ARE AGE-62/63 RETIRED WORKER BENEFICIARIES AT RISK? Eric R. Kingson and Maria T. Brown* CRR WP 2009-13 Revised: July 2009 Released: August 2009 Draft Submitted: April 2009 Center for Retirement Research at Boston College Hovey House 140 Commonwealth Avenue Chestnut Hill, MA 02467 Tel: 617-552-1762 Fax: 617-552-0191 * Eric Kingson is a Professor of Social Work and Public Administration at Syracuse University’s School of Social Work, College of Human Ecology. Maria Brown, LMSW and Hartford Fellow, is a doctoral candidate in Syracuse University’s Social Sciences program at the Maxwell School of Citizenship and Public Affairs. This research was funded under contract with the Boston College Retirement Research Center and conducted under the auspices of the Center for Policy Research at Syracuse University’s Maxwell School of Citizenship and Public Affairs. It was conducted pursuant to a grant from the U. S. Social Security Administration as part of the Retirement Research Consortium. The opinions and conclusions are solely those of the authors and should not be construed as representing the opinions or policies of SSA or any agency of the federal government or of the Center for Retirement Research at Boston College or of the Center for Policy Research at Syracuse University or of any other organization with which the authors are associated. The authors wish to acknowledge the data preparation consultation of Tay McNamara, Research Associate at the Boston College Center on Aging & Work and the advice of Janet Wilmoth, Professor of Sociology at Syracuse University’s Maxwell School of Citizenship and Public Affairs and Director of the University Gerontology Center. © 2009, by Eric R. Kingson and Maria T. Brown. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. About the Center for Retirement Research The Center for Retirement Research at Boston College, part of a consortium that includes parallel centers at the University of Michigan and the National Bureau of Economic Research, was established in 1998 through a grant from the Social Security Administration. The Center’s mission is to produce first-class research and forge a strong link between the academic community and decision makers in the public and private sectors around an 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 a broad audience, trains new scholars, and broadens access to valuable data sources. Center for Retirement Research at Boston College Hovey House 140 Commonwealth Avenue Chestnut Hill, MA 02467 phone: 617-552-1762 fax: 617-552-0191 e-mail: crr@bc.edu www.bc.edu/crr Affiliated Institutions: The Brookings Institution Massachusetts Institute of Technology Syracuse University Urban Institute Abstract This paper provides a longitudinal view, spanning 10 to 12 years, of persons first accepting retired worker benefits at ages 62 or 63 in 1994 or 1996. Using HRS data, matched to Social Security administrative files, we present: 1) findings of variation in income, wealth, health insurance coverage and employability, along such dimensions as race, Hispanic ethnicity, gender, reported health status and functional ability; 2) findings of health and survival outcomes in 2006 for the 1994/1996 pooled sample, paying special attention to variations within the sub-sample of persons who accepted Social Security early retirement benefits by 1996; and 3) estimates of the proportion of persons accepting such benefits who are at risk. The findings indicate that persons first accepting Social Security retired worker benefits at ages 62 and 63 experience varying degrees of risk to their well being at these ages, and that these risks condition their well-being in retirement and survival probabilities. The major policy implication is that consideration should be given to providing a health insurance option for persons first accepting retired worker benefits prior to age 65. The major research implication is that retirement researchers should consider utilizing a range of measures – as opposed to a singular and potentially narrow measure -- of risk when assessing the magnitude of risks existing for those accepting retired worker benefits at early ages. INTRODUCTION For many years, the implications for persons accepting reduced Social Security retired worker benefits at early ages have figured prominently in policy discussions and research about retirement income security and Social Security reform. Those accepting actuarially reduced early retired worker benefits are a varied group, including many in good health, with significant pension or other retirement savings, employment opportunities and health insurance coverage; and others with work-limiting health problems, few employment options, no significant retirement savings and no health insurance. From an adequacy perspective, the most compelling question, and the one that this paper examines, concerns the extent to which persons accepting early benefits are at significant economic, health and employment risk. In assessing the implications of recent and proposed changes in Social Security entitlement age policies, it is important to understand the circumstances of persons when they first accept reduced benefits, especially those with limited incomes and assets and health or employment limitations. Equally important, is the need to understand what happens to these Social Security beneficiaries over time: whether, as they age, there is more or less variance in their circumstances, and whether the problems of those at risk intensify with time. This paper provides a longitudinal view, spanning 10 to 12 years, of persons first accepting retired worker benefits at ages 62 or 63 in 1994 or 1996. The findings presented in this paper have implications for several important retirement age questions. Specifically, whether serious consideration should be given to:  Phasing-in later ages of first eligibility for Social Security retired worker benefits – the “Early Eligibility Age” (EEA);  Phasing-in increases, beyond those contained in current law, in the age of eligibility for full retirement benefits – the “Full Retirement Age” (FRA), also called the “Normal Retirement Age” (NRA);  Providing a Medicare buy-in option for older persons (e.g. 55 to 64) who lack health insurance, as President Clinton proposed in 1998 and as Senate Finance Chair Max Baucus proposed in November 2008; and  Initiating reforms in Supplemental Security Income (SSI), Social Security Disability Insurance, or Unemployment Insurance designed to off-set the negative consequences for some older persons whose health or employment limitations essentially lead them to accept substantially reduced retired worker benefits. ORGANIZATION OF PAPER The main body of this paper presents findings of research funded by the Social Security Administration (SSA) and administered by the Boston College Center for Retirement Research. This project 1) assesses the risks to the economic well-being of persons accepting reduced retired worker benefits, and 2) provides further basis for identifying the implications of potential changes 1 in the early and full retirement ages for such populations. Using HRS data, matched to Social Security administrative files, we present: 1) findings of variation in income, wealth, health insurance coverage and employability, along such dimensions as race, Hispanic ethnicity, gender, reported health status and functional ability; 2) findings of economic, health and survival outcomes in 2006 for the 1994/1996 pooled sample, paying special attention to variations within the sub-sample of persons who accepted Social Security early retirement benefits by 1996; and 3) estimates of the proportion of persons accepting such benefits who are at risk. As background, and after briefly clarifying several important terms, we summarize Social Security entitlement age provisions, followed by a discussion of the policy questions driving our interest in this topic and then a review of related research literature. Following the review, we discuss methods, present and discuss preliminary findings and then discuss potential policy implications and questions for further research. CLARIFICATION OF TERMS The descriptor “early retirees” is applied to a number of different, though over-lapping concepts, including persons accepting private pension benefits and leaving work in advance of an employer’s “normal” retirement age and persons leaving their work of many years prior to “normal” retirement ages, independent of whether they receive any public or private benefits. The term is also used to describe persons who are the subject of this study: individuals accepting early Social Security retired worker benefits. To distinguish this form of “early retirement” from the others, in so far as possible, we term such persons as “Early Acceptors.” Moreover, many who “retire” in the sense of accepting benefits or leaving their employer of many years do not necessarily exit the labor force, and some who do, return at a later time, to full- or part- time employment. Hence, it is important to recognize that many “Early Acceptors” remain actively engaged in the labor force. Also, it is worth noting that, while it is common to refer to the age at which individuals accept retired worker benefits as “early,” “normal” (now “full”) or “delayed,” there is really no “standard” are of retirement within the Social Security program. Social Security provides a full range of ages, 62 to 70, at which benefits can be accepted and, in the aggregate, the value of benefits groups of workers accept at different ages is relatively neutral. SOCIAL SECURITY ENTITLEMENT AGE PROVISIONS Workers covered under Social Security – Old-Age, Survivors and Disability Insurance (OASDI) – can accept reduced retired worker benefits as early as age 62, the EEA, full benefits at the FRA or larger benefits if benefit receipt is postponed past the FRA. The 1983 Amendments to the Social Security Act included provisions gradually raising the age of eligibility for full benefits for people born in 1938 or later, over 27 years, and simultaneously changing monthly benefit amounts for early retirees and for retirees postponing benefit receipt past 2 the FRA. The FRA (age 65) for persons born prior to 1938, inclusive of all those in our sample, remains unchanged. However, beginning with persons born in 1938, the FRA is increased gradually, reaching 66 for persons born from 1943-1954 and 67 for those born in 1960 or later. Although the EEA remains unchanged, for persons born in 1938 or later, the 1983 amendments reduced the value of benefits for the earliest age-62 retirees. Benefits continue to be reduced by five-ninths of a percent for those born prior to 1938 for each month of benefit receipt prior to age 65, to 80% of a full benefit for those accepting benefits at the earliest possible age, the first month of their 62nd birthday. However, as older FRAs are phased in, the benefit reduction factors change and the value of early benefits declines, for example, to 75% of a full benefit for persons born from 1943-1954 and to 70% of a full benefit for those born in 1960 or later. The 1983 amendments also initiated changes in the value of monthly benefits for those delaying receipt until age 70. Benefits are increased for each month that receipt of benefits is postponed past the FRA from 124% of a full benefit for those born in 1960 or later, to 127.5% of a full benefit for persons born prior to from 1933 -1934, and 132% for persons born from 1943-1954. POLICY QUESTIONS While researchers and policymakers disagree about the size of the problem as well as the control various groups have over the decision to accept benefits at early ages, virtually all agree that some early acceptors are at significant disadvantage with respect to economic status, health and the ability to work. John Turner notes (August 2007, p. 4) and others concur (Haveman et. al, 2003; Leonesio et. al, 2000; Zhivan et. al., February 2008) that “the primary criticism of policies that encourage working longer is that doing so places an unfair burden on certain vulnerable groups who have relatively short life expectancies, who are unable to work at older ages because of physical limitations or the physical demands of their jobs and lack early retirement pensions, or who become unemployed at older ages and are unable to find other jobs. In this regard, several policy questions surround Social Security entitlement age policy. Do Increases in the Age of Eligibility for Full Benefits Systematically Disadvantage Economically-Vulnerable Older Persons? The first, a continuation of the discussion that preceded the 1983 Amendments to the Social Security Act, inquires as to whether recent and scheduled changes in the FRA are systematically disadvantaging such persons. Retirement age increases (existing and proposed) are essentially benefit cuts (Munnell et. al., January 2007), justified by past and projected increases in life expectancies at older ages, improved health of older populations in the past 25 years (Turner, August 2007), and the need to address projected financial shortfalls in Social Security. This policy change imposes some disproportionate costs on low-income workers, raising important adequacy and equity concerns. Lower-income workers are generally more financially dependent on Social Security to maintain prior living standards than those with higher lifetime earnings, and thus experience larger reductions in their replacement rates (Munnell et. al, January 2007), as do older persons whose ability to work are constrained by health problems or limited employment options. 3 Should the Age of Earliest Eligibility for Retired Worker’s Benefits Be Raised? Independent of the question of whether the FRA should be further increased once age 67 is fully phased-in, the second question concerns whether the age of earliest (first) eligibility for retired worker benefits, the EEA, should be increased because recent and scheduled FRA changes affect adequacy of retirement income benefits for early acceptors. In 2006, the benefits of three-quarters of new retired worker beneficiaries in 2006 – 1.5 million out of 2 million new retirees – were reduced for early acceptance of benefits. Today, and only slightly lower than the norm in the mid- 1980s, about one-half – somewhat more women (50.2%) than men (45.6%) – accept benefits at age 62 and another 7.6% at age 63. In contrast to early acceptors born prior to 1938 who experienced at most a permanent 20% reduction for early receipt of benefits, the benefits of those accepting benefits as they turn age 62 in 2008 are permanently reduced by 25%. Under current law, those born after 1960 will experience a 30% reduction if they accept retired worker benefits at age 62. Such reductions may be acceptable for those who choose retirement leisure over continuing work for a few more years, but for those who cannot, these reductions represent a major compromise in the program’s adequacy goals. Ironically, proposals to increase the EEA and proposals to maintain the existing EEA can, and are, both justified and opposed on the basis of adequacy. An increase in the EEA may serve to encourage persons to work longer and save more toward their retirement. But such an increase may also pull out an important safety net, albeit less than adequate, for some workers who have few meaningful income alternatives. Herein lies the quandary, that some propose to address through compensatory changes such as lowering the age-65 eligibility age for Supplemental Security Income (Burkhauser, 1996), changing disability eligibility criteria, creating an “elastic” EEA such that it remains at age 62 for workers with very low average lifetime workers with scaled increases in the EEA for others (Zhivan et. al., February 2008), or eliminating the early retirement penalty at a later age (e.g., 70) for early retirees with below-poverty incomes at that age (Haveman et. al., 2003). Proposals to raise the EEA also interact with questions about whether the FRA should be raised beyond age 67 as scheduled in current law. An EEA fixed at age 62 makes it more difficult to increase the FRA, since such increases would further erode the value of benefits for persons accepting them at age 62. Should Medicare Coverage Be Extended to Persons Accepting Retired Worker Benefits Prior to Age 65 who do not have Health Insurance? The third question concerns the availability of health insurance for older working-age persons, especially low- and moderate-income persons with health problems. Richard Johnson (February 2007, p. 2) notes, that “28 percent of adults aged 55 to 64 with incomes below the federal poverty level lacked health insurance in 2004” and that nearly twice as many of those reporting poor or fair health were uninsured – 16 percent compared to 9 percent – as those reporting good or excellent health. Similarly, analysis by John Sheils and Ying-Jun Chen (February 2001) utilizing the March 1996 and 1997 Current Population Survey data find 864,400 uninsured persons ages 62 to 64, with one-half (52%) having incomes below 150% of poverty threshold. With health care costs continuing to rise and private retiree health insurance benefits contracting, it will be increasingly 4 difficult for early acceptors to maintain health insurance until age 65, when Medicare eligibility begins. Thus, the call by some to allow older working-age persons to buy into Medicare. LITERATURE REVIEW Two streams of retirement research and scholarship inform our approach to this study, the first concerned with identifying the extent to which persons who accept early retired worker benefits experience significant income and health security risks; the second with the control various groups have over the decision to accept benefits at early ages. Heterogeneity and Risk Among Early Retirees The retirement literature provides a basis for identifying the diversity of circumstances among early acceptors, including their reasons for accepting retired worker benefits and their economic and social circumstances during their retirement years. The at-risk groups are more likely to include persons with work-limiting health conditions, African Americans, other minorities, persons leaving work prior to age 62 (very early retirees), unmarried women, and persons without employer pensions. Age of withdrawal from the labor force may be closely associated with the retirement circumstances of age-62 early retired worker beneficiaries. Labor force withdrawal prior to age 62, when Social Security benefits are first available, must be funded with pension or other asset income. Plainly, among persons leaving work prior to age 62, many are likely to be among the most high-income retirees. But some studies also suggest that some of these very early retirees are among the poorest, lacking employment income or other resources to maintain their standard of living (Schulz, 2000). Cori Uccello's analysis (1998) of Survey of Income and Program Participation (SIPP) data finds that "the younger a worker is when he or she leaves the job, the more likely it is that the departure is involuntary--either laid-off or discharged.” Uccello (1998) also notes that unmarried women, 55 to 64 and 65 to 70, are more likely to work than married women and potentially at greater risk. Because women continue to earn 25% fewer worker benefits than their male counterparts, women, particularly unmarried women, face up to twice the risk of poverty in retirement than do men (Harrington Meyer & Herd, 2007). These inequalities between men and women have remained consistent over the past 50 years, while the inequalities among women between Whites and other racial and ethnic groups have been growing. As a result, Black and Hispanic women experience higher levels of risk for poverty in old age than do White men or White women, regardless of marital status (Harrington Meyer & Herd, 2007). Burkhauser, Couch and Phillips (1996) point out that in the aggregate, relatively little distinguishes the "Takers" (what we call “Acceptors”) from the "Postponers" with regard to health status or economic well-being (when pension wealth and other assets are included in the measure) (Burkhauser, Couch and Phillips, 1996; Phillips 1997). However, they also note that "relatively small differences between the median early Social Security Takers and Postponers mask large differences within the groups" (Burkhauser, Couch and Phillips, 1996, p. 797). That is, as would 5 be expected, there is much heterogeneity of circumstances within, for example, male Takers or female Takers. While finding that the large majority of male and female Takers are in good health, Phillips also finds that "health is a better indicator of economic well-being than taking early retirement benefits" (1997). Phillips’ analysis also suggests that for black as compared to white men, health may have a stronger influence on the retirement decision, although, as he notes, the sample size is small. The framework that we are using in this study to assess the magnitude of the at-risk population (and sub-groups) of people accepting retired worker benefits at age 62 builds on this previous work. Traditional measures of risk (e.g., poverty, occupational pension income, other assets, health status) are used. Additionally, as the literature suggests, risk is also potentially associated with health insurance coverage, employment capacity, and employment opportunity. In particular, we draw on the work of Karen Holden and Timothy Smeeding (1990). They suggest that in addition to using such traditional measures as poverty thresholds, measures of economic risk should take into consideration the security of income and assets "relative to the economic and health-related hazards that the elderly are likely to face." Holden and Smeeding (1990) also suggest that certain elderly persons --"tweeners" -- with incomes (including the value of food stamps) between 100 and 200 percent of poverty are at risk. These "tweeners" often have limited health insurance protection, unusually large living expenses due to disability, and/or inadequate resources to address possible financial risks associated with long term care (also see Schulz, 2000). Estimates of the size of the at-risk early retiree population vary by the definition of "risk" that is applied (Kingson and Arsenault, 1999; Smith 1999). Assuming adequate demand for older workers, Turner suggests that three to slightly over 10 percent of persons age 62 are at risk if the EEA is raised (August 2007a; August 2007B). Burkhauser, Couch and Phillips (1997) define "risk" as having work-limiting health conditions and being solely dependent on Social Security for pension income. Their analysis concludes that about 10 percent of men and 20 percent of women accepting Social Security benefits at ages 62-63 had work-limiting health problems and reside in houses in which no one receives employer pension income. A Congressional Budget Office study substantiates their finding, but also highlights how sensitive the assessment of "risk" is to the criteria used to define it, noting that: On the basis of either a simple poverty measure alone or the absence of a pension alone, roughly one-quarter to one-third of the early beneficiaries were dependent. Basing dependency on the presence of a work-limiting disability or the absence of a high school education provides a similar range of estimates. But if dependency is determined on the basis of being poor and having a work-limiting disability, its incidence falls to about one in 10 (Smith, 1999). Using a measure of health status drawing on Census Bureau and Social Security Administration measures, Leonesio and colleagues (2000) find that almost one-half of “early retirees” – persons in their study who receive reduced retired worker, spouse or widows benefit – experience health problems, with 22 percent of these retirees reporting impairments that limit or prevent work. Other research, upon which we build in this paper, suggests that narrow conceptions of risk may fall short of fully identifying the distributive consequences of retirement age changes (Kingson and 6 Arsenault, 1999). In this earlier study, when risk is defined exclusively as having both below poverty incomes in 1991 and self-reported poor or fair health, only 3.1 percent of Acceptors meet this standard. In contrast, the most inclusive measure of risk suggests that 52 out of 100 Acceptors have incomes below 200 percent of poverty or do not have liquid assets in excess of $30,000. How Much Choice Do Early Acceptors Have? A rich retirement research literature contributes to the understanding of the extent to which older persons choose when they first accept benefits and exit or otherwise change their labor force attachments. While recognizing that older workers leave work--or in many cases continue to work--for a variety of reasons and in a variety of ways, the literature suggests that most older workers choose leisure over paid employment, oftentimes, primarily incented by comfortable private and public-employee pension benefits (Burkhauser Couch & Phillips, 1996; Quinn & Burkhauser, 1990; Smith, 1999). Social Security and economic incentives, especially the availability of occupational pension income (occupational pension wealth), emerge as the strongest factors influencing the decision to leave employment and accept benefits at ages traditionally defined as early. For individuals covered by occupational pensions, such pensions as well as special early retirement incentives may exert the strongest influence on their labor force withdrawal and pension acceptance decisions (Schulz, 2000). In recent years, researchers have also identified another important economic variable, health insurance coverage, as exerting significant influence on the receipt of benefits and labor force exit (see Karoly & Rogowski, 1994; Leonesio, Vaughan, & Wixon, 2000; Madrian, 1994; Uccello, 1998; Johnson, Davidoff & Perese, 2003). The health of potential “retirees” is not unimportant, but in contrast to the retirement decision literature of the 1950s, 1960s and 1970s, the literature of the last 25 years clearly places health as a secondary influence, relative to pension income (Ruhm, 1989; Quinn, Burkhauser & Myers, 1990; Schulz, 2000). Even so, all studies agree that health remains an important variable influencing the labor force participation and retirement decisions of many older workers (Burtless, 1987; Quinn, Burkhauser & Myers, 1990; Schulz, 2000; Smith, 1999; Uccello, 1998; Pransky, Benjamin & Savageau, 2005; McGarry, 2004). Other factors -- including physically demanding employment, job dissatisfaction (Pransky, Benjamin & Savageau, 2005; Siegrist et. al, 2006), and mental health problems (Buxton, Singleton & Melzer, 2005; Harkonmaki, 2006) -- also exert significant influence on factors leading to labor force exit and benefit receipt, but rarely is the magnitude of the effect as great as pension and related economic incentives or even as physical health status. The retirement literature also examines the extent to which different groups of older workers face qualitatively different decisions. Low-income workers -- a disproportionate number of whom are minorities -- are less likely to be eligible for occupational pension benefits (or, if eligible, for significant benefits). For such workers, the Social Security benefit formula -- which provides for higher replacement rates for low income persons -- may result in Social Security playing a stronger role in their early labor force withdrawal decisions (Bondar, 1993). Coleman (1993) suggests that financial need provides, on average, more motivation for older African Americans to continue to work. However, this is offset by factors such as health status, lower educational levels, effects of past and current discrimination, and occupational requirements for more strenuous activity among current cohorts of older African Americans. Assuming continued health and employment 7 differences between minority and non-minority populations, the early entitlement and other entitlement age increases may have, on average, a disproportionately negative effect on older minorities (Phillips, 1997). For single women, the determinants of the retirement decision seem to be comparable to men (Hurd, 1990), although single women are less likely to have access to employer pensions than men, regardless of work history, race, access to other economic resources, or whether or not they have had children (Harrington Meyer & Herd, 2007). For married women, the husband's retirement status and retirement income appear to exert substantial influence (Hurd, 1990). Also, because women are more likely to experience discontinuities in their labor force participation, they are less likely than men to be vested in employer pensions (Lumsdaine, 1996). Moreover, the pay and occupational differences between men and women also result in differences in the availability and value of such pensions. Hence, occupational pensions can be expected to exert smaller influence on women’s retirement decisions, especially married women. METHODS The main purpose of this paper is to examine the differing circumstances and differing outcomes over 10 to 12 years of persons first accepting Social Security retired worker benefits at ages 62 or 63 in 1994 or 1996. The analysis:  Examines variation in income, wealth, health status, health insurance coverage and employability, along such dimensions as race, Hispanic ethnicity, gender, reported health status, functional ability and work experience; , paying special attention to variations  Identifies health and survival outcomes in 2006 for the 1994/1996 pooled sample; and  Applies measures of risk developed in an earlier study to assess the proportion of at-risk persons who first accept benefits at ages 62 or 63. The Data The analysis uses data from the 1992 (wave 1), 1994 (wave 2), 1996 (wave 3), 2004 (wave 7) and 2006 (wave 8) Health and Retirement Survey (HRS). These data are matched to selected Social Security Administration (SSA) earnings and benefits records for 9472 wave 1 HRS respondents who gave permission in 1992 to access these data for HRS-related research, and to 5247 wave 7 HRS respondents who gave similar permission in 2004. Drawn from SSA’s administrative records, these data provide detailed information on benefits received over the course of life, termination of benefits, quarters of coverage, insured status, earnings in covered employment, and month and year of birth and death. As noted, a substantial minority of HRS respondents did not give permission to join SSA administrative data to their HRS interview data. For this and other reasons, we used both sources to build the sample and construct some variables.1 1 As a reliability check, we frequently cross-checked variables representing similar concepts (e.g., SSA data about receipt of disability benefits and self-reports of disability benefits). The results indicated that the HRS self-reports closely paralleled the information provided in the SSA data, an outcome that increases our confidence in the sample (and variables we created using both types of data). When differences arose, whenever possible we used the SSA data to make sample inclusion decisions. In general, we tried to err on the side of excluding any persons who seemed likely to have received disability benefits as well as persons whose work history (as measured by quarters of coverage and self-reports) were not likely to be eligible to receive retired worker benefits 8 HRS interviews were conducted with 11,539 non-institutionalized persons aged 51 to 61 in 1992, resulting in a nationally representative sample of households (n=7,608) in which at least one member was age 51 to 61 in 1992. African-Americans, Hispanics and Florida residents were over- sampled. Data were collected in subsequent interview years on newly married spouses of survey respondents. If these new spouses were born between 1931 and 1941, they were added to the HRS sample of individuals. This project has also used the RAND Corporation's HRS public use files for waves 2, 3, 7 and 8. RAND has simplified the data structure for researchers. It has "re-coded" the HRS data, assigning household information to each individual respondent. The Sample – Early Benefit Acceptors The sample employed in this paper (n=1129) was drawn from data representing persons who are 1) first interviewed in 1992 and alive at the time of the 1996 interview; 2) elect to receive retired worker benefits at ages 62 or 63 in 1994 or 1996; and 3) do not receive or do not have a history of receiving SSI or Social Security disability benefits. We term persons first accepting Social Security retired worker benefits at ages 62 or 63, “Early Accepters,” and they are the focus of this paper. We began by identifying non-institutionalized HRS respondents who were 1) ages 62 or 63 at the time of the 1994 HRS interview and alive at the time of the 1996 HRS interview or 2) ages 62 or 63 at the time of the 1996 interview. We then restricted the sample to 1994 and 1996 HRS respondents who were eligible or potentially eligible to receive retired worker benefits at ages 62 or 63 in 1994 or 1996. Next, we removed persons who received Social Security or SSI disability benefits, and persons whose work history suggests they would not be eligible – whether applied for or not -- for retired worker benefits at ages 62 or 63. We also excluded like-aged persons (i.e., new spouses) who were added to the HRS sample after wave 2 and persons who were not receiving disability benefits in 1994 or 1996, but subsequently did. Hence, we excluded:  Persons identified in the 1992 or 2004 Social Security permissions data as ever receiving Social Security or SSI disability benefits; and  Persons reporting in the 1994, 1996, 2004 or 2006, HRS interviews that they receive or received SSI disability or Social Security disability insurance benefits. To identify persons eligible to receive retired worker benefits at age 62, we excluded:  Persons identified in the 1992 or 2004 SSA data as having less than 40 quarters2 of coverage in 1991;  Persons who report in the 1996 HRS that they have worked for five or fewer years;  Persons not included in the 1992 sample; and  Persons reporting in the HRS data that they have not worked in 1984 or later. at age 62. 2 For the cohorts we are studying, eligibility for retired worker benefits requires that a worker be “fully insured;” that is, have received at least one quarter of coverage for each year following her/his 21st birthday and prior to age 62. In other words, 40 quarters of coverage – a maximum of 4 earned in any year – are needed by the cohorts we are studying to be defined as “fully covered” and therefore eligible to receive retired worker benefits at age 62. 9 Relying heavily on Rand HRS variables that identify the age and month that respondents and their spouses (if present) first report receiving Social Security benefits,3 we then exclude persons who postpone benefit receipt past age 63. We also used the available SSA benefits data to further refine the sample. A small number of discrepancies were found between the Rand data variables based on self-reports and the SSA administrative benefits variables. In those cases, we relied on the SSA data when determining whether to classify respondents as Acceptors. Structure of Preliminary Analysis In this paper, we present findings describing variations in 1996 and again in 2006 by gender, gender/marital status, race, Hispanic origin, and health status among early acceptors across selected demographic, economic, employment and variables. 1996 weighting factors are used to adjust for any over-representation of African-Americans, Hispanics and Florida residents. We begin by summarizing the commonalties and differences across these groups, and determine whether there are identifiable differences in outcomes in 2006. Next we present different estimates of the size of the at-risk Acceptor population and assess the extent to which risks vary by gender, gender/marital status, race, Hispanic origin, and health status. Definitions of Risk There are many reasonable ways of defining which age-62 retired worker beneficiaries are at risk. The Bureau of the Census poverty thresholds provide an important, although arguably highly restrictive, indication of economic risk. Social Security and pension replacement rates, the present value of pensions, and the amount of available assets can serve as the basis for establishing other measures of retirement income adequacy. Definitions of risk can also properly take into account non-cash assets (e.g., housing) and benefits (e.g., food stamps, health insurance), or the extent to which the decision to leave work is in the control of the individual as opposed to being influenced by a lack of employment opportunity and/or work-limiting health conditions. This study uses multiple definitions of risk. The most restrictive definition assumes that Acceptors are at risk of poor retirement only if they meet three conditions - they must have below poverty income, no health insurance and work limiting health conditions. The more inclusive definitions use expanded notions of income inadequacy (e.g., 200 percent of poverty) and draw on the following concepts to define risk:  Inadequate occupational pension income as operationalized by respondent report that no occupation pension income is received in their household;  Absence of any health insurance at age 62 or 63 as operationalized by respondent report;  Inability to afford medical or other financial emergencies as operationalized by respondent reporting liquid assets valued at less than $30,000 in 1991 or total assets below $40,000;  Limited employment opportunity as operationalized by self report in 1996 of year last worked 3 RASSAGEM and SASSAGEM. These variables are constructed from questions asked in the 1992 through 2006 HRS waves about the date and month that respondents and spouses begin receiving Social Security income. 10  Limited employment opportunity as operationalized by having less than a high school education;  Health-related limits on employment opportunity as operationalized by self report that health limits or prevents work; and  Unhealthy as operationalized by self-report of poor or fair health in 1992 and 1996. RESULTS This first section presents findings concerning the health and economic status in 1996 and ten years later in 2006, of the pooled sample of 1994/1996 early acceptors. Special attention is paid to discussion of variation within and between different groupings of early acceptors and to survival outcomes in 2006. Estimates are presented, based on differing measures of risk, of the extent to which early acceptors, as a whole and sub-groups thereof, can be defined as being at risk in 1996. Demographic, Health, Financial and Employment Status in 1996 and 2006 This section highlights variations within and between different groupings of early acceptors – unhealthy-healthy; male-female; married-unmarried; black-white; survivors in 2006-deceased -- with respect to measures of demographic, health, financial and employment status within the sample at each time point, Demographic Variables: The full sample of early acceptors (n=1129)4, is composed of a slightly larger proportion of males (51.7%) as opposed to females (48.3%). The sample is predominantly white (90%) and married (81.3%), and most members of the sample have 12 or less years of education (38.4% high school diploma, 24% no high school diploma). In terms of self-reported health in 1996, 83.7% of the sample are classified as healthy (excellent, very good, or good health) and 16.3% as unhealthy (fair or poor health). As the sample ages, some proportions change. A little more than four-fifths (82%) of the original sample are alive in 2006. Among these survivors, 62.9% are classified as healthy in 2006 compared to 83.7% in 1996. In other words, by 2006, about 41% of the original sample is either deceased or classified as unhealthy. In contrast to the 1996 sample, the surviving sample in 2006 (n=866) are more skewed towards female, White, better educated and married acceptors. In other words, the data suggest that among early acceptors men, African-Americans, the unmarried and those with less education are more likely to die by 2006. The data presented in Table 1 also suggest that men, African Americans and unmarried persons are more likely to be classified in the unhealthy sub-sample in 1996 and 2006. Table 1. Demographic Characteristics of Early Acceptors Characteristics All Age-62 Sample in 1996 Sample in 2006 Vital Status 2006 4 Sample size may vary based on the measure being analyzed. For example, not all respondents replied to the question on self-rated health in 1996 (n=1097), and vital status was not available for all respondents in 2006 (n=1055). 11 Acceptors Not Not Healthy Healthy Alive Dead Sample Healthy Healthy (n=918) (n=587) (n=866) (n=189) (n=1129) (n=179) (n=218) Gender (%) Female 48.3 48.8 45.8 51.5 44.6 49.7 40.5 Male 51.7 51.2 54.2 48.5 55.4 50.3 59.5 Race (%) White 90.0 91.3 41.5 91.4 88.4 90.4 87.3 Black 8.3 7.1 14.6 6.4 10.4 7.7 11.1 Other 1.7 1.6 2.4 2.2 1.1 1.9 1.6 Percent Hispanic 3.7 3.2 6.2 3.1 4.4 4.0 2.0 Education % no high school diploma 24.0 19.5 48.1 18.4 30.6 23.4 30.4 % high school diploma 38.4 39.6 31.7 40.1 36.5 38.8 33.7 % some college 37.6 40.8 20.2 41.6 33.0 37.8 35.9 Marital Status, 1992 (%) Married/Partnered 81.0 83.0 79.9 82.7 79.7 82.6 77.9 Divorced/Separated 9.3 7.5 10.1 8.4 9.8 8.4 12.3 Widowed 7.0 7.3 7.8 5.6 8.6 6.3 7.0 Never Married 2.7 2.2 2.2 3.3 1.8 2.7 2.9 Health Variables: The proportion of all early acceptors that reported being unhealthy in 1992 (i.e., fair or poor self-reported health) remained relatively stable between 1992 and 1996 (between 13% and 16%). These proportions varied based on individuals’ demographic characteristics. In general, unmarried women and men were in poorer health than married women and men at most points in time, as were Blacks and Hispanics. As would be expected (see Table 2), poor self- reported health in 1992 is strongly associated with poor self-reported health in 1996 (gamma 0.8814, p<.0001). and is also associated with failing to survive until 2006 (gamma 0.4824, p<.0001). Table 2. Self-Reported Health, 1996 (%) Acceptors in 1996 Vital Status in 2006 (N=1097)^ (N=1055)^ Healthy Unhealthy Self-reported health 1992 % (n=918) (n=179) Alive Dead Excellent 25.1 28.5 7.1 27.4 14.8 Very good 32.7 36.1 14.1 33.8 25.5 Good 29.1 29.3 28.0 28.6 34.9 Fair 10.7 5.8 37.2 9.1 18.0 Poor 2.4 0.3 13.6 1.2 6.8 % in fair or poor health 1992 13.1 6.2 50.8*** 10.3 24.8*** ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 In terms of self-rated health, as seen in Table 3, we continue to see heterogeneity among early acceptors. Married women reported better health than unmarried women (gamma -0.2595, p<.0001), and married men reported slightly better health than unmarried men (gamma 0.0137, 12 p<.01). Whites (gamma -0.4293, p<.0001) fare better than Blacks (gamma 0.4487, p<.0001) and Hispanics (gamma 0.4192, p<.01). Table 3. Self-Reported Health of Early Acceptors, 1992 (%) Acceptors Married Unmarried Married Unmarried 1996 1996 Not Black White Hispanic in 1992 Female Female Male Male Healthy Healthy (n=147) (n=964) (n=52) (n=1129) (n=390) (n=148) (n=534) (n=57) (n=918)^ (n=179)^ Excellent 25.1 27.4 23.4 24.4 21.1 12.7 26.4 24.2 28.5 7.1 Very good 32.7 36.1 28.7 29.9 41.7 20.5 33.9 19.4 36.1 14.1 Good 29.1 24.7 30.5 32.9 23.6 41.8 27.9 30.1 29.3 28.0 Fair 10.7 10.3 12.3 10.6 10.5 19.9 9.7 23.8 5.8 37.2 Poor 2.4 1.5 5.2 2.2 3.0 5.1 2.2 2.5 0.3 13.6 % in fair or poor 13.1 11.8*** 23.4*** 12.8** 13.6** 25.0*** 11.9*** 26.3* 6.2*** 50.8*** health 1992 ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 Financial Variables: Median household income in 1991(in nominal dollars) was $37,200, somewhat larger than the $33,304 median income of all households headed by a person ages 55-64 in that year (Census Bureau, 2008). Not surprising, but interesting none-the-less, are the substantial differences in mean (p<.01) and median household incomes in 1991 between those early acceptors who survive until 2006 and those who are deceased by then (see Table 4). As data in Table 4 suggest, early acceptors with household income less than $20,000 in 1991 are less likely to live until the 2006 HRS survey. Household income distributions vary considerably in 1991 by gender and 1992 marital status, race and ethnicity, 1996 health status, and survival to 2006 (see Table 4). Mean household income in 1991 was significantly related to: race (p<.0001), Hispanic ethnicity (p<.05), and survival to 2006 (p<.01). Household income in 1991 was also significantly related to gender and marital status, with the households of married men and women receiving significantly higher incomes than unmarried men (p<.0001) and women (p<.0001). Table 4. Distribution of Early Acceptor Household Income, 1991 (%) Acceptors Unmarried Married Surviving Deceased Black White Hispanic in 1992 Female Male to 2006 by 2006 (n=147) (n=964) (n=52) (n=1129) (n=148) (n=534) (n=866) (n=189) <$10,000 7.1 17.6 4.3 49.0 7.0 3.1 6.4 9.3 $10,000-$19,999 13.4 29.7 10.8 20.3 12.7 28.0 13.0 17.0 $20,000-$29,999 17.7 23.3 15.4 21.0 16.8 27.8 17.3 18.9 $30,000-$49,999 27.3 20.5 28.1 22.6 28.5 26.9 26.9 29.1 $50,000-$74,999 20.8 4.3 25.2 17.0 20.9 10.8 21.8 17.9 $75,000 and over 13.7 4.6 16.3 5.7 14.2 3.4 14.7 8.0 Median Income, 1991 $37,200 $21,000 $42,402 $27,500 $28,000 $26,000 $38,300 $32,960 Mean Income, 1991 45626.34 $27,671*** $50,002*** $32,555*** $33,236 $31,338* $46,880** $39,870** * p<.05, **p<.01, ***p<.001 Although the proportion of the sample living below the Federal poverty level in 1991 ($7,086 for an individual under age 65 and $9,165 for a couple under age 65) was relatively low (4.8%), 16.4% of the respondents were near poor, defined as living within 200% of the Federal poverty 13 level (Table 5). Indeed, there is much variation here, with 32.3% of unmarried females, 22.3% of unmarried men, 29.1% of African-Americans, and 26.8% of unhealthy acceptors in 1996 having incomes below 200 percent of poverty in 1991. Importantly, the findings show that among the sample surviving until 2006, 13.5% of the healthy, compared to 24.3% of the unhealthy ones, had incomes below 200 percent of poverty in 1991, and that fully 22.6% of those deceased by the 2006 survey fell below this threshold. Table 5. Poverty and Near Poverty, 1991 (%) Acceptors Unmarried Married Healthy Unhealthy Black White Hispanic in 1992 Female Male 1996 1996 (n=147) (n=964) (n=52) (n=1129) (n=148) (n=534) (n=918) (n=179) Poverty, 1992 4.8 10.7 3.3 9.2 4.5 3.1 4.2 7.8 Below 150% of Poverty, 9.1 18.5 6.3 16.5 8.5 4.9 8.2 14.0 1991 Below 200% of Poverty, 16.4 32.3 12.6 29.1 15.5 23.6 15.0 26.8 1991 Eighty percent of the sample had health insurance in 1992, and the majority had only one employer-provided insurance policy (68.2%). Smaller proportions of unmarried women, Blacks, Hispanics, and individuals reporting fair or poor health in 1996 reported having health insurance in 1992. Most notable, perhaps, 30.6% of Hispanics, 25.8% of unmarried women, 23.3% of those classified as “Not Healthy in 1996,” and 25.4% of those deceased by 2006 reported in 1992 that they did not have health insurance. When interviewed at age 62 or 63, nearly one-quarter of all acceptors report that they did not have health insurance coverage (Table 6). Most notable, almost one third of unmarried female (31%), and 36% of those we classify as unhealthy in 1996 did not report coverage when interviewed at age 62 or 63. Table 6. Health Insurance Coverage Acceptors Unmarried Married Healthy Not Healthy Black White Hispanic in 1992 Female Male 1996 1996 (n=147) (n=964) (n=52) (n=1129) (n=148) (n=534) (n=918)^ (n=179)^ Percent without health 20.4 25.8 17.7 23.6 20.2 30.6 19.8 23.3 insurance, 1992 Percent without health 24.5 31.0 19.3 24.9 24.6 47.6 22.3 36.4 insurance at ages 62 and 63 Number of employer-provided policies, 1992 (%) 1 68.2 60.1 70.0 60.0 68.9 63.1 68.7 65.2 2 7.4 7.8 6.9 9.3 7.3 6.7 8.2 3.2 3 0.7 1.5 0.7 0.7 0.7 0.0 0.8 0.0 4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 In terms of receipt of employer-pension or annuity income in 1991, 19.7% received these pensions, with a mean benefit among those who did, being equal to $2,611. Among those receiving pension and annuity income in 1992 (Table 7), the amount received varied significantly by marital status among women only (p<.0001), and by 1996 health status (p<.01). Table 7. Distribution of Pension and Annuity Income, 1992 (%) Acceptors Unmarried Married Black White Hispanic Healthy Not Healthy 14 in 1992 Female Male (n=147) (n=964) (n=52) 1996 1996 (n=1129) (n=148) (n=534) (n=918)^ (n=179)^ $0 80.3 76.7 49.2 84.5 80.3 83.1 79.0 87.3 $1-$10,000 9.3 13.5 23.0 7.1 9.5 5.4 9.4 8.9 $10,000-$39,999 9.8 9.3 24.7 8.4 9.6 11.6 10.9 3.8 $39,999-$99,999 0.6 0.5 3.0 0.0 0.7 0.0 0.7 0.0 $99,999-$299,999 0.0 0.0 0.2 0.0 0.0 0.0 0.0 0.0 $299,999+ 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Mean $2,611 $2,326*** $3,870 $1,894 $2,612 $2,621 $2,925* $902* Median $0 $0 $0 $0 $0 $0 $0 $0 ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 Median net worth for the full sample of early acceptors in 1991 was $156,000 and the median value of non-housing or liquid assets was $69,536. Net worth (Table 8) also varies according to specific demographic characteristics, including gender and marital status (p<.0001), race (p<.0001) and ethnicity (p<.01), and health in 1996 (p<.0001). Married men, Whites, and respondents who are healthy in 1996 reported higher mean and median net worth in 1991. We see similar heterogeneity in terms of liquid assets in 1991 (Table 9). Unmarried women (p<.0001), Blacks (p<.0001) and Hispanics (p<.0001), and respondents reporting fair or poor health in 1996 (p<.0001) reported significantly lower liquid assets in 1991, and we can see that their assets were distributed among lower value ranges than married men, Whites, and respondents who were healthy in 1996. Table 8. Distribution of Early Acceptor Net Worth, 1991 (%) Acceptors Unmarried Married Healthy Not Healthy Black White Hispanic in 1992 Female Male 1996 1996 (n=147) (n=964) (n=52) (n=1129) (n=148) (n=534) (n=918)^ (n=179)^ $0 2.4 7.9 0.5 10.0 1.8 10.8 1.8 5.6 $1-$10,000 4.0 8.5 3.3 12.4 3.1 13.2 3.3 7.4 $10,000-$29,999 4.1 6.4 2.8 8.3 3.7 5.0 3.6 6.6 $29,999-$49,999 8.5 11.4 8.1 16.9 7.9 8.0 8.2 10.1 $49,999-$99,999 15.9 17.2 15.6 20.7 15.6 20.4 14.8 22.3 $99,999-$199,999 25.7 24.4 28.3 22.2 25.7 21.8 25.5 26.8 $199,999-$299,999 14.6 11.6 13.2 1.2 16.1 5.3 16.0 7.4 $299,999 and over 24.8 12.6 28.2 8.4 26.2 15.6 26.8 13.8 Median $156,000 $97,400 $162,000 $55,700 $165,000 $83,500 $168,500 $93,000 Mean $268,523 $151,603*** $300,522*** $98,373*** $280,936*** $141,304** $283,572*** $186,547*** ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 Table 9. Distribution of Early Acceptor Liquid Assets, 1992 (%) Acceptors Unmarried Married Healthy Not Healthy Black White Hispanic in 1992 Female Male 1996 1996 (n=147) (n=964) (n=52) (n=1129) (n=148) (n=534) (n=918)^ (n=179)^ $0 4.2 11.9 1.9 17.7 3.1 15.3 3.3 9.2 $1-$10,000 11.3 20.2 11.0 26.3 10.0 29.6 9.3 22.2 $10,000-$29,999 14.3 14.5 12.5 16.8 14.2 16.0 13.8 17.1 $29,999-$49,999 11.5 9.8 12.3 7.9 11.8 10.6 11.5 11.1 $49,999-$99,999 16.6 12.9 17.6 15.0 16.5 5.0 16.3 18.2 $99,999-$199,999 18.6 20.8 17.7 11.2 19.5 14.2 20.1 10.3 $199,999-$299,999 7.7 3.9 8.5 1.2 8.4 9.2 8.6 2.7 $299,999 and over 15.84 6.1 18.4 4.0 16.6 0.0 17.1 9.2 Median $69,536 $44,668 $74,000 $13,000 $77,100 $13,500 $82,000 $35,700 Mean $189,459 $91,000*** $215,779*** $55,313*** $198,898*** $65,861*** $203,037*** $115,495*** ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 15 Employment Variables: Seven in ten early acceptors (71.5%) in the current sample were still working in 1992 although by 1996, nearly 60% consider themselves completely retired and another 14% partially retired. On average, early acceptors had worked over 20 years in at least one job, and had accrued an average of 119 qualifying quarters of Social Security earnings in their careers prior to the 1992 interview. Not surprisingly, married men averaged longer job tenure (p<.05) than unmarried men, and married women averaged shorter job tenure (p<.01) and fewer qualifying quarters (p<.0001) than unmarried women (Table 10). Race and ethnicity were not significantly related to either job tenure or quarters worked. A little more than one-half of African-American, Hispanics and persons classified in 1996 as unhealthy reported holding jobs in 1992 that require “lots of physical effort” all or most of the time compared to about 40% of the other groups. Importantly, more than four times (35.7% versus 7.8%) as many acceptors classified as unhealthy in 1996 reported work impairments in 1992, compared to those classified as healthy in 1996. Table 10. Differences in Employment-Related Factors, 1992 Acceptors Unmarried Married Healthy Not Healthy Black White Hispanic in 1992 Female Male 1996 1996 (n=147) (n=964) (n=52) (n=1129) (n=148) (n=534) (n=918)^ (n=179)^ Longest job tenure prior to 20.3 18.1** 24.3* 20.8 20.7 17.0 20.6 19.3 1992 (mean) Eligible quarters worked 118.7 114.4*** 137.0 117.6 119.1 109.8 119.3 115.6 (mean) Percent Working, 1992 71.5 78.1*** 76.6*** 69.4*** 71.9*** 67.1*** 77.3*** 71.4*** Percent with Work 12.1 12.9*** 13.3*** 10.3*** 12.2*** 4.7*** 7.8*** 35.7*** Impairment, 1992 Job requires Lots of Physical Effort, all or Most 39.7 40.1*** 42.2*** 50.3*** 38.6*** 53.9*** 36.9*** 54.0*** of the time (%) Retirement Status, 1996 58.6 52.5*** 66.5*** 63.1*** 57.8*** 66.8*** 58.1*** 61.3*** Completely Retired Partly Retired 14.0 8.6 12.9 12.8 14.2 7.9 14.1 13.6 Working/Not Retired 27.4 38.9 20.6 24.0 28.0 25.2 27.8 25.1 ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 The data presented in Tables 11 and 12 highlight associations between a number of health-related, employment-related, economic status and employment both with mortality and health status in 2006. Survival to 2006 was significantly associated with a number of factors (Table 11)—including reported work limitations in 1991, jobs that require physical strength, spouse working full-time in 1996, lack of health insurance in 1992 & 1996, less income, more wealth and self report of being retired in 1996. However, the degrees to which these factors were associated with surviving or dying were fairly weak, with gammas ranging from -0.0193 (no health insurance at 62 or 63) to 0.3255 (reported work-related limitation in 1992). A couple of words of explanation are needed about Table 12. It presents data for essentially three 16 different groups – 1) the sample of all acceptors in 1996 subdivided by those who we define as healthy versus unhealthy based on self-reports in 1996; 2) the subsample of acceptors surviving until 2006 subdivided by those defined as healthy versus unhealthy based on self-reports in 2006; and 3) the subsample of acceptors deceased by 2006 subdivided by those defined as healthy versus unhealthy based on self-reports in 1996. Living below 200% of poverty in 1992 is not related to any of these outcomes, while self-reported probability of living to 75 is related to poor self-reported health in 2006 and in 1996, whether or not the respondent survived to 2006. Spousal labor force participation in 1996 is related to respondent survival to 2006 (gamma -0.1115, p<.0001) and to fair or poor health in 2006 (gamma 0.0588, p<.0001), and to fair or poor health in 1996 for respondents who died by 2006 (gamma 0.1394, p<.0001). Respondents with a work impairment in 1992 are more likely to report fair or poor health in 1996 (gamma 0.6063, p<.0001) and 2006 (gamma 0.5514, p<.0001). Having a job in 1992 that requires physical strength is associated with better health in 2006 (gamma -0.2659, p<.0001), as is working in 1992 (gamma -0.0713, p<.001) or being retired by 1996 (gamma - 0.0352, p<.0001). Whether or not a respondent is working in 1996 is weakly associated with poorer health in 2006 (gamma 0.0615, p<.0001). When disaggregating those who are deceased by 2006 in terms of those who were defined as healthy or unhealthy in 1996, we find that a substantially higher proportion of the those decedents defined as unhealthy (versus healthy) in 1996 reporting: work impairments in 1992 (49.9% versus 10.9%), smaller probability in 1992 of living to age 75 (71.5% versus 54.6%), incomes below 200% of poverty in 1992 (29.3% versus 18.9%), smaller median net worth in 1991 ($97,400 versus $144,000), absence of health insurance in 1992 (35.7 versus 19.7%) a job that requires physical strength in 1996 (27.2% versus 16.3%) and being fully retired in 1996 (75.9 versus 61.2%). Table 11. Selected Indicators Vital Status, 2006 (%) Acceptor Alive in Acceptor Deceased 2006 (%) by 2006 (%) (n=805) (n=189) Self-reported probability of living until 75 (mean)*** 67.8 65.7 With work impairment in 1992 (%)*** 10.8 18.7 Below 200% of Poverty 1991 (%)*** 15.4 22.6 Median Net Worth, 1991 $162,500 $119,500 Percent without health insurance, 1992*** 19.7 25.4 Percent without health insurance at ages 62 or 63*** 24.4 24.8 Working 1992 (%)*** 73.3 65.4 Probability of working full-time after 62, 1992 (mean)*** 41.1 46.0 Job always requires physical strength 1992 (%)*** 21.7 40.4 Works full-time in 1996 (%)*** 11.9 10.8 Fully retired in 1996 (%)*** 61.1 67.8 Spouse works full-time in 1996*** 19.4 31.5 31.5*** ^32 respondents did not answer this question ^^2 respondents did not answer this question * p<.05, **p<.01, ***p<.001 Table 12. Potential Indicators of Poor Health and Vital Status, 1996 and 2006 (%) Sample in 1996 Surviving sample in 2006 Deceased by 2006 (n=189) 17 Healthy 1996 Not Healthy Healthy 2006 Not Healthy Healthy 1996 Not Healthy (n=918)^ 1996 (n=179)^ (n=587) 2006 (n=218) (n=122)^^ 1996 (n=65)^^ Self-reported probability of living until 70.0* 51.9* 72.1*** 55.9*** 71.5** 54.6** 75 (mean % considering it likely) With work impairment in 1992 (%) 7.8* 35.7* 6.9*** 20.3*** 10.9* 49.9* Below 200% of Poverty 1991 (%) 15.0 24.3 13.5 22.4 18.9*** 29.3*** Median Net Worth $168,500 $93,000 $176,450 $110,200 $144,000 $97,400 Percent without health insurance, 1992 19.8 23.3 20.2*** 20.6*** 22.0*** 31.7*** Percent without health insurance at 22.3 36.4 24.6*** 26.4*** 18.9*** 35.7*** ages 62 or 63 Working 1992 (%) 70.8 75.2 74.3 71.4 62.9*** 69.9*** Probability of working full-time after 62, 42.2 39.7 40.0 43.6 45.9 46.2 1992 (mean) Job always requires physical strength 20.4 28.7 18.4*** 30.6*** 16.3*** 27.2*** 1992 (%) Works full-time in 1996 (%) 11.4 12.2 10.6 16.1 13.3*** 6.1*** Fully retired in 1996 (%) 60.6 69.9 61.7 58.7 61.2*** 75.9*** Spouse works full-time in 1996 (%) 20.7 26.4 19.7*** 23.9*** 35.2*** 25.1*** ^^2 respondents did not answer this question * p<.05, **p<.01, ***p<.001 To summarize the data presented in Tables 1 through 12 indicates that:  There is much diversity substantial among persons accepting Social Security retired worker benefits at ages 62 or 63 in 1994 or 1996 Some groups of early acceptors bear large disproportionate shares of income, health, and insurance coverage risks.  Self-rated health prior to receipt of Social Security retired worker benefits is linked to poorer health during retirement and to earlier death.  Lower income, net worth, and liquid assets are linked to poorer health in retirement and to morality 10-12 years after acceptance of early retired worker benefits.  Unmarried men and women, African Americans, and Hispanics tended to be at greater risk in terms of income and assets, and were also less likely to have health insurance in 1992.  Men, African Americans, and unmarried people were more likely to be unhealthy in later years, regardless of whether or not they survived to 2006.  Living below 200% of the poverty level prior to benefit receipt is linked to a greater risk of dying by 2006.  Those who reported having health insurance prior to early retirement reported being in better health in later years and enjoyed greater longevity – one fifth of unhealthy respondents in 2006, and a full quarter of those who died by 2006, had reported in 1992 that they did not have health insurance. Finally, and most importantly, the findings highlight that a significant proportion of early acceptors – one quarter of the full sample – do not have health insurance when they first accept Social Security retired worker benefits at ages 62 or 63 in 1994 or 1996. Indeed, a quarter of African- Americans, almost a third of unmarried women and men, over a third of those classified as unhealthy in 1996, and almost half of Hispanics are not covered by any health insurance. 18 Risk of a Poor Retirement The magnitude of risk experienced by early acceptors varies substantially by the definition of risk being evaluated. The more inclusive the definition of risk, the larger the proportion of acceptors likely to be affected (Table 13). Regardless of how risk is defined, incidence of risk varies among different groups of acceptors. Table 13. Alternative Estimates of Risk as a Percentage of Age-62 Retired Worker Beneficiaries (Adjusted sample n=1129) % At Risk Restrictive Definitions of Risk Below poverty in 1991 and in fair or poor health 1.5 Below 150% of poverty in 1991 and in fair or poor health 1.9 Below 150% of poverty in 1991 and did not finish high school 3.6 Below poverty in 1992 4.8 Moderate Definitions of Risk (Below poverty) or (between 100 and 200 percent of poverty in 1991 and no health insurance) 10.2 Below 150% of poverty in 1991 9.1 No household pension and health limits work - 1992 11.0 No household pension and in fair or poor health - 1992 9.4 (Below poverty) or (between 100 and 200 percent of poverty in 1991 and either does not have 11.6 health insurance or health limits work) (Below 150% poverty) or (between 150 and 200 percent of poverty in 1991 and either does not 13.8 have health insurance or health limits work) Below 200% of poverty and has less than $30,000 in liquid assets 16.4 Below 200 percent of poverty in 1991 16.4 Inclusive Definitions of Risk Below 200 percent of poverty in 1991 or (between 200 and 300 percent of poverty in 1991 and 24.5 either does not have health insurance or has less than high school education) Below 200 percent of poverty in 1991 or less than $40,000 in total assets 26.6 Below 200 percent of poverty in 1991 or less than $30,000 in liquid assets 37.4 On most measures roughly twice as many Black early acceptors as compared to Whites were defined as being at risk (Table 14) For example, in 1991 48.3% of Black compared to 24.9% of White acceptors reported incomes that were below 200% poverty in 1991 or total assets (including home equity) of less than $40,000. Table 14. Incidence of Selected Risks by Race (%) All Age-62/63 Black White Characteristics Acceptors (n=147) (n=964) (n=1129) Below poverty, health limits work, no health insurance 3.0 6.2*** 2.8*** No pension in 1991 and fair or poor health 11.0 21.2*** 10.1*** No pension in 1991 and health limits work 9.4 7.9*** 9.6*** Tweeners 13.8 23.0*** 13.1*** Below 200% poverty in 1991 or less than $40,000 in total assets 26.6 48.3*** 24.9*** Below 200% poverty or less than $30,000 in liquid assets 37.4 63.5*** 35.4*** ^32 respondents did not answer this question * p<.05, **p<.01, ***p<.001 19 In general, unmarried men and women were more at risk than married men and women, and women in both groups were at higher risk than their male counterparts when applying more inclusive definitions of risk (Table 15). Table 15. Incidence of Selected Risks by Sex and Marital Status (%) All Age- Married Unmarried Married Unmarried 62/63 Female Female Male Male Characteristics Acceptors 1992 1992 1992 1992 (n=1129) (n=390) (n=148) (n=534) (n=47) Below poverty in 1991, health limits work, no 3.0 1.9*** 8.4*** 1.7*** 7.8*** health insurance No pension in 1991 and fair or poor health 11.0 10.6*** 16.3*** 10.3*** 7.3*** No pension in 1991 and health limits work 9.4 9.1*** 10.3*** 9.9*** 6.0*** a Tweeners 13.8 12.9*** 25.7*** 10.1*** 22.3*** Below 200% poverty in 1991 or less than 26.6 23.1*** 48.6*** 21.3*** 42.0*** $40,000 in total assets Below 200% poverty in 1991 or less than 37.4 34.0*** 57.2*** 32.5*** 53.4*** $30,000 in liquid assets ^32 respondents did not answer this question a Below 150% poverty) or (between 150% and 200% of poverty in 1992 and either does not have health insurance or health prevents work. * p<.05, **p<.01, ***p<.001 Risk indicators based on data collected in 1992 are associated, 14 years later, with health outcomes. A higher proportion of acceptors defined as unhealthy in 2006 were defined as being at risk in 1992 as measured by each indicator in Table 16. Table 16. Incidence of Selected Risks by Health Status in 2006 (%) Acceptors Surviving Until 2006 All Age- Healthy Unhealthy Characteristics 62/63Acceptors 2006 2006 (n=1129) (n=587) (n=218) Below poverty, health limits work, no health insurance 3.0 2.4*** 2.8*** No pension in 1992 and fair or poor health 11.0 4.2*** 23.7*** No pension in 1991 and health limits work 9.4 6.0*** 15.5*** a “Tweeners” 13.8 10.8*** 17.7*** Percent without health insurance at ages 62/63 20.4 20.2*** 20.6*** Below 200% poverty in 1992 or less than $40,000 in total 26.6 22.3*** 37.1*** assets Below 200% poverty in 1992 or less than $30,000 in liquid 37.4 30.6*** 49.5*** assets Poor or fair health in 1992 13.1 4.6*** 26.26*** _____________________ a Below 150% poverty) or (between 150% and 200% of poverty in 1992 and either does not have health insurance or health prevents work. * p<.05, **p<.01, ***p<.001 Survival to 2006 was significantly related to several definitions of risk laid out by Kingson and Arsenault (2000). When applying restrictive definitions of risk, we see that respondents who were living below poverty in 1991, whose health limited their ability to work, and who had no health insurance (gamma 0.6007, p<.0001), those who were near-poor and unhealthy in 1992 (gamma 20 0.4243, p<.0001), and those who had no household pension in 1991 and whose health limited their ability to work in 1992 (gamma 0.5148, p<.0001) were at greater risk for dying by 2006 (Table 17). Survival to 2006 is significantly related to all four moderate definitions of risk listed in Table 17, as well as to being below 150% of poverty in 1991. Finally, survival to 2006 is also significantly related to the inclusive definitions of risk listed in Table 18, as well as to living below 200% of poverty in 1991. Table 17. Incidence of Selected Risks by Survival in 2006 (%) Acceptors All Age-62/63 Acceptors Deceased by Acceptors Alive in 2006 2006 (n=1129) (n=805) Restrictive Definitions of Risk (n=189) Below poverty in 1991, health limits work, no 3.0 2.5*** 5.13*** health insurance Below 150% of poverty in 1991 and unhealthy 1.9 1.3*** 3.8*** in 1992 Below poverty in 1991 4.8 4.0*** 6.7*** Moderate Definitions of Risk No household pension and health limits work 9.4 8.5*** 15.1*** in 1992 Below 150% of poverty in 1991 9.1 7.7*** 10.4*** Poor or (near-poor in 1992 with no health 10.2 8.9*** 15.6*** insurance) Poor or (near poor in 1992 (without insurance 11.6 10.4*** 17.5*** or health limits work)) Near poor or (near poor and (no health 13.8 12.4*** 21.1*** insurance or health limits work)) Below 200% of poverty in 1991 16.4 15.4*** 22.6*** Inclusive Definitions of Risk Percent without health insurance at age 24.1 24.4*** 24.8*** 62/63 Below 200% of poverty or less than $40,000 26.6 25.6*** 33.2*** in total assets Below 200% of poverty in 1992 and less than 37.4 35.7*** 48.8*** $30,000 in liquid assets * p<.05, **p<.01, ***p<.001 As mentioned previously, the magnitude of risk experienced by early acceptors varies substantially by the definition of risk being evaluated, with the larger proportions of acceptors being affected by more inclusive definitions of risk.  The most restrictive risk measures, those relying heavily on poverty (or 150% of poverty), define, at most, 5 percent of age 62/63 acceptors as being at risk.  The more moderate measures lead to estimates of roughly 9 to 16 percent being at risk when they accept benefits.  The more inclusive definitions of risk, which incorporate variously incorporate health insurance coverage and modest assets into the definition and a higher poverty ratio (200% of poverty), define one-quarter to two-fifths of early acceptors as at risk. Very importantly, the incidence of risk varies among different groups of acceptors, regardless of how risk is defined. These findings support previous literature on retirement and early retirement: 21  Black early acceptors were more at risk than Whites in a variety of ways, bearing almost twice as much risk on every measure. For example, where ten percent of Whites met the risk criterion of being in poor or fair health and also not having a pension in 1991, 21% of Blacks met this criterion. When risk is defined as either having incomes below 200% poverty or less than $30,000 in liquid assets, 35% of Whites and 63% of African- Americans are defined as at risk.  Similarly, and only somewhat less dramatically, unmarried men and women were generally more at risk than their married counterparts. Women were also generally at higher risk than men.  Unhealthy respondents who survived until the 2006 wave were more likely to be at risk than healthy survivors.  Survival to 2006 was significantly related to several definitions of risk, from the most restrictive to the most inclusive, and many of the indicators of fair or poor health and vital status defined in this project are significantly associated with fair or poor health in early retirement and risk of earlier death. CONCLUSIONS: DISCUSSION AND POLICY IMPLICATIONS This paper asks, “Are age-62/63 retired worker beneficiaries at risk?” The short answer, fully consistent with the literature reviewed at the beginning of the paper, is “Some are, and some are not.” One in four (24.5%) persons first accepting retired worker benefits at ages 62 or 63 in 1994 or 1996 lack health insurance; three in ten (29.8%) report liquid assets of less than $30,000 in 1992; one in six (16.4%) report household incomes below 200 percent poverty in 1991; one in eight (12.1%) report work impairments in 1992; and one in six (16.3%) that that their health is poor or fair in 1996 as opposed to good, very good or excellent. The data also show that these and other indicators of risk are associated with indicators of well-being, including probabilities of surviving until 2006. Moreover, consistent with previous studies, African-American, Hispanic, low-income, unmarried female and unmarried male acceptors carry substantially greater risk. By contrast, the data also show that a majority of early acceptors reported in 1992 that their health is very good or excellent in 1992 (57.8%); more than one third (34.5%) reported household incomes of at least $50,000 in 1991; eight out of nine (87.9%) reported that they did not have a work impairment in 1992; seven out of ten (71.5%) were working in 1992 and eight out of ten (82%) survive until 2006. In short, as the literature suggests, persons first accepting retired worker benefits at ages 62/63 in 1994/96 are a heterogeneous group. In terms of policy implications, first we interpret the findings as a basis for caution with regards to proposals that would further increase the age of eligibility for full retired worker benefits and increase the early entitlement age (EEA). Increasing the full retirement age (FRA) beyond what is already scheduled in the law would result in additional benefit and replacement rate reductions. In the context of recent declines in the value of retirement savings and home equity and increasing levels of unemployment, raising the FRA would be an increasingly questionable policy change, even for middle-income households. Such changes would fall most heavily on those low-income 22 older workers, who are not well-positioned to continue work. If a scheduled increase in the FRA is considered as part of a solution to Social Security’s projected financing problem, the findings point to the need for off-setting changes in Social Security disability eligibility and SSI disability and age eligibility criteria, perhaps lowering the SSI age to 62. From an adequacy perspective, conflicting arguments can be marshaled with respect to raising the age of first eligibility for benefits, commonly called the early entitlement age (EEA), currently set at age 62. The 1983 Amendments to the Social Security Act increased the actuarial reduction for persons first accepting benefits at age 62 from 20% to 25% as the current FRA (age 66) was phased in. This reduction will increase to 30% as the FRA of age 67 is phased in for persons first accepting benefits at age 62 in 2022. From one point of view, it is important to encourage older workers to build up retirement savings and avoid accepting benefits at early ages that result in large permanent reductions in retired worker benefits. With this in mind, some argue that the EEA should be increased, perhaps to age 64 so that benefits will not be reduced by more than 20% for early acceptors. From another point of view, acceptance of early benefits is a life-line for many. It is not ideal for benefits of future retired workers to be reduced by as much as 30%, but from this point of view it would be even less acceptable to remove this economic life-line as an option for those older workers whose employment prospects are very limited. As a matter of strategy, those who believe the FRA should be increased would reasonably consider an increase in the EEA as a necessary condition to furthering that objective. Those opposing a change in the FRA should find strategic benefit in holding the EEA at age 62. An EEA of age 62 arguably serves as ballast with respect to raising the FRA. If the FRA were to be further increased while maintaining the existing EEA, at some point the actuarial reduction at the earliest ages would be so large that the income adequacy goals of the program would be undermined. For instance, if the FRA is raised to age 70, the monthly benefits for persons first accepting benefits at the EEA might be reduced to as little as 55% of the full benefit. Thus, absent raising the EEA, it becomes increasingly difficult to justify an increase in the FRA, given the desire to maintain adequacy goals of Social Security. We are less circumspect with respect to the findings as it relates to the question of whether the age- eligibility criteria for Medicare should be lowered or some other means utilized to assure health insurance coverage of early acceptors. The findings point to a need to provide access to health insurance benefits for a substantial portion of early acceptors in the years immediately prior to the age-65 eligibility for Medicare benefits. One fifth (20.4%) of age 62/63 early acceptors reported in the 1992 interview that they did not have health insurance. More concerning, one in four (24.5%) early acceptors, including almost one-half of Hispanic acceptors (47.8%), reported that they did not have health insurance when interviewed in 1994 or 1996 at ages 62 or 63. To conclude, our research suggests that persons first accepting Social Security retired worker benefits at ages 62 and 63 in 1994 or 1996 experience varying degrees of risk to their well being at these ages, and we find rough evidence that these risks condition their well-being in retirement and survival probabilities. The major policy implication of the findings presented in this study is that consideration should be given to providing a health insurance option for persons first accepting retired worker benefits prior to age 65. 23 Finally, in terms of research implications, the findings reinforce a conclusion reached in previous research “that narrow conceptions of risk may fall short of fully identifying the distributive consequences of retirement age changes, especially for African Americans, Hispanics, low- income, unmarried individuals and unhealthy early retirees” (Kingson and Arsenault, 2000). As noted, depending on measure used, one can conclude that as few as 1.5% or as many 39.9% of early acceptors are at risk. This large range implies that policy researchers should be careful not to seize on only one measure, drawing instead on several measures when seeking to identify the magnitude of risks existing for those accepting retired worker benefits at early ages, 62 and 63. 24 References Bondar, J (1993). "Beneficiaries Affected by the Annual Earnings Test, 1989," Social Security Bulletin, vol. 56, no. 1 (Spring 1993), pp. 20-28. Burkhauser, R. (1996). Touching the third rail: Time to return the retirement age for early Social Security benefits to 65. The Gerontologist, 36 (6): 726-727. Burkhauser, R. V., Couch, K. A., and Phillips, J. W. (1996). Who takes early Social Security benefits? The economic and health characteristics of early beneficiaries. The Gerontologist, 36 (6): 789-799. Burtless, G. (1987). Work, Health and Income Among the Elderly (Studies in Social Economics). Washington, DC: Brookings Institution Press. Burtless, G. and Quinn, J. (2003) "Is Working Longer The Answer For An Aging Workforce?" Issues in Brief #11. Chestnut Hill, MA: Center for Retirement Research. Buxton, J. W., Singleton, N., and Melzer D. (2005). The mental health of early retirees: National interview survey in Britain. Social Psychiatry and Psychiatric Epidemiology, 40: 99-105. Census Bureau, United States (2008). Table H-10. Age of Head of Household: All Races by Median and Mean Income: 1967 to 2007. Current Population Survey, Annual Social and Economic Supplements. Retrieved online, April 8, 2009: http://www.census.gov/hhes/www/income/histinc/h10AR.html. Coleman, L. M. (1993). The Black Americans Who Keep Working. In J. S. Jackson, L. M. Chatters & R. J. Taylor (Eds.), Aging in Black America. Newbury Park, CA: Sage Publications Feldman, D. C. (1994). The decision to retire early: A review and conceptualization. The Academy of Management Review, 19 (2): 285-311. Friedberg, L. (March 2007). “The Recent Trend Towards Later Retirement”. WOB #9. Chestnut Hill, MA: Center for Retirement Research at Boston College. Harkonmaki, K., Lahelma, E., Martikainen, P., Rahkonen ,O., and Silventoinen, K. (2006). Mental health functioning (SF-36) and intentions to retire early among ageing municipal employees: The Helsinki health study. Scandinavian Journal of Public Health, 34: 190-198 Harrington Meyer, M. and Herd, P (2007). Market Friendly or Family Friendly? The State of Gender Inequality in Old Age. New York, NY: Russel Sage Foundation. Haveman R, Holden K., Wilson K, and Wolfe B. (2003). Social Security, age of retirement, and economic well-being: Intertemporal and demographic patterns among retired-worker beneficiaries. Demography, 40 (2): 369-394. 25 Holden, K. C. and Smeeding, T. (1991). The Poor, the Rich, and the Insecure Elderly Caught in Between. Milbank Quarterly, 68 (2): 191-219. Hurd, M. D. (1990). Research on the Elderly: Economic Status, Retirement, and Consumption and Saving, Journal of Economic Literature, American Economic Association, 28(2): 565-637. Johnson, R. W. (February 2007). “What Happens to Health Benefits After Retirement?” WOB #7. Chestnut Hill, MA: Center for Retirement Research at Boston College Johnson, R.W., Davidoff, A. J., and Perese, K. (2003). Health insurance costs and early retirement decisions. Industrial and Labor Relations Review, 56 (4): 716-729. Karoly, L.A. and J.A. Rogowski (1994). "The Effect of Access to Post-Retirement Health Insurance on the Decision to Retire Early," Industrial and Labor Relations Review, 48 (1): 103- 123. Kingson, E. R. and Arsenault, Y. (2000). “The diversity of risk among age-62 retired worker beneficiaries.” Working Paper 2000-08. Chestnut Hill, MA, Center for Retirement Research at Boston College. Leonesio, M.V., D.R. Vaughan, & B. Wixon (2000). Early Retirees Under Social Security: Health Status and Economic Resources. Social Security Bulletin, 63 (4): 1-16. Lumsdaine, R. .L.(1996). Factors affecting labor supply decisions and retirement. In E.A. Hanushek & N. L. Maritato, eds., Assessing Knowledge of Retirement Income Behavior. Panel on Retirement Income Modeling, Committee on National Statistics, National Research Council,. Washington, D.C.: National Academy Press. Madiran, B. C. (February 1994). Employment-based health insurance and job mobility: Is there evidence of job-lock? Quarterly Journal of Economics, 109 (1): 27-54. McGarry, K. (2004). Health and retirement - Do changes in health affect retirement expectations? Journal of Human Resources, 39 (3): 624-648. Mosisa A. and Hipple S. (2006), Trends in labor force participation, in the United States, Monthly Labor Review, pp. 35 – 57. Retrieved online, February 23, 2009 http://www.bls.gov/opub/mlr/2006/10/art3full.pdf Munnell, A. H., Golub-Sass, F., Perun, P., and Webb, A. (2007) “Households ‘at Risk’: A Closer Look at the Bottom Third. Issues in Brief #7-2. Chestnut Hill, MA, Center for Retirement Research at Boston College. Phillips, J. W. (May 16, 1997). The Well-Being of Early Social Security Beneficiaries by Race, Health, and Employer Pension Eligibility: Evidence from the Retirement History Survey. Dissertation, Syracuse University, Syracuse; Chapter 1.Pransky, Benjamin & Savageau, 2005 26 Quinn, J. F. and Burkhauser, R. (1990). “Economic Incentives and the Labor Force Participation of Older Workers”, pp. 159-179. In Research in Labor Economics, Vol. 11, Bassi, L. and Crawford, D. (editors). Quinn, J. F., Burkhauser, R. V. & Myers, D. A. (1990). Passing the Torch: The Influence of Economic Incentives on Work and Retirement. Kalmazoo, MI: W.E. Upjohn Institute for Employment Research. Quinn, J. F. and Burtless, G. (December 2002). “Is Working Longer the Answer for an Aging Workforce?” Issue Brief #11, Chestnut Hill, MA: Center for Retirement Research at Boston College. Ruhm, C. J. (June 1989). Why Older Americans Stop Working. The Gerontologist, 294-299. Schulz, J. (2001). The Economics of Aging. Westport, CT: Auburn House. Sheils J. and Chen Y-J (2001). Medicare Buy-In Options: Estimating Coverage and Costs. The Lewin Group, Inc., The Commonwealth Fund. Retrieved online, February 23, 2009: http://www.commonwealthfund.org/usr_doc/sheils_buy-inoptions_441.pdf?section=4039 Siegrist, J., Wahrendorf, M., von dem Knesebeck, O., Jurges, H., and Borsch-Supan, A. (2006). Quality of work, well-being, and intended early retirement of older employees - baseline results from the SHARE study. European Journal of Public Health, 17 (1): 62-68. Smith, R. (January 1999). Raising the Earliest Age for Social Security Benefits. Washington, DC: Congressional Budget Office. Turner, J. A. (2006). Work at Older Ages: Is Raising the Early Retirement Age an Option for Social Security Reform? Work Opportunities for Older Americans Series. Chestnut Hill, MA Center for Retirement Research at Boston College. Uccello, C. E. (1998). Factors Influencing Retirement: Their Implications for Raising Retirement Age. Document #9810. Washington, DC: AARP Public Policy Institute. Zhivan, N., Sass, S. A., Sapozhnikov, M., and Haverstick K. (February 2008). An “Elastic” Earliest Eligibility Age for Social Security. Issues Brief #8-2. Center for Retirement Research, Boston College, Boston, MA. 27 RECENT WORKING PAPERS FROM THE CENTER FOR RETIREMENT RESEARCH AT BOSTON COLLEGE Taxes and Pensions Peter Diamond, May 2009 How Much Do Households Really Lose By Claiming Social Security at Age 62? Wei Sun and Anthony Webb, April 2009 Health Care, Health Insurance, and the Relative Income of the Elderly and Nonelderly Gary Burtless and Pavel Svaton, March 2009 Do Health Problems Reduce Consumption at Older Ages? Barbara A. Butrica, Richard W. Johnson, and Gordon B.T. Mermin, March 2009 Financial Hardship Before and After Social Security’s Early Eligibility Age Richard W. Johnson and Gordon B.T. Mermin, March 2009 Rising Tides and Retirement: The Aggregate and Distributional Effects of Differential Wage Growth on Social Security Melissa M. Favreault, February 2009 Accounting for the Heterogeneity in Retirement Wealth Fang Yang, January 2009 Labor Supply Elasticity and Social Security Reform Selahattin Imrohorolu and Sagiri Kitao, January 2009 Evaluating Micro-Survey Estimates of Wealth and Saving Barry P. Bosworth and Rosanna Smart, January 2009 Portfolio Choice in Retirement: Health Risk and the Demand for Annuities, Housing, and Risky Assets Motohiro Yogo, January 2009 The Disappearing Defined Benefit Pension and its Potential Impact on the Retirement Incomes of Boomers Barbara A. Butrica, Howard M. Iams, Karen E. Smith, and Eric J. Toder, January 2009 All working papers are available on the Center for Retirement Research website (http://www.bc.edu/crr) and can be requested by e-mail (crr@bc.edu) or phone (617-552-1762).