Paper #

Abstracts

1998

WPS 98-24

Elo, Irma T. and Samuel H. Preston

The Demography of the African-American Population: 1930-1990.

This paper describes results of a project to reconstruct African American demography during the period from 1930 to 1990. The purpose of the project is to construct a consistent set of age and sex-specific death and birth rates and estimates of age and sex distribution of the population at census dates. These series are designed to correct data errors present in vital statistics and censuses. The paper draws together results of analysis on age misreporting in death statistics, estimation of African American populations for census years from 1930 to 1990, estimates of life tables for five year periods from 1935-40 through 1985-90, and estimates of age-specific birth rates and total fertility rates from 1935 through 1990.

WPS 98-05

Dunn, Thomas A. and John W. Phillips

The Division of Parental Financial Assets Among Adult Children.

Economic estimates of intergeneration transfers in 1986 total nearly $140 billion. Given the magnitude and prevalence of these wealth flows, researchers are trying to better understand the underlying factors that induce transfers from parents to their children. In this study, we examine financial transfers from parents to their adult children using the Asset and Health Dynamics Among the Oldest Old Survey. Previous research on financial transfers have focused separately on cash transfers or bequests. This research is the first to combine these two types of transfers as well as deeds to property, life insurance policies, and trust funds to determine if a child’s income influences the distribution of parental assets. We examine the pattern of transfers both within and across families using two different estimation techniques. The findings both support and extend the previous literature on transfers. Parents appear to target inter vivos transfers of cash to the relatively poorer children in a family, while transfers made at death tend to be made to all children without regard for income differences among children. Together, the results imply that parental asset transfers reduce income inequality among siblings.

WPS 98-04

Phillips, John W.

Intrafamily Resource Allocation: Joint Determination of Multiple Transfer Outcomes.

Economists have found that intergenerational transfers of wealth influence the distribution of income and in the process of household wealth accumulation.  Interest regarding the effect that government tax and transfer policies have on these wealth flows has inspired researchers to examine how parents distribute economic resources among their children.

Research in this area has primarily focused on one of three types of transfers:  cash transfers, bequests, or intergenerational coresidence.  Recent literature on cash transfers has found that parents tend to give to their relatively poorer children more frequently and in larger amounts.  In contrast, the literature on bequests has found that parents tend to divide their estates equally regardless of income differences between their children.  The coresidence literature suggests that coresidence arrangements in the home of the parent are usually the result of the financial needs or circumstances of a child.  While these analyses have provided information on the distribution of a single transfer type, they do not analyze the distribution of combinations of transfers.  If relationships exist among transfer types, then receiving one type of transfer could have an influence on the likelihood of receiving other types of transfers.  These relationships could determine the degree to which transfers to poorer children influence the wealth distribution.

In this study, I jointly model five transfers (cash, coresidence, deeds, life insurance, and wills) to identify correlations among transfer types.  This is accomplished by modeling ten bivariate probits of all pairwise transfer combinations.  I find that significant linkages do exist between transfers and these relationships are complementary with the exception of cash and coresidence transfers, which are substitutes.  The estimated parameters from the first stage are used to produce minimum-distance estimates of explanatory variables for each transfer equation.  These estimates are used to simulate the probabilities of children receiving various “bundles” of transfers.

I find that, holding all other characteristics of the sample constant, poorer children are more likely to receive “bundles” of all transfers (inter vivos and at-death) than relatively richer children in a family.  I also find that poorer children are less likely not to receive any transfer from parents.  The inter vivos component appears to drive the overall results.  The targeting of inter vivos transfers to relatively poorer children in conjunction with positive linkages between inter vivos and at-death transfers makes the overall distribution of parental resources favor poorer children.  A comparison of the results from the joint estimations with results from independent equation estimates shows that the independent estimates understate the probabilities of poorer children receiving all combinations of transfers.

WPS 98-03

Elo, Irma T.

Childhood Conditions and Adult Health: Evidence from the Health and Retirement Study.

The paper examines the effects of childhood socioeconomic and health conditions on adult health using data from the Health and Retirement Study (HRS).  In Wave 3 of the HRS, an experimental module was included which collected information on socioeconomic characteristics of the respondent’s family of origin, living arrangements, and health status in childhood. These data, together with information from the full survey, are employed to examine the association between childhood conditions and health status in adulthood within a multivariate framework.  Respondents who grew up in less well-off households report lower self-assessed health status in adulthood. The effects of childhood social class are substantially attenuated when respondent’s own education is controlled.  Poor health in childhood has a significant positive association with poor health in adulthood. This association persists when adult characteristics are controlled. The results are consistent with the hypothesis that poor health in childhood leads to adverse health outcomes in later life.

WPS 98-02

Dwyer, Debra Sabatini and Olivia S. Mitchell

Health Problems as Determinants of Retirement: Are Self-Rated Measures Endogenous?

We explore alternative measures of unobserved health status in order to identify effects of mental and physical capacity for work on older men’s retirement. Traditional self-ratings of poor health are tested against more objectively measured instruments. Using the Health and Retirement Study (HRS), we find that health problems influence retirement plans more strongly than do economic variables. Specifically, men in poor overall health expected to retire one to two years earlier, an effect that persists after correcting for potential endogeneity of self-rated health problems. The effects of detailed health problems are also examined in depth.

Keywords: Measurement of health, retirement, ADL/IADL, Self-Reported and Objective health

WPS 98-01

Preston, Samuel H., Irma T. Elo and Quincy Stewart

Effects of Age Misreporting on Mortality Estimates at Older Ages.

This study examines how age misreporting typically affects estimates of mortality at older ages. We investigate the effects of three patterns of age misreporting age overstatement, age understatement, and symmetric age misreporting on mortality estim ates at ages 40 and above. We consider five methods to estimate mortality: conventional estimates derived from vital statistics and censuses; longitudinal studies where age is identified at baseline; variable-r procedures based on age distributions of th e population; variable-r procedures based on age distributions of deaths; and extinct generation methods. For each of the age misreporting patterns and each of the methods of mortality estimation, we find that age misstatement biases mortality estimates downwards at the oldest ages.

1997

WPS 97-12

Behrman, Jere R. and Mark R. Rosenzweig

Ability" Biases in Schooling Returns and Twins: A Test and New Estimates.

(Published in Economics of Education Review forthcoming.)

Identical twins long have been used to control for “ability” in efforts to obtain unbiased estimates of the earnings impact of schooling and of biases in estimates that do not control for earnings endowments. This study (1) presents new estimates of schooling returns and of “ability” bias using a new twins sample, (2) develops and applies a test of the significance of that bias, and (3) demonstrates that there may be “ability” bias even if the genetically-endowed component of ability does not affect schooling decisions directly as long as this component of ability is correlated with other family characteristics such as income that do affect schooling and that it is not possible to identify separately these individual components of “ability” bias. The basic empirical result is that, net of measurement error, upward “ability” bias is statistically significant in OLS estimates, causing an overestimate of the schooling impact of 12%.

WPS 97-03

Preston, Samuel H., Mark E. Hill and Greg L. Drevenstedt

Childhood Conditions that Predict Survival to Advanced Ages among African Americans.

(Publication Informatin: Social Science & Medicine, 47(9), November 1998:1231-1246.)

This paper investigates the social and economic circumstances of childhood that predict the probability of survival to age 85. It uses a unique study design in which survivors are linked to their records in U.S. Censuses of 1900 and 1910. A control group of age and race-matched children is drawn from Public Use Samples for these censuses. It concludes that the factors most predictive of survival are farm background, having literate parents, and living in a two-parent household. Results support the interpretation that death risks are positively correlated over the life cycle.

KEYWORDS:
COHORT MORTALITY; LONGEVITY; AFRICAN AMERICANS; SOCIOECONOMIC FACTORS; GEOGRAPHIC FACTORS; OLDEST OLD

WPS 97-02

Gragnolati, Michele, Irma Elo and Noreen Goldman

New Insights Into the Far Eastern Pattern of Mortality.

The Far Eastern pattern of mortality, first identified in 1980, is characterized by some of the largest sex differentials at adult ages to be found anywhere in the world. These atypically high levels of excess male mortality were present in several Far Eastern populations during the 1960s and 1970s and have progressively disappeared since that time. This study uses cause of death data to determine the diseases responsible for the existence and attentuation of these sex differences in Hong Kong, Singapore and Taiwan. The analysis focuses primarily on two hypotheses – regarding the roles of respiratory tuberculosis and liver diseases associated with hepatitis B infection – which were proposed to explain the Far Eastern pattern but were never tested. The results of our analysis indicate that respiratory tuberculosis is the single most important cause underlying the existence and attenuation of the Far Eastern pattern, that the role of liver diseases is far from clear cut, and that other causes (such as cardiovascular diseases) are important as well. Some of the risk factors which may underlie these exceptional mortality patterns are identified.

WPS 97-01

Merli, M. Giovanna

Mortality in Vietnam, 1979-1989.

This paper employs the only official data on mortality available from the first two modern censuses of Vietnam to estimate mortality in Vietnam during the 1979-1989 period. During this period, massive outflows of refugees, population redistribution policies, and a highly mobile population as a result of the loosening of control under the economic reforms seriously undermine the effort to accurately measure mortality from two census age distributions. In this paper, I take steps to minimize bias from these sources. First, I correct census age distribution for the effects of intercensal emigration. To deal with Vietnam's population departures from stability, I apply methods that relax the assumption of stability by use of the age-specific growth rates from two census age distributions. I then propose a way to reduce the impact of errors in the growth rates caused by differential census enumeration completeness and "residual" emigration. The reconciliation of results from different methods of mortality estimation allows one to identify the most consistent, accurate measurement of mortality levels prevailing in Vietnam in the 1979-1989 period. 

1996

WPS 96-06

Preston, Samuel H., Irma T. Elo and Lynn Gale

Using Successive Censuses to Reconstruct the African-American Population, 1930-1990.

This paper develops a new procedure for estimating census undercounts. It takes advantage of the fact that a cohort appears multiple times in population censuses and that differences in its true size at successive censuses are known from death records. An age/period/cohort model of census counts is developed. It is applied to data on African Americans from 1930 to 1990. Results are quite satisfactory for males, judging from extraneous information, but less so for females. 

WPS 96-05

Elo, Irma T.

Adult Mortality Among Asian Americans and Pacific Islanders: A Review of the Evidence.

Mortality estimates have consistently pointed to a sizable health advantage for Asian Americans and Pacific Islanders compared to white Americans, but a question remains as to whether mortality estimates for Asian/Pacific Islanders are reliable. This paper presents mortality estimates for Chinese, Japanese, Filipinos, Other Asian and Pacific Islanders, all Asian and Pacific Islanders combined, and for white Americans in 1989-91 based on vital statistics and census data, and for Asian and Pacific Islanders and whites based on the National Longitudinal Mortality Survey. The paper reviews evidence on data quality and discusses possible biases in estimated death rates. It ends with a brief discussion of cause-specific mortality differentials. Relative to whites, Asian and Pacific Islanders are found to have lower mortality at ages 25 and above. Lower death rates from heart disease and cancer among Asian/Pacific Islanders than white Americans account for most of the all cause differentials at ages 45+. Substantial uncertainty remains, however, about the exact level of mortality among Asian Americans and Pacific Islanders residing in the United States.

WPS 96-04

Hussey, Jon M. and Irma T. Elo

Consistency of Age Reporting By Cause of Death Among Elderly African-American Decedents.

This paper examines the quality of age reporting, by major causes of death, on death certificates of elderly African-Americans. We utilize a sample of death certificates linked to early census records and to social security administration records. For eight of the eleven causes examined, the rate of agreement between death certificate age and corrected age is lower for females than for males. Educational background is identified as an important predictor of agreement between death certificate age and corrected age. Finally, we show that cause-specific death rates for African-Americans estimated from uncorrected vital statistics data are generally too low at ages 85 and above. 

WPS 96-03

Mitchell, Olivia S., Jan Olson and Thomas Stenmeier

Construction of the Earnings and Benefits File (EBF) for Use With the Health and Retirement Survey.

Analysts using the Health and Retirement Survey (HRS) often require information on earnings, labor market attachment, and social security benefits, in order to better understand the factors affecting retirement and well-being at older ages. To this end, several derived variables were constructed and documented in the Earnings and Benefits File (EBF) described here. The EBF provides a set of summary earnings, employment, and social security wealth measures for a subset of HRS respondents in Wave 1 of the survey, for whom administrative records are available. The EBF, a restricted data file, is available from the University of Michigan's Institute for Social Research for matching with only with versions of the HRS containing geographic detail no finer than the Census Division level. Interested users should contact hrsquest@umich.edu by email for further information on access to the data. Olivia Mitchell, The Wharton School, University of Pennsylvania and NBER. Jan Olson, Office of Research, Evaluation, and Statistics, Social Security Administration. Thomas Steinmeier, Department of Economics, Texas Tech University.

WPS 96-02

Koropeckyj-Cox, Tanya
Loneliness and

Depression in Middle and Old Age: Are the Childless More Vulnerable?

This study examines the relative circumstances of community-dwelling childless and parents in middle and old age (50-84 years old), using data from the 1988 National Survey of Families and Households, in order to update and test earlier findings of negative consequences related to childlessness in later life. Results indicate that net of other effects, both loneliness and depression are significantly related to childlessness for women but not men. Childless women are 46% more likely to report high depression compared to mothers. Among both men and women, being formerly married is related to greater loneliness and depression. These results demonstrate the greater salience of childlessness for women compared to men. The findings are discussed in the context of the changing norms regarding marriage, divorce, childlessness, and gender roles experienced by the newly emerging cohorts of the middle-aged and elderly.

WPS 96-01

Advisory Council on Social Security Technical Panel on Assumptions and Methods

1994-95
Advisory Council on Social Security Technical Panel on Assumptions and Methods, Final Report.

The Charter provided to the Panel was to: [a]ssist the 1994-95 Advisory Council by reviewing the assumptions and methodology used to project the future financial status if the old age, survivors, and disability insurance (OASDI) programs including, if necessary, measures of the financial soundness of these programs. This report presents the Panel's findings.

1995

WPS 95-06

Mitchell, Olivia S and Joseph F. Quinn.

1994-95 Advisory Council on Social Security Technical Panel on Trends and Issues in Retirement Saving Final Report.

The charge of the Technical Panel on Trends and Issues in Retirement Savings (TIRS) was to "assist the 1994-95 [Social Security] Advisory Council with respect to its charge to analyze the relative roles of the public and private sectors in the provision of retirement income, particularly how underlying policies of public and private programs, including relevant tax laws, affect retirement decisions and the economic status of the elderly." This report presents the Panel's findings.

WPS 95-05

Hill, Mark E., Samuel H. Preston, Irma T. Elo and Ira Rosenwaike

Age-Linked Institutions and Age Reporting among Older African Americans.

With economic and technological development, numerical age became an important dimension of social differentiation in the United States. The vast majority of Americans now have the ability to report their own age and the ages of relatives with accuracy. Nevertheless, studies have found that age misreporting remains substantial for older African Americans. This paper describes levels of age misreporting and investigates the determinants of age reporting accuracy on the death certificates of a national sample of native-born African Americans aged 65+. Consistent with previous studies, levels of age misreporting are found to be high. When checked against childhood census records, only 53% of the death certificate ages are found to be correctly reported; slightly over 10% are misstated by five years or more. Multivariate results provide compelling evidence that the quality of age reporting critically depends on interaction with age-linked institutions.

WPS 95-04

Rosenwaike, Ira and Mark E. Hill.

The Accuracy of Age Reporting among Elderly African Americans: Evidence of a Birth Registration Effect.

This paper expands on previous research that has documented relatively high levels of inconsistency in age information for elderly African Americans. Drawing on a sample of death certificates for Maryland-born African Americans purportedly aged 65-79 at death in 1985, the validity of age data in both death certificates and social security records is examined by linkage to a birth record. The commonly assumed relationship between availability of birth registration and quality of age reporting also is investigated. Among matches to a birth record, age on social security records is significantly more accurate than on death records. Age agreement between matched death and social security records closely reflects age validity as determined from birth records. Findings based on logistic regression analysis support the hypothesized birth registration effect: controlling for demographic characteristics, persons with a birth certificate exhibited greater age agreement on linked death certificates and social security records (odds ratio = 2.3).

WPS 95-03

Elo, Irma T., Samuel H. Preston, Ira Rosenwaike, Mark Hill and Timothy P. Cheney

Consistency of Age Reporting on Death Certificates and Social Security Administration Records among Elderly African-American Decedents.

This paper investigates the consistency of age-at-death reporting on death certificates and the Social Security Administration (SSA) records among elderly African-American decedents. At death certificate ages 65+ in 1985, only about 63% of ages at death in the two sources agreed in a national sample of African-American decedents. Age agreement was significantly higher among decedents born in Maryland (73%), a state with better than average birth registration and level of African- American literacy during the early decades of this century. When ages disagreed, the most common pattern was for the death certificate age to be younger than the age based on the SSA record. Birth record availability and the literacy level of the population were key predictors of age agreement. A third data source, US Censuses of 1900, 1910 and 1920, shows greater age agreement with Social Security Age than with death certificate age at death. Key words: African-Americans; death certificate; age reporting; vital statistics, Social Security Administration.

WPS 95-01

Preston, Samuel H. and Irma T. Elo

Are Educational Differentials In Mortality Increasing in the United States?

Two recent studies have compared the size of educational mortality differentials among adults in the 1980's to estimates for 1960. Both studies have concluded that educational differentials have increased for males. One study also finds a similar increase for females. We reconsider this question by introducing a data source that is better suited to estimating recent differentials than either of the two that have been employed. We also evaluate the quality of the 1960 baseline estimate and introduce broader measures of inequality. We conclude that educational inequalities have widened for males but contracted for working age females. For both sexes, inequality trends are more adverse for persons aged 65+ than for persons aged 25-64. The role of national health insurance in shaping these trends is briefly considered.