March 2007, Number 2007-3 WILL PEOPLE BE HEALTHY ENOUGH TO WORK LONGER? By Alicia H. Munnell and Jerilyn Libby* Introduction Health and Work As recently as the mid-1960s, the median retirement Intuitively, people’s health affects their ability and age for men — the age at which half of all men are desire to work. Poor health can make work difficult no longer in the labor force — was 66. Today, it is and unpleasant, leading people to withdraw from the 63. But given the scheduled decline in Social Secu- labor force. Poor health can reduce people’s produc- rity replacement rates, increased longevity, and the tivity, leading to lower wages, and lower wages reduce relatively low balances in 401(k) accounts, Americans the incentive to work. Poor health can make people risk serious income shortfalls, especially at older less attractive to employers and therefore less likely to ages, if they continue to retire at age 63. A rational be hired. In 1969, intuition carried the day, as shown response is to move the average retirement age back by the following observation by the day’s leading to 66 or even older. A key consideration is whether experts: people will be healthy enough to work longer. This “… that the labor force status of an individual will brief compares the health status of older people today be affected by his health is an unassailable proposi- with those forty years ago and explores what happens tion [because] a priori reasoning and casual observa- to people’s health as they age. tion tell us it must be so, not because there is a mass The bottom line is that the health of older people of supporting evidence.”1 (those 65 and older), as opposed to older work- Today, we have the evidence. In the last 35 years, ers (those 50 to 64), showed little improvement in research into the impact of health on labor force activ- the 1970s, mixed results in the 1980s, and marked ity has become a major industry, and virtually all stud- improvement since the 1990s. The marked improve- ies show that poor health reduces the likelihood of ment for older workers most likely began earlier, in being in the labor force and the expected retirement the 1980s. Today, the health of older workers appears age, as well as hours worked and wages.2 to be at least as good as it was forty years ago. Thus, if The size of the effect of health on work, however, half of the male population were then healthy enough is sensitive to the measurement of health status. to work until age 66, the same percentage should be Most studies use a measure of self-assessed health able to do so today. Two important issues not ad- (very good, good, fair, bad, or very bad) or whether dressed in this brief are whether the jobs will be there respondents report health limitations that affect their for older workers and the challenge presented by the ability to work. Researchers have also used objective 15 to 20 percent of the older population for whom measures such as whether the person has a problem work will be impossible. with activities of daily living or the presence of a chronic or acute condition. * Alicia H. Munnell is the Director of the Center for Retirement Research at Boston College (CRR) and the Peter F. Drucker Professor of Management Sciences at Boston College’s Carroll School of Management. Jerilyn Libby is a research associate at the CRR. The authors would like to thank Richard Burkhauser, Dora Costa, and David Cutler for valuable comments on an earlier draft of this brief. 2 Center for Retirement Research It turns out that self-reported health status is actu- Figure 1. Life Expectancy in Years at Age 50 for ally a pretty good indicator of a person’s medically de- Men, 1900-2030 termined health status. These self reports, however, are sensitive to other parts of the employment picture, Actual which create problems for researchers attempting to establish relationships between health and work. Projected For example, people who like their work downplay their health problems and work longer, while those who dislike their work emphasize health issues and retire sooner. Similarly, people who have cut back on their work are more likely to report a health problem, either because they want to justify their decision or because they may be eligible for government benefits if they are unhealthy.3 Despite these possible biases, using self-reported health information may be the best approach to determining how health affects Source: U.S. Social Security Administration, Life Tables for work. While the “justification” phenomenon tends to Males, 1900-2030. bias the estimated effect of health on work upward, measurement error biases the results downward, and the two biases may well cancel each other out. In Although longer life spans generally imply im- contrast, objective measures of health are often not provements in health, keeping less healthy people very good indicators of whether people can work or alive could actually increase the percent of the popula- not — for example, difficulty walking up stairs may tion with disabilities. Thus, for a time, researchers have little effect on a person’s ability to work as a referred to the “failure of success.”6 Therefore, it computer programmer — and simply bias the effect is important to check on the health of the surviving toward zero. population and determine the extent to which disabili- A huge body of literature has confirmed that ties may prevent them from working. health affects work. Originally, researchers simply added some measure of poor health to an equation explaining labor force participation and found nega- What is a Disability? tive effects. Increasingly, the studies have become more sophisticated to address the biases discussed Disablement is generally defined as a process. It above. Regardless of the approach and the measure- begins with a “pathology,” a change in a person’s body ment of health and work activity, the studies provide caused by disease, infection, or some other factor.7 overwhelming evidence that poor health reduces the An example is hypertension, whereby high blood likelihood of work.4 Therefore, it is important to de- pressure stretches the walls of the arteries. A pathol- termine whether older people will be healthy enough ogy can then lead to an impairment, which makes to work. it difficult for a person to function. For example, hypertension can lead to angina, which causes chest pains upon exertion, or to heart attack or stroke. The Life Expectancies at Age 50 impairment can then lead to an inability to perform work or household tasks. Finally, the functional limi- One starting point for exploring the health of older tations can lead to dependence. workers is to look at trends in life expectancy at age For older people, dependence usually means the 50. Figure 1 shows life expectancy at age 50 for males person has difficulty with basic activities of daily liv- over the 20th century. Interestingly, life expectancy ing (ADLs), such as eating, bathing, or dressing, or at older ages rose very slowly at the beginning of difficulty with instrumental activities of daily living the century and then accelerated sharply toward the (IADLs), such as doing light housework, shopping, or end of the century. In fact, life expectancy at 50 was preparing meals. ADL disability is generally consid- not very different in 1960 than in 1900 — 24 years ered the most severe because it is generally associated versus 21 years. After 1960, however, life expectancy with long-term care needs.8 at 50 took off, rising to 29 years in 2000 and is pro- jected to increase to 32 years by 2030.5 Issue in Brief 3 With a focus on work, the key question is the ex- Figure 2. Percent of Older Americans with Any tent to which older people have disabilities that might Kind of Disability, by Age, Selected Years 1984- limit their labor force activity. Our primary concern is 2004/2005 with older workers, people age 50 to 65. But we first look at trends in disability among the population 65 and over, because substantial research has been con- ducted for this age group and presumably a healthier group of retirees would imply a healthier cohort of older workers. The following section then reports the more limited survey results for those 50 to 65. Trends in Disabilities among Those 65 and over As noted above, the relationship between improve- ment in mortality and the health of the older popu- Source: Manton, Gu and Lamb (2006) from Duke Uni- versity, National Long-Term Care Survey (NLTCS), 1984- lation is theoretically ambiguous. For example, 2004/2005. if the reduction in mortality were due to keeping more stroke victims alive, the health of the surviv- ing population could decline since stroke survivors tions have remained virtually unchanged since the are often quite disabled. On the other hand, if the beginning. improvement in mortality came from better treatment Figure 2 shows the percent of the total 65-and-over for hypertension, the overall health of the popula- population that was dependent, where dependent is tion would improve since many would not develop defined as having an ADL or IADL difficulty or resid- an impairment. Indeed, trends during the 1970s led ing in a nursing home. Between 1984 and 2004, the a number of researchers to conclude that increased share of the elderly that lacked the ability to function longevity had led to increased frailty among the sur- independently with ease declined from 26.2 percent viving population, but since the early 1990s it seems to 19.0 percent.11 The figure also shows the preva- irrefutable that the health of the older population has lence of dependency by age. Although dependency been improving. rises sharply as people get older, the pattern of de- In 2002, a technical working group examined creasing dependency was evident for all age groups.12 trends in disability for older Americans across five (See Box for a discussion of the decline in the nursing major national surveys.9 The group concluded that, home population). when standardizing for the definition of disability, Experts have cited a number of reasons for this time period, and consistent inclusion or exclusion of improvement in the health of those 65 and over. the nursing home population, all five surveys showed Since the change has occurred so recently and so consistent downward trends for two common dis- rapidly, environmental factors — as opposed to ability measures — difficulty with daily activities and genetic or evolutionary developments — must clearly help with daily activities — beginning in the early to play a major role.13 The usual suspects include better mid-1990s. The evidence for change in the 1980s medical care, reduced exposure to childhood diseases, and for a third measure of disability (the use of help improved lifestyles, fewer occupational hazards, and or equipment with daily activities) remained mixed. increased education and income. Understanding the The most consistently positive findings regarding source of the improvements for those 65 and over the health of those 65 and over come from the Na- should shed light on whether and when older work- tional Long-Term Care Survey (NLTCS). The NLTCS, ers would also be expected to enjoy better health. For a longitudinal survey of the Medicare-enrolled example, to the extent that most of the improvements population aged 65 and over, has been conducted in for those 65 and over was the result of Medicare- 1982, 1984, 1989, 1994, 1999, and 2004. In each driven improved medical care, those under 65 would survey, approximately 20,000 people are screened be less likely to benefit. But to the extent that it was for chronic limitations in activities of daily living and the result of other factors such as reduced exposure instrumental activities of daily living.10 Researchers to early childhood disease, improved working condi- put a lot of weight on this survey because the ques- tions, etc., the improvement should be evident in those under 65, but show up roughly a decade earlier. 4 Center for Retirement Research Better medical treatments. Twenty years ago, one contributed to the reduced disability of today’s 65 and of the major reasons that older people had problems over population. with walking and shopping was arthritis.14 The major Improved lifestyles. In addition to having healthier developments that required nursing home care were childhoods, the current elderly also evidenced bet- cognitive impairment, followed by heart disease and ter behavior as adults. In 1960 about 40 percent stroke. The medical profession has been able to al- of adults were regular smokers; today only about leviate many of these health problems. The debilitat- 25 percent of the population smokes. Smoking is ing effects of arthritis have been substantially con- the leading risk factor for heart disease, stroke, and trolled by the use of anti-inflammatory drugs.15 Joint respiratory diseases — all precursors to a disability.19 replacements, which roughly doubled from the 1980s People have also reduced the intake of salt and fats in to the 1990s, have also been a major innovation. In their diets, which may have reduced the incidence of terms of heart disease and stroke, the use of hyperten- atherosclerosis and hypertension. Diabetics are tak- sion medication also rose in the late 1970s and early ing better care of themselves relative to earlier cohorts 1980s, which may explain the decline in incidence in terms of consumption of alcohol and foods with of stroke in recent years. It appears that much of the sugar.20 On the other hand, the growing trend toward improvement has come from earlier diagnosis and obesity and rising incidence of diabetes are examples improved treatment of those who develop the condi- of unfavorable developments.21 tion rather than from a reduced onset of conditions in Reduced occupational hazards. The nature of work the first place, although the age of onset remains an has also become less physically demanding and less unsettled question.16 hazardous. First, employment has shifted from man- Reduced exposure to infectious disease in childhood. ual jobs to white-collar work. This is an important The current elderly were also less exposed to disease development because economists have documented in childhood. The medical and epidemiological that, even controlling carefully for education and literature provides many examples of the possible income, those in manual occupations have worse self- linkage between early life infectious disease and reported health and experience more rapid declines in chronic disease and cognitive disorders late in life.17 their health with age than their white-collar counter- For example, individuals who had acute rheumatic parts.22 Manual workers also have less control over fever as a child were likely to experience a recurrence their work schedules, face repetitive tasks, and hold of attacks following a streptococcal infection. Other jobs with low prestige, which can cause psychological infectious diseases, such as measles, syphilis, typhoid problems. Second, within manual jobs, regulations fever and malaria, can also cause heart problems in have substantially reduced occupational hazards by later life.18 The decline of infectious diseases likely limiting workers’ exposure to dust, fumes, and gases that can cause lung diseases. Figure 3. Percent of Population Age 65 and THE DECLINE IN THE NURSING over in Nursing Homes, Selected Years HOME POPULATION 1984-2004/2005 Over the past twenty years, the percentage of the population over age 65 in nursing homes has also declined dramatically for all age groups (see Figure 3). Some of the decline may reflect a shift to home-based care and assisted living facili- ties. The number of residential care and assisted living beds increased by 97 percent from 1990 to 2002.23 Changes in Medicare, specifically a clarification of eligibility criteria for home health care, also increased the percentage of the elderly population receiving care at home.24 Some of the decline, however, probably reflects a generally healthier older population. Source: Manton, Gu and Lamb (2006) from the 1984- 2004 NLTCS. Issue in Brief 5 Higher educational attainment. The improvement Trends in Disabilities among in educational attainment among those 65 and over could also have led to improved health. More edu- the Working-Age Population cated people have a 50 percent lower disability rate than the less educated.25 The share of the elderly The fact that the health of older Americans has with a college degree more than doubled from 1980 improved would lead one to conclude that the health to 2005.26 Some contend that education inevitably of the older working-age population was also getting stands for more than years in the classroom. That is, better. But for a long time, such a conclusion was not it is a broad measure that reflects access to medical obvious. The major survey that tracked disabilities care, patterns of medical care use, as well as exercise, among the working-age population — the National diet, and smoking patterns, and access to devices Health Interview Survey (NHIS) — showed the percent when disability does occur.27 A recent study, however, of this population with disabilities increasing from attempts to disentangle education from these other the mid-1960s through the early 1980s (see Figure 4). factors, and finds that even controlling for income and wealth as well as other reasons why educa- Figure 4. Percent of Men Age 45 to 64 with tion might matter — past health behaviors such as Activity Limitation, 1967-2004 smoking and drinking, job-related hazards, early life economic environment, and parental education and 25% health — education remains an important explana- 20% tory variable.28 One possible reason is that more educated people will follow what can be complicated 15% regimens and better manage their diseases.29 This discipline may reflect an improved understanding 10% of how current actions can affect future events that comes with more education. In short, now that we 5% have eliminated the huge disparities between rich and poor in terms of exposure to infectious diseases and 0% even in terms of food and shelter, the impact of edu- 91 95 98 67 71 79 75 83 87 02 cation on health has become increasingly important.30 19 19 19 19 19 19 19 19 20 19 The improvement of the condition of older Americans has been both recent and dramatic. The Note: From 2002-2004, the figure shows work limitation explanation for the timing may be two-fold. First, for all persons instead of males only. improvements in medical care, reductions in occu- Source: Authors’ calculations from the National Center for Health Statistics (1967-2004). pational stress, changes in lifestyles, and increases in education and income all occurred in a short period of time. On the medical side, Medicare, which was Decennial census data also showed an increase in the enacted in 1965, may well have encouraged treatment fraction of both men and women unable to work dur- innovations for the elderly through teaching hospitals ing the 1970s. Skeptics of the increasing disability and clinical research.31 Second, the life experiences of story contend that the trend during the 1970s may, at different population groups differed significantly. As least in part, reflect social factors such as earlier detec- described by Costa (2005): tion and diagnosis of chronic diseases and greater “Those who were 70 in 1980 were born in 1910 availability of disability insurance.32 Thus, the trend when infectious disease rates were still high and in the prevalence of disabilities during the 1970s when incomes were low and spent their prime years remains controversial. Since the early to mid-1980s, in relatively dangerous jobs. In contrast, those who however, it is clear that the percent of men with an were 70 in 2000 were born in 1930 when infectious activity limitation has declined.33 disease rates, while still high by today’s standards, had The NHIS is an annual cross-sectional survey of fallen. They enjoyed higher incomes, ate a more bal- 100,000 non-institutionalized civilians conducted by anced diet, acquired more education, worked in less the National Center for Health Statistics. Unfortu- dangerous jobs and had access to improved medical nately, the survey questions have been revised every care.” 6 Center for Retirement Research 10 to 15 years, making it impossible to construct a Figure 5 also includes data from the NHIS for pur- series over a long period of time.34 Nevertheless, con- poses of comparison. The average level of work-based sistent data are available from 1967-1982, 1983-1996, disability was higher in the NHIS than in the CPS. and 1997-2004. For the period 1983-1996, the survey The NHIS might elicit a higher rate of reported dis- asked “Does any impairment or health problem now ability because it is a health-based survey.36 But both keep [person] from working at a job or business? Is surveys show a downward trend in the 1980s and [person] limited in the kind or amount of work [per- early 1990s. It makes sense that improved disability son] can do because of any impairment?” A person trends would show up earlier among those 50 to 64 who answers yes to either question is considered to than for those 65 and over, since the younger cohort have a work limitation.35 As Figure 4 shows, the per- was less exposed to infectious diseases in childhood, cent of those 45 to 64 with a disability declined from worked in less hazardous jobs, and enjoyed higher the early 1980s and through the mid-1990s. Between education and incomes. 1997 and 2004, a similar question produced a more One note of caution may be in order. Some stable trend. But the general conclusion emerging researchers have recently raised concerns about in- from the NHIS data is one of declining disability creased disability among younger people, most likely among older working-age individuals to a level at least due to the increases in obesity.37 Obesity often results comparable to that in the mid-1960s. in diabetes, and rates of diabetes are on the rise. If Another source of data on work limitations is the these trends hold, the story of improving health for Current Population Survey (CPS). The CPS is also a older workers could reverse for the younger genera- large annual cross-sectional survey (about 150,000 tion. non-institutionalized civilians). Unlike the NHIS, the CPS was not designed to track health trends but rather to gather employment and income data for the Conclusion U.S. population. Nevertheless, beginning in 1981 the March Supplement asks a question about work limita- Numerous studies have shown that health and work tions: “Does anyone in this household have a health are related. Those reporting poor health are less problem or disability which prevents them from likely to work than those in good health. Although working or which limits the kind of work they can the trends in the 1970s remain controversial, the do? [If so,] who is that? Anyone else?” And unlike the NHIS data indicate a rise in work limitations among NHIS, the survey question has remained unchanged men age 45 to 64 from the mid-1960s to the mid- for the last 25 years. The percent of men with a 1980s. This period was when the average retirement work-limitation-based disability is shown in Figure age for men fell from 66 to 63. The expansion of the 5. The trend since the early 1980s is one of declining nation’s retirement income system — Social Security, disabilities. Medicare, and employer pensions — clearly contrib- uted to this decline in the average retirement age. But declining health could be part of the explanation. Figure 5. Percent of Men Age 50 to 64 with a Now that the retirement income system is con- Work Limitation-Based Disability, NHIS and CPS, tracting, workers need to remain employed longer 1981-2005 to gain the same level of retirement income security. 25% The evidence suggests that the health of older work- NHIS ers is at least as good today as it was forty years ago. 20% CPS Moreover, jobs are much less physically demanding than they were in the past. Thus, physical limita- 15% tions should not inhibit the ability of the bulk of older Americans to work at least until their mid-sixties. 10% Important questions still remain concerning whether the jobs will be there for older workers. And 5% the data also make clear that, despite a positive trend, 15 to 20 percent of people in their late fifties and 0% sixties will find work virtually impossible. Moreover, many of those who need to work longer — particu- 99 81 89 91 93 95 97 83 85 87 01 03 05 19 19 19 19 20 19 19 19 20 20 19 19 19 larly low-wage workers dependent on Social Security Sources: Authors’ calculations from the U.S. Census Bureau, — are precisely the individuals who have onerous Current Population Survey (CPS), 1981-2005 and the Na- jobs that stress their health and who lack the educa- tional Center for Health Statistics, National Health Interveiw tion to manage their care. Thus, the working longer Survey, 1983-1996. prescription must be administered with care. Issue in Brief 7 Endnotes 1 Bowen and Finnegan (1969). 12 Wolf, de Leon, and Glass (2007) explore the dynamics of the declining population level disability 2 For a survey of the literature, see Currie and prevalence and find that, on the positive side, dis- Madrian (1999); an update can be found in Deschry- abilities are occurring later but, on the negative side, vere (2005). people are taking longer to recover. 3 Also, people are more likely to report a health prob- 13 For an extensive discussion of this issue, see Fogel lem if they have sought treatment. Since people with and Costa (1997). higher incomes and more education use more medi- cal care, they may be more likely to report certain 14 See Cutler (2001). conditions. 15 See Fries et al. (1996). 4 Health alone may not be the sole determinant of whether someone is able to work. Nagi (1976) views 16 Cutler (2001) argues that the onset of conditions disability as the interaction between the individual’s has not receded, while Fogel (2003) asserts that the disability and the demands presented by the social average age of onset of chronic disease occurred more and physical enviornments. Consequently, as Jette than a decade later at the end of the 20th century than and Badley (2000) lay out, varying levels of accomo- at the beginning. dation as well as an individual’s own personality and characteristics can affect the likelihood of working 17 For a review of the literature, see Cutler, Deaton, with a disability. Burkhauser, Butler, Kim, and Weath- and Lleras-Muney (2006). ers (1999) and Burkhauser, Butler, and Gummus (2004) find that following the onset of a work-limit- 18 See Costa (2005) and Cutler (2001). ing condition, employer accomodation delays the time between onset and claiming Social Security Disability 19 Cutler (2001). Insurance benefits. 20 See Costa (2005). 5 The same pattern is evident in the probability of 50-year-olds surviving to 65 — very little change until 21 See Lakdawalla, Bhattacharya, and Goldman 1970 and then a surge thereafter. (2004). 6 See Waidmann, Bound, and Schoenbaum (1995). 22 See Case and Deaton (2003). 7 The following discussion is based on a process 23 See Harrington et al. (2005). described by Nagi (1976) and also presented in Cutler (2001). 24 See Bishop (1999). 8 See Freedman, Martin, and Schoeni (2002). 25 See Manton and Gu (2001). 9 See Freedman et al. (2004). The five surveys 26 The percent of the elderly with a college degree included the Health and Retirement Study (HRS), increased from 5 percent in 1980 to 12 percent in the Medicare Current Beneficiary Survey (MCBS), the 2005 (U.S. Census Bureau, Current Population Survey, National Health Interview Survey (NHIS), the National 1980 and 2005). Long Term Care Survey (NLTCS), and the Supplements on Aging (SOAs). 27 See Schoeni, Freedman, and Wallace (2001). 10 The sample consists of 15,000 who were sur- 28 See Smith (2004). veyed on previous surveys and 5,000 who passed age 65 since the previous survey. See Manton and Gu 29 Goldman and Smith (2002) found that in a (2001). randomized trial in which one group of diabetics was placed in a group with enforced treatment, the biggest 11 Manton, Gu, and Lamb (2006). beneficiaries were those with the least education. 8 Center for Retirement Research 30 Insight from correspondence with Dora Costa. 35 The NHIS also asks directly about certain impair- ments (deaf in both ears, blind in both eyes, etc.) of 31 See Cutler and Meara (2001). a subset of survey respondents. This practice has allowed researchers to explore the people who have 32 Waidmann, Bound, and Schoenbaum (1995). similar impairments but report no work limitations. From the mid-1960s until the mid-1970s, Social Security Disability Insurance benefits rose while 36 See Burkhauser, Daly, Houtenville, and Nargis eligibility requirements became less strict. Until the (2002) for an assessment of the limitations of the Social Security Administration and Congress started CPS for measuring the portion of the population with to tighten these requirements in 1976, the availability disabilities. In fact, both surveys may understate the of disability insurance may have influenced workers’ percent of the population with impairments, because view of their health and ability to work. having an impairment, even a serious one, does not necessarily mean the individual will not work. For 33 Cutler, Liebman, and Smyth (2006) recently com- example, according to the 1996 NHIS, 31 percent of pared the health status in the 1960s/1970s with today those blind in both eyes reported no work limitation; and found significant improvement. They used two 26 percent of those with cerebral palsy reported no measures that are consistent over time: 1) the share of work limitation. Therefore, both estimates exclude people in the last two years of their life (a period when those sufficiently integrated into the workforce that disability is high); and 2) the share of people who they do not report a work limitation. For any given report themselves in fair or poor health. The reported person, the likelihood of employment depends on the data, however, did not provide a clear indication of interaction of state of health, functional capacity, the what happened during the 1970s. nature of the work, and the possibilities for work ac- commodation (see Chan, Tan, and Koh (2000)). 34 The National Center for Health Statistics rede- signed the NHIS questionnaire format in 1982 and 37 Lakdawalla, Bhattacharya, and Goldman (2004). again in 1997. The NHIS asks all adult respondents whether they are unable to perform their major activity because of health problems; limited in their ability to perform their major activity; and limited in any activity. Prior to 1982, men were asked these questions in regards to paid work, while women who identified their major activity as “keeping house” were asked about their ability to perform housework. Start- ing in 1982, the question which asked respondents to identify their “major activity” changed to give men and women the same set of choices (working, keeping house, going to school, or something else). Addi- tionally, regardless of what respondents identified as their major activity, all those under age 70 were asked about their ability to work. Those who did not report their major activity as working were asked a set of follow-up questions from which a work limitation response could be constructed. Changes to the survey in 1997 include changes to the wording, structure, and context of questions as well as a shift from paper to laptop computers for the collection process. Issue in Brief 9 References Bishop, Christine E. 1999. “Where Are the Missing Cutler, David, Angus Deaton, and Adriana Lleras- Elders? The Decline in Nursing Home Use, 1985 Muney. 2006. “The Determinants of Mortality.” and 1995.” Health Affairs 18(4): 146-155. Working Paper 11963. Cambridge, MA: National Bureau of Economic Research. Bowen, William and T. Finnegan. 1969. The Econom- ics of Labor Force Participation. Princeton, NJ: Cutler, David, Jeffrey B. Liebman, and Seamus Princeton University Press. Smyth. 2006. “How Fast Should the Social Security Eligibility Age Rise?” NBER Retire- Burkhauser, Richard V., John S. Butler, and Gulcin ment Research Center Working Paper NB04-05. Gumus. 2004. “Dynamic Programming Model Cambridge, MA: National Bureau of Economic Estimates of Social Security Disabililty Insurance Research. Application Timing.” Journal of Applied Economet- rics 19: 671-685. Cutler, David and Ellen Meara. 2001. “Changes in the Age Distribution of Mortality Over the 20th Burkhauser, Richard V., John S. Butler, Yang-Woo Century.” Working Paper 8556. Cambridge, MA: Kim, and Robert R. Weathers II. 1999. “The National Bureau of Economic Research. Importance of Accommodation on the Timing of Disability Insurance Applications: Results from Deschryvere, Matthias. 2005. “Health and Retirement the Survey of Disability and Work and the Health Decisions: An Update of the Literature.” ENEPRI and Retirement Study.” The Journal of Human Research Report No. 6. Belgium: European Net- Resources 34(3): 589-611. work of Economic Policy Research Institutes. Burkhauser, Richard V., Mary C. Daly, Andrew J. Duke University. National Long-Term Care Survey, Houtenville, and Nigar Nargis. 2002. “Self-Re- 1984-2004/2005. Durham, NC. ported Work-Limitation Data: What They Can and Cannot Tell Us.” Demography 39(3): 541-555. Fogel, Robert. 2003. “Changes in the Process of Aging During the Twentieth Century: Findings Case, Anne C. and Angus Deaton. 2003. “Broken and Procedures of the Early Indicators Project.” Down By Work and Sex: How Our Health De- Working Paper 9941. Cambridge, MA: National clines.” Working Paper 9821. Cambridge, MA: Bureau of Economic Research. National Bureau of Economic Research. Fogel, Robert and Dora L. Costa. 1997. “A Theory of Chan, Gregory, V. Tan, and David Koh. 2000. “Age- Technophysio Evolution with Some Implications ing and Fitness to Work.” Occupational Medicine- for Forecasting Population, Health Care, and Pen- Oxford 50 (7): 483-491. sion Costs.” Demography 31(1): 49-66. Costa, Dora L. 2005. “Causes of Improving Health Freedman, Vicki, Eileen Crimmins, Robert Schoeni, and Longevity at Older Ages: A Review of the Brenda Spillman, Hakan Aykan, Ellen Kramarow, Explanations.” Genus 61(1): 21-38. Kenneth Land, James Lubitz, Kenneth Manton, Linda Martin, Diane Shinberg, and Timothy Currie, Janet and Brigitte C. Madrian. 1999. “Health, Waidmann. 2004. “Resolving Inconsistencies in Health Insurance, and the Labor Market.” In Orley Trends in Old-Age Disability: Report from a Tech- C. Ashenfelter and David Card, eds. Handbook of nical Working Group.” Demography 41 (3): 417-441. Labor Economics. Volume 3C. Amsterdam: Elsevier Science Publishers BV. Freedman, Vicki, Linda Martin, and Robert Schoeni. 2002. “Recent Trends in Disability and Function- Cutler, David. 2001. “Declining Disability among the ing Among Older Adults in the United States — A Elderly.” Health Affairs 20 (6): 11-27. Systematic Review.” JAMA — Journal of the Ameri- can Medical Association 288 (24): 3137-3146. 10 Center for Retirement Research Fries, James F., Catherine A. Williams, Dianne National Center for Health Statistics. 1967-2004. Morfeld, Gurkirpal Singh, and John Sibley. 1996. Current Estimates from the National Health Inter- “Reduction in Long-Term Disability in Patients view Survey. Washington, DC: U.S. Centers for With Rheumatoid Arthritis by Disease-Modifying Disease Control and Prevention. Antirheumatic Drug-Based Treatment Strategies.” Arthritis and Rheumatism 39(4):616-622. National Center for Health Statistics. National Health Interview Survey, 1981-1996. Public Use Sample, Goldman, Dana and James P. Smith. 2002. “Can Pa- Documentation, and Codebook. Washington, DC: tient Self-Management Help Explain the SES U.S. Centers for Disease Control and Prevention. Health Gradient?” Proceedings of the National Acad- emy of Sciences USA 99(16): 10929-10934. Schoeni, Robert F., Vicki A. Freedman, and Robert B. Wallace. 2001. “Persistent, Consistent, Wide- Harrington, Charlene, Susan Chapman, Elaine Miller, spread, and Robust? Another Look at the Trends in Nancy Miller, and Robert Newcomer. 2005. Old-Age Disability.” Journal of Gerontology: Social “Trends in the Supply of Long-Term-Care Facilities Sciences 56B (4):S206-S218. and Beds in the United States.” The Journal of Ap- plied Gerontology 24(4):265-282. Smith, James P. 2004. “Unraveling the SES Health Connection.” In Aging, Health, and Public Policy: Jette, Alan M. and Elizabeth Badley. 2000. “Concep- Demographic and Economic Perspectives,” Supple- tual Issues in the Measurement of Work Disabil- ment to Population and Development Review 30: ity.” In Nancy Mathiowetz and Gooloo S. Wun- 108-132. derlich, eds. Survey Measurement of Work Disability: Summary of a Workshop. Washington, DC: Nation- U.S. Census Bureau. Current Population Survey, 1980- al Academy Press. 2005. Washington, DC. Lakdawalla, Darius, Jayanta Bhattacharya, and Dana U.S. Social Security Administration. Life Tables for Goldman. 2004. “Are the Young Becoming More Males, 1900-1930. Unpublished data. Disabled?” Health Affairs 23(1): 168-176. Waidmann, Timothy A., John Bound, and Michael Manton, Kenneth and XiLiang Gu. 2001. “Changes Schoenbaum. 1995. “The Illusion of Failure: in the Prevalence of Chronic Disability in the Trends in the Self-Reported Health of the U.S. United States Black and Non-Black Population Elderly.” Milbank Quarterly 73(2): 253-287. Above Age 65 from 1982 to 1999.” Proceedings of the National Academy of Sciences USA 98(11): 6354- Wolf, Douglas A., Carlos F. Mendes de Leon, and 6359. Thomas A. Glass. 2007. “Trends in Rates of One- set of and Recovery from Disability at Older Ages: Manton, Kenneth, XiLiang Gu, and Vicki Lamb. 1982-1994.” Journal of Gerontology 62B(1): S3-S10. 2006. “Change in Chronic Disability from 1982 to 2004/2005 as Measured By Long-Term Changes in Function and Health in the U.S. Elderly Popula- tion.” Proceedings of the National Academy of Sci- ences USA 103(48): 18374-18379. Nagi, Saad. 1976. “An Epidemiology of Disability Among Adults in the United States.” Milbank Memorial Fund Quarterly: Health and Society 54(4): 439-467. About the Center Affiliated Institutions The Center for Retirement Research at Boston Col- American Enterprise Institute lege was established in 1998 through a grant from the The Brookings Institution Social Security Administration. The Center’s mission Center for Strategic and International Studies is to produce first-class research and forge a strong Massachusetts Institute of Technology link between the academic community and decision Syracuse University makers in the public and private sectors around an Urban Institute issue of critical importance to the nation’s future. To achieve this mission, the Center sponsors a wide variety of research projects, transmits new findings to Contact Information Center for Retirement Research a broad audience, trains new scholars, and broadens Boston College access to valuable data sources. Since its inception, Hovey House the Center has established a reputation as an authori- 140 Commonwealth Avenue tative source of information on all major aspects of Chestnut Hill, MA 02467-3808 the retirement income debate. Phone: (617) 552-1762 Fax: (617) 552-0191 E-mail: crr@bc.edu Website: http://www.bc.edu/crr © 2007, by Trustees of Boston College, Center for Retire- The research reported herein was supported by The Pru- ment Research. All rights reserved. Short sections of text, dential Foundation and by The Atlantic Philanthropies. The not to exceed two paragraphs, may be quoted without ex- opinions and conclusions expressed are solely those of the plicit permission provided that the authors are identified and authors and should not be construed as representing the full credit, including copyright notice, is given to Trustees of opinions or policy of The Prudential Foundation, The Atlantic Boston College, Center for Retirement Research. Philanthropies or the Center for Retirement Research at Boston College.