Changing Stays? Duration of Supplemental Security Income Participation by First-Time Child Awardees and the Role of Continuing Disability Reviews

Introduction

The growing long-term role of the Supplemental Security Income (SSI) program in serving children in low-income families has been the subject of intense policy interest (National Academies of Sciences, Engineering, and Medicine 2018). SSI, which is administered by the Social Security Administration (SSA), provides cash payments to eligible low-income children and working-age adults with disabilities—and to aged persons—who meet certain income and asset criteria.

Although program enrollment has declined in recent years, the number of children who receive SSI has risen dramatically since 1996 despite the absence of major changes in eligibility criteria. The demographic, impairment, and geographic characteristics of these children have also changed, signaling a possible shift in how SSI serves children in low-income families. The increase in the SSI child caseload may indicate a growing role for the program in supporting youths with disabilities, a role potentially magnified by program participation that can continue over long periods. For example, previous studies of earlier cohorts of child SSI awardees show that payment receipt can last well into adulthood (Rupp, Hemmeter, and Davies 2015; Davies, Rupp, and Wittenburg 2009).

A better understanding of the growing role of SSI in supporting children in low-income families depends on knowing how the experiences of recent SSI awardees compare with those of previous awardees. Prior research has documented some shifts in length of program participation, but it has not compared cohorts of awardees over this period of large caseload growth.

Another key to understanding SSI's long-term role is to know the extent to which administrative processes might affect the length of program participation. Children must meet strict disability, income, and asset criteria to enroll in SSI and, later, to continue receiving payments. After an SSI award, SSA is required to conduct continuing disability reviews (CDRs) to verify the ongoing eligibility of child recipients. CDRs are meant to occur at varying intervals, depending on the “diary” that SSA creates for the recipient, which is based on the individual's likelihood of medical improvement. If medical improvement is expected, SSA generally conducts a CDR within 6 to 18 months of award. If a child's impairment is considered nonpermanent and medical improvement is deemed possible, SSA generally conducts a CDR every 3 years. For children whose impairment is not expected to improve, SSA is supposed to conduct CDRs at least every 7 years.

Some child recipients receive an SSI award for low birth weight. In those cases, SSA is required to conduct a special low birth weight CDR when the recipient reaches 12 months of age. If SSI eligibility continues after a low birth weight CDR, the child is subject to the other, ongoing childhood CDRs described above.¹ The number of other childhood CDRs SSA conducts varies over time, depending on caseload size, administrative priorities, and budgets.²

In addition, eligibility redeterminations are required for all children who are still receiving SSI when they reach age 18. Not surprisingly, children whose eligibility continues into adulthood after an age-18 redetermination have much longer average stays in the program than those whose participation ceases at age 18 (Hemmeter, Mann, and Wittenburg 2017).

Since SSA initiated child CDRs in fiscal year 1994, fluctuations in their annual volumes have been substantial. At times, SSA has focused on conducting CDRs as part of a broader program-integrity initiative. For example, in fiscal year 1999, SSA conducted more than three times as many CDRs (including age-18 redeterminations) as it did in fiscal year 2006. In the last few years, SSA has again substantially increased the number of child CDRs and age-18 redeterminations—particularly since fiscal year 2015, when the number first exceeded 300,000 per year (SSA 2019a).

This article provides new evidence on the changing role of SSI for children in low-income families since 1997. We present findings for three cohorts of first-time child awardees (1997, 2007, and 2012) for which we have at least 5 years of follow-up data to measure program outcomes; namely, average periods of participation—hereafter, “program stays” or “payment durations”—and average cumulative payment amounts. The 1997 cohort represents the first group of child SSI awardees who were subject to the current SSI eligibility criteria through their entire potential tenure in the program. The 2007 and 2012 cohorts represent SSI child recipients whose awards occurred after the major growth in child caseloads had begun. Hence, comparing these latter cohorts to the 1997 cohort will provide evidence on how compositional changes might affect average program stays. For each cohort, we track outcomes through 2017, and we examine whether payment duration varies by selected demographic, impairment, and program-participation characteristics. We also show how program stays change for youths after their age-18 redetermination to illustrate the experiences of children receiving SSI into adulthood.

We find substantive variation in the payment-duration trends for first-time SSI child awardees. The average durations for the 2007 and 2012 cohorts were much longer than that of the 1997 cohort, underscoring the growing importance of SSI's long-term role for low-income families. Cross-cohort differences in program stays occurred among all demographic and impairment groups, suggesting that changes in cohort composition cannot explain the findings. CDRs conducted during childhood were an important factor in these long-term program dynamics, as members of the 1997 cohort faced CDRs more frequently than the later cohorts did. We find that removing SSI recipients from the rolls at earlier ages because of more frequent CDRs might explain as much as half of the total differences between cohorts. Although program stays have risen over the long term, the recent increases in CDRs, particularly since 2015, might reverse this trend for future cohorts.

Background

This section provides information about SSI eligibility requirements, discusses changes in the SSI caseload, and highlights related literature on children's program stays. This information provides context on the factors that could drive differences in the length of program participation between the awardee cohorts we studied.

Since 1996, the child SSI caseload has grown dramatically despite the absence of changes in eligibility requirements. It reached a peak of over 1.32 million children in 2013, compared with about 880,000 children in 1997. The caseload has since declined, reaching 1.15 million in 2018 (SSA 2019a). Despite the recent decline, the SSI child caseload rose by 30 percent from 1997 to 2018.

The factors driving this growth are not well understood, though there is strong evidence of changes in both the impairment distribution and the geographic composition of the caseload (Government Accountability Office 2012; Aizer, Gordon, and Kearney 2013). Since 1997, the number of youths diagnosed with mental disorders has greatly increased (National Academies of Sciences, Engineering, and Medicine 2015). The growth in the caseload also has varied by state; Wittenburg and others (2015) found that more than half of this growth was concentrated in four large states: Texas, Pennsylvania, Florida, and California. Schmidt and Sevak (2017) found that substantial variations in state-level factors such as poverty rates also might have influenced these trends.

Initial SSI Eligibility Requirements

To qualify for SSI payments, a child must meet eligibility criteria related to disability, income, and assets. To meet the disability criteria, the child must be younger than 18 and have

a medically determinable physical or mental impairment which results in marked and severe functional limitations, and which can be expected to result in death or which has lasted or can be expected to last for a continuous period of not less than 12 months (42 U.S.C. §1382c[C][i]).

The nonmedical criteria consist of limits on total income and assets, including the child's income and any parental income and resources “deemed” to the child (that is, treated as the child's own).

Children who qualify for SSI are eligible for a cash payment. In 2020, the federal maximum SSI payment was $783 per month, and 23 states provided an optional supplemental payment to children with disabilities.³ Most child SSI recipients are also automatically eligible for Medicaid. In addition, their limited incomes indicate that many of these youths live in families eligible for other means-tested supports, such as the Supplemental Nutrition Assistance Program (Romig 2017; Bailey and Hemmeter 2015).

Although the SSI medical eligibility requirements for children changed significantly in the years after the program's inception in 1974, there have been no major changes since 1996. Berkowitz and DeWitt (2013) documented the evolution of SSI, noting how several important legislative changes and Supreme Court decisions transformed the eligibility requirements. The most recent major changes were enacted as part of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996. This legislation partially reversed changes to SSI in the early 1990s by establishing the current childhood disability definition, which introduced the concept of “marked and severe functional limitations” to replace individualized functional assessments for determining a child's eligibility. The legislation also required a redetermination of eligibility at age 18 using the adult criteria for disability.

Ongoing Eligibility Requirements: CDRs and Redeterminations

Unlike some other cash supports such as Temporary Assistance for Needy Families (TANF), SSI payments are not time-limited. Recipients can receive payments indefinitely, as long as they continue to meet the eligibility requirements. SSA reassesses the medical eligibility of recipients during CDRs, which are mandatory at 12 months of age in most cases for which low birth weight was a factor in the SSI award and at various regular intervals during the program stay, depending on the recipient's diary type; and during age-18 redeterminations, for all children still receiving SSI at that point. As noted above, in the age-18 redetermination, SSA uses the adult eligibility criteria.⁴

The frequency of childhood CDRs other than those for low birth weight depends on funding availability and other factors. The principal factor is the recipient's likelihood of medical improvement. SSA categorizes cases based on whether medical improvement is expected, possible, or not expected, depending on the type of impairment. SSA prioritizes cases with expected medical improvement for reviews. Unlike the age-18 redetermination—for which SSA assesses the disability anew—CDRs require medical improvement to have occurred since the last favorable eligibility decision before SSA can remove the child from the program for medical reasons.

The number of CDRs SSA conducts has varied substantially since 1996, which may have affected program stays. Chart 1 shows large fluctuations in child CDRs, driven primarily by wide swings in the number of other childhood reviews (those not tied to low birth weight or reaching age 18). Notably, from fiscal year 2005 to fiscal year 2014, these CDRs dropped below 100,000 per year, with fewer than 10,000 (representing less than 0.1 percent of children receiving SSI) in fiscal years 2007 and 2008. However, the incidence of other childhood CDRs has recently increased sharply, to more than 200,000 per year since fiscal year 2015, representing about 16 percent of children receiving SSI each year (SSA 2019a). Below, we show that the likelihood of cessation of SSI disability payments after a recipient's first-time award differs meaningfully by the year of award, which could drive cross-cohort differences in the average length of program participation.

Administrative funding levels affect SSA's ability to conduct CDRs. For example, in its fiscal year 2015 annual report on CDRs, SSA (2019b) noted fluctuations in the number of CDRs over time, with the agency completing more cases in recent years because Congress provided additional funding for them. The historical fluctuation in CDRs is notable because although SSA schedules CDRs at regular intervals, the ability to conduct all scheduled CDRs depends on capacity. Times of high demand for administrative staff time, such as periods with increased application volume, can lower that capacity.

The fluctuations in CDRs have important implications for SSA's age-18 redetermination volume and outcomes. Specifically, the number of CDRs that recipients undergo before age 18 can affect the composition of the caseload that remains on the rolls at age 18. With SSA conducting relatively few other childhood CDRs from fiscal year 2005 to fiscal year 2011 (Chart 1), many child SSI recipients in that period did not have their cases reviewed until age 18; correspondingly, the cessation rate resulting from age-18 redeterminations rose over that period (Chart 2).

In addition, a recent proposed policy change could make CDRs even more frequent for some SSI recipients. As mentioned earlier, SSA uses three CDR diary types—Medical Improvement Expected, Medical Improvement Possible, and Medical Improvement Not Expected—to schedule CDRs. In November 2019, SSA proposed adding a new category—Medical Improvement Likely—to this list.⁵ If implemented, this policy change would alter the frequency of CDRs. Recipients in the Medical Improvement Likely category would have reviews scheduled more often than those classified as Medical Improvement Possible or Not Expected, but less often than those classified as Medical Improvement Expected. This policy proposal brings to the forefront the importance of analyzing patterns in CDRs.

Disability Program Interactions

Another factor that could affect SSI program stays is the possible concurrent receipt of Social Security Disability Insurance (DI) benefits. The DI medical eligibility criteria are the same as those for adult SSI recipients, although DI calculates benefit amounts based on previous covered earnings, with higher earnings leading to higher benefits. A child SSI recipient can receive concurrent DI benefits as the dependent or survivor of a DI beneficiary. In addition, unmarried adults who experience disability onset before age 22 can become eligible for DI as “disabled adult children” if at least one of their parents qualifies for DI or Old-Age and Survivors Insurance benefits, although this is less common.⁶ The potential for concurrent receipt of SSI and DI benefits is important because DI benefits can provide additional income to current and former SSI recipients and, equally important, can provide access to Medicare.

Related Studies on Program Stays

Several papers have examined SSI program stays, with many of them using administrative data from SSA to assess outcomes for different subgroups of awardees. Rupp and Scott (1995, 1998) provided one of the earliest and most comprehensive analyses of length of program participation, examining 1974–1982 first-time SSI awardees. They found that 36 percent of SSI recipients who were aged 0–17 at award stayed continuously on the rolls for at least 10 years after first award. Rupp and Scott also used projections to simulate lifetime program stays for first-time awardees. They estimated that child SSI awardees would remain on the rolls for an average of 27 preretirement-age years.

Davies, Rupp, and Wittenburg (2009) compared program stays for 1980 and 1997 SSI child awardees to examine how durations for those cohorts differed over time. They found substantial differences between the two cohorts, particularly at age 18, which is consistent with the introduction of the age-18 redetermination in 1996. In both cohorts, children who stayed in the program for 5 years were likely to stay on in adulthood. The authors argued that a lifecycle framework is needed to examine the long-term program outcomes of children receiving SSI.

Rupp, Hemmeter, and Davies (2015) extended the previous studies by explicitly modeling the interaction between SSI and DI in accounting for total duration of participation in both programs. Their analysis included cohorts of child SSI awardees from 1980 through 2000. They found that a large share of former child SSI recipients receives either SSI or DI, which the authors classified under a definition of “any disability benefit” receipt. As we describe in more detail below, we adopt that definition for this article.

Rupp, Hemmeter, and Davies (2015) also found that accounting for DI participation is important because it raises the observed rate of participation in either or both of SSA's disability programs, especially as recipients reach adulthood. The authors also found differences in program stays between the cohorts from 1980 through 2000, which likely reflects the major changes to SSI program rules that occurred in 1996 and prior years, noted above. Importantly, a nontrivial portion of child SSI awardees died, although mortality generally decreased across successive cohorts. Rupp, Hemmeter, and Davies found suggestive evidence that program stays initially increased after the 1996 welfare reform. For example, they found that the percentage of 2000 child awardees receiving benefits 10 years after their first award date was higher than that of 1997 awardees (51 percent versus 46 percent).

Besides these studies, several other papers have examined long-term changes in length of program participation stemming from SSA policies. Hemmeter, Mann, and Wittenburg (2017) tracked the outcomes of child SSI recipients from their age-18 redeterminations through age 24. Not surprisingly, the authors found that the rates of receipt of any SSA disability-program benefits were much lower among recipients whose eligibility ceased after the age-18 redetermination than for continuing recipients (18 percent versus 86 percent). Likewise, Deshpande (2016) used administrative data from SSA to track long-term outcomes (well into adulthood) of former child SSI recipients whose payments ceased after an age-18 redetermination. The author found that most youths who are removed from the SSI rolls have low earnings and minimal earnings growth over time. She projected that an 18-year-old removed from SSI, relative to one who remained on the rolls, would have lower annual SSI payments (by $7,900), lower annual DI benefits (by $600), and higher annual earnings (by $3,000). In addition, Levere (2019) looked at long-term patterns in SSI receipt for cohorts of child SSI awardees who began receiving payments after the 1991 relaxation of medical eligibility rules that stemmed from the 1990 Sullivan v. Zebley Supreme Court decision. He found that those who receive payments for a longer time in childhood have longer periods of SSI receipt during adulthood, suggesting that long-term payment receipt tends to persist over time.

Our article builds on this literature in several ways. First, we include more recent cohorts of child SSI awardees (2007 and 2012), which enables us to observe whether program stays identified in Rupp, Hemmeter, and Davies (2015) continued to change in the period of rapid caseload growth and during the Great Recession. Second, we examine whether the large changes in the number of CDRs affected the duration of payment receipt. This analysis is particularly important for assessing whether the results from previous cohorts could help to predict outcomes for current cohorts. Finally, we estimate program stays by recipient characteristics to assess how changes in the composition of the cohort might affect payment duration overall. We also use regression models to assess whether these changes can explain differences between cohorts.

Data and Methods

We use administrative records from SSA to identify first-time awardees and track their SSI and DI participation histories, including their CDRs and age-18 redeterminations (if applicable). We use SSA's primary systems for tracking benefits: the Supplemental Security Record for SSI payments and the Master Beneficiary Record and Payment History Update System for DI benefits. We track benefit receipt (durations and total amounts received) for all beneficiaries in this sample until 2017, and we inflation-adjust all amounts to 2017 dollars using the Consumer Price Index for All Urban Workers (CPI-W).

We examine the descriptive patterns in durations and payment amounts over time for three cohorts of first-time SSI child awardees: 1997, 2007, and 2012. For each cohort, we track participation through 2017, which allows a 5-year follow-up for all cohorts and a 10-year follow-up for the 1997 and 2007 cohorts.⁷

The number of first-time SSI child awardees rose across the first three cohorts (Table 1). Growth in the number of child SSI awardees far exceeded growth in the U.S. child population; from 1997 to 2012, the number of new SSI child awards increased by about 60 percent while the U.S. child population rose by 4 percent (Federal Interagency Forum on Child and Family Statistics 2019). Table 1 also includes the 2017 cohort to show how its characteristics compare with those of the earlier cohorts, which is important for assessing whether recent compositional changes might have affected cross-cohort differences. The population of first-time child awardees increased by 42 percent from 1997 to 2007, followed by another 14 percent increase from 2007 to 2012 and a decline of almost 15 percent from 2012 to 2017. The patterns between 2007 and 2017 likely reflect cyclical factors stemming from the Great Recession, such as lower employment and lower wages, which made more families eligible for SSI (Maestas, Mullen, and Strand 2018), and the subsequent economic restabilization.

The composition of the caseload differed across the three study cohorts, which might partly reflect changes in the medical community's use of certain diagnoses over time, such as the increase in autism diagnoses (Shattuck 2006). Compared with the 1997 cohort, the later cohorts included increasing shares of first-time child awardees diagnosed with autistic disorders, developmental disorders, and childhood and adolescent disorders not elsewhere classified. However, from 1997 to 2017, the percentage of first-time child awardees diagnosed with intellectual disability fell by nearly 80 percent, from 26.5 percent to 6.3 percent. The percentages of children with other primary diagnoses remained fairly stable over this period. In addition, the share of first-time child awardees who are male increased slightly, from 61.6 percent to 65.5 percent. In all four cohorts, about two-thirds of the first-time child awardees were younger than 8 at the time of award.

To track the duration and dollar amounts of benefits received for each cohort, we used the definitions for SSI, DI, and “any disability benefit” used in Rupp, Hemmeter, and Davies (2015). This means we report the SSI and DI benefits actually received by each person in a given month, rather than the amount due, which may vary based on factors that only become known several months or years later. Although we present findings for SSI, DI, and combined SSI and DI (any disability benefit), our analysis focuses on SSI payments. Because our statistics represent the full population of first-time child awardees, we do not present standard errors or significance tests in our descriptive comparisons.

Within each cohort, we examine whether benefit durations and amounts differ by demographic characteristics and impairment. This helps us understand whether differences in the size and make-up of cohorts shown in Table 1 contributed to the aggregate patterns of benefit receipt. Besides providing descriptive statistics, we also estimate regressions that control for the composition of the caseload, as described in our Results section. The dependent variable is either duration of stay or total payment amount, and the independent variables are the demographic and impairment characteristics in Table 1. In addition, we include cohort-specific dummies that account for differences across cohorts after controlling for demographics and impairments.

Age-18 redeterminations and CDRs can also play an important role in SSI receipt. Because the redetermination applies only to those who have reached age 18, we split our sample by age. We refer to those who reached 18 within 10 years of benefit award as “older” child recipients, which includes anyone who was aged 8 or older at the time of award, and we refer to those who did not reach age 18 within 10 years of benefit award as “younger” child recipients (aged 0–7 at award). About two-thirds of both cohorts for which we have 10 years of follow-up data are younger awardees (Chart 3).

We stratify the program-stay and benefit-amount trajectories by the result of the low birth weight CDR for younger children and by the result of the age-18 redetermination for older youths. We further divide both the younger and older groups into three subgroups, defined by their status as of the end of the observation period:

1. Youths who did not have a low birth weight CDR or an age-18 redetermination because, respectively, low birth weight was not a factor in their award or they left SSI before age 18.
2. Youths whose payments continued after the low birth weight CDR or age-18 redetermination.
3. Youths whose payments ceased because of the low birth weight CDR or age-18 redetermination.

We define the result of a low birth weight CDR or an age-18 redetermination as the final decision after all levels of appeal were completed. We use data from the Office of Continuing Disability Review Support in SSA's Office of Operations to categorize youths into each redetermination-status subgroup.

Results

In this section, we present results related to duration of payments and SSI payment amounts.

2007 and 2012 Cohorts Had Longer Benefit Duration Than 1997 Cohort

Members of the 1997 cohort received SSI payments for a shorter time and had lower total payment amounts in the 5 years after first award than the members of the later cohorts (Table 2). In the 1997 cohort, the average duration of payments among all first-time child SSI awardees was 44.7 months, about 5 months shorter than that of the 2007 cohort (50.1 months) and the 2012 cohort (48.9 months). Similarly, the average cumulative SSI payment amounts were more than $2,000 lower for members of the 1997 cohort than for those in the 2007 and 2012 cohorts ($31,911 versus $34,773 and $34,156, respectively). This relationship did not change when we added DI benefits to account for all disability-program benefits paid (only a small share of each cohort also receives DI benefits).

The 2007 and 2012 cohorts have similar program stays and cumulative payment amounts, particularly when contrasted with the 1997 cohort, suggesting that the Great Recession did not meaningfully affect patterns of longer-term payment receipt by cohort.⁸ Thus, for brevity, the rest of this section focuses on the comparison between the 1997 and 2007 cohorts 10 years after the initial award. For completeness, the appendix presents tabulations showing analogous results for 5 years after award for all three cohorts.

After 10 years, the relative and aggregate differences in payment durations and amounts between the 1997 and 2007 cohorts increased (Table 2). SSI payment duration was 10.5 months shorter for members of the 1997 cohort than for those in the 2007 cohort (74.8 months versus 85.3 months). Furthermore, SSI payment amounts were $6,714 lower for youths in the 1997 cohort than for those in the 2007 cohort ($51,397 versus $58,111). These differences underscore the importance of examining long-term outcomes, given the lengthy program stays of child SSI recipients.⁹ For example, if the 158,534 awardees in the 2007 cohort had the same SSI payment durations that those in the 1997 cohort did, the total payments over that period would have been about $1 billion lower (158,534 × $6,714).

The following subsections explore possible reasons for the differences in payment amounts received between cohorts. First, we assess whether the different composition of the cohorts, shown in Table 1, is an important factor. Next, we explore whether the youths in the 1997 cohort were more likely to exit the program before reaching age 18. Finally, we explore the narrower question of whether differences in CDR frequency played a role in the prevalence of youths leaving the program before age 18.

Differences in Payments Received Are Consistent for All Demographics and Impairments

One potential driver of the differences in cumulative per-recipient payments received between cohorts is the differing case mix. As shown in Table 1, the cohorts vary considerably in their demographic and impairment characteristics. Below, we document descriptive patterns for these characteristics then use a regression model to explore whether observable differences in cohort composition can explain the variances in program stays and payment amounts.

Mean SSI payment durations are longer and the amounts received are higher for youths in the 2007 cohort than for those in the 1997 cohort across all age, sex, and impairment subgroups (Table 3). This finding indicates a categorical shift upward in program stays across all groups.¹⁰ By contrast, there are no notable differences between the 2007 and 2012 cohorts in average duration or cumulative payments by age, sex, or impairment in the 5-year postaward period available for comparison (Appendix Table A-1).

In Table 3, for both the 1997 and 2007 cohorts, payment receipt patterns differ dramatically by age. Recipients who were younger than age 1 at the time of award have shorter mean durations and lower total payment amounts than awardees of other ages, likely because low birth weight recipients are generally subject to a special CDR by age 1. Youths first awarded at ages 13 to 17 also have relatively shorter program stays, likely because they are subject to an age-18 redetermination relatively sooner after award. Below, we discuss the possible effect of these redeterminations and additional CDRs on the differences in payment receipt between cohorts.

Comparing results by impairment, mean SSI payment durations and amounts for 1997 awardees were highest for those with intellectual disabilities (92.5 months and $63,906) and autistic disorders (87.8 months and $56,488). Other impairments with notably long durations include schizophrenic and other psychotic disorders, diseases of the blood and blood-forming organs, and organic mental disorders. The categories with notably shorter durations include neoplasms and “other” impairments.

Even after controlling for variation in cohorts' characteristics, our regression results indicate differences between cohorts in mean length of benefit receipt and mean cumulative amounts received (Table 4).¹¹ The regression-adjusted differences between cohorts are similar to the unadjusted descriptive statistics; after 10 years, the 2007 cohort had received SSI payments for about 10.6 months longer than the 1997 cohort amounting to $6,869 more received. Differences in DI receipt were small.

The similarity between the regression-adjusted differences and the descriptive statistics suggests that the case mix did not strongly affect the aggregate differences in payment receipt.¹² This means that something other than differences in recipient characteristics drives the differences in payment receipt over time. We next explore patterns by age, focusing on the role that the low birth weight CDRs and age-18 redeterminations may play.

Longer Periods of SSI Receipt Among Younger Child Awardees Drive Cohort Differences

Younger awardees in the 2007 cohort had persistently longer program stays than their peers in the 1997 cohort, regardless of whether they had a low birth weight CDR (Chart 4, Panel A). For younger awardees who were not subject to a low birth weight CDR, SSI payment duration was 15 percent longer for members of the 2007 cohort than for the 1997 cohort (87.6 months versus 76.2 months). Because younger awardees did not reach age 18 within 10 years of award, those who left SSI must have done so independent of an age-18 redetermination. Most of the younger awardees in both cohorts were not subject to a low birth weight CDR (86.1 percent in 1997 and 86.6 percent in 2007; Table 5). For younger awardees whose low birth weight factored into program entry, we also find patterns of longer participation for the 2007 cohort than for the 1997 cohort.¹³

The patterns for older child awardees also suggest that long-term differences in payment receipt center on youths who leave the SSI rolls before reaching age 18 (Chart 4, Panel B). Among those who left SSI before turning 18, the 2007 cohort received payments for 68.7 months on average, about 18 percent longer than did the 1997 cohort (58.4 months). However, among the older youths who had an age-18 redetermination, the differences between the 1997 and 2007 award cohorts in the duration of payments were much smaller. For example, among the awardees who continued receiving payments after age 18, mean SSI duration for the 2007 cohort was only 2 percent longer than that of the 1997 cohort (110.8 months versus 108.5 months). Therefore, a key factor in the aggregate differences between cohorts is that youths who left the program tended to do so more quickly if they were in the 1997 award cohort.

Further underlying the difference between the cohorts is that a larger share of older youths in the 1997 cohort left SSI before age 18 (46.8 percent) than did so in the 2007 cohort (37.5 percent; Table 6). In comparison, the share of first-time awardees whose payments ceased as the result of an age-18 redetermination, after all appeals, was more than 1.5 times higher in the 2007 cohort than in the 1997 cohort, 31.7 percent versus 19.1 percent.

Taken together, these results indicate that the differences between cohorts in SSI receipt stem mostly from differences in outcomes that occur before reaching age 18. Youths in the 1997 cohort were more likely to leave the rolls—and to leave more quickly—than were those in the 2007 cohort.¹⁴ One reason youths leave the rolls is because of child CDRs; administrative patterns discussed earlier suggest that CDRs were more likely to be initiated for youths in the 1997 cohort than for those in the 2007 cohort. We next explore the effect of those CDRs.

Timing and Quantity of Other Childhood CDRs Play Major Roles in Program Stays

The frequency and timing of other childhood CDRs (those not related to low birth weight) strongly affect the broad patterns of program payment duration and amounts received over time. For example, a CDR resulting in cessation could occur in the first year after an award for a recipient in one award cohort, and not until 5 years after award for a recipient with a similar impairment in another cohort. Because these differences compound over time, both the timing and the quantity of CDRs have important implications for payment duration, as the comparisons between the 1997 and 2007 cohorts illustrate.

We estimate that CDRs account for about half of the overall difference in SSI payment duration between the 1997 and 2007 cohorts. To generate this estimate, we first calculated the cumulative percentage of youths in each cohort whose payments ceased after a CDR in the 10 years following the initial award; the results appear in Chart 5.¹⁵ Similar shares (around 11 or 12 percent) of both the 1997 and 2007 cohorts had had their payments ceased after a CDR by the end of the period, but a much greater share of the cessations occurred earlier for the 1997 cohort. The cessation rate from a child CDR in the first 5 years was more than four times greater for the 1997 cohort than for the 2007 cohort (9.9 percent versus 2.3 percent). These large cessation-rate differences reflect changes over time in the frequency with which SSA conducts other childhood CDRs, as discussed earlier.

Greater cessation rates, particularly in the first years after award, have direct implications for some of the differences between the 1997 and 2007 cohorts in payment duration shown in the preceding tables. To estimate the potential magnitude of these differences, we simulate for each cohort the duration of potential payment receipt that a CDR cessation negates (Chart 6). For example, if SSA ceased a recipient's payments in year 1, we assume that he or she lost 9 potential years of additional payments.¹⁶ This assumption represents an upper bound on potential payment durations, although it is a credible estimate given the long durations shown in Chart 4. Additionally, Hemmeter and Bailey (2015) found that less than 10 percent of children whose participation ceased as the result of a CDR returned to the SSI rolls before age 18, which suggests that reenrollments would not substantially reduce that upper bound.

If the 1997 cohort had experienced the same lower cessation rate that the 2007 cohort did, their average payment duration would have increased substantially. Specifically, over the full 10-year period, the higher rates of early CDR cessations for those in the 1997 cohort might have reduced their average potential payment durations by almost 5.2 months.¹⁷ This potential 5.2-month increase would explain half of the 10.5-month difference between the 1997 and 2007 cohorts in mean SSI payment duration (Table 2).¹⁸

Our analysis shows that the timing of the cessations differed between the cohorts. For example, cessations in the first 3 years after award were substantially higher for the 1997 cohort than the 2007 cohort. The initial 3-year periods contribute the most to the estimated difference between the two cohorts. These differences persist despite the narrowing of the gap in cessation rates between the two cohorts in years 7 through 10. We estimate that the cumulative difference between the cohorts in payment amounts resulting from CDR cessations might have been more than $400 million over a 10-year period.¹⁹

Later Cohorts Had Higher Age-18 Redetermination Cessation Rates

Differences in payment receipt and CDR cessations could have important implications for age-18 redetermination volumes and outcomes. Specifically, in the 2007 and 2012 cohorts, payment durations before age 18 increased, and fewer recipients underwent childhood CDRs; this led to compositional changes for the group that undergoes an age-18 redetermination. These changes are reflected in a higher redetermination cessation rate among older youths in the 2007 cohort (50 percent) than in the 1997 cohort (about 35 percent; not shown). However, the difference between cohorts may be an upper bound; some members of the 2007 cohort may eventually have a cessation overturned on appeal.²⁰ This difference is consistent with previous research finding that youths without a childhood CDR are more likely to have payments ceased during the age-18 redetermination, and those whose payments are ceased in a redetermination are more likely to return to the program within 10 years (Hemmeter and Bailey 2015).

The recent pattern of higher cessations resulting from age-18 redeterminations is consistent with broader trends revealed in administrative records from SSA. Because the number of other childhood CDRs was sharply lower during the period from fiscal year 2005 to fiscal year 2013 (Chart 1), youths reaching age 18 in the early 2010s were less likely to have been subject to a child CDR than those reaching age 18 in the early 2000s had been. Among youths reaching age 18 in the early 2010s, those who had the least severe disabilities—who might have been removed from the rolls if they had undergone a CDR—would therefore reach age 18 still enrolled. Consistent with this trend, the cessation rate for age-18 redeterminations was much higher in the early 2010s than the early 2000s (Chart 2). Future research could explore whether patterns in the timing and frequency of redeterminations and CDRs (1) only affect the timing of removal for recipients who would otherwise have their payments ceased by age 18 or (2) actually change who is removed from the rolls, which would have longer-term implications for payment receipt.

Robustness Check

We tested whether choosing 1997 as the year to start the analysis drove the observed changes over the study period. We sought to provide a degree of consistency with other research such as Rupp, Hemmeter, and Davies (2015). Because 1997 was the first full year following landmark welfare-reform legislation, which included mandating the age-18 redetermination, its awardee cohort may differ from subsequent cohorts in ways that affect outcomes. Additionally, the 1997 cohort was notably smaller than subsequent cohorts. Comparing the 1997 cohort to later cohorts might therefore lead to misinterpretations of observed changes.

Table 7 shows SSI payment mean durations and cumulative amounts in the 5 years and the 10 years after first award for not only our study cohorts but also the 1999, 2000, 2001, and 2003 cohorts. Payment durations and payment amounts in the 10 years after award increased consistently between the 1997 and 2003 cohorts, indicating that our primary findings are not an artifact of the choice of 1997 as the base-cohort year. Each successive cohort saw continued growth in the number of new SSI recipients, program stays, and payment amounts. We see similar growth in the 5-year statistics for those cohorts as well as the 2010 cohort. Payment durations and amounts increased between the late 1990s cohorts to early 2010s cohorts. We therefore believe the intrinsic effects of our chosen starting year are minimal.

Discussion

We find evidence that SSI program stays among children are generally longer for more recent award cohorts (2007 and 2012) than for the 1997 cohort. Although the cohorts in our study differed in terms of their demographic and impairment characteristics, our regression results suggest that controlling for these factors does not affect the basic patterns of payment durations. The increase in length of payment receipt for recent cohorts coincided with increases in the size of the child SSI caseload.

We estimate that if the 2007 cohort had had the same mean payment duration as the 1997 cohort, the accumulated SSI amounts paid over 10 years would have been about $1 billion lower. Whether the amounts to be paid in the coming decade to members of the 2007 award cohort will be larger by similar proportions than the amounts paid to the 1997 cohort remains to be seen. For recipients whose participation continued after an age-18 redetermination, the duration of payment receipt was nearly identical between the two cohorts. This implies that the $1 billion amount noted above might not expand further.

Although the duration of SSI receipt lengthened over the study period, the recent increase in childhood CDRs might shorten program stays for future SSI award cohorts. We find evidence that ongoing CDRs play a key role in payment duration, possibly explaining about half of the difference in mean duration between the 1997 and 2007 cohorts. Because the frequency of CDRs rose sharply between fiscal years 2015 and 2018 (Chart 1), the trend of increasing mean durations may reverse. Additionally, SSA's proposal to create a new “Medical Improvement Likely” category for disability program participants could, if implemented, increase the frequency of medical CDRs for some SSI recipients.

Mean program stays for future SSI award cohorts might therefore differ from those of the 2007 and 2012 cohorts, particularly because of the large surge in childhood CDRs beginning in fiscal year 2015. Although the major driver of the longer payment durations for the 2007 and 2012 award cohorts was the relatively low number of CDRs SSA conducted during fiscal years 2006–2014, other factors (such as the shifting geography of SSI recipients and the economy) might also have contributed to these trends.²¹

The fluctuations in CDR policy might also affect the number of age-18 redeterminations. Because Congress has authorized and funded SSA to conduct more child CDRs in recent years, the size and composition of future caseloads subject to age-18 redeterminations may change. Hence, it is important to understand how CDR timing and frequency ultimately affect continuation and cessation rates for age-18 redeterminations, particularly when tracking program outcomes. For example, although the cessation rate for age-18 redeterminations increased through fiscal year 2015, the trend began to reverse in more recent years, and this reversal might continue as discretionary CDRs conducted before age 18 increase; some youths whose participation might have continued until cessation at age 18 could instead be removed from the rolls earlier. It is possible that SSA's proposed changes—including revising the CDR diary types, requiring a medical review after 2 years on the rolls, and mandating reviews at ages 6 and 12—will, if implemented, accelerate any changes in observed patterns of program participation at age 18.

This article cannot address how changes to patterns of SSI receipt affect youth outcomes. Earlier research indicates that the income sources of former SSI recipients tend to be unstable after the cessation of program payments (Deshpande 2016; Hemmeter, Kauff, and Wittenburg 2009; Hemmeter 2011). Given the large fluctuations in program stays, it is especially important to understand how well families are prepared for CDRs and the age-18 redetermination and, for those whose participation ceases, how able they are to replace the SSI payment.

For youths who exit SSI following a CDR or redetermination, outcomes—such as employment or connections to other programs—are an important consideration. Additional research looking into the efficiency of CDRs in identifying youths who can engage in substantial gainful activity, and whether observed patterns change, could reveal ways to serve youths as they leave SSI. SSA's fiscal year 2021 budget proposes a project identifying the services and supports needed to improve the self-sufficiency of individuals who exit DI because of a medical CDR. Evidence from that study might also provide suggestions about the needs of former child SSI recipients.

Appendix A