Demographic and Economic Characteristics of Children in Families Receiving Social Security

by
Social Security Bulletin, Vol. 65, No. 2, 2003/2004

Each month, over 3 million children receive benefits from Social Security, accounting for one of every seven Social Security beneficiaries. This article examines the demographic characteristics and economic status of these children using Social Security administrative records matched to the 1996 Survey of Income and Program Participation. Most child beneficiaries receive benefits based on the earnings of a deceased or disabled parent, and nearly two-thirds live in female-headed families. The families of child beneficiaries rely about equally on earnings and income from Social Security for economic support. On average, the family income of child beneficiaries was 25 percent lower than that of all children, but there was no statistically significant difference in the poverty rates of the two groups.


The author is with the Division of Policy Evaluation, Office of Research, Evaluation, and Statistics, Office of Policy, Social Security Administration.

Acknowledgments: Thanks to David Weaver, Howard Iams, James Sears, Kelly Olsen, and Theresa Wilson for their helpful comments and to Thuy Ho and Melissa Koenig for their assistance in processing Survey of Income and Program Participation and administrative data files. Tabulations from an unpublished version of this article have appeared in Hill and Reno (2003).

The findings and conclusions presented in the Bulletin are those of the authors and do not necessarily represent the views of the Social Security Administration.

Summary

The Social Security system provides an important source of support to many children. Currently, just over 3 million children under the age of 18 receive Social Security benefits, accounting for 6.5 percent of all individuals receiving Social Security and $1.2 billion in monthly benefit payments. Fewer than 300,000 of these beneficiaries are children of retired workers, with the remainder almost evenly divided between children of disabled and deceased workers (Social Security Administration 2003a). In addition, over 2 million children who do not receive benefits themselves live in the family of a beneficiary.

This article examines the family insurance benefits that Social Security provides to children of retired, disabled, or deceased workers. The analysis uses Social Security Administration records matched to the U.S. Census Bureau's Survey of Income and Program Participation, thereby combining the accuracy of program records with the wide-ranging and diverse information available in a large sample survey. The following are the major findings of the analysis (all group differences highlighted in this summary were tested using 2-tailed t-tests, and all were statistically significant at the 0.01 level, except as noted.):

Introduction

The nexus between Social Security and children's economic status is often overlooked by researchers or policymakers who are interested in one or the other of these issues. Discussions about the Social Security system tend to center on retirees and disabled adults, while consideration of programs that provide economic support to children rarely includes Social Security. Yet children represent an important group of Social Security beneficiaries, and for many of these children, the program provides a critical source of financial support.

This article uses Social Security Administration (SSA) records that were matched to the 1996 Survey of Income and Program Participation (SIPP) to examine the demographic characteristics of child beneficiaries and their families and the importance of Social Security benefits to their financial well-being. This analysis builds on previous work by Kearney and others (1994, 1995) using the 1990 SIPP panel.1 Examining family-level data, those authors found that the Old-Age, Survivors, and Disability Insurance (OASDI) program was an important income source for beneficiary families but that family earnings were also crucial to economic well-being. The present analysis moves beyond the family level to focus on individual characteristics of child beneficiaries, including demographic characteristics, sources of economic support, reliance on Social Security benefits, and poverty status. It also examines important differences between all children, child beneficiaries, and nonbeneficiary children who live in beneficiary families, as well as differences across benefit type.

Social Security Eligibility Requirements for Children

Under rules for the OASDI program (referred to here as Social Security), a child is eligible to receive a monthly payment from Social Security if a parent has enough quarters of coverage to be insured for benefits and has died or become entitled to retirement or disability benefits (Social Security Administration 2003b).2 If a worker retires or becomes disabled, each of the worker's children may be entitled to a benefit of up to 50 percent of the worker's primary insurance amount (PIA), the basic benefit amount payable to the worker prior to certain adjustments. If an insured worker dies, each of the children may be entitled to a monthly benefit of up to 75 percent of the worker's PIA. For the purposes of Social Security eligibility, the adult-child relationship is defined fairly broadly.3 The biological child of a worker is always eligible for benefits on that parent's record, as long as maternity or paternity has been established. Legally adopted children are eligible on the record of their adoptive parent or parents, and, in many cases, stepchildren are eligible. Children may also be eligible for benefits on the record of a grandparent if both parents are deceased or disabled or if the children have been legally adopted by a grandparent and there is no parent living in the household and making substantial contributions to their support. To simplify the discussion, however, these various types of adult caretakers are generally referred to as "parents" in this article, and child beneficiaries are referred to as the "children" of those on whose earnings record they receive benefits.

Children can continue to receive benefits until they turn 18, or until they turn 19 if they are still in high school.4 A disabled child can continue to receive benefits indefinitely as long as the disability began before age 22. For the purposes of the present analysis, however, only children under the age of 18 are defined as child beneficiaries. If a child beneficiary marries, the child's benefits generally will be terminated.5 A child's benefits may also be partially or completely withheld if the child (or a parent receiving retired-worker benefits) has earnings above the annual exempt amount.6

Children also benefit from the Social Security payments made to other members of their families. The surviving spouse who is providing care for the child (who is under the age of 16 or is disabled) of a deceased worker may be entitled to a benefit of up to 75 percent of the deceased worker's PIA. Similarly, the spouse of a retired or disabled worker with a child in care may be eligible for a benefit of up to 50 percent of the worker's PIA. Earnings above the exempt amount reduce these benefits for nondisabled parents. Disabled parents are not subject to the earnings test, but their benefits are suspended for each month of substantial earnings, if they have more than 9 months of substantial earnings within the prior 60-month period. If the parent is a disabled-worker beneficiary, the benefits received by other members of their family will also be suspended.

The total benefits that the members of a family can draw from the record of a single worker are capped at a certain multiple of that worker's PIA. This maximum family benefit varies according to the amount of the worker's PIA and the type of benefit, but it is not affected by the size of the worker's family. For families receiving retirement or survivors benefits, the maximum family benefit varies from 150 percent to 188 percent of the worker's PIA. In disability cases, the maximum family benefit is the lower of 150 percent of the PIA or 85 percent of the worker's average indexed monthly earnings (but not lower than 100 percent of the PIA). This amount is generally lower than the maximum family benefit for a family receiving retirement or survivors benefits. Benefits to the worker's entitled children (and spouse, if applicable) are proportionately reduced to fit under the maximum family benefit, which is indexed to inflation so that families subject to the cap are not denied the benefit of the annual cost-of-living adjustment.

In summary, children can receive Social Security benefits as dependents of beneficiary parents or as survivors of insured workers. Specific program rules limit the maximum benefits to family members and decrease benefits for earnings above the annual exempt amount.

Data and Method

This analysis uses the 1996 Survey of Income and Program Participation, which has been matched to Social Security Administration benefit records. The resulting data set combines the accuracy of SSA program records with the wide-ranging information and representative sample provided by the SIPP. The 1996 SIPP is a U.S. Census Bureau survey of about 35,000 households designed to represent the noninstitutionalized U.S. population. The Census Bureau interviewed respondents every 4 months over a period of 48 months, collecting a wealth of information on demographic and social characteristics and monthly sources of income. The SIPP is the best source of survey data for this analysis, not just because of its broad coverage and detailed monthly data but also because it is the only data set containing information on child Social Security benefits that has been matched to Social Security Administration data files.

The analysis focuses on data pertaining to March 1996, the first month for which data were obtained from the full SIPP sample, to avoid the problems associated with survey attrition. All analyses pertaining to March 1996 were weighted using the SIPP person weights for that month developed by the Census Bureau. To obtain annual income data for poverty calculations, the first four SIPP waves were combined to create estimates for the period from March 1996 through February 1997. To avoid censoring effects, this "annual" file included only individuals who were present in the SIPP for all months during this period. For all analyses using this annual file, the weight of each respondent was increased proportionally to align with the weighted population count for March 1996.7

Social Security Administration analysts have found that the SIPP contains significant misreporting of benefits from the Social Security and Supplemental Security Income (SSI) programs (Huynh, Rupp, and Sears 2001). SIPP respondents often misreport the amount of their benefits and sometimes misreport SSI payments as Social Security benefits. To correct for benefit misreporting, this analysis used information from SSA's benefit records for SIPP respondents who provided a valid Social Security number.

The three Social Security administrative data files used in this analysis were the Master Beneficiary Record, the Supplemental Security Record, and the Summary Earnings Record. Of the three, the Master Beneficiary Record was by far the most important to the analysis: it provided information on months of OASDI receipt, benefit type, and benefit amounts that were substituted for survey-reported data.8 The Supplemental Security Record provided similar information on SSI receipt and payment amounts, useful information given the misreporting of SSI as OASDI described above.9 The Summary Earnings Record was used to identify SIPP respondents for whom the Census Bureau had obtained a valid Social Security number.10 This allowed beneficiaries to be matched to the other administrative data files and permitted the accurate identification of nonbeneficiaries.

Administrative information was not available for survey respondents who did not provide a matchable Social Security number. In March 1996, 21 percent of children and 16 percent of adults lacked an administrative match. To deal with this problem, the study used self-reported survey data in lieu of administrative information for these unmatched cases.11 The adjusted OASDI and SSI benefit amounts were then aggregated over all members of the family and substituted for families' total self-reported benefits in the determination of family income. Families were defined, in accordance with SIPP methodology, as persons who lived in the same household and were related by blood, marriage, or adoption.

Characteristics of Child Beneficiaries

According to the administratively matched SIPP data described above, approximately 2.78 million children received Social Security benefits in March 1996 (Table 1). This is about 6 percent below the estimate for December 1995 contained in the Annual Statistical Supplement, 1996 (Social Security Administration 1996). Compared with these figures, the SIPP estimates slightly understated benefit receipt and the average benefit amount of children and adults. Some degree of underestimation was expected given the differences between the two populations and the underreporting of benefits by SIPP respondents.12 Despite these discrepancies, the relative distribution of child beneficiaries by benefit type was quite similar for these two data sources. Because of the limited availability of demographic and income data in the Supplement and in other purely administrative data sources, the remainder of this analysis will rely solely on the administratively matched SIPP data.

Table 1. Comparison of data from the matched SIPP and theAnnual Statistical Supplement
  March 1996 Matched SIPP Data a Annual Statistical Supplement, 1996 b
Adults
Number of beneficiaries 37,538,430 40,422,550
Beneficiaries as a percentage of all adults 19.4 20.0
Average monthly benefit (dollars) 672 673
Children
Number of beneficiaries 2,784,052 2,957,890
Beneficiaries as a percentage of all children 3.9 4.1
Average monthly benefit (dollars) 313 318
Type of benefit c (percent)
Child of deceased worker 46.6 46.9
Child of disabled worker 46.1 45.0
Child of retired worker 7.2 8.1
SOURCE: Data are from the March 1996 Survey of Income and Program Participation (SIPP) and the Annual Statistical Supplement (Social Security Administration 1996).
a. The data were weighted using the March 1996 Survey of Income and Program Participation person weights. The total unweighted count of adult beneficiaries was 13,481. The total unweighted count of child beneficiaries was 1,116.
b. Data are from December 1995, Table 5.A1. Social Security area population figures for proportions are from Table 4.C5, with interpolation for 18- and 19-year-olds.
c. The breakdown by benefit type for the matched 1996 SIPP data includes only children with an administratively identified benefit (total unweighted count, 975).

About 3.9 percent of all children received Social Security benefits in March 1996, with an average benefit of $313 (Table 2). Since most child beneficiaries lived with others who also received Social Security benefits, the average family benefit of child beneficiaries was considerably larger, just over $850. About 47 percent of child beneficiaries were children of deceased workers, nearly equal to the 46 percent who were children of disabled workers.13 Only about 7 percent were children of retired-worker beneficiaries. The average benefit received by children of deceased workers was almost three times that of children of disabled workers and more than one and a half times that of children of retired workers. The gap in average benefits was considerably smaller when all family benefits were accounted for, but the average family benefit for children of disabled workers was still substantially below that of the other two groups. Two main factors account for this differential. First, families of disabled workers are subject to a more stringent maximum family benefit, as described above.14 Second, disabled workers tend to have lower primary insurance amounts than retired or deceased workers because of lower yearly earnings (Social Security Administration 1996, Tables 5.C1, 5.E1, and 5.F7). For these and other reasons to be discussed later, the average family incomes of children of disabled workers were also lower than those of children in the other two benefit categories.

Table 2. Selected characteristics of child Social Security beneficiaries
Characteristic Participation rate of children in category (percent) Percentage of all child beneficiaries in category Average benefit (dollars) Average family benefit (dollars) Average family income of child beneficiaries (dollars)
By characteristics of child
Total a 3.9 100 313 852 2,845
Type of benefit b
Child of deceased worker . . . 46.6 460 917 2,932
Child of disabled worker . . . 46.1 164 768 2,596
Child of retired worker . . . 7.2 293 892 2,876
Sex
Male 4.0 52.5 316 866 2,861
Female 3.8 47.5 310 837 2,828
Race and ethnicity
White, non-Hispanic 3.8 63.9 342 917 3,155
Black, non-Hispanic 5.2 20.5 271 652 2,160
Other, non-Hispanic 3.1 3.8 264 1,070 3,807
Hispanic 3.2 11.8 250 777 2,033
Age
Under 5 1.2 8.6 231 769 2,250
5–9 3.0 21.4 280 828 2,816
10–14 5.5 38.0 311 917 2,860
15–17 7.7 31.9 360 813 3,009
By characteristics of family head
Sex
Male 2.3 36.1 274 914 3,755
Female 6.3 63.9 335 817 2,331
Marital status
Married 2.5 45.5 273 968 3,885
Widowed 49.7 21.2 464 1,067 2,402
Divorced or separated 5.3 21.4 270 582 1,902
Never married 3.8 12.0 274 512 1,368
Age
18–29 2.2 11.3 299 651 1,746
30–39 3.2 36.5 283 761 2,658
40–49 4.7 35.3 340 986 3,307
50–64 9.0 12.8 359 859 3,367
65 or older 28.1 4.1 243 1,039 1,922
Education
Less than high school 5.7 23.7 240 753 1,526
High school 4.1 33.6 302 865 2,581
Some college 4.1 30.7 342 873 3,134
College graduate 2.1 11.9 409 984 5,626
By characteristics of family
Number of children in family
One 5.3 33.1 402 799 2,693
Two or three 3.4 53.0 287 894 3,026
Four or more 3.9 13.9 202 817 2,519
By characteristics of primary adult beneficiary
Primary adult beneficiary
Parent . . . 47.3 233 1,046 2,582
Other adult
Parent present in family . . . 2.8 409 1,499 4,949
No parent in family . . . 2.5 368 1,268 2,004
No adult beneficiary
Parent present in family . . . 42.0 392 612 3,146
No parent in family . . . 5.4 335 435 1,807
Type of benefit c
Retired worker . . . 15.3 297 1,143 3,152
Disabled worker . . . 62.0 174 1,024 2,772
Spouse or widow(er) . . . 22.3 387 1,127 1,735
Adult disabled in childhood . . . 0.4 d d d
SOURCE: Data are from the March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTE: All benefit and income amounts are monthly values.
. . . = not applicable.
a. The total unweighted count of child beneficiaries was 1,116.
b. The breakdown by benefit type for the matched 1996 SIPP data includes only children with an administratively identified benefit (total unweighted count, 975).
c. The breakdown by benefit type of the primary adult for the matched March 1996 SIPP data includes beneficiary children in the family of an administratively identified adult beneficiary (total unweighted count, 515).
d. Data on average benefits and income could not be presented because of sample size constraints.

An analysis of children's demographic characteristics shows little gender difference but large variations across race and ethnicity and age groups (Table 2). The majority of child beneficiaries were non-Hispanic whites, but the prevalence of Social Security receipt was lower among white children than it was among non-Hispanic black children. This difference resulted from black adults' greater likelihood of death or disability during their working lives (U.S. Census Bureau 1996, Table B-1; Diez 2000). Hispanic children and non-Hispanic children of some other race were less likely to receive benefits than were non-Hispanic blacks and, to a lesser extent, non-Hispanic whites.15 The parents of Hispanic children tend to be somewhat younger than parents of non-Hispanic ethnicity; thus they face lower risks of disability and mortality. Also, because they are more likely to immigrate as adults, they tend to have worked for fewer years in employment covered by Social Security, making them less likely to meet Social Security's eligibility requirements (Liebman 2001). Because of their parents' lower lifetime earnings, black and Hispanic children received lower average benefits than did white children, at the individual and family levels. Black and Hispanic child beneficiaries also had lower average family incomes than whites did. Non-Hispanic children of other races had an average benefit similar to that of blacks and Hispanics, but their average family benefit and their average family income were higher than those of any other group.

The likelihood that a child would receive Social Security benefits increased dramatically with age: just 1.2 percent of children under 5 received benefits compared with 7.7 percent of 15- to 17-year-olds. One reason for this pattern is that the parents of older children tend to be older themselves, and the likelihood of death or disability increases with age. At the same time, the cumulative probability of a parent's death or disability rises as a child gets older. Average child and family benefits and family income were also higher for children in older age groups, reflecting the higher average lifetime earnings of the older workers on whose records these benefits were based. One exception was the lower average family benefit of 15- to 17-year-olds, which could have resulted from the termination of parental child-in-care benefits when the youngest child reached the age of 16. In the aggregate, however, these lower benefits did not translate into lower average family income.

The probability that a child would receive Social Security as well as family benefits and income varied in important ways according to the characteristics of the family head (Table 2).16 Nearly two-thirds of all child beneficiaries lived in families headed by a woman, reflecting the higher incidence of benefit receipt among children in female-headed families. Although these children received higher average individual benefits than children in male-headed families received, their average family benefit was lower, and their average family income was considerably smaller, primarily because of lower family earnings.

Slightly more than 45 percent of all child beneficiaries lived with a married family head, yet the prevalence of child benefit receipt was actually lower among these children than it was among those living with nonmarried family heads. The prevalence of Social Security receipt was highest (nearly 50 percent) among children in families with widowed family heads, not surprising given the likelihood that children in these families would qualify for survivor benefits.17 Child beneficiaries in widow(er)-headed families received the largest average individual benefits, in part because of the higher share of the insured worker's PIA payable to family members of decedents. Child beneficiaries with married or widowed family heads received a far larger average family benefit than did children with divorced or never-married family heads, and those with married family heads also had much higher average family income than did any of the other three groups.

As would be expected, when looking at the relationship between age, mortality, and the incidence of disability, the share of children receiving Social Security increased with the age of the family head. The share was highest among children living in families headed by someone over the age of 65, but relatively few children lived in such senior-headed families. There did not appear to be a consistent relationship between average benefit levels and the age of the family head, but average family income was higher for child beneficiaries with older family heads, except for those in families headed by seniors.

The prevalence of benefit receipt among children was inversely related to the educational attainment of the family head. Here again, patterns of mortality and disability provide a likely explanation. Although more highly educated adults are more likely to be insured under Social Security, they are less likely to die or become disabled during their child-rearing years (Lleras-Muney 2002). Child beneficiaries' average benefits increased with the educational attainment of the family head because of the greater lifetime earnings of those with more education.

About 47 percent of child beneficiaries lived with a parent receiving Social Security. An almost identical share did not live with any adult beneficiary, and the remaining 5 percent lived with an adult beneficiary other than a parent.18 What can account for the large proportion of child beneficiaries who did not live with an adult beneficiary, especially given that most of these children were living with a parent? There are a number of programmatic reasons why benefits to children may be payable while benefits to adult beneficiaries are not. Earnings above the annual exempt amount may lead to benefit suspension for these adults, while remarriage or no longer having a child under the age of 16 can lead to the cessation of benefits. Among child beneficiaries living with parents who did not receive benefits, 13 percent did not have anyone under the age of 16 in their family, 10 percent had a parent whose benefit was withheld because of earnings above the annual exempt amount, and 22 percent received survivors benefits but lived with a married parent (data not shown).19 Together, these three factors explain the lack of parental benefits for 45 percent of child beneficiaries living with nonbeneficiary parents.

Child beneficiaries who lived with a parent beneficiary had a lower average individual benefit than did children living with another adult beneficiary or with no adult beneficiary, but their average family benefit was higher than that of the latter group. Among children who did not live with an adult beneficiary or who lived with an adult beneficiary who was not their parent, average family income was considerably higher if there was a parent present in the family. It is interesting to note that among children living with a parent, family income was higher if the parent was not receiving benefits.

Among child beneficiaries who lived with an adult beneficiary, 62 percent lived with a disabled-worker beneficiary.20 Of the remaining child beneficiaries, most lived with spouse or widow(er) beneficiaries or retired workers. The average family benefit was similar for the three groups, but average family income was lower among children living with a spouse or widow(er) beneficiary than it was among the other two groups.

The characteristics of beneficiary children varied markedly by the type of benefit they received (Table 3). Non-Hispanic whites made up a relatively smaller share of children of retired workers than they did in the other two groups, while Hispanic children made up a relatively larger share. The disparities in benefits across racial and ethnic groups were smaller among the children of disabled workers than they were among the children of deceased workers, mainly because of lower average benefit levels among white children in the former group.

Table 3. Selected characteristics of child beneficiaries, by type of benefit
Characteristic Children of deceased workers Children of disabled workers Children of retired workers a (percent)
Percent Average family benefit (dollars) Average family income (dollars) Percent Average family benefit (dollars) Average family income (dollars)
Total 100.0 917 2,932 100.0 768 2,596 100.0
By characteristics of child
Sex
Male 51.9 932 2,981 53.4 781 2,535 54.1
Female 48.1 902 2,880 46.6 753 2,667 46.0
Race and ethnicity
White, non-Hispanic 64.7 1,026 3,223 67.3 791 2,889 61.0
Black, non-Hispanic 19.2 637 2,096 19.7 644 1,939 19.0
Other, non-Hispanic 4.7 b b 3.4 b b 0.0
Hispanic 11.4 807 2,246 9.5 765 1,934 20.0
Age
Under 5 6.7 756 2,083 10.4 716 1,981 6.5
5–9 18.1 930 3,244 21.8 708 2,182 21.8
10–14 39.48 988 2,822 38.8 830 2,823 32.4
15–17 35.72 863 3,055 29.0 748 2,826 39.3
By characteristics of family head
Sex
Male 24.36 792 4,111 42.9 909 3,305 69.6
Female 75.64 958 2,553 57.1 662 2,063 30.4
Marital status
Married 27.86 918 4,801 59.5 941 3,305 68.1
Widowed 41.95 1,088 2,469 2.1 b b 2.8
Divorced or separated 17.09 763 2,156 27.1 469 1,630 20.3
Never married 13.1 573 1,457 11.3 508 1,319 8.8
Age
18–29 11.76 655 1,734 10.6 604 1,753 10.6
30–39 37.74 923 2,948 40.1 635 2,370 8.2
40–49 35.74 1,024 3,286 38.2 917 2,877 11.2
50–64 12.82 859 3,212 10.0 850 3,376 29.4
65 or older 1.95 b b 1.2 b b 40.8
Education
Less than high school 23.69 816 1,589 25.8 712 1,584 31.0
High school 32.37 903 2,596 31.9 777 2,540 33.4
Some college 31.07 1,005 3,273 32.2 758 2,851 25.0
College graduate 12.87 996 5,782 10.1 916 4,558 10.6
By characteristics of family
Number of children in family
One 34 739 2,582 30.9 810 2,750 55.5
Two or three 51.78 986 3,242 52.2 781 2,527 35.1
Four or more 14.21 1,093 2,645 17.0 651 2,530 9.4
By characteristics of primary adult beneficiary
Parent 27.05 1,143 1,964 66.8 990 2,778 72.1
Other adult
Parent present in family 3.87 b b 0.5 b b 1.9
No parent in family 3.17 b b 1.3 b b 5.6
No adult beneficiary
Parent present in family 56.78 805 3,403 29.1 261 2,236 14.7
No parent in family 9.12 516 1,910 2.3 b b 5.7
SOURCE: Data are from the March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTES: All benefit and income amounts are monthly values.
This table includes only children receiving an administratively identified benefit. The unweighted sample count was 460 for children of deceased workers, 70 for children of retired workers, and 445 for children of disabled workers.
a. Because of sample size constraints, average family benefits and income could not be included for children of retired workers.
b. Data on average benefits and income could not be presented because of sample size constraints.

The most striking differences across benefit type involved the characteristics of the beneficiary's family head. About 76 percent of the children of deceased workers had a female family head, compared with 57 percent of children of disabled workers and just 30 percent of children of retired workers. Particularly in female-headed families, children of disabled workers had markedly lower average family incomes than did children of deceased workers. The majority of children of retired and disabled workers lived with a married family head, while the largest proportion of children of deceased workers (42 percent) lived in families headed by a widow(er). Children of retired workers tended to have family heads who were much older and who were less likely to have completed high school than did children in either of the other two groups. The majority of children of retired workers were the only child in the family, while among children of deceased and disabled workers, there was at least one other child in the majority of cases. Among children of deceased workers, average family benefits increased with the number of children in the family, but average family benefits were lower for children of disabled workers when there were more children in the family.

Only 27 percent of children of deceased workers lived with a parent beneficiary, compared with 72 percent of children of retired workers and nearly 67 percent of children of disabled workers (Table 3). Average family benefits were higher for children living with a parent beneficiary, but this was not necessarily true of family income. Among children of deceased workers, average family income was highest when there was a parent present who was not receiving benefits.

Characteristics of Nonbeneficiary Children in Beneficiary Families

The data presented so far on child Social Security beneficiaries do not encompass the program's full impact on children. In addition to the 2.8 million child beneficiaries described above, another 2.2 million children had a family member who received Social Security benefits even though the children did not receive benefits themselves. These nonbeneficiary children in beneficiary families constituted 3.1 percent of all children.

The demographic characteristics of nonbeneficiary children living in beneficiary families are summarized in Table 4. Comparing Tables 2 and 4 reveals that child beneficiaries and these nonbeneficiary children generally had similar personal and family characteristics. Some degree of overlap might be expected, given that about 30 percent of nonbeneficiary children in beneficiary families had a child beneficiary in their family. However, there were a few notable exceptions to the general comparability of the two groups: black children were more likely to live with a beneficiary than to receive benefits themselves, as were children under age 10 and children living in families headed by someone over the age of 65. In contrast, children in female-headed families—especially those in families headed by a widow—were much more likely to receive benefits directly than to be nonbeneficiary children in a beneficiary family. These nonbeneficiary children tended to have smaller average family Social Security benefits but lived in families with larger average family incomes than did child beneficiaries. Among black children, however, average family income was higher for child beneficiaries than it was for children living in beneficiary families who did not receive benefits themselves. This was also true for children with widowed family heads and family heads over the age of 50.

Table 4. Selected characteristics of nonbeneficiary children in (Social Security) beneficiary families
Characteristic NBCBFs as a percentage of all children in category Percentage of all NBCBFs in group Average family benefit (dollars) Average family income of NBCBFs (dollars)
Total a 3.1 100.0 626 3,505
By characteristics of child
Sex
Male 3.4 54.9 618 3,500
Female 2.9 45.1 637 3,511
Race and ethnicity
White, non-Hispanic 2.5 51.3 715 4,575
Black, non-Hispanic 6.4 31.6 555 1,870
Other, non-Hispanic 2.5 3.8 501 4,102
Hispanic 2.9 13.2 490 3,083
Age
Under 5 2.4 22.0 623 3,147
5–9 3.1 27.6 600 3,301
10–14 3.6 31.6 656 3,778
15–17 3.6 18.8 619 3,766
By characteristics of family head
Sex
Male 2.3 43.7 669 4,917
Female 4.4 56.3 593 2,410
Marital status
Married 2.5 56.5 673 4,774
Widowed 22.0 11.7 749 1,663
Divorced or separated 3.5 17.5 498 1,876
Never married 3.6 14.4 500 1,997
Age
18–29 1.7 10.6 474 2,042
30–39 2.6 36.9 533 3,379
40–49 2.7 25.2 688 5,475
50–64 8.8 15.6 711 2,860
65 and older 63.5 11.6 814 1,833
Education
Less than high school 5.3 27.8 578 1,841
High school 3.4 34.1 587 2,522
Some college 2.6 24.6 655 3,925
College graduate 1.9 13.5 776 8,660
By characteristics of family
Number of children in family
One 3.2 25.0 694 3,026
Two or three 2.7 52.4 611 3,902
Four or more 5.1 22.6 586 3,115
By characteristics of primary adult beneficiary
Primary adult beneficiary
Parent . . . 19.5 574 2,370
Other adult
Parent present in family . . . 31.8 677 5,443
No parent in family . . . 24.1 747 2,136
No adult beneficiary
Parent present in family . . . 23.7 471 3,308
No parent in family . . . 1.0 b b
Type of benefit c
Retired worker . . . 44.7 744 4,999
Disabled worker . . . 31.3 660 2,674
Spouse or widow(er) . . . 20.1 662 2,937
Adult disabled in childhood . . . 3.9 b b
SOURCE: Data are from the March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTES: All benefit and income amounts are monthly values.
. . . = not applicable.
a. The total unweighted count of nonbeneficiary children in beneficiary families (NBCBFs) was 885.
b. Data on average benefits and income could not be presented because of sample size constraints.
c. The breakdown by benefit type of the primary adult for the matched March 1996 SIPP data includes nonbeneficiary children in the family of an administratively identified adult beneficiary (total unweighted count, 543).

How can so many nonbeneficiary children live in beneficiary families, especially given that Social Security eligibility rules generally entitle children to benefits when they have a parent receiving them? Part of the answer lies in the fact that, for approximately 56 percent of these children living in beneficiary families who do not receive benefits themselves, an adult other than their parent—such as an aunt or grandparent—was the Social Security beneficiary in their family (Table 4). In this context it is important to bear in mind that we used a relatively broad definition of "family" that included all individuals who lived together and were related by blood, marriage, or adoption. Unfortunately, the data do not provide an explanation for the lack of benefits to the 20 percent of nonbeneficiary children who lived with a beneficiary parent or the 25 percent for whom another child was the only beneficiary in their family. However, we are able to rule out earnings above the exempt amount as a significant factor for a lack of benefits to these children.

Among nonbeneficiary children who lived with an adult beneficiary, the largest proportion (45 percent) lived with a retired-worker beneficiary (Table 4).21 For nonbeneficiary children living with nonparental adult beneficiaries, the share of those living with a retired-worker beneficiary was 55 percent, whereas for 70 percent of nonbeneficiary children living with a parent beneficiary, that parent was receiving disabled-worker benefits (data for both not shown). Nonbeneficiary children in beneficiary families who are living with retired workers had by far the largest average family incomes (Table 4).

Sources of Economic Support

Major sources of economic support (including food assistance and health insurance) differed between child Social Security beneficiaries, nonbeneficiary children in beneficiary families, and all children (Table 5). The average family income of child beneficiaries was 25 percent below that of all children, and the average family income of nonbeneficiary children in beneficiary families was 10 percent lower. Earnings were the dominant source of family income for the average child, accounting for almost 80 percent of family income.22 For child beneficiaries and nonbeneficiary children in beneficiary families, however, earned income accounted for less than 50 percent of their family's income, on average. This difference may reflect the generally diminished earnings capacity of families receiving Social Security. Nevertheless, close to 70 percent of child beneficiaries and nonbeneficiary children in beneficiary families lived in families with some earned income. In the case of child beneficiaries, many parents were eligible for benefits as the spouse of a deceased or disabled worker but did not collect them, because they had earnings above the Social Security exempt amount. Other sources tended to account for a relatively small share of the family income of children in all three groups. These other sources, particularly Supplemental Security Income and Temporary Assistance for Needy Families, were most significant among nonbeneficiary children in beneficiary families.

Table 5. Sources of economic support of all children, child beneficiaries, and nonbeneficiary children in beneficiary families
Source All children Child beneficiaries Nonbeneficiary children in beneficiary families
Average share of family income a (percent) Percentage receiving Average amount among recipients (dollars) Average share of family income a (percent) Percentage receiving Average amount among recipients (dollars) Average share of family income a (percent) Percentage receiving Average amount among recipients (dollars)
Family income
Total 100.0 . . . 3,831 100.0 . . . 2,845 100.0 . . . 3,505
Earnings 78.8 85.7 4,109 43.0 68.0 2,444 47.6 70.2 3,391
Social Security 2.7 7.0 752 43.1 100.0 852 31.9 100.0 626
Supplemental Security Income 1.9 4.5 520 2.7 10.5 363 5.5 22.0 414
Temporary Assistance for Needy Families 6.5 9.7 409 2.6 8.7 268 5.2 18.6 314
Pensions 0.5 1.8 951 1.8 7.1 902 3.0 17.5 604
Child support 2.5 10.7 379 0.8 5.6 264 1.8 12.4 378
Other 7.1 52.1 256 6.1 47.7 398 5.1 49.0 398
Food assistance
Food Stamps . . . 15.3 259 . . . 19.8 205 . . . 25.2 256
Women, Infants, and Children Program . . . 10.4 49 . . . 8.6 44 . . . 13.3 45
Health insurance
Private . . . 68.6 . . . . . . 54.0 . . . . . . 49.9 . . .
Medicaid . . . 18.1 . . . . . . 26.3 . . . . . . 32.8 . . .
Uninsured . . . 13.2 . . . . . . 19.8 . . . . . . 17.3 . . .
SOURCE: Data are from March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTES: All benefit and income amounts are monthly values.
The unweighted sample count was 26,966 for all children, 1,116 for child beneficiaries, and 885 for nonbeneficiary children in beneficiary families.
. . . = not applicable.
a. Income shares were calculated as the average share across individuals, not as shares of aggregate income.

Child beneficiaries were somewhat more likely than all children to live in families receiving Food Stamps, but they were slightly less likely to receive support from the Women, Infants, and Children Program (WIC; or Special Supplemental Nutrition Program for Women, Infants, and Children Program), perhaps because of the low prevalence of Social Security receipt among very young children. Nonbeneficiary children in beneficiary families were more likely than were children in either of the other two groups to live in families receiving Food Stamps or WIC.

The death, disability, or retirement of a family's primary earner may create serious difficulties in maintaining health insurance coverage for children, since Social Security receipt does not provide children with any entitlement to health insurance. Indeed, data in Table 5 indicate that child beneficiaries and nonbeneficiary children in beneficiary families were less likely than all children to be covered by health insurance in March 1996.23 Their greater likelihood of receiving Medicaid coverage did not fully offset their lower likelihood of private coverage.24 Nonbeneficiary children in beneficiary families were slightly less likely to lack insurance than were child beneficiaries, because of their greater likelihood of receiving Medicaid.

Child beneficiaries' sources of support varied by the type of benefit they received (Table 6). Children of retired workers received on average a smaller share of their family income from earnings than was the case with the other two groups, while children of disabled workers received a smaller share from Social Security. The lower average family income of children of disabled workers (relative to the other two groups of child beneficiaries) resulted mainly from their families' lower average benefits and earnings levels. Children of retired and disabled workers relied more heavily for support on income from public assistance programs than children of deceased workers did. Unlike the other two groups, children of retired workers received a substantial share of their family income from pensions. The greater prevalence of income from public programs among children of disabled and retired workers extended to Food Stamps and Medicaid as well. Though children of deceased workers were more likely to have private health insurance coverage than were the other groups, they were nonetheless more likely to be uninsured because of their lower rate of Medicaid receipt.

Table 6. Sources of economic support of children of deceased, disabled, and retired workers
Source Children of deceased workers Children of disabled workers Children of retired workers
Average share of family income (percent) Percentage receiving Average amount among recipients (dollars) Average share of family income (percent) Percentage receiving Average amount among recipients (dollars) Average share of family income (percent) Percentage receiving Average amount among recipients (dollars)
Family income
Total 100.0 . . . 2,932 100.0 . . . 2,596 100.0 . . . 2,876
Earnings 43.7 69.4 2,543 43.0 68.5 2,103 30.6 55.1 2,602
Social Security 46.1 100.0 917 39.9 100.0 768 46.6 100.0 892
Supplemental Security Income 1.9 8.1 342 3.4 12.5 378 5.8 21.2 326
Temporary Assistance for Needy Families 1.6 6.8 252 3.2 9.7 280 3.4 14.7 248
Pensions 1.1 4.5 809 1.6 6.7 829 7.9 21.6 1,218
Child Support 0.5 4.5 348 1.1 6.2 227 0.8 8.8 216
Other 5.1 52.1 296 7.8 44.5 542 4.9 39.5 411
Food assistance
Food Stamps . . . 12.5 202 . . . 25.3 221 . . . 30.4 182
Women, Infants, and Children Program . . . 6.7 45 . . . 8.8 47 . . . 5.4 31
Health insurance
Private . . . 55.8 . . . . . . 51.3 . . . . . . 52.6 . . .
Medicaid . . . 23.2 . . . . . . 28.9 . . . . . . 32.1 . . .
Uninsured . . . 21.0 . . . . . . 19.8 . . . . . . 15.3 . . .
SOURCE: Data are from the March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTES: All benefit and income amounts are monthly values.
This table includes only children receiving an administratively identified benefit. The unweighted sample count was 460 for children of deceased workers, 445 for children of disabled workers, and 70 for children of retired workers.
. . . = not applicable.

The importance of Social Security as a source of family income for child beneficiaries is analyzed in more detail in Table 7. Overall, there was a high degree of variability: for 36 percent of child beneficiaries, Social Security accounted for less than a quarter of their family income, but for almost 35 percent, it accounted for more than half, including 10 percent for whom Social Security was their sole source of family income. Children in families headed by women (particularly widows), by those 65 or older, and by those with less education relied most heavily on support from Social Security.25 Nearly 50 percent of poor child beneficiaries relied on Social Security for more than half of their family income, compared with just 10 percent of those with incomes above 200 percent of poverty. While 13 percent of poor child beneficiaries were entirely reliant on Social Security, none of the child beneficiaries with family income above 200 percent of poverty were. Although these findings indicate the importance of Social Security to poor child beneficiaries, another way to interpret them is that child beneficiaries tended to be poor if their families did not have supplementary sources of income.

Table 7. Percentage of child beneficiaries receiving specified share of family income from Social Security, by selected characteristics
Characteristic Social Security benefits as a percentage of family income
Under 25% 25–50% 50–75% 75–99% 100%
Total 36.2 29.4 14.7 9.6 10.1
By characteristics of child
Type of benefit a
Child of deceased worker 36.3 25.0 15.5 12.2 11.0
Child of disabled worker 37.8 32.0 14.5 7.0 8.8
Child of retired worker 22.6 39.9 21.9 10.0 5.7
Sex
Male 35.5 30.5 15.0 9.4 9.7
Female 37.0 28.1 14.4 9.9 10.6
Race and ethnicity
White, non-Hispanic 36.5 29.6 14.7 10.1 9.2
Black, non-Hispanic 38.8 26.0 15.4 7.5 12.3
Other, non-Hispanic 33.8 55.2 3.7 0.0 7.3
Hispanic 30.8 25.7 17.3 14.1 12.2
Age
Under 5 34.0 31.0 10.2 6.9 17.8
5–9 37.7 26.5 16.0 11.4 8.4
10–14 33.5 31.8 15.0 10.8 9.0
15–17 39.0 28.0 14.8 7.8 10.5
By characteristics of family head
Sex
Male 43.2 29.8 12.8 8.1 6.1
Female 32.6 29.2 15.7 10.4 12.0
Marital status
Married 41.9 33.0 11.9 6.6 6.5
Widowed 18.8 25.9 21.3 18.6 15.4
Divorced or separated 40.8 26.3 13.9 8.9 10.1
Never married 34.2 24.6 16.8 6.4 17.9
Age
18–29 36.4 25.2 17.1 7.6 13.7
30–39 43.0 24.9 12.8 7.4 11.8
40–49 30.6 35.9 14.4 10.4 8.7
50–64 40.2 26.5 16.3 8.5 8.5
65 and older 7.6 26.9 27.3 32.2 5.9
Education
Less than high school 20.1 25.0 25.0 16.0 14.0
High school 33.2 31.2 15.1 7.5 13.0
Some college 42.2 33.6 7.6 9.3 7.4
College graduate 58.6 21.1 13.8 5.1 1.5
By characteristics of family
Number of children in family
One 35.5 27.1 16.0 7.9 13.6
Two or three 36.8 29.3 14.5 11.0 8.5
Four or more 35.6 34.9 12.6 8.8 8.2
Ratio of income to poverty threshold b
Under 100% 29.1 26.5 8.3 23.4 12.8
100–150% 30.5 28.1 15.2 23.6 2.7
150–200% 50.5 24.3 21.5 3.3 0.4
Above 200% 58.1 31.2 8.4 2.4 0.0
SOURCE: Data are from the March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTE: The unweighted count of child beneficiaries was 1,116.
a. The breakdown by benefit type for the matched 1996 SIPP data includes only children with an administratively identified benefit (total unweighted count, 975).
b. These data are from the annual file covering the period March 1996 to February 1997. The unweighted count of child beneficiaries was 1,090.

The independent effect of each demographic factor on the share of child beneficiaries' family income coming from Social Security was tested statistically using an ordinary least squares regression model (Table 8).26 Although Table 7 focused on percentage cutoffs, the share of income from Social Security was entered into the model as a continuous variable.27 The results indicate that children whose family heads were widowed, divorced, or never married depended on Social Security for a significantly greater share of their income than did children with married family heads. Children with more-educated family heads were significantly less reliant on Social Security. Although they appeared potentially important in the descriptive analysis, benefit type, race and ethnicity, female headship, and age of family head did not have a significant impact in the multivariate model.

Table 8. Regression models of share of family income from Social Security and poverty status of child beneficiaries
  Share of family income from Social Security (OLS model) Poverty status (Logit model)
Coefficient p-value Coefficient p-value Odds ratio
By characteristics of child beneficiary
Type of benefit
Child of deceased worker -0.03419 0.1137 -0.6085 0.0076 0.544
Child of disabled worker . . . . . . . . . . . .
Child of retired worker -0.02814 0.523 -0.2876 0.5304 0.75
Sex
Male . . . . . . . . . . . .
Female 0.00776 0.6687 -0.2519 0.1886 0.777
Race and ethnicity
White, non-Hispanic . . . . . . . . . . . .
Black, non-Hispanic -0.04562 0.0678 0.6043 0.0109 1.83
Other, non-Hispanic -0.06637 0.1693 2.1913 a 8.947
Hispanic -0.02958 0.3515 0.421 0.1886 1.523
Age -0.00260 0.2314 -0.0499 0.0173 0.951
By characteristics of family head
Sex
Male . . . . . . . . . . . .
Female 0.03097 0.1703 0.2923 0.259 1.34
Marital Status
Married . . . . . . . . . . . .
Widowed 0.21999 a 0.6507 0.0781 1.917
Divorced or separated 0.05251 0.0408 1.6646 a 5.284
Never married 0.14138 0.0002 2.2404 a 9.397
Age -0.00423 0.3899 -0.0527 0.2535 0.949
Age squared 0.00008868 0.0996 0.000614 0.2343 1.001
Education -0.02199 a -0.2751 a 0.759
By characteristics of family
Number of children in family 0.0024 0.7553 0.3263 a 1.386
Adjusted R2/Quasi R2 0.1477 0.2331
Log likelihood -479.479
SOURCE: Data are from the March 1996 Survey of Income and Program Participation, matched to administrative records from the Social Security Administration.
NOTES: Number of observations (unweighted) for the OLS model were 967 and 921 for the Logit model. Eight cases were excluded from the regression models because of missing values on the education variable, resulting in slightly lower sample counts than in the associated univariate tabulations.
OLS = ordinary least squares regression; . . . = not applicable.
a. Less than 0.0001.

Poverty Status

Poverty status—obtained by dividing family income by the poverty threshold—provides one measure of income relative to basic needs. Poverty status was measured using the annual file of those who remained in SIPP for the entire period from March 1996 through February 1997.28 These poverty estimates are not directly comparable to official Census Bureau estimates, which are based on the March Supplement to the Current Population Survey and cover a January–December calendar-year period. The official estimate of the child poverty rate for 1996 was 20.5 percent, whereas the SIPP estimate was 21.6 percent for the March 1996 to February 1997 period.29

The poverty rate for child beneficiaries was somewhat higher than that for all children and was higher still among nonbeneficiary children in a beneficiary family, even though they had a higher average family income than child beneficiaries had (Tables 5 and 9). Relative to all children, a disproportionate share of child beneficiaries and nonbeneficiary children in beneficiary families also had family incomes between 100 percent and 150 percent of the poverty threshold. The poverty rate was slightly lower among children who received benefits for the entire year than it was among those who began benefits during the year (data not shown), suggesting that families may suffer more hardship in the period just preceding benefit receipt than after benefits begin.

Table 9. Percentage distribution of children by ratio of family income to federal poverty threshold
Ratio of family income to poverty threshold All children Child beneficiaries Nonbeneficiary children in beneficiary families
Below 100% of threshold 21.6 23.4 29.8
100–150% of threshold 11.8 19.3 18.4
150–200% of threshold 11.4 16.0 12.9
Above 200% of threshold 55.2 41.3 38.9
SOURCE: The data are from the annual file of those who remained in the Survey of Income and Program Participation from March 1996 through February 1997.
NOTE: The total unweighted count was 22,479 for all children, 1,090 for child beneficiaries, and 740 for nonbeneficiary children in beneficiary families.

The differences in poverty status across demographic groups are shown in Table 10. The poverty rate was markedly higher among children of retired and disabled workers than it was among children of deceased workers. The large differences in poverty rates across racial and ethnic groups found in the general population of children were also apparent among child beneficiaries and nonbeneficiary children in beneficiary families, but the gap between blacks and whites was somewhat smaller among child beneficiaries. Also, the poverty rate was actually lower among black child beneficiaries than it was among all black children. A pattern of higher poverty rates among younger children (less than 10 years of age) was evident across all three groups, but poverty was particularly severe among the youngest child beneficiaries (under the age of 5).

Table 10. Poverty status, by selected characteristics (in percent)
Characteristic All children Child beneficiaries Nonbeneficiary children in beneficiary families
By characteristics of child
Total 21.6 23.4 29.8
Benefit type a
Children of deceased workers . . . 17.9 . . .
Children of disabled workers . . . 24.7 . . .
Children of retired workers . . . 28.8 . . .
Sex
Male 21.6 24.6 29.4
Female 21.5 22.1 30.3
Race and ethnicity
White, non-Hispanic 12.5 14.6 14.0
Black, non-Hispanic 44.0 37.0 50.1
Other, non-Hispanic 22.9 43.2 14.6
Hispanic 42.2 41.7 44.0
Age
Under 5 26.6 41.5 31.8
5–9 22.4 24.2 33.4
10–14 19.0 19.7 30.5
15–17 15.8 21.1 21.2
By characteristics of family head
Sex
Male 10.2 15.6 12.7
Female 41.1 27.7 44.0
Marital status
Married 10.3 12.2 13.2
Widowed 35.2 18.6 54.4
Divorced or separated 39.2 31.4 43.2
Never married 67.5 54.8 54.7
Age
18–29 47.0 41.3 44.0
30–39 19.1 24.3 30.2
40–49 10.5 16.8 16.8
50–64 15.5 23.0 38.2
65 or older 28.8 18.5 32.0
Education
Less than high school 57.0 42.4 49.2
High school 23.4 22.6 35.2
Some college 14.8 13.5 16.8
College graduate 2.1 8.9 0.8
By characteristics of family
Number of children in family
One 20.7 20.1 23.6
Two or three 18.2 20.8 27.1
Four or more 38.5 39.9 43.6
SOURCE: The data are from the annual file of those who remained in the Survey of Income and Program Participation from March 1996 through February 1997.
NOTE: Poverty status is obtained by dividing family income by the poverty threshold. The total unweighted count was 22,479 for all children, 1,090 for child beneficiaries, and 740 for nonbeneficiary children in beneficiary families.
. . . = not applicable.
a. The breakdown by benefit type for the matched 1996 SIPP data includes only children receiving an administratively identified benefit (total unweighted count 929).

Children in female-headed families were much more likely to live in poverty than were their counterparts in male-headed families, mainly because of lower family earnings, but the poverty differential was much smaller among child beneficiaries than it was among all children or nonbeneficiary children. In fact, among children living in female-headed families, child beneficiaries were much less likely to be in poverty than were nonbeneficiary children. Looking at marital status of the family head, the data show that children with married family heads were by far the least likely to live in poverty, whether or not their families received Social Security. Among children living with widowed, divorced or separated, or never-married family heads, those who received Social Security tended to have lower poverty rates than their nonbeneficiary peers had. Children with widowed family heads in particular were better off if they received benefits themselves than they were if they lived with a beneficiary and were not receiving benefits.

Just as poverty was most common among the youngest children, it was also most prevalent among children in families with heads under the age of 30. Among these children, the poverty rate was somewhat lower among child beneficiaries and nonbeneficiary children in beneficiary families than it was among all children. Children whose family heads were age 18 to 29 had the highest rates of poverty. Poverty rates were lowest among children whose family heads were between the ages of 30 and 50, increasing again among those whose family heads were over the age of 50. Among all children, poverty was higher among children in families headed by seniors (65 or older) than it was among those whose family heads were between the ages of 50 and 64, but the opposite was true among child beneficiaries and nonbeneficiary children in beneficiary families. Not surprisingly, a child's likelihood of living in poverty was inversely related to the educational attainment of the family head. The differential was slightly smaller among child beneficiaries than it was among all children. In cases in which the family head lacked a high school degree, the poverty rate was lower among child beneficiaries compared with the rate for all children, but the opposite was true when the family head possessed a college degree. Among all three groups of children, poverty rates were highest among children living in families with four or more children.

Returning to Table 8, one can see that many of the factors discussed above were significant predictors of poverty among child beneficiaries in the multivariate context.30 The poverty rate for child beneficiaries was significantly lower among children of deceased workers than it was among children of disabled workers. It was significantly higher among black children, younger children, and children with divorced, separated or never-married family heads. The poverty rate was also higher for child beneficiaries with less-educated family heads and among those with more children in their families.

Conclusion

The findings of this study show clearly that the Social Security system is an important source of support for many children. The death or disability of a parent—the primary reasons for children's benefit receipt—often leaves families in a financially precarious position. Thus, even though its family insurance benefits are not specifically targeted toward the economically disadvantaged, Social Security provides benefits to many poor or near-poor children with limited alternative means of support. At the same time, this analysis indicates that in most cases the earnings of other family members were at least as important to these children as their families' Social Security benefits. However, these earnings were substantially lower than those of nonbeneficiary families, and in most cases Social Security benefits were not sufficient to make up the difference.

One important group highlighted in this analysis is nonbeneficiary children living in beneficiary families. These children are not identified in official program statistics, but for many of them Social Security is a crucial source of economic support. There are important differences between child beneficiaries and nonbeneficiary children in beneficiary families and between children of deceased, disabled, or retired workers. For average family income and poverty status, child beneficiaries tended to be better off economically than were children living in beneficiary families who do not receive benefits themselves, and children of deceased workers tended to be better off economically than were children of retired or disabled workers.

Notes

1. Other analyses of child beneficiaries have been conducted by the National Academy of Social Insurance (Hill and Reno 2003) and the Urban League (Rawlston 2000).

2. For details on the attainment of insured status for retirement, disability, or survivors benefits, see chapter 2 of the Online Social Security Handbook.

3. In this context, the term "eligibility" does not refer to entitlement but rather to the satisfaction of a necessary criterion for potential entitlement.

4. To maintain benefits, the 18-year-old must be in full-time attendance at an educational institution that provides courses at the secondary-school level.

5. The child's benefits will not be terminated upon marriage if the child is disabled and the child's spouse is also a Social Security beneficiary.

6. In 1996, the year of analysis for this article, the annual exempt amount was $8,280 for individuals age 64 and under and $12,500 for individuals age 65 to 69. Above that threshold, $1 of benefits was withheld for every $2 of earnings for the former group and for every $3 of earnings for the latter, with the amount withheld to be paid to the beneficiaries in months when they did not have earnings. The earnings test has since been repealed for beneficiaries who have reached the full retirement age.

7. Approximately 17.5 percent of March Survey of Income and Program Participation (SIPP) respondents did not remain in the sample for the entire annual period and were thus excluded from the annual file. The adjustment to the sample weights for the annual file was not intended to remedy the potential bias from sample attrition in the survey over the 12-month period.

8. The Monthly Benefit Credited was used to determine each individual's benefit amount, if they were in current-payment status. The Master Beneficiary Record reflects the amount the Social Security Administration computed for a month, rather than the actual payment amount received in a month. These amounts may differ because of adjustments to actual payments for past underpayments or overpayments.

9. The Supplemental Security Record provides data on all federally administered SSI payments, which include both federal payments and federally administered state supplements, but it does not capture state-administered state supplements.

10. The fact that a valid Social Security number was obtained did not ensure that it was the number belonging to the respondent. Cases in which the SIPP-reported and administratively determined age differed by more than 2 years for children and more than 6 years for adults were considered to be invalid matches for the purposes of this study.

11. The determination of benefits for children who did not have a valid Social Security number was complicated by the lack of data on individual child benefit amounts in the SIPP. In such cases, the total amount of child benefits reported for the family was divided equally among beneficiary children. (For a more detailed description of the method used to determine child benefits in these cases, contact the author at chad.newcomb@ssa.gov.)

12. The populations covered by the SIPP and the Annual Statistical Supplement are somewhat different. Although the SIPP only includes the noninstitutionalized population of the 50 states and the District of Columbia, the Supplement also includes institutionalized beneficiaries and beneficiaries in U.S. territories, as well as overseas military personnel and citizens living abroad.

13. To simplify matters, children receiving benefits on the record of a deceased worker are often referred to as "children of deceased workers" in the remainder of the text. Similarly, children receiving benefits on the record of a retired or disabled worker are referred to as "children of retired workers" and "children of disabled workers," respectively.

14. In addition, the disability benefits payable to a family may be reduced because of circumstances such as the receipt of workers' compensation payments.

15. The "Other" category of race and ethnicity includes American Indians, Aleuts, Eskimos, Asians, and Pacific Islanders. Information about these groups could not be presented separately because of sample size limitations. Since the unweighted count of child beneficiaries in the "Other" category varied between 33 and 45, depending on the variable being analyzed, the results for this group should be interpreted with caution.

16. The family head was defined as the adult in the family with the highest personal income. For 97.5 percent of children, the family head was either a parent or a guardian.

17. Although there are a number of reasons that a child in a widow(er)-headed family might not qualify for Social Security benefits, by far the most common was that the widow(er) was not the child's parent. This was true of 52 percent of children in widow(er)-headed families who did not receive benefits.

18. In cases in which there is both a parent beneficiary and another adult beneficiary, the parent beneficiary is always selected as the primary adult beneficiary. The primary adult beneficiary may be a different person than the family head whose characteristics were described above.

19. These calculations only include children with an administratively identified benefit and at least one parent providing a valid Social Security number (n = 413).

20. Since data on benefit type were available only from administrative sources, these results include only child beneficiaries in families in which at least one adult had an administratively identified benefit.

21. Here the sample is restricted to nonbeneficiary children whose families included at least one adult with an administratively identified benefit.

22. Income shares are calculated as averages across individuals rather than as percentages of aggregate income. This approach gives equal weight to each child rather than representing children in proportion to their family income.

23. Current levels and patterns of health insurance coverage for children may differ from those described here because of the substantial changes that have taken place in the intervening years. Of particular significance is the enactment in 1997 of the State Children's Health Insurance Program (S-CHIP); this program allows states to expand health insurance coverage to more low-income children, either through Medicaid or a separate state program.

24. A child beneficiary can receive private insurance coverage in a number of ways: through the employer-provided insurance of a family member who is working, through the individual insurance market, or through continuation of coverage under the employer-based plan of a deceased or disabled parent (according to the provisions of the Consolidated Omnibus Budget Reconciliation Act of 1985).

25. These observations are based on differences in the overall distribution of Social Security as a share of family income, not on differences in the share of children receiving 100 percent of their income from Social Security.

26. Since type of benefit is included in the model, the sample is restricted to child beneficiaries identified through an administrative match. Similar results were obtained for all child beneficiaries when benefit type was omitted from the model. The income-to-poverty ratio was not included in the model because of endogeneity.

27. A logit model was also specified predicting whether a child beneficiary received more than half of his or her family income from Social Security. The results were quite similar to those of the ordinary least squares regression model presented in Table 8.

28. This calculation was performed using the 12-month total of monthly family incomes and poverty thresholds, accounting for any changes in family membership over the 12 months.

29. The results presented here are not representative of current conditions. Census Bureau figures indicate that child poverty declined substantially during the late 1990s, especially among minority children.

30. For this model, logistic regression was used rather than ordinary least squares regression. As with the model for reliance on Social Security, the sample was restricted to administratively identified child beneficiaries. Here again, inclusion of all child beneficiaries in the model had virtually no effect on the results.

References

Diez, Greg. 2000. Earnings of Black and Nonblack Workers Who Died or Became Disabled in 1996 and 1997. Research and Statistics Note 2000-01. Washington, DC: Social Security Administration, Office of Policy, Division of Earnings Statistics and Analysis. Available at http://www.ssa.gov/policy/docs/rsnotes/rsn2000-01.pdf.

Hill, Catherine, and Virginia Reno. 2003. "Children's Stake in Social Security." Social Security Brief No. 14. Washington, DC: National Academy of Social Insurance. Available at http://www.nasi.org/publications2763/publications_show.htm?doc_id=146170.

Huynh, Minh, Kalman Rupp, and James Sears. 2001. The Assessment of Survey of Income and Program Participation (SIPP) Benefit Data Using Longitudinal Administrative Records. 2001 FCSM Conference Papers. Washington, DC: Office of Management and Budget, Federal Committee on Statistical Methodology Research. Available at http://www.fcsm.gov/events/papers2001.html.

Kearney, John R., Herman F. Grundmann, and Salvatore J. Gallicchio. 1995. "The Influence of OASDI and SSI Payments on the Poverty Status of Families with Children." Social Security Bulletin 58(3): 3–18.

_______. 1994. "The Influence of Social Security Benefits and SSI Payments on the Poverty Status of Children." Social Security Bulletin 57(2): 27–43.

Liebman, Jeffrey B. 2001. Redistribution in the Current U.S. Social Security System. NBER Working Paper No. 8625. Cambridge, MA: National Bureau of Economic Research.

Lleras-Muney, Adriana. 2002. The Relationship Between Education and Adult Mortality in the United States. NBER Working Paper No. 8986. Cambridge, MA: National Bureau of Economic Research.

Rawlston, Valerie A. 2000. The Impact of Social Security on Child Poverty. Washington, DC: National Urban League, Research and Policy Department.

Social Security Administration. 2003a. OASDI Monthly Statistics, December 2003. Available at http://www.socialsecurity.gov/policy/docs/statcomps/oasdi_monthly/index.html.

_______. 2003b. Online Social Security Handbook. Available at http://www.socialsecurity.gov/OP_Home/handbook/ssa-hbk.htm.

_______. 1996. Annual Statistical Supplement to the Social Security Bulletin. Washington, DC: Social Security Administration, Office of Policy.

U.S. Census Bureau. 1996. Population Projections of the United States by Age, Sex, Race, and Hispanic Origin: 1995 to 2050. PD25-1130. Washington, DC: U.S. Government Printing Office.