ORES Working Papers
Correlation Patterns between Primary and Secondary Diagnosis Codes in the Social Security Disability Programs
This paper addresses impairment comorbidity among participants in programs that provide disability benefits. Comorbidity in this context is defined as the simultaneous presence of a primary and a secondary medical diagnosis. The author fits a high-dimensional Bayesian multivariate probit model with a 10% random sample of 2009 initial claimants (disabled workers, including individuals concurrently applying for Disability Insurance and Supplemental Security Income). The resulting correlation estimates provide evidence of strong impairment comorbidity patterns at the initial-claim level. Many of the findings mirror the epidemiological evidence, such as associations of diabetes with chronic renal failure, open wounds of a lower limb, peripheral neuropathies, and blindness/low vision. Other results are surprising. For instance, the correlation estimates defy the presumption of high positive association between mental and musculoskeletal-system diagnoses.
A Multidisciplinary Review of Research on the Distributional Effects of Raising Social Security's Early Entitlement Age
When estimating potential adversity caused by an increase in the early entitlement age (EEA), findings from both the EEA literature and the broader public health literature do not suggest that the Social Security–covered worker population can be easily separated into two groups—an unaffected or low-risk group and an easily identifiable vulnerable or high-risk group. This evidence appears largely supportive of the conclusions reached by the retired-worker benefit's original designers and may suggest implementation difficulties for proposals that seek to raise the EEA, while protecting groups deemed by the proposers to be adversely affected by that increase. Because the risks insured against by the retired-worker benefit are not limited to an easily identifiable segment of the population, the universality of Old-Age Insurance under current law may better match the underlying exposure to risk in the insured population than a targeted or needs-based alternative.
This paper evaluates the out-of-sample performance of two stochastic models used to forecast age-specific mortality rates: (1) the model proposed by Lee and Carter (1992); and (2) a set of univariate autoregressions linked together by a common residual covariance matrix (Denton, Feavor, and Spencer 2005).
Social Security as a Retirement Resource for Near-Retirees, by Race and Ethnicity, Nativity, Benefit Type, and Disability Status
This paper analyzes Social Security benefits as a retirement resource for selected subgroups of current and recent cohorts of near-retirees. The paper examines the distribution of benefits among (1) several race-ethnic subgroups, (2) the native-born and the foreign-born, (3) worker, spouse, and survivor beneficiaries, and (4) the disabled and the nondisabled. We use improved data (actual earnings history data) to produce more accurate measures of benefits. We look at how the average values of several benefit measures such as Social Security wealth and earnings replacement rates differ among the selected subgroups and discuss reasons for these differences. We find that substantial differences in earnings levels and/or mortality levels among these subgroups interact with Social Security program provisions to produce sizable differences in the values of our benefit measures.
Trends in Mortality Differentials and Life Expectancy for Male Social Security–Covered Workers, by Average Relative Earnings
This study presents an analysis of trends in mortality differentials and life expectancy by average relative earnings for male Social Security–covered workers aged 60 or older. Mortality differentials, cohort life expectancies, and period life expectancies by average relative earnings are estimated. Period life expectancy estimates for the United States are also compared with those of other Organisation for Economic Co-operation and Development (OECD) countries. In general, for birth cohorts spanning the years 1912–1941 (or deaths spanning the years 1972–2001 at ages 60–89), the top half of the average relative earnings distribution has experienced faster mortality improvement than has the bottom half. The sample is expected to be selectively healthier than the general population because of a requirement that men included in the sample have some positive earnings from ages 45 through 55. This requirement is expected to exclude some of the most at-risk members of the U.S. population because of the strong correlation between labor force participation and health.
New Evidence on Earnings and Benefit Claims Following Changes in the Retirement Earnings Test in 2000
In April 2000, Congress enacted the Senior Citizens Freedom to Work Act of 2000, which removed the retirement earnings test for individuals at the full retirement age and older. This paper examines the labor force activity of workers aged 65–69 relative to older and younger workers in response to the removal of the earnings test. We use the 1 percent sample of Social Security administrative data that covers the period from 4 years before to 4 years following the removal of the test. Quantile regression methods allow us to identify the earnings levels of workers who change their work effort.
This paper analyzes Social Security benefits as a retirement resource for near-retirees. It looks at how the average values of several measures of benefits such as Social Security wealth and earnings replacement rates have changed from earlier cohorts to today's near-retiree cohort, examines differences among demographic and socioeconomic groups within cohorts, and discusses reasons for these changes and differences. The paper uses greatly improved data (actual earnings histories) to produce more accurate measures of benefits; it also uses some new benefit measures. Key findings include the following: (1) average real Social Security wealth increases markedly for successive age cohorts, primarily because of increases in average real earnings; (2) replacement rates fall for recent cohorts, primarily because of the phase-in of increases in the age of eligibility for full benefits; and (3) median Social Security wealth is much higher for women than for men because women live longer.
Conventional wisdom holds that the majority of early retirees are in good health and that only a minority are in poor health. This wisdom is based on examinations of levels of health among the early retiree population. In contrast, this paper looks at both the health and mortality risk of early retirees relative to the health and mortality risk of age 65 retirees. This paper finds substantial heterogeneity among early retirees in health and mortality risk related to the age at which they are entitled to Social Security benefits. Early retirees consist of a group in extremely poor health, a group with health equal to age 65 retirees, and a group with health in between. The majority of early retirees are in poorer health and have higher mortality risk than age 65 retirees, and only a minority have health and mortality risk as good as that of age 65 retirees.
This paper analyzes the predisability earnings of Social Security Disability Insurance (DI) applicants using yearly pools of applicants from 1977 through 1997 constructed from SSA program data that are matched to multiple panels of the Survey of Income and Program Participation (SIPP). Results of this study show that the predisability earnings of workers denied DI benefits are significantly lower, by $4,518 per year, than the earnings of those allowed and that the influx of workers with low predisability earnings coincides with the recent rapid growth in applications for DI benefits. Average predisability earnings are the highest for applicants during 1981–1983 (when benefit eligibility was tightened) and are the lowest for applicants during the 1990–1994 period, which included the 1990–1991 recession.
Upon a worker's death, Social Security pays benefits to each minor or disabled child and to the worker's widow(er), provided that a child of the worker is in his or her care. Although remarriage has no effect on a child's eligibility for benefits, the benefit going directly to the widow(er) terminates if he or she remarries. This paper examines the termination provision and discusses possible effects of the provision on the young widow(er) population.
Historical Redistribution Under the Social Security Old-Age and Survivors Insurance and Disability Insurance Programs
This study is the third in a series of studies that use comprehensive Social Security administrative data on past earnings and benefits by year, age, gender, and race to analyze historical redistribution across those characteristics under the Social Security program. It examines historical lifetime redistribution to date across and within cohorts born through 1927, combining and extending the results of the previous two studies, for which less historical data were available. Redistributional estimates incorporating the additional data confirm the results of the earlier studies relative lifetime redistributional outcomes to data under the DI program have generally been much more favorable for "Other Races" than for "Whites;" have generally been more favorable for females than for males in most, but not all, of the cohorts considered; and accumulated benefits have generally exceeded accumulated taxes by substantial margins for all but the earliest birth cohort groups. In contrast to outcomes under the OASI program, accumulated net transfers to date for very early birth cohorts have generally been negative under the DI program taken by itself, although the size of these negative net transfers is relatively small.
This study uses Social Security administrative data on past earnings and benefits by year, age, sex, and race to analyze historical redistribution under the Old-Age and Survivors Insurance program across and within cohorts born through the year 1922. The results generally support the findings of closely related previous research, confirming that early cohorts have received large accumulated net transfers to date, that females, as a group, have experienced substantially higher accumulated benefit/tax ratios and internal rates of return than their male counterparts in these cohorts, and that the "Other Races" group fared better by these measures than the "White" race group in most of the cohorts considered. Differences by race in the accumulated benefit/tax ratios estimated in this analysis are sensitive to the choice of the interest rate series and cohort grouping, however, and differ sharply between males and females under some of the interest rate assumptions.
In the 2001 report of the President's Commission to Strengthen Social Security, the commission states that blacks "on average have both lower incomes and shorter life expectancies than other Americans." This paper examines the extent to which the shorter life expectancies of blacks are explained by differences between their average socioeconomic status and that of other Americans.
Estimates in this paper for men aged 25 to 64 show that about half of the difference in risk of death between blacks and all other races was explained by education level—the measure of socioeconomic status employed. At ages 65 to 90, black men were not found to have a significantly higher risk of death than men of all other races.
The Social Security Administration (SSA) operates two programs that provide disability benefits: Social Security Disability Insurance (DI) and Supplemental Security Income (SSI). The Social Security Act and the regulations that implement it establish uniform national criteria for determining whether someone who applies for disability benefits under either program is disabled. However, an agency of the state in which the claimant lives makes the initial determination under contract to SSA and using SSA guidelines. Historically, states have allowed initial disability claims at rates that vary from one state to another, in some cases widely. This study estimates the amount of variation in allowance rates that is related to certain economic and demographic differences among states.
In a 2001 working paper, Links Between Early Retirement and Mortality (ORES Working Paper No. 93), the author used cross-sectional Current Population Survey (CPS) matched to longitudinal Social Security administration data and found that men who retire early die sooner than men who retire at age 65 or older. Estimates of relative mortality risk control for current age, year of birth, education, marital status in 1973, and race, and the sample is restricted to men who have lived to at least age 65.
This paper uses the 1982 New Beneficiary Survey and a 1 percent extract of the Social Security Administration's year 2000 Master Beneficiary Records to test whether the mortality differentials reported in the author's earlier work can be replicated in other independent data sets.
Social Security Benefit Reporting in the Survey of Income and Program Participation and in Social Security Administrative Records
The quality of Social Security benefit reporting in household surveys is important for policy research on the Social Security program and, more generally, for research on the economic well-being of the aged and disabled populations. This is particularly true for the aged among whom receipt of Social Security benefits is nearly universal and reliance on such benefits is considerable. This paper examines the consistency between Social Security benefit amounts for May 1990 as reported in the Survey of Income and Program Participation and given in the Social Security Administration's administrative records for the respondent.
There are large differences in wealth across racial and ethnic groups, much of which remain unexplained even after controlling for income and demographic factors. This paper studies the issue of whether differences in saving behavior and rates of return on assets are a possible source of the differences in wealth. It uses data from the Health and Retirement Study to examine the differences in various components of aggregate wealth (including nonhousing equity, housing equity, financial assets, and risky assets) and to inspect differences in portfolio choices by race and ethnicity.
Descriptive tabulations of components of aggregate wealth and portfolio choices shown here point to differences between white and minority households in their saving behavior and choice of assets. These findings suggest that some of the large differences in wealth across racial and ethnic groups that remain unexplained even after controlling for income and demographic factors, may be attributable to the smaller participation in financial markets by minority households.
This paper estimates effects on elderly poverty rates of a steady growth in incomes for 50 years. It assumes that the poverty threshold continues to be adjusted for inflation but not for increases in real incomes. Simulations with the March 1998 Current Population Survey indicate that if Social Security and Supplemental Security Income (SSI) benefit rules are not changed and if earnings and other incomes grow by 1 percent per year (the growth rate in earnings assumed in the Social Security Trustees' Report intermediate scenario) in an otherwise unchanging population, poverty among the elderly will decrease from 10.5 percent to about 7.7 percent in 2020 and to 4.8 percent in 2047. Those projected poverty rates are quite sensitive to the earnings growth rate assumption and to the assumption that benefits are not further reduced to maintain solvency. The paper quantifies the sensitivity to these assumptions and discusses several other aspects that might affect future poverty rates—changes in other income components like SSI, earnings, and pensions; changes in longevity and marital patterns; and changes in the distribution of earnings.
In this paper, the author uses the 1973 cross-sectional Current Population Survey (CPS) matched to longitudinal Social Security administrative data (through 1998) to examine the relationship between retirement age and mortality for men who have lived to at least age 65 by 1997 or earlier. Logistic regression results indicate that controlling for current age, year of birth, education, marital status in 1973, and race, men who retire early die sooner than men who retire at age 65 or older. A positive correlation between age of retirement and life expectancy may suggest that retirement age is correlated with health in the 1973 CPS; however, the 1973 CPS data do not provide the ability to test that hypothesis directly.
The widow(er)'s limit provision of Social Security establishes caps on the benefit amounts of widow(er)s whose deceased spouse filed for early retirement benefits. Currently, 33 percent of Social Security's 8.1 million widow(er) beneficiaries have lower benefits because of that provision. This paper describes the widow(er)'s limit provision and evaluates proposed changes to it. The proposals considered range from the modest (allowing widow(er)s to receive adjustments to the capped amounts by delaying receipt of benefits) to the substantial (abolishing the widow(er)'s limit).
This paper summarizes the work completed by SSA, with substantial assistance from the Brookings Institution, RAND, and the Urban Institute, for the Modeling Income in the Near Term (MINT I) model. In most cases, several methods of estimating and projecting demographic characteristics and income were researched and tested; however, this appendix describes only those methods eventually used in the MINT I model.
Counting the Disabled: Using Survey Self-Reports to Estimate Medical Eligibility for Social Security's Disability Programs
This paper develops an approach for tracking medical eligibility for the Social Security Administration's (SSA's) disability programs on the basis of self-reports from an ongoing survey. Using a structural model of the disability determination process estimated on a sample of applicants, we make out-of-sample predictions of eligibility for nonbeneficiaries in the general population. This work is based on the 1990 panel of the Survey of Income and Program Participation. We use alternative methods of estimating the number of people who would be found eligible if they applied, considering the effects of sample selection adjustments, sample restrictions, and several methods of estimating eligibility/ineligibility from a set of continuous probabilities. The estimates cover a wide range, suggesting the importance of addressing methodological issues. In terms of classification rates for applicants, our preferred measure outperforms the conventional single variable model based on the "prevented" measure.
Under our preferred estimate, 4.4 million people—2.9 percent of the nonbeneficiary population aged 18–64—would meet SSA's medical criteria for disability. Of that group, about one-third have average earnings above the substantial gainful activity limit. Those we classify as medically eligible are similar to allowed applicants in terms of standard measures of activity limitations.
In this paper we focus on an age restriction for remarriage in the Social Security system to determine if individuals respond to economic incentives for marriage. Aged widow(er) benefits are paid by the federal government to persons whose deceased spouses worked in Social Security covered employment. A widow(er) is eligible to receive benefits if she or he is at least age 60. If a widow(er) remarries before age 60, she or he forfeits the benefit and, therefore, faces a marriage penalty. Under current law, there is no penalty if the remarriage occurs at 60 years of age or later. The Social Security rules on remarriage have changed over time. Only since 1979 have widow(er)s been allow to marry at or after age 60 and not face reductions in benefit amounts.
We investigate whether the age-60 remarriage rule affects the timing of marriage and whether the elimination of the marriage penalty in 1979 encouraged widows 60 or older to marry. For this study, we primarily use Vital Statistics data from the National Center for Health Statistics.
Our major findings are as follows. In 1979, there was an increase in the marriage rate of widows 60 or older. This suggests many widows in this age group chose not to marry until the marriage penalty they faced was removed. Also, in the post-1979 period, there was a drop in marriage rates immediately prior to age 60 and an increase after this age. We do not observe this pattern in the period before 1979, and we do not observe it for divorced women, who generally are not subject to the age-60 remarriage rule. These findings suggest that the age-60 remarriage rule affects the timing of marriage and has the most influence on women who are very close to age 60.
One aspect of the current debate about changing the Social Security program concerns how new rules might affect elderly women, many of whom have low income. This paper examines three possible changes: (1) a reduction in spousal benefits combined with a change in the computation of the survivor benefit, (2) a redefined minimum benefit, and (3) a 5 percent increase in benefits for persons aged 80 or older. The paper assesses the cost, distributional consequences, and antipoverty impact of each option.
Although the Social Security program has substantially reduced poverty among older Americans, 17.3 percent of nonmarried elderly women (widowed, divorced, or never married) are living in poverty today. This paper explores several policy options designed to reduce poverty by enhancing Social Security widow(er)'s benefits, Supplemental Security Income benefits, and Social Security's special minimum benefit. Depending on the option, 40 percent to 58 percent of the additional federal spending would be directed to the poor or near poor.
Some proposals to change the Social Security program to ensure long-run solvency would reduce or eliminate benefits to some early retirees. To what extent might those benefit reductions cause hardship for individuals with precarious financial circumstances and whose health appears to limit their ability to offset reductions in Social Security income through increased earnings? Our research is intended to identify the size and characteristics of the population that might be at risk as a consequence of such changes.
The central finding is that over 20 percent of early Social Security retirees have health problems that substantially impair their ability to work. In fact, among those aged 62–64 who are severely impaired, there are as many Old-Age and Survivors Insurance beneficiaries as there are beneficiaries under SSA's two disability programs. The retirement program functions as a substantial, albeit unofficial, disability program for this age group. Moreover, the majority of the most severely impaired early retirees would not qualify for Disability Insurance benefits.
Who Is "62 Enough": Identifying Eligibles for Social Security Early Retirement in the Health and Retirement Study
Either the normal retirement age (NRA) or the earliest eligibility age (EEA) for Social Security retirement benefits would be increased under many proposals for Social Security reform. As a consequence, research interest in who retires at early ages and the potential effects of an increase in the NRA or EEA has grown. This note discusses how well researchers can do using data from the Health and Retirement Study in identifying the pool of respondents who could have received early Social Security retirement benefits.
The Health and Retirement Study (HRS) is a major longitudinal study designed for scientific and policy researchers for study of the economics, health, and demography of retirement and aging. The primary HRS sponsor is the National Institute of Aging, and the project is being conducted by the Survey Research Center of the Institute for Social Research at the University of Michigan. Several agencies, including the Social Security Administration, are supporting the project. This is the second paper describing SSA's data support for the HRS. It describes the data from SSA records that have been released for linking to HRS data, linkage rates resulting from the consent process, and subgroup patterns in linkage rates.
Employer pensions that integrate benefits with Social Security have been the focus of relatively little research. Potentially this is an important omission given the current Social Security reform debate. Since changes in Social Security benefit levels and other program characteristics can affect the benefit levels and other features of integrated pension plans, it is important to know who is covered by these plans. This paper uses data from the Health and Retirement Survey to examine the characteristics of individuals who are covered under integrated pension plans by comparing them with people covered by non-integrated plans and those with no pension plan. The results show that individuals who are female, white, non-unionized, or do not have postgraduate education are significantly more likely to be in an integrated employer pension plan.
In this paper, the author uses large Social Security administrative data sets to examine changes in earnings distributions in the United States over the 1980s through the mid-1990s. Because the earnings information contained in these data sets comes directly from the W-2 forms filed by employers, the self-reporting errors and top-coding problems common in other data used for this type of analysis are minimized. Previous research has documented an increase in overall earnings inequality during the 1970s and the 1980s. The author finds that this upward trend in overall earnings inequality continues in the mid-1990s, despite a period of nearly constant or slightly decreased earnings inequality from 1988 through 1992. The data also suggest that between-group earnings inequality, whether dividing the sample into groups by age group or by birth cohort, is increasing. Despite the increase in between-group earnings inequality over the period examined, however, within-group earnings inequality remains by far the largest contributor to overall earnings inequality.
This paper provides a brief overview of the more important studies of lifetime redistribution under the Old-Age and Survivors Insurance (OASI) and Disability Insurance (DI) programs. Studies are categorized into two types: those that focus on redistribution across successive cohorts of workers or typical members of those cohorts, and those that focus on the distribution of results across characteristics of interest within particular cohorts of workers. A list of related studies is provided at the end for those interested in additional reading.
The Accuracy of Survey-Reported Marital Status: Evidence from Survey Records Matched to Social Security Records
Many researchers have concluded that, in surveys, divorced persons often fail to report accurate marital information. In this paper, I revisit this issue using a new source of data—surveys exactly matched to Social Security data. I find that divorced persons frequently misreport their marital status, but there is evidence that the misreporting is unintentional. A discussion of possible improvements in surveys is presented. Implications for the study of differential mortality and the study of poverty among aged women are discussed.
This study examines the United Kingdom's retirement income security system from the American perspective. It addresses issues that most concern U.S. analysts: how the United Kingdom has kept its future public pension costs at a manageable level, the extent to which privatization of public pensions has contributed to these savings, the popular appeal of individual pension accounts, and the impact of privatization on retirement income. These issues are best understood in the context of the U.K. pension program's particular institutional structure and policies, two of which—"contracting out" of public pensions and strong reliance on means-tested benefits—have been largely rejected in the evolution of U.S. policy to date.
Particular use is made of recently available data on coverage rates for public and private pension programs over the total working population and administrative records on inactive personal pension accounts.
This paper asks whether information about immigrants other than their age, education, and years since migration can be productively used to project their earnings. Although many factors could affect immigrants' earnings, what is most useful for Social Security modeling purposes is relevant information that is readily available on a continuous basis. Country of origin is a good candidate as it is regularly and readily available from several administrative and survey data sources.
In this paper, microdata samples from the 1960–1990 censuses are used to examine the relationship between country of origin and the earnings of immigrants. By following cohorts of immigrants over 10-year intervals, we learn how country of origin affects the initial earnings of immigrants and how the relationship between country of origin and immigrants' earnings changes as immigrants live in the United States. The paper also presents theoretical insights and empirical evidence about the underlying causes of the link between country of origin and immigrants' earnings.
This study uses Social Security administrative data on historical taxes and benefits by year, age, gender, and race for an ex post analysis of redistribution under the Disability Insurance program. The relationship between the taxes paid and benefits received to date under the program is described for successive cohorts as a whole and for specific race and gender groups both within cohorts and across time.
In this paper I use large, Social Security administrative data sets to examine changes in earnings distributions in the U.S. over the 1980s and early-1990s. Because the earnings information contained in these data sets comes directly from the W-2 forms filed by employers, the self-reporting errors and top-coding problems common in other data used for this type of analysis are minimized. Previous research has documented an increase in overall earnings inequality during the 1970s and the 1980s. While I too find that overall earnings inequality generally increased during the early- to mid-1980s, I find that this upward trend in earnings inequality might have slowed, or reversed, during the late-1980s and early 1990s. I also find that within-group inequality for various race and gender subgroups of the population generally increased over the period examined, confirming the results of others and extending those findings into the early 1900s. Finally, I find that women's earnings increased relative to men's earnings over the entire period and that the earnings of black males declined relative to the earnings of the other groups examined.
Many employer-provided pension plans explicitly account for Social Security in their benefit formulas—a practice known as integration. Because integrated pensions are directly linked to Social Security, both the incidence and design of explicitly integrated plans are likely to be affected by changes in the current Social Security program. While integration has been mentioned as an important issue in discussions of Social Security reform, researchers have largely ignored the concept of pension integration. This paper provides basic information about pension integration and addresses, in general terms, the relationship between Social Security reform and pension integration.
This paper examines the financial prospects of the baby boomers in their elderly years. The paper primarily attempts to draw together and summarize results found by other researchers, but a few new estimates are presented. The consensus of the research appears to be the following. Up to this point, the baby boom generation as a whole has a higher economic status than their parents' generation did at the same ages, but this does not hold for some subgroups. When it becomes elderly, the baby boom generation as a whole probably will have a higher economic status than their parents' generation has and will have at those ages, but, again, this may not hold for some subgroups. It is uncertain whether the baby boom generation as a whole will have enough resources in retirement to maintain their preretirement standard of living without increasing their saving or retiring later, but some subgroups will be able to maintain their living standard without changing their behavior.
The Economic Well-Being of Social Security Beneficiaries, with an Emphasis on Divorced Beneficiaries
There are numerous types of benefits paid under the Social Security programs of the United States, with each type of benefit having its own set of eligibility rules and benefit formula. It is likely that there is an association between the type of benefit a person receives and the economic circumstances of the beneficiary. This paper explores that association using records from the Current Population Survey exactly matched to administrative records from the Social Security Administration. Divorced beneficiaries are a particular focus of this paper.
Type of benefit is found to be a strong predictor of economic well-being. Two large groups of beneficiaries, retired-worker and aged married-spouse beneficiaries, are fairly well off. Other types of beneficiaries tend to resemble the overall U.S. population or are decidedly worse off. Divorced-spouse beneficiaries have an unusually high incidence of poverty and of serious health problems. A proposal to increase benefits for these beneficiaries is evaluated. Results indicate that much of the additional government expenditures would be received by those with low income.
We estimate a multistage sequential logit model reflecting the structure of the disability determination process of the Social Security Administration (SSA), as implemented by state Disability Determination Services (DDS) agencies. The model is estimated using household survey information exactly matched to SSA records on disability adjudications from 1989 to 1993. Information on health, activity limitations, demographic traits, and work is taken from the 1990 Survey of Income and Program Participation. We also use information on occupational characteristics from the Directory of Occupational Titles, DDS workload pressure, and local area economic conditions from unpublished SSA sources. Under the program provisions, different criteria dictate the outcomes at different steps of the determination process. We find that without the multistage structural approach, the effects of many of the important health, disability, and vocational factors are not readily discernible. As a result, the split-sample predictions of overall allowance rates from the sequential model performed considerably better than the conventional approach based on a simple allowed/denied logit regression.
In this paper we focus on the relationship between a woman's economic status earlier in life and her poverty status in old age. Previous research on the determinants of poverty among aged women has documented the socioeconomic and demographic correlates of the poor and has examined the financial impact of adverse late-life events such as widowhood, deterioration of health, and loss of employment. Using data from the National Longitudinal Survey of Mature Women, we find that most women who experience these types of adverse events in their later years do not become poor and that a large majority of older NLSMW respondents who were poor in 1991–92 were poor earlier in their adult lives. Whether women are impoverished by adverse late-life events depends on their economic resources just prior to the event. But the financial resources available in old age, in turn, depend very much on their long-term economic status throughout much of their adult lives. This article underscores the fact that for most older women these adverse events do not appear to precipitate poverty spells—at least not within the first couple of years—and directs attention at longer-term circumstances that make some women more vulnerable.
The economic status of the elderly and the economic status of children are analyzed using a comprehensive definition of income that takes selected types of noncash income and taxes into account. Estimates are presented for detailed age groups over the entire age range and for socioeconomic classifications within the elderly subgroup and within the subgroup of children. The paper finds that children and the elderly are less well off than the middle age groups. This result is obtained using median incomes and the percentage of the group that has low income, as defined here. When results obtained with the measures presented in this paper are compared with results obtained with more commonly used measures, there are important differences for both the elderly and for children. For both groups, the composition of the low-income population differs in important ways from the composition of the official poverty population.
Immigrant cohorts have varied over time in many ways that have important implications for projecting the contributions immigrants make to the Social Security system. Using immigrant cohorts in the 1970, 1980, and 1990 decennial censuses, we find that immigrant men experience faster earnings growth than native-born men and that there has been a large increase over time in immigrant earnings growth rates. Thus, recent reductions in immigrant entry earnings are significantly compensated for by faster immigrant earnings growth.
This paper examines the money incomes of the elderly and the nonelderly. The economic status of the elderly is put in perspective by discussing changes in real incomes since 1967 and the income of the elderly relative to the incomes of other age groups. Detailed age groups within both the elderly and nonelderly groups are examined. The paper finds that the economic status of the elderly in 1992 was substantially better than in 1967 but was about the same as in 1984. The real median income of the elderly rose from 1967 to 1989 but fell from 1989 to 1992. The ratio of the income of the elderly to that of the nonelderly was higher in 1992 than in 1967, but the 1992 ratio was below the 1984 ratio. Large increases in mean Social Security benefits were important in the increase in the total income of the elderly since 1967.
This paper discusses some of the major issues associated with the question of whether workers receive their money's worth from the Social Security program. An effort is made to keep the discussion as nontechnical as possible, with explanations provided for many of the technical terms and concepts found in the money's worth literature. Major assumptions, key analytical methods, and money's worth measures used in the literature are also discussed. Finally, the key findings of money's worth studies are summarized, with some cautions concerning the limitations and appropriate usage of money's worth analyses.
This paper provides a nontechnical explanation of the basic ideas that underpin economists' thinking about work and retirement decisions and discusses and elaborates on the basic economic model of retirement. The paper begins with a simple economic model of an individual's work decision, to explain the construction and logic of this model, and to show how the model can be used to make basic predictions about factors that might plausibly affect the timing of retirement. From this starting point—which essentially describes the economic retirement models before the late 1970s—the paper then explains how the model has been extended during the past 2 decades. The increasing sophistication and complexity of the models reflect scientific progress in which new retirement research incorporates the findings of previous efforts, the desire to incorporate more realism into the models, and the availability of improved data. The progress in economic modeling is emphasized as the contributions of various influential studies are reviewed.
This paper explores the extent to which occupational experience is responsible for the adverse effect of low income and education on mortality. Using Current Population Survey data on education and disability matched to Social Security data on earnings, disability, and mortality, this question is pursued by examining how the estimated effects of income and education are affected once occupational experience is included in the mortality model. The inclusion of various occupational experience variables, as measured in the Dictionary of Occupational Titles and the National Occupational Hazards Survey, has virtually no effect on the estimated effects of income and education on mortality. These findings suggest that the high mortality of low-income and poorly educated persons is not due to characteristics of their employment but to other aspects associated with poverty.
Between 1970 and 1983, the rate at which the elderly were hospitalized grew by over 40 percent, but the rate of hospitalization for the younger population was fairly stable. Past attempts to explain the different patterns among the young and the old have focused on technology, insurance, health status, and the supply of hospital services. These attempts have been largely unsuccessful. In this paper, I examine other possible explanations, namely, that the elderly, who experienced a decline in the rate of participation in the labor force and an increase in income over this period, used increases in available time (i.e., nonwork time) and increases in income to seek out and receive greater amounts of health care.
Using an empirical strategy that adequately controls for the health status and insurance status of the subjects under study, I analyze small area data from the state of North Carolina. This approach yields results that indicate labor force status and income are important determinants of hospital use among the elderly.
The economic well-being of subgroups of the population usually is measured by comparing resources and needs. The measure of resources often includes noncash income. Equivalence scales are used to adjust for differential needs. Little attention, however, has been paid to the desirability of consistency between the specifications of the resources and the equivalence scales in these comparisons. This exploratory paper suggests that a lack of consistency between the definitions used on the income and the needs sides can be important for the assessment of the economic well-being of subgroups when some types of noncash income are included in the definition of income. The measured economic status of the aged in the United States when Medicare noncash income is included in the definition of income is used as an example of this consistency problem. Some previous estimates have used equivalence scales that probably understated the relative needs of the aged by omitting needs associated with Medicare. The measured economic well-being of the aged relative to that of other age groups could be overestimated substantially as a result of this consistency problem. The basic problem is not confined to the treatment of Medicare or to the United States, but is much broader in nature.
Covariance Estimates for Regression Parameters from Complex Sample Designs: Application of the Weighted Maximum Likelihood Estimator to Linear and Logistic Regression Analysis in Which Observations Might Not be Independent
Statistical methods of variance estimation are presented in this paper for the analysis of survey data involving complex sample designs. With certain complex sample design, estimation of the covariance matrices in linear and logistic regression is not straightforward. The design may be complex because of disproportionate sampling of strata, necessitating the use of weights, or because the observations are not independent, or possibly both. Examples are given from projects at the Social Security Administration, and computer programs written in SAS (Statistical Analysis System) are provided.
This paper reviews the economic literature on the work and retirement decisions of older women. Economic studies generally find that married women respond to the financial reward for work (for example, wages) in making their work and retirement decisions, but that they do not respond to unearned income and wealth (for example, the value of lifetime Social Security benefits). Unmarried women are found to respond to all type of financial variables. Most economic studies find that the family plays only a limited role in the work and retirement decisions of women. The retirement status of the husband does influence the wife's retirement decision, but the health status of the husband does not. The presence of dependents in the household, regardless of whether they are children or parents, is not found to influence work and retirement among women. The relevance of these results to Social Security policy is discussed.
There are a number of reasons to be cautious about the results. The literature to date is small; it is based on data that are deficient in some respects, and it contains studies that have methodological problems. These problems are discussed and prospects for future research are explored.
Each year the Social Security Administration forecasts the financial status of the Old-Age, Survivors, and Disability Insurance (OASDI) programs by projecting trends in key variables such as the labor force participation and earnings of the U.S. population. In the difficult task of projecting the long-term financial status of Social Security, assumptions are made concerning the relationship of immigrants to Social Security. An important aspect of that relationship is the emigration of immigrants.
This paper describes the general assumptions related to the level and timing of emigration that underlie projections of Social Security's financial status and examines how closely these assumptions fit research findings based on a variety of data sources. Previous trends in emigration and factors that may affect current and future levels of emigration are described. The paper also presents theoretical expectations and empirical evidence concerning the timing of emigration.
This paper develops estimates of lifetime net transfers across cohorts under the Social Security Old-Age and Survivors Insurance (OASI) program. Estimates are developed both from the perspective of individual cohorts, indicating the extent to which each cohort has received or can expect to receive its money's worth from the program, and from the perspective of the OASI program, indicating the extent of redistribution across cohorts. This paper also contrasts intercohort redistribution under the present OASI program with the redistribution that would have occurred under two counterfactual pay-as-you-go programs that incorporate different implicit standards of fairness. The data sources and techniques employed in this analysis provide a more accurate and extensive description of the treatment of different cohorts under the OASI program than has been available to date. Estimates based on past or projected data are presented for all cohorts participating in the OASI program since its inception through the cohort born in 2050.
This paper examines the family income and the household wealth and income of old old persons. Subgroups of the old old are compared, and the old old are compared with the young old. When the old old group is separated into three subgroups—widows living alone, other females, and males—the economic status of widows living alone is substantially below that of the other two subgroups. This difference is found when income, wealth, and combined income-wealth measures are used. When the old old group is compared with the young old group, the economic status of the old old is substantially lower for all measures examined. When the three subgroups within both the old old and young old groups are compared, the economic status of each subgroup is lower for the old old for most measures. Income data from the March 1991 Current Population Survey and wealth and income data from the 1984 Survey of Income and Program Participation are used.
Statistical Methodology for a Comparison of the Individual Characteristics and Death Rates of Disabled-Worker Beneficiaries Entitled in 1972 and 1985
This paper contains the technical details about the statistical methodology used in the article, "A Comparison of the Individual Characteristics and Death Rates of Disabled-Worker Beneficiaries Entitled in 1972 and 1985," published in the Fall 1992 issue of the Social Security Bulletin, vol. 55, no. 3. Logistic regression techniques were used to test for differences between the covariate distribution of the 1972 and the 1985 entitlement cohorts. Survival analysis techniques were used to model the death rates of the two cohorts.
Simulating the Long-Run Aggregate Economic and Intergenerational Redistributive Effects of Social Security Policy
This paper reports on the status of a long-run simulation model of the U.S. economy and its relationships with the Social Security program that was designed with these considerations in mind. The model was developed specifically to analyze the potential equity and efficiency effects of alternative Social Security policies in a long-run context.
This paper discusses what is known about the economic status of the aged. Numerous complexities involved in the assessment of the economic status of the aged are discussed. Compared with most other recent assessments, this study shows a less favorable status for the aged relative to other age groups. The focus is on an examination of detailed age groups, rather than summary aged and nonaged groups, thus providing a more complete picture of age differences. More than most other assessments, this study stresses uncertainty about the relative status of the aged and emphasizes what we do not know. The need for better adjustments for differences in needs among age and other subgroups of the population is stressed. The need for consistency between the definition of resources and the specification of needs is also emphasized. The vulnerability of the aged to economic risks is discussed.
This working paper includes two interrelated papers presented at the annual meeting of the American Statistical Association in August 1991. The papers outline the central ideas and the progress to date associated with the development of a new microsimulation model for program analysis at the Social Security Administration (SSA). The first paper, Rationale for a SIPP-Based Microsimulation Model of SSI and OASDI, relates the analytical potential of the proposed model to data development efforts intended to overcome specific information gaps. It also suggests areas in which the model can enrich SSA's ability to address issues specifically related to either the Supplemental Security Income or Old-Age, Survivors, and Disability Insurance programs or issues requiring comparative analysis of both programs. The second paper, Implementing an SSI Model Using the Survey of Income and Program Participation, describes progress on a preliminary version of the model focusing on the SSI program. It includes a brief description of the model, presentation and discussion of initial results, and comparisons with other studies.
Many observers have noted that the long-term decline in labor force participation by older Americans may reflect the evolution of social institutions that effectively discourage work. Often-cited factors include employer discrimination against older workers, private pension plans that penalize continued employment, and the Social Security system. Various policies, such as eliminating Social Security's retirement test, have been proposed with a view to eliminating or lessening employment barriers.
This paper summarizes the economic evidence that addresses the role played by the Old-Age and Survivors Insurance (OASI) programs in retirement decisions. OASI is shown to have statistically significant effects on both the timing of retirement and the amount of post-retirement work; however, the influence is not large relative to the many other factors that determine the labor-supply decisions of older workers. Consequently, changes in Social Security policy of the type and magnitude that are politically feasible are unlikely to result in large changes in retirement behavior.
Despite extensive study of the work and retirement decisions of older individuals, the nature of employers' demand for older workers remains relatively unexplored. This paper investigates the plausibility, pervasiveness, and causes of limited employment opportunities for older workers by examining age discrimination, long-term employment relationships, and partial-retirement work options. The central theme is that much of the differential treatment of older workers that persists over time is likely to be part of a privately efficient, economic equilibrium. Provisional implications for Social Security and age-work policy choices are drawn.
In recent years there has been great interest in the economic status of the aged, especially in connection with the debates about the appropriate level of Social Security benefits and Medicare coverage and financing. The economic status of the aged relative to other age groups has been of particular interest in these debates. This paper examines changes in the before-tax cash income of the aged and of other age groups from 1984 to 1989. Earlier research found that the real income of the aged rose substantially, both absolutely and relative to the income of the nonaged, from about 1970 to the mid-1980s. It is shown here that from 1984 to 1989 the real income of the aged rose slowly, and fell slightly relative to the income of the nonaged. The different rates of income growth for different aged groups are explored in this paper, with the emphasis on differences between the aged and nonaged. This paper also serves as an update of an earlier paper that contained estimates for the 1967–1984 period. The estimates in this paper generally are consistent with those presented in the earlier article.
As a result of the buildup of the Old Age, Survivors, and Disability Insurance (OASDI) trust funds, the supply of U.S. securities to the public by the second and third decades of the next century might become extremely limited. While this increase in Federal savings would lower real interest rates and stimulate investment, the buildup would create a difficulty: it would force Federal Reserve open market operations to be conducted in assets other than Treasury securities. It is important to know whether monetary policy would continue to be effective under this new modus operandi. To answer this question it is necessary to have evidence concerning the transmission mechanism through which monetary policy affects the economy. Obtaining such evidence is especially important now since many economists argue that monetary policy works through a black box which we do not understand. Evidence demonstrating one channel though which monetary policy works is presented here. It is demonstrated that news of increases (decreases) in the Federal Reserve's target for the federal funds rate during the 1974–1979 period lowered (raised) stock prices. This period was unique because the Federal Reserve controlled its operating instrument, the federal funds rate, so closely that market participants were able to discern a change in the target on the day the target changed. This evidence supports the arguments of Tobin and Brunner and Meltzer that the stock market is an important link in the monetary transmission mechanism. The results indicate that if the OASDI trust funds purchased most or all Treasury securities, open market operations conducted using other assets would still be efficacious through this channel. By affecting bank reserves and thus the federal funds rate, these operations would influence stock prices and economic activity.
A Mathematical Demonstration of the Pareto Optimality of Pay-As-You-Go Social Security Programs in a Closed Economy
A 1989 article by Breyer concludes that it is impossible to compensate pensioners in the transition from a pay-as-you-go public pension system to a privatized or funded system without making at least one later generation worse off; Breyer reaches this conclusion in the context of a simple overlapping generations model of a closed economy under the assumption that the transition results in increased saving by workers. Although this conclusion is correct under the increased saving assumption in the relevant domain of the production function, the proof that Breyer provides is not sufficient to establish that fact. This note extends Breyer's analysis to provide a sufficient proof.
Some proponents of the privatization of the Social Security program in the United States have suggested that, because privately available rates of return exceed the internal rate of return implicit in that program, it may be possible to find Pareto-superior privatization schemes. In a similar vein, Townley (1981) argues that, so long as the government can incur debt, a Pareto-superior scheme can always be found to convert a dynamically inefficient pay-as-you-go Social Security program to a fully funded basis. This note uses Townley's own model to demonstrate analytically that Pareto-superior schemes to reverse a dynamically inefficient pay-as-you-go social security program do not exist, either through privatization or through conversion of the program to a fully funded basis.
In recent years, a number of proposals have been advanced for privatizing all or part of the Social Security program in the United States. These proposals range from the immediate abolition to the gradual phasing-out of Social Security taxes and benefits. This paper evaluates several premises that often underlie privatization proposals—that rates of return in the private sector exceed those implicit in the Social Security program, that privatization would lead to an increase in national saving, and that privatization could somehow improve the lifetime welfare of all affected generations. The paper first considers whether rates of return in the private sector actually exceed those implicit in the Social Security program and discuss the conditions required for privatization to lead to an increase in national saving. The paper then demonstrates theoretically that an existing, well-managed, pay-as-you-go social security program is Pareto optimal in an economy with exogenous factor prices, regardless of the extent to which privately available rates of return exceed those implicit in the pay-as-you-go program; i.e., no privatization scheme can be found that benefits at least one present or future generation without harming at least one other generation, and no scheme can be found that allows the winners from privatization to compensate the losers and still come out ahead. The analysis is extended to incorporate the assumption of endogenous factor prices and the possibility that pay-as-you-go social security programs reduce private saving. The theoretical conclusions are illustrated by using a long-run economic projection model to simulate the aggregate economic and intergenerational redistributive effects of two stylized privatization schemes.
Simulating Aggregate and Distributional Effects of Various Plans for Modifying the Retirement Earnings Test
Social Security's retirement test continues to receive considerable attention among policymakers. During the past several years a variety of proposals have been advanced that would modify or eliminate the test for persons aged 65–69. In January 1989, we completed a study report, prepared for SSA internal use, that examined several of these proposals, analyzing their effect on earnings, taxes, and benefits in the first year of implementation, assumed to be 1990. The analysis included both aggregate estimates and estimates for selected population subgroups.
Although the specific proposals for modifying the retirement test have changed somewhat during the past 2 years, continued congressional interest has prompted the release of this initial version of our research for public discussion. Because we are in the process of revising the report for final publication, readers are cautioned that numbers and interpretations contained in this paper are subject to change.
This bibliography is a by-product of preparing a review of the economic literature on the effect of Social Security's retirement program on the labor supply of older workers. In the course of organizing a set of scribbled notes, the outline of the current document began to take shape. Several colleagues found earlier, incomplete drafts of these notes to be of some value in their own work, and encouraged me to offer them to a wider audience.
These notes are intended to provide a helpful overview of the models, data sources, and statistical procedures used by economists in recent years to investigate the work-retirement decision.
This study attempts to determine how persons aged 65–69 would respond to eliminating the earnings test by looking at the changes in labor market behavior of 70- and 71-year-olds whose earnings test coverage was eliminated beginning in 1983. In particular, it tries to determine whether 70- and 71-year-olds increased their labor force participation and earnings once the earnings test was removed. This issue is important because proposals to eliminate the earnings test for persons aged 65–69 generally assume that a portion of the additional benefit expenses would be recovered by income and payroll taxes generated by increased work effort among this age group.
This study examines the work history and economic circumstances of wives soon after receiving Social Security retirement benefits. Findings are based on a nationally representative sample of married women, aged 62 or over, who received their first benefit either as retired workers or as spouses of retired workers between mid-1980 and mid-1981.
Most analyses of economic status use only income as the measure of resources. It is clear, however, that wealth also plays an important role in economic well-being. The existence of both income and asset tests for eligibility purposes in several government transfer programs (e.g., Supplemental Security Income, Aid to Families with Dependent Children, food stamps) suggests the importance of both wealth and income. Units of the same age, income, and needs are not equally well off if they have different amounts of wealth. A fully satisfactory way of taking differences in wealth into account in a combined income-wealth measure is not available. Particularly controversial is the comparison of different age groups when such measures are used. This exploratory paper examines the use of income-wealth measures for the analysis of the distribution of economic well-being for age groups in the current period.
This note discusses the net revenue estimates in the report "Paying People Not to Work: the Economic Cost of the Social Security Retirement Earnings Limit," by Aldona Robbins and Gary Robbins.
This paper reports on an analysis of the consumption decisions of individuals. A consumption function is developed that can be viewed as an extension of the traditional life cycle-permanent income specification, with consumption determined as an age-specific proportion of current and prospective wealth. Special attention is focused on the degree of substitutability between current and prospective wealth and on the differential effects of the various types of prospective income flows on the consumption decision.
Reflections on the Income Estimates from the Initial Panel of the Survey of Income and Program Participation (SIPP)
The Survey of Income and Program Participation (SIPP) represents a major effort on the part of the Federal statistical community to improve the quality and comprehensiveness of information on the economic resources of the household sector and to permit a more accurate portrayal of the impact of government tax and transfer programs on the economic status of the population.
This paper will not offer a comprehensive and definitive statement on the quality of SIPP income data. Neither the time nor resources available to the author, nor indeed, the state of SIPP data products, would permit making such a statement. However, enough information is available to offer a tentative interpretation of important aspects of the income data available from the first SIPP panel. Two broad themes will be touched upon. Since it is generally believed that the major technical defect of income surveys is the substantial tendency to underidentify the sources and amounts of income received by the population, the issue of the completeness of the SIPP money income estimates will be the central issue. A second important aspect of income data has to do with its suitability for analytic purposes.
Pension Coverage Among Private Wage and Salary Workers: Preliminary Findings from the 1988 Survey of Employee Benefits
Pension coverage is declining in the United States, and most of the decline can be attributed to decreasing coverage among younger workers. In addition, it appears that the types of pension coverage are shifting, with a decline in traditional pension plans and an increase in 401(k) plans.
These are perhaps the most important findings from a 1988 survey of American workers, similar to pension surveys in 1972, 1979, and 1983. The 1988 survey collected data from a sample representing 114 million workers who were currently employed. This paper examines patterns of pension coverage among all private wage and salary workers, but focuses on those working full time.
Statistical Methods for the Estimation of Costs in the Medicare Waiting Period for Social Security Disabled Worker Beneficiaries
This paper presents the statistical methods used to estimate Medicare costs in the waiting period that were presented in text tables 2–3 of Bye and Riley (1989). The first part describes the development of Medicare utilization equations for each Social Security Disability Insurance (DI) program status group. The second part describes how these equations were used to predict expected costs per month and how the monthly estimates were aggregated to yield estimates of costs in the full 2-year waiting period and in the second year only. Finally, there is a brief discussion of the accuracy of the predictions.
This paper discusses and illustrates the use of wealth data for the analysis of the economic status of households. Selected estimates of wealth for 1984 from the Survey of Income and Program Participation (SIPP) are used as illustrations. The particular focus is on the wealth of age groups, with a special interest in the aged. Comparisons of the amounts and composition of wealth of the aged and nonaged (and of more detailed age groups) are presented. The emphasis is on the economic resources available to households other than the very wealthy. The degree of concentration of wealth, the subject that wealth data traditionally have been used to examine, is not discussed. Thus, this paper reflects a somewhat different perspective on the use of wealth data.
In recent years there has been a substantial amount of discussion about the economic status of the aged. There is a widely accepted view that the status of the aged has improved relative to the nonaged. This view has affected the debate on modifications to the Social Security system and other retirement plans. This paper discusses changes in the economic status of the aged during the past several years, in terms of the real income of the aged and in terms of the income of the aged relative to the income of the nonaged. The analysis uses detailed age groups within both the aged and nonaged groups. This detail is important because summary age groups are not homogeneous. Income change at different income levels within each age group is also examined. Income is adjusted for size of family unit and, in some cases, age of head.
Evidence on the Effects of Payroll Tax Changes on Wage Growth and Price Inflation: A Review and Reconciliation
The Social Security payroll tax rate is scheduled to increase by almost 1 percent for both employees and employers between now and 1990. One of the major elements of the recently adopted Social Security package was an acceleration of the timing of this increase. A number of economists have recommended that as an anti-inflationary policy scheduled increases be avoided or even that the current rates be rolled back.
This paper examines the economic well-being of age groups in the U.S. using data on both income and wealth. Although income will be discussed, we will focus on wealth in order to exploit relatively current data on wealth that have become available recently.
This exploratory paper examines the role of age in the distribution of family income in several countries. Unlike most papers that compare the distribution of income across countries, the primary concern in this paper is not with comparisons of the overall degree of inequality. Instead we are more interested in two aspects of the cross-section relationship between age and income. First, we are interested in the relative economic well-being of income recipient units in different age (of head) groups in several developed countries. In the U.S. in recent years, in connection with modifications to the social security system, there has been considerable discussion of the "fair" level of income of the aged population. That discussion has led us to a particular interest in the relative economic well-being of the aged population in other developed countries. Where the data allow, the aged (age 65 and over) group is split into 65–69 and 70 and over age groups as at least partial recognition that economic well-being can differ markedly among subgroups of the aged population. (Other important characteristics such as labor force participation, sex, and the receipt of government retirement income could not be examined.) This paper attempts an initial look at the very complex subject of the relative economic well-being of different age groups in several countries.
In a recent article in the Journal of Political Economy (Leimer and Lesnoy 1982), we presented new time series evidence that cast considerable doubt on earlier evidence presented by Martin Feldstein (1974) which implied that social security had a large and statistically significant negative effect on personal saving in the United States. Our results may be summarized as follows: First, the social security wealth variable used by Feldstein was seriously flawed as a result of a computer-programming error. Simply correcting this error substantially changes the estimated effect of social security on saving. Second, the statistical evidence depends upon assumptions which are embedded in the construction of the social security wealth variable. These assumptions relate, first, to how individuals form their expectations about the social security benefits they expect to receive and the social security taxes they expect to pay and, second, to estimates of the number of workers, dependent wives, and surviving widows who will receive benefits. Adopting reasonable assumptions that differ from those used by Feldstein leads to generally weaker estimates of the relationship between social security and saving. Finally, the estimated relationship between social security and saving is acutely sensitive to the period of estimation examined. We concluded that the time series evidence simply does not support the hypothesis that social security has substantially reduced personal saving in the United States.
To be awarded Disability Insurance benefits, an individual must have an objectively determinable, severe medical condition or impairment that, according to Social Security regulations, is serious enough that it can be presumed to keep the individual from working. We know, however, that some people who have medical conditions serious enough to qualify them for disability benefits are nevertheless able to continue working, while others who consider themselves unable to work do not have a serious enough impairment to qualify them for benefits. Whether or not a seriously impaired individual files for Social Security Disability Insurance benefits (SSDI) will depend, in part, on his or her own self-assessment of his ability to work, i.e., whether he considers himself to be severely disabled. This self-assessment depends upon many factors in addition to the actual severity of the individual's medical condition. These factors, therefore, become important elements in the decision to apply for SSDI benefits. This report examines how the relationship between measures of actual individual functional capacity and individual self-assessments of work capacity vary by age and other important job-related attributes.
Testing the Predictive Power of a Proportional Hazards Semi-Markov Model of Postentitlement Histories of Disabled Male Beneficiaries
In the Disability Amendments of 1980 (P.L. 96-265), Congress mandated that certain experiments be carried out which are designed to encourage disabled beneficiaries to return to work and save trust fund monies. A research plan has been developed which would offer alternative program provisions, experimentally, to different samples of beneficiaries. An observation period of three to four years will be possible before a report to Congress must be written. However, a period of this length is not sufficient to observe, fully, the postentitlement experience of disabled beneficiaries. In order to estimate the long run effects of the experiments, a method is needed which can project postentitlement behavior beyond the observation period.
This paper tests the ability of proportional hazards semi-Markov model to make accurate predictions in this type of setting. The data are divided into two segments: the first 14 calendar quarters and the last 16 quarters. Various types of rate functions including proportional hazards rate functions are estimated on the first segment, then projected over the entire 30 quarters and compared to the actual data. The proportional hazards rate functions are then used in a simulation to estimate monthly benefit cost to the social security disability trust fund over the last 16 quarters, using an age-dependent, absorbing, semi-Markov model. The model does a very good job of capturing the dynamics of the process and should prove quite useful as one of the major components in an analysis of the Work Incentive Experiments.
A Note on Maximum Likelihood Estimation of Discrete Choice Models from the 1978 Survey of Disability and Work
This paper demonstrates an alternative maximum likelihood procedure for estimating discrete choice models in retrospective samples, such as a model of SSA disability beneficiaries or application status in the 1978 Survey of Disability and Work.
This report examines the impact of local labor market characteristics on three steps in the disability process: The perception of oneself as disabled; the decision to apply for benefits under the social security disability insurance program (SSDI); and the determination of disability status under SSDI. The research attempts to determine whether the elements of an individual's local economic environment play a role in the various steps of the disability process specifically above and beyond his or her own demographic characteristics and economic motivations. Among the key variables used to measure the local economic environment are the unemployment rate, the percent of families below the low income (poverty) level, rural location, occupational diversity and the percent of the unemployed exhausting their unemployment benefits. With the exception of the last variable, which is measured on a statewide basis, all variables pertain to the county of residence.
The results contradict earlier findings which were based on aggregated data. No significant effect on any of the three elements in the disability process was found for either variable measuring the dimensions of the unemployment problem. With few exceptions, results from the other labor market variables were sketchy at best. One surprising result is noted with respect to the benefit replacement ratio, the variable intended to measure the relative attractiveness of SSDI benefits.
Estimation of Disability Status as a Single Latent Variable in a Model with Multiple Indicators and Multiple Causes
In this paper, we are concerned with the underlying structure of self-definitions of disability. Our purpose is to identify the contribution of exertional and nonexertional impairment and the contributions of such nonmedical factors as age, sex, and education to the individuals' assessment of their own situations. On a statistical level, we seek to accomplish a substantial reduction of a large number of data items into a form that can be used conveniently in subsequent behavioral analyses.
This paper presents analysis of the distributional and other effects of a change from the existing income tax exclusion of Social Security benefits to the proposed 50 percent inclusion. In emphasizing the differences between these two policies, very limited attention will be given to other policy alternatives.
It is well-known that for most purposes income size distribution data collected in household surveys are far from ideal. The problems with those data can be separated into two types: the data items that are collected, and the accuracy of the data collected. Usually, although there are important exceptions, the income data collected are confined to cash income before taxes, thus ignoring the effects of both taxes and noncash income of all types. Also, the income estimates usually are for one year, which often is not the best accounting period for analysis. Furthermore, there usually is a lack of adequate detail by income type, and the data ordinarily are not sufficiently detailed to adjust for changes in the composition of the family unit during the income accounting period.
The data for this study are drawn mainly from the Consumer Expenditure Survey conducted by the Bureau of Labor Statistics during 1972–73. The respondents are divided into five income classes and two age groups. The focus of this analysis is placed on the consumption-type value-added tax.
The purpose of this paper is to consider several alternative specifications of the consumer expenditure function.
The purpose of this paper is to investigate the effect of the social security system on the labor supply of aged men using U.S. time series data for the period 1947 to 1975. The specific phenomena to be explained is the dramatic decrease in the labor supply of aged men during this period. Between 1947 and 1975, the annual labor force participation rate of men 65 and over decreased from 47.8 percent to 21.7 percent—a decrease of 55 percent. In terms of annual hours worked per capita for men 65 and over, there was a decrease from about 880 hours to 312 hours during this period—a decrease of 65 percent. The specific focus of the analysis will be on the relative importance of social security in explaining this decrease in labor supply.
Social Security has sizable obligations to workers who contributed and made savings decisions in the anticipation of future benefits, and the assessment of future options must explicitly account for impacts on these as well as future participants. To this end, our paper develops cohort-specific, general-equilibrium comparisons of concrete policy alternatives.
Social Security and Private Saving: A Reexamination of the Time Series Evidence Using Alternative Social Security Wealth Variables
In an important article in the Journal of Political Economy , Martin Feldstein estimated that the introduction of the social security system had reduced personal saving by 50 percent, with serious consequences for capital formation and output. His conclusion was based on a consumer expenditure function estimated with U.S. time series data and incorporating a social security wealth variable of his construction.
The original intent of this paper was to examine the sensitivity of Feldstein's conclusions to certain assumptions underlying his construction of the social security variable. In particular, we wanted to examine the implication of his assumptions concerning how individuals perceive future benefits and taxes.
An Example of the Use of Statistical Matching in the Estimation and Analysis of the Size Distribution of Income
This paper discusses the use of statistical matching in the estimation and analysis of the size distribution of family unit personal income. Statistical matching is a relatively new technique that has been used to combine, at the single observation level, data from two different samples, each of which contains some data items that are absent from the other file. In a statistical match, the information brought together from the different files ordinarily is not for the same person but for similar persons; the match is made on the basis of similar characteristics. In contrast, in an "exact" match, information for the same person from two or more files is brought together using personal identifying information.
Empirical evidence suggests that Social Security causes many individuals to retire earlier than otherwise. An important policy question is whether the program should be designed to lessen or eliminate this induced retirement effect. This paper proposes a framework for analyzing the socially desirable relationship between Social Security and retirement. Two common rationales for the program, forced saving and retirement insurance, are examined. If importance is attached to either of these rationales, then it is shown that retirement neutrality should probably not be a feature of Social Security.
Pension Coverage and Vesting Among Private Wage and Salary Workers, 1979: Preliminary Estimates from the 1979 Survey of Pension Plan Coverage
This paper examines pension coverage and vesting in 1979 among private wage and salary workers aged 14 and older in the employed labor force. Coverage and vested status are examined in relation to personal and current job characteristics in order to provide a profile of workers protected and not protected under the private retirement system. The data are derived from the 1979 Survey of Pension Plan Coverage, a supplement to the May 1979 Current Population Survey.
Three major findings emerge from the analysis. First, coverage rates among full-time workers increased slightly between 1972 and 1979, and vested rates increased substantially during the same period. Second, although coverage rates were moderate to high for certain groups of workers, many workers were not in these groups. Third, women were much less likely than men to be covered by a retirement plan and to have acquired vested rights to their benefits.
In 1971, 44 percent of workers who had been currently entitled to social security disability insurance benefits (SSDI) for 1 year or more received benefits from at least one income source in addition to SSDI. These recipients of multiple benefits (RMB's) were found to have average benefits from SSDI which were greater than the average SSDI benefit for those who did not receive income from these additional sources. On the average, total benefits to RMB's were double the benefits paid to those who received only SSDI. The combined benefits for overlappers produced median replacement rates for nonoverlappers. The rate of receipt of replacement rates in excess of 80 percent of predisability earnings was 70 percent larger for persons who were RMB's than for those who were not.
Based on the present research, consideration of replacement rates based solely on SSDI benefits substantially understates the extent to which benefits from public and private programs actually replace predisability earnings. Since replacement rates based solely on SSDI benefits are generally higher for persons receiving only SSDI than for persons who receive multiple benefits, employing policies which cap replacement rates based only on SSDI benefits may only serve to increase the differential in the total replacement of predisability earning which exists between those who receive multiple benefits and those who do not. Increasing this differential could be considered undesirable from both the adequacy and equity viewpoints.
This paper examines the effect of inflation on private pension saving. The role that private pensions can or should play in providing income in old age in the current inflationary environment is an important policy issue. A number of studies have discussed the effect of inflation on pensions. This study extends the existing analysis and presents the first empirical estimates. Inflation is seen to have a large negative effect on this aspect of retirement saving by workers.
In the recent economic literature on social security, much attention has been focused on its welfare implications (e.g., Samuelson ), and its impacts on individual retirement decisions (e.g., Boskin ), Sheshinski , Diamond and Mirrlees ) and capital accumulation (e.g., Feldstein , Munnell , and Kotlikoff ). In all these works, the level of social security is assumed to be exogenous although it is often determined in the real world by the desire of the majority of voters and thus is an endogenous variable of the economic system. While Browning  and Hu  did consider the determination of social security by a majority-voting process, they used the partial-equilibrium approaches in the sense that wages and the interest rate were assumed exogenous and independent of social security. The present paper constructs a simple three-period life-cycle model in which social security is determined by the majority-voting process, and the rate of interest by the demand for and supply of capital. In this framework, the tax rate voted by each person depends on the market rate of interest, which in turn is affected by the prevailing tax rate. It is assumed that social security is financed by a pay-as-you-go plan.
One-period models predict that a substantial welfare gain would result from removing the Social Security earnings test. In this paper we show that such models overestimate the size of potential gains.
If one uses instead a two-period model, which captures intertemporal effects, the net result of removing the earnings test is ambiguous. In the presence of a personal income tax, workers who reduce their labor supply in the first period create a welfare loss that must also be considered. We use a present-value model to estimate the change in lifetime welfare. We find that the net potential gain from removing the earnings test is probably small, especially when compared with the alternative of an increased personal income tax.
In late 1977, the U.S. Congress passed Social Security legislation that included a series of increases in the payroll tax. These increases, which began in 1979 and carry on into the 1980s, substantially raise the projected levels of the Social Security trust funds. Since the amendments were passed, there has been some discussion and several proposals to roll back part of the tax. It is highly likely that additional rollback proposals will be made in the near future. The purpose of this paper is to shed some light on some of the macroeconomic effects of a payroll tax rollback.
The role of time as an input into the utility maximization process has long been recognized in the labor/leisure decision. Expanded research has dealt with this input in a family context. Assuming a joint utility maximization model, the resulting labor supply functions can be determined for both spouses.
The model presented here is an extension of previous models by its incorporation of the effects of disabling conditions of the husband on the labor supply decisions of both spouses.
Because hours worked takes on a lower limit of zero, the standard simultaneous equation techniques would yield estimates lacking the ideal properties. Instead, the model is estimated using a simplification of a simultaneous TOBIT technique, which yields consistent estimates.
For the past few years, the Division of Disability Studies has been using simple random and stratified random sampling procedures for many of its studies. The beneficiary sample for the 1978 Survey of Disability and Work was a stratified random sample drawn from the Master Benefit Record. The samples used in the Study of Consistency and Validity of Initial Disability Decisions and the Trial Work Period Folder Study also used simple random sampling procedures. Simple random subsampling has been used to enable multivariate analysis to be performed on files that would otherwise have been too large for existing software.
Because of the Division of Disability Studies' wide use of simple and stratified random sampling designs, software was developed to efficiently accomplish these sampling schemes. This paper describes the algorithm and presents the computer programs that are currently being used in the division.
The quest for suitable indices to summarize the inequality between two groups has lagged behind the effort to obtain summary coefficients of within-group inequality. Numerous measures of within-group inequality were proposed, and their merits and shortcomings debated. Yet, apparently, at the same time, there was little exploration of alternative indices to the ratio-of-medians and ratio-of-means for measuring differences between groups.
This paper reports on estimates of federal income tax and Social Security tax liabilities of family units in 1972 and summarizes the methods used to make the estimates. Distributions of income both before and after subtracting those liabilities are shown. Several microdata files were combined using both "exact" and "statistical" matching of individual observations in the process of making these estimates.
There is empirical evidence that in the recent past the Old-Age Insurance portion of the Social Security program has acted as a net wage subsidy. In addition, the program had significant intragenerational redistributive effects. Our purpose is to demonstrate how these findings alter conventional views of the labor supply effects of Social Security. Our method is the analysis of a labor supply model that is extended to include empirically significant operational components of the program. We show that the analyses of others are special cases of our more general approach.
Markov models have been widely used for the analysis and prediction of shifts in population distribution over time. The point of departure for most of these analyses has been the finite state, time stationary Markov chain. The usual Markov chain model has, however, been shown to be inadequate for most social science applications.
This paper presents a particular kind of discrete time nonstationary Markov chain. Such chains will be built using a mathematical quantity called a causative matrix.
A long-term disability reflects the interaction between a continued physical or mental impairment that limits functioning and restrictions and requirements of the social environment. Impairments and functional limitations are, however, central to any disability, and the Social Security Administration is constructing measures to assess the impact these factors have on the development of disability.
In particular, SSA is interested in work disability or loss of or reduction in the ability to work. The functional capacity index presented in this paper was developed for that purpose. Based on a model prevalent in the literature, the index is an attempt to represent the underlying medically related aspects of disability in contrast to other factors such as the person's age, educational level, or work history.
This paper presents the social and demographic characteristics of those disability claimants whose cases go to hearing. Particular attention is given to how these characteristics may be related to (1) the individual decision to contest a denial or accept it; (2) the general increase in disability claims and contested applications in recent years; and (3) the high proportion of reversals in hearings.
In recent years, the number of workers awarded disability insurance benefits has rapidly increased, while there has been no corresponding increase in the numbers leaving the rolls for recovery. Concern has been expressed that cash benefit payments may be leading to disincentives to beneficiaries to return to work after medical improvement
To examine this question, a comparative analysis was made of the demographic, disability, and benefit characteristics of a sample of disabled workers who left the benefit rolls for recovery in contrast to the characteristics of those who remained on the rolls after award of disability benefits in 1972. Characteristics related to greater recovery included younger age, higher education, disability due to traumatic injury, residence in western states.
This paper attempts to make two contributions to this research. The first one is expositional. A simple overlapping generation's model is developed and used to reinvestigate the wealth and endowment redistribution effects from the introduction of pay-as-you-go social security. Our second contribution is substantive and extends the analysis of the endowment redistribution effect. Finally, perspective is offered on the relationship between pay-as-you-go social security and private saving.