5*** .0 82*** .0 79 *** .0 37*** .033*** .026* .0 33*** .0 32*** .033**

Fair/ Poor Self-Rated Memory (19 9 8) .0 84*** .0 58*** .0 49 ** .017* .015 .023* .0 19 ** .0 19 ** .0 14 Any Chronic Diseases (19 9 8) .10 3*** .105*** .059 *** .0 51*** .0 51*** .0 34*** .0 27*** .026** .0 0 6

3 Or More Drinks Per Day (19 9 8) -.002 -.017 -.0 12 -.019 -.0 20 -.0 12 -.019 -.020* -.008

Vigorous Exercise (19 9 8) -.040*** -.0 38*** -.006 -.0 24*** -.023** -.0 0 5 -.0 16 * -.014* -.0 0 9

Smoking (19 9 8) .0 70 *** .056*** .038* .018* .017 .016 .0 25*** .023** .0 13

Age (20 0 0 ) -.001 -.003 .001 .001 .0 0 1 .0 0 0

Male (2000) .0 42*** .025 .001 .010 .0 0 6 .017*

White (20 0 0 ) -.06 0*** -.050* -.0 0 3 .003 -.005 -.0 0 9

H ispanic (20 0 0 ) .056* .019 .0 11 .005 -.016 -.0 20

Married (2000) -.026* .003 -.013 .010 -.009 -.0 0 5

Years of Education (2000) -.011*** -.0 10 *** -.0 0 1 .000 -.001 .0 0 0

H ousehold Incom e (20 0 0 ) -.020* -.013 -.0 0 1 .001 -.0 15* -.0 0 6

Pseudo R² .0 329 .29 31 .30 89 .2444 .0 320 .2710 .2725 .19 84 .0 358 .2349 .2398 .1652

Number of Observations 6251 6251 6 251 256 3 6 251 6 251 6 251 256 3 6 251 6 251 6 251 2563

Source: Author’s calculations (probit estim ation reporting discrete and m arginal effects) using data from the 19 9 8, 20 0 0 and 20 0 2 waves of the H ealth and Retirement Study.

*Note: *p<.0 5, **p<.0 1, ***p<.0 0 1

(12)

Variables Mood Indicators (20 0 2) (Negative vs. Positive)

Mortality (20 0 2) (Dead vs. Alive)

Self-Rated Mem ory (20 0 2) (Fair-Poor vs. Excellent-Good) Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4 Working for Pay (20 0 0 ) -.0 75*** -.026 ** -.0 18* - -.0 26 *** -.0 14*** -.0 0 9 ** - -.0 9 0 *** -.032** -.0 23 -

Enjoy Working (20 0 0 ) - - - -.049** - - - -.0 0 3 - - - -.0 13

Dem anding Job (20 0 0 ) - - - .0 0 4 - - - .0 0 2* - - - .0 0 4

Fair/ Poor Self-Rated H ealth (19 9 8) .065*** .0 55*** .0 32* .0 38*** .0 0 4*** .0 14** .124*** .0 9 1*** .0 10 ***

Any ADL Lim itation (19 9 8) .0 83*** .077*** .0 33 -.0 13** .012** .003 .0 38 .032 .0 25

Any IADL Lim itation (19 9 8) -.0 21 -.0 23 -.0 35 .0 15* .0 14* -.0 0 3 .0 38 .029 .0 0 6

Negative Mood Indicators (19 9 8) .26 5*** .243*** .19 0 *** .0 0 3 .003 -.002 .0 6 0 *** .0 54** .058*

Fair/ Poor Self-Rated Mem ory (19 9 8) .0 45*** .038*** .0 23 -.0 0 6 -.0 0 6 -.002 .413*** .382*** .36 0 ***

Any Chronic Diseases (19 9 8) .0 27* .0 25* .0 28* .0 15** .0 13** .0 0 6 * .026 .0 28 .021

3 Or More Drinks Per Day (19 9 8) .0 0 1 -.0 18 -.0 20 .0 0 8 .0 0 1 .0 0 2 .0 40 .0 16 .0 0 0

Vigorous Exercise (19 9 8) -.023** -.0 19 * -.0 18 -.0 10 ** -.0 0 9 ** -.0 0 3 -.0 19 -.0 21 -.0 32*

Sm oking (19 9 8) .0 48*** .0 37*** .0 27 .0 23*** .020*** .0 17*** .0 20 .0 0 7 -.0 26

Age (20 0 0 ) .0 0 0 .0 0 1 .002*** .0 0 1** .0 0 0 .0 0 1

Male (20 0 0 ) -.0 38*** -.0 19 .0 16 *** .0 10 *** .06 7*** .075***

White (20 0 0 ) .0 0 2 .0 0 5 -.0 0 4 .0 0 1 -.0 48** -.032

H ispanic (20 0 0 ) -.0 16 .0 15 -.0 0 4 -.0 0 5 .0 19 .0 14

Married (20 0 0 ) -.0 30 *** -.0 46 *** -.0 12** -.0 12*** .0 0 8 -.0 0 1

Years of Education (20 0 0 ) -.007*** -.0 0 6 ** .000 .000 -.0 21*** -.019 ***

H ousehold Incom e (20 0 0 ) .0 0 0 .0 0 1 -.0 0 7 -.0 0 4 -.017 -.0 31*

Pseudo R² .0 152 .1859 .19 9 6 .16 6 5 .0187 .0 9 0 9 .1182 .19 52 .0 0 9 4 .19 35 .2178 .210 1

Num ber of Observations 6 0 6 0 6 0 6 0 6 06 0 6 0 6 0 6 473 6 473 6473 26 0 8 6 0 53 6 0 53 6 0 53 2513

Source: Author’s calculations (probit estim ation reporting discrete and m arginal effects) using data from the 19 9 8, 20 0 0 and 20 0 2 waves of the H ealth and Retirement Study.

*Note: *p<.0 5, **p<.0 1, ***p<.0 0 1.

(13)

APPENDIXFIGURE2. EFFECTS OFPAIDWORK ATOLDERAGES ON TH EPROBABILITY OFREPORTINGANYIADL LIMITATION(20 0 2)

Source: Author’s calculations (probit estim ation reporting discrete and m arginal coefficients) using data from the 19 9 8, 20 0 0 , and 20 0 2 waves of the H ealth and Retirement Study.

Exercise (20 0 0 )

White (20 0 0 )

Bad health (19 9 8) ADL lim it (19 9 8) IADL lim it (19 9 8) Neg. m ood (19 9 8) Bad m em ory (19 9 8) Diseases (19 9 8) Sm oking (20 0 0 ) Age (20 0 0 ) Male (20 0 0 ) H ispan ic (20 0 0 )

Drin ks (20 0 0 ) Work (20 0 0 )

Exercise (20 0 0 )

White (20 0 0 ) Education (20 0 0 ) Incom e (20 0 0 ) Married (20 0 0 ) Education (20 0 0 ) Incom e (20 0 0 ) Married (20 0 0 )

Drin ks (20 0 0 ) Work (20 0 0 )

Percen t

-25 -20 -15 -10 -5 0 5 10 15 20 25

Statistically significant Statistically insignificant Statistically significant Statistically insignificant

(14)

Source: Author’s calculations (probit estim ation reporting discrete and m arginal coefficients) using data from the 19 9 8, 20 0 0 , and 20 0 2 waves of the H ealth and Retirement Study.

ADL lim it (19 9 8) Bad m em ory (19 9 8)

Age (20 0 0 ) White (20 0 0 ) H ispan ic (20 0 0 )

Education (20 0 0 )

Bad health (19 9 8)

IADL lim it (19 9 8) Neg. m ood (19 9 8)

Diseases (19 9 8) Sm oking (20 0 0 ) Male (20 0 0 )

Exercise (20 0 0 )

Incom e (20 0 0 ) Married (20 0 0 )

Drin ks (20 0 0 ) Work (20 0 0 )

Percen t

APPENDIXFIGURE4. EFFECTS OFPAIDWORK ATOLDER AGES ON THEPROBABILITY OFDYING(20 0 2)

Source: Author’s calculations (probit estim ation reporting discrete and m arginal coefficients) using data from the 19 9 8, 20 0 0 , and 20 0 2 waves of the H ealth and Retirement Study.

Exercise (20 0 0 )

White (20 0 0 )

Incom e (20 0 0 )

Bad health (19 9 8) ADL lim it (19 9 8) IADL lim it (19 9 8) Neg. m ood (19 9 8)

Married (20 0 0 ) Bad m em ory (19 9 8)

Diseases (19 9 8) Sm oking (20 0 0 )

H ispan ic (20 0 0 )

Education (20 0 0 ) Age (20 0 0 ) Male (20 0 0 ) Drin ks (20 0 0 ) Work (20 0 0 )

Percen t

Statistically significant Statistically insignificant Statistically significant Statistically insignificant

-25 -20 -15 -10 -5 0 5 10 15 20 25

(15)

© 20 0 5, by Trustees of Boston College, Center for Retirem ent Research. All rights reserved. Short sections of text, not to exceed two paragraphs, m ay be quoted without explicit perm ission provided that the author is identified and full credit, including copyright notice, is given to Trustees of Boston College, Center for Retirem ent Research.

The research reported herein was supported by The Atlantic Philanthropies. The opinions and conclusions expressed are solely those of the author and should not be construed as representing the opinions or policy of The Atlantic Philanthropies or the Center for Retirem ent Research at Boston College.

www.bc.edu/ crr/

Abo u t t h e Cen t er

The Center for Retirem ent Research at Boston College was established in 19 9 8 through a grant from the Social Security Adm inistration. The Center's m ission is to produce first-class research and forge a strong link between the academ ic com m unity and decisionm akers in the public and private sectors around an issue of critical im portance to the nation's future. To achieve this m ission, the Center sponsors a wide variety of research projects, transm its new findings to a broad audience, trains new scholars, and broadens access to valuable data sources. Since its inception, the Center has established a reputation as an authoritative source of in form ation on all m ajor aspects of the retirem ent incom e debate.

Af f il iat ed In st it u t io n s

Am erican Enterprise Institute The Brookin gs In stitution

Cen ter for Strategic an d In tern ation al Studies Massachusetts Institute of Technology Syracuse Un iversity

Urban Institute

Co n t act In f o r mat io n

Center for Retirem ent Research Boston College

Fulton H all 550

Chestnut H ill, MA 0 246 7-380 8 Phone: (6 17) 552-176 2

Fax: (6 17) 552-0 19 1 E-m ail: crr@bc.edu

Website: http:/ / www.bc.edu/ crr