Characteristics of Noninstitutionalized DI, SSI, and OASI Program Participants, 2016 Update

Introduction

The Social Security Administration (SSA) produces several statistical publications based on the data used to administer the Old-Age, Survivors, and Disability Insurance (OASDI) and Supplemental Security Income (SSI) programs. Although these data are extensive, they do not capture many of the economic and demographic characteristics of program participants. To better understand those beneficiary populations, SSA matches information from its administrative records with data collected in surveys conducted by the Census Bureau and presents the results in notes such as this. DeCesaro and Hemmeter (2008) contains tables describing the characteristics of SSI and Social Security Disability Insurance (DI) participants based on 2002 Survey of Income and Program Participation (SIPP) data. Bailey and Hemmeter (2014) updates those tables with 2010 SIPP data, and Bailey and Hemmeter (2015) updates them with 2013 SIPP data. This note updates the tables for 2016 using data from the March 2017 Annual Social and Economic Supplement (ASEC) of the Current Population Survey (CPS). It expands on previous notes by also including statistics on Social Security Old-Age and Survivors Insurance (OASI) beneficiaries.

Data and Methodology

The CPS is a household survey of the noninstitutionalized resident population of the United States. The ASEC focuses specifically on various sources of household, family, and individual income. We use CPS ASEC data for this update for two reasons. First, the CPS ASEC has a larger sample than the SIPP does. This larger sample size allows us to report statistics for racial groups and other subpopulations that could not be covered in previous versions of this note. Second, the CPS ASEC is collected on an annual basis with relatively few changes between survey years, which allows us to compare beneficiary characteristics across single years.

We match 2017 CPS ASEC data to Social Security administrative records for 2016 based on the respondent's validated Social Security number. We include only individuals with matched records, who constitute roughly 85 percent of the CPS ASEC sample. We then adjust the survey weights to match our subsample to the full survey population. Using logit modeling, our approach identifies a combination of observable factors that predict whether a survey respondent has an administrative-data match in SSA's Numerical Identification System (Numident) file.

We run a series of logit models, the most robust of which includes age, race/ethnicity, nativity, education, family income, and the householder's marital status as predictive factors. Using this model, we assign a propensity score to each CPS ASEC respondent that indicates the likelihood, based on that individual's set of observable characteristics, that the Numident contains a matching record. We multiply the survey weights of individuals with matching Numident data by the inverse of this propensity score, so that individuals with lower propensity scores receive higher adjustments. For instance, individuals with lower levels of education or who were born outside the United States are found to be less likely to have matching administrative records and thus receive a higher adjusted weight. The interquartile range of the adjusted survey weights is from 109 percent to 119 percent of the original survey weights. This approach produces Social Security beneficiary and SSI recipient counts in line with administrative records for noninstitutionalized populations. Unlike the weights in previous versions of this note, ours are not adjusted to account for Social Security beneficiaries or SSI recipients residing in institutions.

We identify individuals as Social Security beneficiaries and SSI recipients if administrative records indicate that they received a payment in any month in 2016. To identify Social Security beneficiaries, we use SSA's Master Beneficiary Record (MBR) and Payment History Update System (PHUS). For this note, if an individual has an OASDI payment recorded in the PHUS and the MBR indicates that the person qualifies for the benefit because of disability, we classify that individual as a disabled Social Security (DI) beneficiary. This definition includes disabled workers, disabled widow(er)s, and disabled adult children.¹ If the MBR indicates that an individual qualifies for an OASDI benefit based on age rather than because of a disability, we classify that person as an aged Social Security (OASI) beneficiary. That sample includes covered workers aged 62 or older, individuals aged 62 or older who are eligible as the spouse or surviving parent of a covered worker, and individuals aged 60 or older who are eligible as the widow(er) of a covered worker; it excludes individuals who receive Social Security benefits as dependents or survivors of covered workers but are not in those age groups.

In this note, Social Security payments for 2016 reflect the sum of all payments recorded in the PHUS for that year. In some cases, disabled beneficiaries receive lump-sum retroactive DI payments covering the period between the first month they were eligible for payments and the month benefit payments commenced after the disability determination process. Because these retroactive payments generally occur only once during the lifetime of a disabled beneficiary, they do not reflect the monthly Social Security income that beneficiaries typically receive. We account for these retroactive lump sums by censoring monthly OASDI payments at the 99^th percentile for that month, which averaged $7,400 in 2016.²

To identify SSI recipients, we use the Supplemental Security Record (SSR). For individuals who received any SSI payments in 2016, we sum the monthly amounts to calculate the annual SSI amount received. Some disabled and aged Social Security beneficiaries—15.5 percent and 2.6 percent, respectively—receive concurrent SSI payments. Thus, our sample includes overlap between SSI recipients and Social Security beneficiaries.

We measure family income and poverty for calendar year 2016. We define “family” as two or more persons residing together who are related by blood, marriage, or adoption. When counting family Social Security income and SSI payments, we use administrative data values for all family members with matching records. For family members who do not have an administrative-data match, we use the income amounts they reported in the CPS ASEC. Likewise, for earnings, we use administrative records from SSA's Master Earnings File for individuals with an administrative-data match, and self-reported earnings in the CPS ASEC for individuals without a match. Where we find administrative-data matches for respondents' family members, we recalculate family income by subtracting self-reported OASDI benefit amounts, SSI payment amounts, and earnings, then substituting the respective values from their administrative records.³ We base all other demographic characteristics on self-reported CPS ASEC data for March 2017, the month the survey was conducted.

Although most of the variables in the CPS ASEC mirror those from the SIPP, there are some differences relating to public assistance. For instance, 44 states and the District of Columbia supplement the federal payments for SSI recipients. SIPP respondents report federal SSI payments and state SSI supplementation payments separately, while the CPS questionnaire does not differentiate between the two payment types. Previous versions of this note define SSI as federal payments only and classify state SSI supplementation payments as a separate form of public assistance. This update instead includes both federal and state supplementation payments in the computation of SSI amounts received.

Of the states that provide SSI supplementation, some administer those payments themselves and others agree to have SSA administer them.⁴ For the latter, we use the state supplementation amounts recorded in the SSR. However, the SSR does not record the supplementation amounts for the states that administer their own payments. Where possible, we impute those state supplementation payments based on SSI recipient characteristics such as living arrangement and dependent status. Using the most recent documentation of state supplementation rules and payment amounts available at SSA and state agency websites, we approximate a recipient's eligibility for state SSI supplementation and the monthly amount that would be received. Appendix A describes our imputation process.

Other forms of public assistance measured in the CPS include Temporary Assistance for Needy Families (TANF); Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) benefits; Supplemental Nutrition Assistance Program (SNAP) benefits (formerly known as food stamps); state and local general assistance; financial assistance from individuals; energy assistance; public housing; rental subsidies; and other welfare.⁵ Consistent with the approach used in previous versions of this note, we recode the CPS Medicaid variable to reflect the automatic Medicaid coverage of SSI recipients in certain states under Section 1634 of the Social Security Act.⁶ In this version of the note, we also recode the CPS Medicare variable to include both self-reported Medicare coverage (of any kind) and Medicare Part B coverage reflected in the PHUS. Recoding these variables using administrative records may reduce the effects of underreported insurance coverage (or misreported coverage type) in the CPS ASEC.⁷

Our samples comprise 4,670 disabled Social Security beneficiaries, 3,931 SSI recipients, and 20,952 aged Social Security beneficiaries. The disabled Social Security beneficiary and SSI samples are nearly twice the size of those covered in previous notes because the CPS sample is larger than the SIPP's. As noted earlier, we use survey weights adjusted for the likelihood that a respondent has a matching record in the administrative data. All household surveys, including the CPS ASEC, are subject to some degree of measurement error; however, the CPS ASEC is a source of valuable data. It provides richly detailed information on population characteristics and, despite deficiencies in measuring some income sources, it captures most sources, including earnings and Social Security benefits, with reasonable accuracy. Readers should nevertheless be aware of the limitations of CPS data and consider them when interpreting results.

Discussion of the Estimates

Tables 1–5 present estimates for disabled Social Security beneficiaries, Tables 6–10 present estimates for SSI recipients, and Tables 11–12 present estimates for aged Social Security beneficiaries. Each table presents estimates for the full noninstitutionalized population of disabled Social Security beneficiaries, SSI recipients, or aged Social Security beneficiaries, as applicable.

Although the tables in this update resemble the tables in previous editions of this note in structure, they differ in six significant ways. First, as noted above, this update introduces new tables on aged Social Security beneficiaries (Tables 11 and 12). However, because of concerns about the accuracy of survey-reported income for aged respondents, the new tables omit estimates of income by source and income relative to the poverty threshold. Bee and Mitchell (2017) and Dushi and Trenkamp (2021) show that the CPS ASEC consistently understates the retirement income (comprising pension benefits and retirement account distributions) of aged respondents. In both studies, the authors link the CPS ASEC data with administrative records from SSA and the IRS, with which they validate data for income sources including earnings, Social Security benefits, SSI payments, interest and dividend income, and retirement income. Both studies find significant underreporting of retirement income among individuals aged 65 or older across a wide range of demographic characteristics. Retirement income is an important component of overall income for aged Social Security beneficiaries, and because it is underreported in the CPS ASEC, we omit statistics on income amounts and poverty for this population.

Second, estimates of the shares of income by category now include detail for retirement income and veterans' benefits for disabled DI beneficiaries (Table 4) and SSI recipients (Table 9). In previous editions of this note, those sources were included in the “other” income category. Because retirement income is not a primary component of the family income of disabled beneficiaries and SSI recipients, its inclusion does not subject the estimates to the same kind of uncertainty that leads us to omit estimates by income category for aged OASI beneficiaries.

Third, this update omits tables addressing the poverty-reduction effects of Social Security benefits and SSI payments (Tables 6A, 6B, and 7 in Bailey and Hemmeter 2015).

Fourth, this update expands the number of race categories from three (White, Black, other) to five: White, Black, American Indian or Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander. Because CPS respondents can select multiple racial identities, each of these categories may be reported either singly or in combination with one or more others; as such, the sum of the racial-category values exceeds the number of respondents in the sample. A sixth category, “two or more races,” accounts for individuals who reported multiple racial identities.

Fifth, this update presents estimates for family income but not for personal or household income. We focus on family income because it serves as the official basis of poverty statistics in the United States. It also accounts more fully than individual income for the resources on which Social Security program populations may draw.⁸

Finally, this update reports educational attainment only for individuals aged 25 or older to reduce the likelihood that our measures include those who have not yet completed their education.

Disabled Social Security Beneficiaries

The DI program provides monthly benefits to individuals with a severe long-term disability that prevents them from engaging in “substantial” work (defined as earnings of $13,560 or more for nonblind individuals in 2016). To qualify for DI benefits, an individual must have worked in a job with Social Security–covered earnings for a specified length of time or be the spouse or dependent of an individual with qualifying earnings. Our sample of disabled Social Security beneficiaries comprises disabled workers, disabled widow(er)s, and disabled adult children.⁹ Because 91 percent of the beneficiaries in this sample are disabled workers, Tables 1–5 present information separately for that subgroup, although we do not discuss those results in the text below.

Table 1 shows that more than one in five disabled Social Security beneficiaries (21.2 percent) identify as Black, 3.4 percent identify as American Indian or Alaska Native, 2.1 percent identify as Asian, 0.5 percent identify as Native Hawaiian or Other Pacific Islander, and 2.5 percent identify as multiracial. The CPS measures ethnicity separately from race, and 11.7 percent of beneficiaries identify as Hispanic. Nearly two-fifths of beneficiaries (38.7 percent) are married, 28.7 percent have never been married, 26.0 percent are divorced or separated, and 6.6 percent are widowed. Fewer than one-fifth of beneficiaries (18.1 percent) aged 25 or older have not acquired a high school diploma or equivalent, while 11.1 percent of beneficiaries have a bachelor's or higher degree. Nearly one in ten beneficiaries (9.2 percent) are veterans.

Although disabled Social Security beneficiaries are eligible for Medicare, coverage generally does not begin until 24 months after the first month that an individual is eligible to receive DI benefits. Roughly three-quarters (75.7 percent) of beneficiaries report Medicare coverage (alone or in combination with other coverage), 44.2 percent report Medicaid coverage, 31.4 percent report private health insurance, and 1.9 percent report having no health insurance (Chart 1). About 29 percent of beneficiaries report joint Medicare and Medicaid coverage and 12.2 percent have only Medicaid coverage (Table 1).

Family income includes SSI payments for about one in five disabled Social Security beneficiaries. Other common sources of family income include earnings (50.6 percent) and property income (42.3 percent).¹⁰ Less common sources are retirement income (16.0 percent) and veterans' benefits (4.4 percent).

DI benefits constitute at least 75 percent of family income for almost two-fifths (37.9 percent) of disabled beneficiaries (Table 2). Social Security tends to account for larger proportions of family income among older beneficiaries. For instance, roughly one-third of beneficiaries aged 60 or older rely on Social Security benefits for at least 90 percent of family income, while 16.2 percent of beneficiaries younger than 40 do. Social Security benefits also tend to account for larger proportions of family income among beneficiaries who did not complete high school, are not married, or do not live with relatives.

Almost half (47.8 percent) of Social Security disability beneficiaries have annual family income of less than $30,000, while 60.4 percent are in homeowning families (Table 3). About 28 percent of beneficiary households receive food assistance through SNAP benefits. More than one in ten receive housing or energy assistance—10.8 percent and 10.6 percent, respectively. About two-thirds of disabled Social Security beneficiaries live with other family members, although only 17.2 percent live with a family member younger than 18. Two in five DI beneficiaries live with another Social Security beneficiary in the household.

Table 4 shows the distribution of family income by source for disabled beneficiaries. The two primary income sources are Social Security benefits (59.1 percent) and earnings (23.7 percent). Social Security benefits constitute a particularly large share of family income for disabled beneficiaries aged 60 or older (63.1 percent), beneficiaries who are not married (63.5 percent if never married, 65.9 percent if widowed, and 72.1 percent if divorced or separated), beneficiaries without a high school diploma (63.8 percent), and those who do not live with relatives (78.4 percent if living with nonrelatives only and 78.5 percent if living alone).

Although 27.8 percent of disabled Social Security beneficiaries have family income at or above 300 percent of the poverty threshold, 20.1 percent are in poverty (Table 5). Beneficiaries are more likely to have family income below the poverty threshold if they are women (23.7 percent), younger than 40 (28.6 percent), Black (27.3 percent), divorced or separated (27.8 percent), or never married (28.5 percent); or if they do not have a high school diploma (31.7 percent).¹¹ In 2016, the federal poverty threshold was $12,486 for a single adult aged 18–64 and $11,511 for one aged 65 or older (Census Bureau 2022). The proportions of beneficiaries with family income above 300 percent of the poverty threshold differ widely by educational attainment. Only 14.9 percent of beneficiaries without a high school diploma have family income exceeding that threshold, compared with 50.1 percent of beneficiaries with a bachelor's or higher degree (Chart 2).

SSI Recipients

SSI provides monthly payments to individuals with low income and few assets who have a disability or are aged.¹² To qualify for SSI payments, an individual cannot have assets exceeding $2,000, and a couple cannot have assets exceeding $3,000 (SSA 2021c). Unlike DI, the SSI program does not require recipients to have past work to qualify for payments.

In this note, the tables show results for three age groups: children (younger than 18), working-age adults (18–64), and the aged (65 or older). The maximum monthly SSI federal payment amount, called the federal benefit rate, can be reduced for individuals with countable earnings and unearned income. In 2016, the federal benefit rate was $733 per month for an individual and $1,100 per month for a couple if both qualified for SSI payments. Some states provide supplementation payments to SSI recipients in addition to federal payments.

Table 6 shows that 30.7 percent of SSI recipients identify as Black, 6.4 percent identify as Asian, 3.8 percent identify as American Indian or Alaska Native, 0.3 percent identify as Native Hawaiian or Other Pacific Islander, and 2.9 percent identify as multiracial. More than one in five (21.3 percent) SSI recipients identify as Hispanic. Recipients aged 65 or older are disproportionately more likely to identify as Asian or Hispanic. The majority of SSI recipients aged 18–64 have never married (55.6 percent), while 25.1 percent are divorced or separated, 14.6 percent are married, and 4.8 percent are widowed. Slightly more than 2 percent of adult SSI recipients are veterans.

Most SSI recipients (93.5 percent) report having Medicaid insurance, while 40.5 percent report having Medicare and 18.0 percent report having private insurance. The most common sources of family income are earnings (45.6 percent of SSI recipients have some family earnings income), Social Security (44.4 percent), and property income (25.6 percent). Retirement income (4.7 percent) and veterans' benefits (1.2 percent) are less common.

Chart 3 summarizes educational attainment patterns. In 2016, 51.2 percent of aged SSI recipients did not have a high school diploma, compared with 32.2 percent of recipients aged 25–64.

Table 7 shows that seven in ten SSI recipients have family income below $30,000. SSI recipients aged 65 or older are more likely to have family income below this level (79.9 percent) than are recipients aged 18–64 (69.2 percent) or recipients younger than 18 (53.8 percent).¹³ About one-sixth of SSI recipients live in public housing (16.1 percent). More than half of SSI recipients (51.7 percent) live in households receiving food assistance through SNAP, 16.8 percent live in households receiving energy assistance, and 23.3 percent receive housing assistance. Nearly two-thirds of SSI recipients (65.8 percent) live with another family member, although most adult SSI recipients (75.8 percent of those aged 18–64 and 88.6 percent of those aged 65 or older) do not live with a family member younger than 18, and very few (3.4 percent and 0.8 percent, respectively) live with a child who also receives SSI.

Table 8 shows that 29.3 percent of adult SSI recipients receive 75 percent or more of their family income from SSI payments. Recipients aged 18–64 are slightly more likely to receive three-quarters or more of their family income from SSI than are recipients aged 65 or older—30.5 percent and 26.4 percent, respectively. Recipients who are divorced or separated, and those who live alone or with nonrelatives only, are more likely than recipients with another marital status or living arrangement to rely on SSI for at least half of their family income.

Table 9 shows that the family income of all recipients consists primarily of SSI payments (44.4 percent), earnings (26.6 percent), and Social Security benefits (21.3 percent). SSI payments constitute particularly large percentages of family income for divorced or separated recipients (54.4 percent) and for those who live alone (63.5 percent) or with nonrelatives only (66.0 percent). Earnings constitute a higher share of family income for the families of child SSI recipients (46.4 percent) than for those of recipients aged 18–64 (23.9 percent). Most family income for SSI recipients aged 65 or older includes a combination of Social Security benefits (32.7 percent) and SSI payments (44.4 percent), although earnings account for another 16.9 percent of income.

Most SSI recipients aged 65 or older (60.1 percent) and aged 18–64 (51.7 percent) have family income below the poverty level, as do a substantial minority of recipients younger than 18 (37.4 percent; Table 10). Female recipients (55.2 percent) and recipients who are widowed (59.8 percent) or divorced or separated (70.5 percent) are also relatively likely to have family income below the poverty level. In a reversal of previous trends, recipients with at least some college are more likely to have family income below the poverty level than are recipients who have not attended college.

Aged Social Security Beneficiaries

OASI provides monthly benefits to individuals aged 62 or older with a history of Social Security–covered work, as well as to dependent and surviving spouses, children, and parents of covered workers. The estimates presented here are for individuals aged 62 or older who receive benefits either based on their own earnings record or as the parent or spouse of a covered worker; and individuals aged 60 or older who receive benefits as the widow(er) of a covered worker.¹⁴

Most aged Social Security beneficiaries (93.5 percent) report Medicare coverage either alone or in combination with other health insurance (Chart 4). Very few beneficiaries report being uninsured (0.8 percent), 52.6 percent report private coverage alone or in combination with other coverage, and 7.2 percent report Medicaid alone or in combination with other coverage. Coverage from Medicare alone is reported by 40.2 percent of aged Social Security beneficiaries (Table 11). Only 1.2 percent of respondents report both Medicaid and private health insurance coverage.

Most aged Social Security beneficiaries live in a two-person household (55.4 percent, Table 12), followed by single-person households (29.3 percent). Homeownership for aged beneficiaries is high, with 82.4 percent residing in owned homes. In addition, the prevalence of household assistance is generally low, with the most common form of assistance being SNAP benefits, reported by 7.2 percent of beneficiaries.

Conclusion

This note updates Bailey and Hemmeter (2015), Bailey and Hemmeter (2014), and DeCesaro and Hemmeter (2008). However, this update differs from earlier versions in a few critical ways. The estimates presented here are based on data from the CPS ASEC, whereas prior versions used the SIPP. Additionally, this note uses survey weights adjusted for the likelihood that an individual survey respondent has matching administrative data in SSA records, which differ slightly from the weights used in prior versions. This note also expands the coverage of previous versions by including statistics on aged Social Security (OASI) beneficiaries in addition to disabled Social Security (DI) beneficiaries and SSI recipients. Although the CPS-based statistics presented here are in some cases not directly comparable to the SIPP-based estimates in prior versions, they provide similar information. Moreover, this update provides a broader context for understanding beneficiary populations by including estimates for OASI beneficiaries.

Appendix A: Imputing State-Administered SSI Supplementation Payments

We use administrative data from the SSR to identify SSI recipients and to determine whether a recipient's spouse was also an SSI recipient in a given month. We then use data from the CPS ASEC to identify the characteristics relevant to whether an SSI recipient qualifies for a state-administered supplementation payment: whether the recipient has a spouse who also receives SSI in that month, and the recipient's dependent and householder statuses. Based on these variables and documentation from SSA and state agencies, we identify recipients who are likely to be eligible for a state-administered supplementation payment and impute a proxy monthly payment amount. When state supplementation applies to couples in which both spouses receive SSI, we impute half the supplementation payment amount to each spouse.

Table A-1 shows the criteria under which SSI recipients are likely to be eligible for state-administered supplementation payments and the proxy monthly supplementation amount. In some states, SSI supplementation payments are available only to individuals residing in institutional settings. Because the CPS ASEC sample does not include these populations, we omit those states from our proxy imputations.¹⁵ In some other states, the CPS ASEC does not include all the variables needed to determine whether an individual qualifies for a supplementation payment; we omit those states from our imputations as well.¹⁶ A further limitation of this imputation approach is that although data from the CPS ASEC provide an individual's marital status and living arrangement as of March 2017, our calculations assume that this information also applied in each month of 2016.