About This Report
This biennial report presents detailed statistical information on the major sources and amounts of income for people aged 55 or older. The tabulations focus on the major sources of total income by age, sex, marital status, race, and Hispanic origin. Several tables describe the economic situation of the aged with varying levels of Social Security benefits. Their poverty status is presented in terms of the income of the families they live with.
Source of Data
Data for this series are from the March Current Population Survey (CPS) of the U.S. Census Bureau.1 The CPS samples a large cross section of households in the United States each year (approximately 60,000 in March 2003). The March supplement gathers detailed information on income and labor force participation of each person 15 years of age or older in the sample households. For this series, the Social Security Administration (SSA) creates a subsample of persons 55 or older arranged in aged units. A separate data record is made for each married couple living together—at least one of whom is 55 or older—and for each nonmarried person 55 or older. Married persons living apart are classified as nonmarried persons.
From time to time, changes have been made in the survey. Although the changes have improved the measurement of income and labor force participation, they have reduced the comparability of estimates made in different years.2
Definition and Explanation of Key Variables
These tabulations cover units aged 55 or older. The age of a married couple is defined as the age of the husband, unless he is under 55 and the wife is 55 or older, in which case it is the age of the wife. Beginning with data for 2001, the Census Bureau has put all persons aged 80 or older into a single age category on the Current Population Survey public-use file. This edition reflects that change.
The major unit of analysis in Social Security surveys of the aged has been the aged unit rather than the construct of families and unrelated individuals used by the Census Bureau. With 55 as the age cutoff, aged units are married couples living together—at least one of whom is 55 or older—and nonmarried persons 55 or older. Persons who are married but not living with a spouse are counted as nonmarried persons.
Certain differences exist between Census Bureau and SSA figures because the units of analysis are not directly comparable. The Census Bureau classifies families by characteristics of the householder—the first person in whose name a home is listed as owned or rented. It therefore classifies aged persons living with a younger relative who is considered the householder as members of nonaged families. Also, SSA treats nonmarried individuals simply as nonmarried persons. In comparison, the Census Bureau counts nonmarried persons living with other relatives as part of a family and nonmarried persons who are living alone or with nonrelatives as unrelated individuals. The Census Bureau's family category includes both married couples and those nonmarried persons who are living with relatives.
Census data show that the number of households with a householder aged 65 or older in 2002 was 22,659,000.3 In comparison, SSA data show 26,219,000 such households. The SSA count generally includes the Census Bureau's aged households plus some aged units living in nonaged households or living with other aged units in the same household. The number of aged households was 86 percent of the number of aged units.
Aged unit income.
The income of an aged unit is either the income of a nonmarried person or the sum of income from both spouses in a married couple. A married couple is counted as receiving income from a particular source if one or both persons are recipients of that source.
For a family, income is calculated as the sum of total money income of all persons related by blood, marriage, or adoption and residing together. Total money income is the same as family income for aged units who live with no other relatives.
Reporting of income.
Income amounts reported by persons in the Current Population Survey are somewhat less than amounts derived from independent sources such as the Bureau of Economic Analysis, Social Security Administration, and Department of Labor. A comparison of aggregates from the March 1996 CPS with independent estimates found that the CPS accounted for 102 percent of wages and salaries, 92 percent of Social Security and Railroad Retirement benefits, 93 percent of private pensions, 81 percent of federal government pensions, 58 percent of military retirement pensions, 57 percent of state and local government retirement pensions, 84 percent of Supplemental Security Income payments, 84 percent of interest, and 59 percent of dividends.4
A matching of 1972 data from the Census Bureau, SSA, and the Internal Revenue Service has provided a rich source for methodological comparisons of record and survey information of individuals.5 In a report from the 1972 match, the adjusted mean income of families headed by a person aged 65 or older was 41 percent higher than that found in the CPS.6
Social Security beneficiary status.
A yes or no response to the question in the CPS on receipt of Social Security benefits is the measure of beneficiary status. Missing answers are imputed by the Census Bureau.
Although Social Security benefits are referred to as retirement benefits in these tabulations, Social Security beneficiaries include not only retired workers but also dependent spouses, dependents or survivors with young children in their care, and persons with a disability. At the end of 2002, SSA records showed that 98 percent of those aged 55–59 with a Social Security benefit were disabled; the remaining 2 percent were parents with young children in their care. At the age of 60, old-age benefits are available to survivors. Men aged 60–61 all received a benefit because of disability, except for a very small number of widowers. The proportion of female beneficiaries aged 60–61 who received a disability benefit was 67 percent; the remaining were aged widows and those with young children in their care. At the age of 62, reduced retired-worker and dependent's benefits are available. Among beneficiaries aged 62–64, 25 percent of men and 17 percent of women had disability benefits. Almost all remaining men in that age group were receiving retired-worker benefits. Only a small number received benefits as a father or widower. The remaining women aged 62–64 received benefits as retired workers, dependents, or survivors.7
Total money income.
The amount of total money income is calculated as the sum of all income received by the aged unit before any deductions, such as those for taxes, union dues, or Medicare premiums. Income may come from any source that was regularly received. The sources include wages and salaries, self-employment income (including losses), Social Security, Supplemental Security Income, public assistance, interest, dividends, rent, royalties, estates or trusts, veterans' payments, unemployment compensation, workers' compensation, private and government retirement and disability pensions, alimony, and child support. Capital gains (or losses) and lump-sum or one-time payments, such as life insurance settlements, are excluded.
Total money income does not reflect nonmoney transfers such as food stamps, health benefits, subsidized housing, payments in kind, or fringe benefits from one's employment.
To reduce the amount of nonsampling error resulting from nonresponses, the Census Bureau has devised procedures to impute work and income data for all persons for whom this information is missing. Amounts assigned to a nonrespondent are those observed for another person with similar demographic and economic characteristics who did respond.8
Measurement of Poverty
The poverty concept, originally developed in 1964 by Mollie Orshansky of the Social Security Administration and revised by federal interagency committees in 1969 and 1981, consists of a set of thresholds that vary by family size and composition. There are 48 thresholds for families composed of one to nine or more persons cross-classified by the presence and number of family members under the age of 18 (from no children to eight or more). One- and two-person families are further differentiated by the age of the family householder (under the age of 65 and aged 65 or older). The poverty index for families of three or more persons is based on the cost of the Department of Agriculture's economy food plan, multiplied by the ratio of income to food costs derived from the 1955 Household Food Consumption Survey. The factors used to derive the poverty index from food costs for one- and two-person families were higher.
These criteria for determining the extent of poverty in the United States have become the basis for the official statistics issued annually by the Census Bureau in Poverty in the United States, Current Population Reports, Series P-60. The poverty levels are adjusted to reflect changes in the annual average consumer price index.
Calculations of the poverty threshold do not consider nonmoney income. When the poverty index was originally developed, public noncash benefits for both nonneedy and needy families were relatively small and unimportant. The index was therefore developed as a measure of the adequacy of money income only.
The official poverty measures used by the Census Bureau compare family total money income with the appropriate thresholds. Families as well as all persons in families with income below the appropriate threshold are considered poor. Tables 8.1 to 8.3 present measures of the poverty status of aged persons based on their family income, consistent with the official measures.
Interest also centers on the number of persons whose income is only slightly above the poverty line. Those groups, sometimes called the near poor, have income between the poverty line and 125 percent of the poverty line. Estimates of proportions of the aged who are poor or near poor are presented in Tables 8.1 to 8.3.
In May 1995, the Committee on National Statistics of the National Academy of Sciences released a report on poverty measurement, titled Measuring Poverty: A New Approach, that contained a number of recommendations for improving the measurement of poverty.9 Among the recommendations were expanding the income definition to include the growing amount of nonmoney income (such as food stamps and subsidized housing) and taking into account such expenses as income and payroll taxes, child care and other work-related expenses, out-of-pocket medical expenses, and payments of child support made to another household. In 1997, the Office of Management and Budget formed a working group, under the auspices of the Interagency Council on Statistical Policy, to conduct a review of the available options for revising the definition of poverty. The group has coordinated with the Census Bureau to develop experimental poverty measures that incorporate the recommendations of the National Academy of Sciences.10
1 For a detailed description of the basic CPS sample design, see U.S. Census Bureau, Current Population Survey: Design and Methodology, Technical Paper No. 63RV (March 2002).
2 These changes are discussed in some detail in U.S. Census Bureau, Current Population Reports, Series P-60, various years.
3 Income in the United States, 2002, Current Population Reports, Series P-60, No. 221 (September 2003), Table 3.
4 U.S. Census Bureau, Money Income in the United States: 1999, Current Population Reports, Series P-60, No. 209 (September 2000), Table D-1.
5 See Social Security Administration, Studies from Interagency Data Linkages, a series of seven reports, including an introductory paper, published between August 1973 and June 1975.
6 Daniel B. Radner, "Distribution of Family Income: Improved Estimates," Social Security Bulletin 45 (July 1982), pp. 13–21.
7 Annual Statistical Supplement to the Social Security Bulletin, 2003, Tables 5.A1.1–5.A1.8 and 5.A10.
8 For a detailed discussion of these imputation procedures, see Emmett F. Spiers and Joseph J. Knott, "Computer Method to Process Missing Income and Work Experience Information in the Current Population Survey," in American Statistical Association, Proceedings of the Social Statistics Section, 1969, New York, August 19–22, pp. 289–297. The CPS imputation technique is described in U.S. Census Bureau, Current Population Survey: Design and Methodology, Technical Paper No. 63 (March 2000). For an overview of imputation techniques and an extensive reference list, see Graham Kalton and Daniel Kasprzyk, "Imputing for Missing Survey Responses," in American Statistical Association, Proceedings of the Section on Survey Research Methods, 1982, Cincinnati, Ohio, August 16–19, pp. 22–31.
9 Constance F. Citro and Robert T. Michael, Measuring Poverty: A New Approach (Washington, D.C.: National Academy Press, 1995).
10. U.S. Census Bureau, Experimental Poverty Measures: 1999, Current Population Reports, Series P-60, No. 216 (October 2001).