The Age-18 Redetermination and Postredetermination Participation in SSI

Introduction and Background

The landmark Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA, Public Law 104-193) had substantial impacts on a number of federal programs, including the Supplemental Security Income (SSI) program. SSI provides monthly, means-tested cash payments to aged, blind, and disabled persons with low income and assets. Among other changes, the PRWORA changed the definition of disability for children who apply for SSI from an impairment (or combination of impairments) of "comparable severity" to one that would disable an adult, to the current and more restrictive "marked and severe functional limitations." This has been interpreted by the Social Security Administration (SSA) as requiring an impairment(s) that meets or medically equals the severity of a listing in SSA's Listing of Impairments (the listings)¹ or that functionally equals the listings. The PRWORA also required SSA to redetermine the eligibility of child SSI recipients who attain age 18 using the adult program rules, in which eligibility is based on the inability to perform substantial gainful activity (SGA).² An unfavorable initial determination, where the child is determined not disabled under the adult standard, can eventually lead to a cessation of payments if upheld through an appeals process. Throughout this analysis, we refer to an initial determination that the youth is not disabled, as defined by SSA's legislative and regulatory requirements,³ as an initial cessation determination or an adverse determination, reflecting the first decision made in the youth's age-18 redetermination.

In this study, we provide a description and analysis of the results of the PRWORA and other regulatory changes, using administrative data to summarize the characteristics of those who go through the age-18 redetermination and the relationship of those characteristics to the initial redetermination decision and later program participation. This will provide a useful baseline for any discussion of the longer-term outcomes of these youth and their transition to adulthood. We find little change in the initial cessation rate for all groups over time, although there are large differences in initial cessation rates between disability type and other characteristics. Consistent with previous studies, the majority of redeterminations result in initial continuances. However, many youth initially found no longer categorically disabled at age 18 successfully appeal that decision, and a nontrivial number who lost eligibility successfully reapply at a later date.

The next section of this article presents the legislative and regulatory history and provides background on the age-18 redetermination process. We then describe the literature on age-18 redeterminations and the implications for the current study. The data, variables, and methodology are detailed next, and the following section provides descriptive statistics for the initially redetermined population. We then discuss the statistics on the initial redetermination outcome, focusing on adverse determinations. The odds of successfully appealing the initial adverse determination or reapplying for SSI are then examined. We conclude with a discussion of the findings, possible policy implications, and suggestions for future work. The Appendix provides additional tables.

Program History and Description

The SSI program provides monthly, means-tested payments to qualifying aged, blind, and disabled individuals. Before 1996, the Social Security Act provided that a child (an individual who has not attained age 18) was categorically disabled if he or she had an impairment that was of comparable severity to an impairment that would disable an adult.⁴ However, from 1974, when the SSI program began, to 1990, a child was determined to be disabled under SSA's regulations only if he or she had an impairment (or a combination of impairments) that met or medically equaled the criteria of a listing in SSA's listings. Because this differed from the adult rules, which provided for an assessment of overall functioning and allowed many adults to qualify with impairments that did not meet or medically equal the listings, the U.S. Supreme Court ruled that SSA's childhood regulations did not properly interpret the comparable severity standard in the law (Sullivan vs. Zebley).⁵

In response, SSA issued two new policies for evaluating disability in children: (1) functional equivalence—a new standard that considered functioning at the listings step of the disability analysis; and (2) Individualized Functional Assessment (IFA)—a new standard for determining disability in children whose impairments did not meet, medically equal, or functionally equal the listings.⁶ The IFA was considered to be one of the primary factors leading to a dramatic increase in the growth of the child SSI program in the 1990s.⁷ This growth, as well as allegations of fraud and other issues (Auxter and others 1999), paved the way for changes in the eligibility rules of the child SSI program to be included in the PRWORA. However, reports of fraud were found to be exaggerated (General Accounting Office 1995).

Among the changes implemented by the PRWORA was a revised definition of disability that removed the comparable severity standard and required individuals who have not attained age 18 to have impairments that result in marked and severe functional limitations. Congress also required SSA to remove the IFA policy from its regulations. Another important change was the addition of a new provision in the Social Security Act requiring the redetermination of medical eligibility within a year after the individual attains age 18 under the definition of disability for adults who file new claims, that is, inability to perform any SGA by reason of any medically determinable impairment that has lasted, or can be expected to last, a continuous period of at least 12 consecutive months or to result in death.⁸ This redetermination is treated as a new application for SSI.

The PRWORA also required SSA to redetermine the eligibility of about 288,000 children who were allowed under the IFA rules or under the listings, which reference "maladaptive behaviors." In all, roughly 100,000 of these children lost eligibility; the remaining were found to have disabilities that met the new definition of disability for children.

In addition to the disability requirement, recipients (both adults and children) must also have limited financial resources, which can include income and assets deemed from the parents of child recipients.⁹ Countable assets must be at or below defined levels, currently $3,000 for couples and $2,000 for individuals. After the first $65 of earned income and $20 of unearned or earned income, SSI payments are reduced $1 for every $2 of earned income and $1 for every $1 of unearned income until payments are reduced to zero.¹⁰ In addition to the federal payment, most recipients are also eligible for Medicaid if they receive SSI. Many states also provide a supplement to the federal SSI payment.

Children eligible for SSI payments in the month before they reach age 18 are required to go through the redetermination process.¹¹ SSA's field offices collect disability and functional reports—including the names and addresses of medical sources for the previous year—and work, education, rehabilitation, and support services received. Completed case files are forwarded to a state agency (the disability determination service, or DDS), charged with making the initial determination for SSA. The DDS obtains evidence and makes the determination whether the individual's condition satisfies the adult definition of disability using SSA's rules. If an individual could not be contacted by the field office and insufficient medical information has been collected to make a decision, payments can be ceased for failure to cooperate (FTC).¹²

If the individual is dissatisfied with the determination, he or she has the right to appeal. There are three levels of appeal within SSA: (1) reconsideration (at the DDS), (2) Administrative Law Judge (ALJ), and (3) Appeals Council. If the individual is still dissatisfied at the end of this process, he or she may appeal through the court system, starting with a federal district court and (potentially) ending with the U.S. Supreme Court. Court appeals are relatively rare, and U.S. Supreme Court appeals are extremely rare. At each level of the appeals process, the individual has 60 days in which to appeal. The individual may also request continuation of benefits at the reconsideration and ALJ-appeal levels, but has only 10 days in which to make this request.¹³

The requirement of the age-18 redetermination was intended to moderate the growth of SSI and ensure only those continuing to meet the medical eligibility for the program remained on the rolls. In 1997, when age-18 redeterminations were extended to the full SSI population,54 percent of age-18 redeterminations resulted in an initial cessation decision.¹⁴ This has since declined to 46 percent of age-18 redeterminations in 2006 when 40,640 young adults underwent an age-18 redetermination. About half of initially ceased determinations are appealed. Overall, about a third of all age-18 redeterminations result in a final cessation decision (SSA 2007b).

Review of Previous Research and Research Hypotheses

Very little is known about the effect these redeterminations have on long-term outcomes, such as future program participation or employment (Social Security Advisory Board 2006). Measurement of many long-term outcomes could be problematic with measures only a few years after the age-18 redetermination. For example, many children (with and without disabilities) may have gone into postsecondary education or have an extended period of secondary education, reducing the likelihood of observing employment before attaining age 22. This makes it difficult to fully understand how the age-18 redetermination will affect these youth. Several studies, however, have analyzed these outcomes in the short and intermediate term. Those studies provide a springboard for the hypotheses we consider in this study.

Previous work by Rogowski and others (2002) analyzing the characteristics of SSI recipients affected by the PRWORA employed an early cohort from Social Security administrative records that was likely not prepared for the changes in program rules brought about by the PRWORA and may have behaved much differently from, or were more adversely affected than, later cohorts. The authors found that about 45 percent of child SSI cases received an initial cessation determination and that there was a relatively low rate of reapplication within 12 months. They did, however, find varying rates of initial cessation determinations by type of disability.¹⁵

Loprest and Wittenburg (2007) and Hemmeter, Kauff, and Wittenburg (2009) have shown that those youth who leave SSI after age 18 have poorer educational backgrounds than those who remain on SSI, potentially reflecting lower opportunities for postsecondary education or employment. As a result, individuals may return to the SSI program because of a combination of worsening health status and their inability to support themselves, that is, inability to work and earn above SGA. However, these authors were estimating the characteristics of SSI participants before and after attaining age 18 in the National Survey of SSI Children and Families and did not have access to information on age-18 redeterminations; thus, they could not attribute post-age-18 outcomes to redetermination decisions.

These studies of short- and intermediate-term outcomes suggest that the cessation of payments may have long-term consequences for many SSI youth. Although the initial determination (and the appeals process) has determined that these individuals are capable of SGA, many of these youth have difficulty finding employment (Loprest and Wittenburg 2007). For some youth, health-care access via Medicaid is lost once they are determined ineligible for SSI (Loprest and Wittenburg 2007). This may result in their health worsening to the point of becoming eligible for SSI payments again.

There is some evidence that the age when a child initially becomes eligible for SSI may be correlated with later education and employment outcomes (Loprest and Maag 2007); this may give rise to different redetermination outcomes. Similarly, those persons initially allowed at earlier levels of the adjudication process or who did not require a consultative examination (CE) may have more clearly identifiable disabilities, and thus may be more likely to be allowed to continue into the adult SSI program. Previous research (Hemmeter, Kauff, and Wittenburg 2009) has shown that the presence of earnings before the age-18 redetermination is correlated with not receiving SSI at age 19. It may also be inversely correlated with return to SSI, since adult eligibility depends on the inability to work, and these youth have demonstrated some ability to work.

Previous work has also shown that judicial and legislative changes, such as the Zebley decision and the PRWORA, have had a significant effect on SSI participation (Kubik 1999; Brady, Seto, and Meyers 1998). Although the redetermination of childhood disability decisions as required by the PRWORA effectively resulted in all children receiving SSI at the age-18 redetermination being eligible under the current definition of disability, there may remain some differences between youth allowed during periods with different eligibility requirements. For example, some children (or their parents) who would not have initially applied under the more strict disability definition may have been induced to apply for SSI during the Zebley era. Once receiving SSI, however, they may not have exited the program rolls for a variety of reasons. Some of these selection differences may result in youth allowed under different eligibility requirements being more or less likely to be ceased as a result of the age-18 redetermination.

We focus on two major changes to child eligibility requirements: the Zebley decision and the PRWORA. Although there were several other changes in the disability regulations over the years, we focus on these two regulations as the major changes in this period. The Zebley decision in 1990 led to more initial allowances of youth to the SSI program, especially among youth with mental disorders. Additionally, SSA altered the way mental impairments were evaluated in 1990, making it more likely for those persons with attention deficit hyperactivity disorder (ADHD) and other developmental and mental disorders to receive SSI. Although these two changes occurred in 1990, they were not fully implemented until 1991, which we use later in identifying the periods under study.

The second major change we consider is the PRWORA, which occurred in 1996. The PRWORA rules were not finalized until 2000; because of this, there may be differences between those allowed under the interim rules and the final rules. It should be noted that the vast majority of children (roughly three-fourths) receiving SSI payments were not affected by PRWORA, that is, there would have been no difference between the eligibility requirements when they first applied compared with other years. Because of this, differences across these cohorts may suggest unobserved factors other than the policies themselves, which have effects on the age-18 redetermination. However, children with the disorders specifically targeted by Zebley and PRWORA during these time periods may be differentially affected by the age-18 redetermination.

Additional Hypotheses

In addition to hypotheses suggested by previous studies and the legislation, we raise a few additional ones here. We believe that less severe disabilities (that is, disabilities that, although still meeting the eligibility criteria, do not cause as much of an interference with activities of daily living) may not have as great an impact on an individual's ability to perform SGA. For example, an individual with ADHD might have a less severe disability than an individual with Down's syndrome. In addition, there may be slight differences in the threshold for a particular disability between the child and adult listings. As a result, there might be more negative determinations for individuals with a specific type of disability.

Many youth have had a continuing disability review (CDR) before attaining age 18, where medical eligibility has been previously reassessed, subject to a medical improvement standard; this might have an effect on the likelihood of future SSI participation. Those individuals who have had a prior CDR have already been determined to have a continuing disability and may be more likely to continue SSI participation or appeal or reapply after an initial determination of not being disabled. Those who have not had a CDR before reaching age 18 may thus be more likely to initially have an adverse determination, all else equal. However, because CDRs are typically conducted on individuals who are likely to recover from their disability, the presence of a prior CDR could indicate a less severe disability, as defined earlier, increasing the likelihood of an adverse determination.¹⁶ On the other hand, those who remain eligible for SSI after a CDR (and are thus in our sample) are likely to be "more" disabled, all else equal. These individuals would thus lower the adverse determination rate for individuals with prior CDRs. These selection issues make it difficult to determine what the effect of a prior CDR will be on the likelihood of an initial cessation determination. Conditional on an initial cessation determination, it is unclear whether having a prior CDR would have a further effect on the likelihood of future SSI participation.

Certain youth may be less likely to cooperate during the age-18 redetermination and therefore their payments are initially ceased for that reason, even though SSA might have found that they were still disabled had they cooperated. Rogowski and others (2002) found that youth with infectious and parasitic diseases, schizophrenia and other psychiatric disorders, and "other" unlisted disabilities had a higher than average rate of initial cessation determinations because of FTC. If such individuals find that they are unable to provide for themselves at a later time, they may appeal the initial decision (within 60 days) or reapply.

Additionally, some youth have disabilities that do not directly correspond to the adult listings, for example, growth impairment. These individuals may be more likely to appeal or reapply for SSI than those who did not meet the adult medical eligibility requirements, if other factors contribute to their inability to perform SGA. Other factors, such as sex, may also impact the initial redetermination decision through the interrelationship between, for example, sex and disability type. Factors such as the year of the redetermination may also affect the youth's postredetermination decision to appeal or reapply for SSI because of secular trends in opportunities.

Our analysis addresses some of the shortcomings earlier research had in addressing these issues. Although our analysis does not include as extensive information on personal characteristics as can be found in the National Survey of SSI Children and Families, we use Social Security administrative data from a long time period and from cohorts that would have had time to adjust to, and prepare for, the changes resulting from the PRWORA. The analysis is broken out by certain individual characteristics, shedding light on how different groups fare during the redetermination, and we show how the initial redetermination experiences of these groups have changed over time using yearly cohorts of redeterminations from 1998 through 2005. Information on appealing the initial redetermination decision and reapplication to the SSI program can give a fuller understanding of how the age-18 redetermination process changes the composition of the SSI caseload, and it can identify groups of youth that might need additional help in becoming self-sufficient as they transition to adulthood.¹⁷

Data Sources and Methodology

The data we use for this project are from Social Security administrative records. The Office of Quality Data Management within the Office of Quality Performance maintains a record of all age-18 redeterminations.¹⁸ For the period under study, from January 1, 1998, through December 31, 2005, we obtained an extract of this file including 409,260 age-18 redetermination decisions.¹⁹ Only redeterminations for which an initial decision has been made are included in this file (and thus in our population). The file contains information on the result of the initial redetermination decision as well as the date of the decision, reason for the decision, and disability diagnosis. Similar information for each appeal through the Appeals Council level is also contained in the file.²⁰ We matched these records to SSA's Numident file to obtain date of birth, date of death, and sex. Longitudinal data from the Supplemental Security Record, which contains administrative data on SSI participation, was merged to these records to obtain age at first SSI receipt. If the sex of the person was missing from the Numident file, we used the sex designation from the Supplemental Security Record.

Additionally, we merged information from SSA's Disability Research File, maintained by the Office of Disability Programs, into the data. The file combines data from multiple administrative sources and contains information on applications and appeals for SSI and DI benefits. Detailed earnings records from the Master Earnings File were also merged into the file. Beginning in 1978, earnings information from an individual's W-2 Form is provided each year to SSA, with the most recent year's complete data (at the time of this research) from 2006. Because some individuals may be self-employed or have covered earnings not taxed under the Federal Insurance Contributions Act (FICA), we use the total earnings reported on the W-2—including noncovered and self-employment income, not just FICA-taxable earnings—in our analysis. Each year of earnings data is associated with the age an individual turns in a given year because W-2s reflect yearly information. If an individual turned 17 in 2003, then the 2003 earnings data are associated with age 17.

Statistical Method and Approach

In addition to descriptive statistics about the sample population, we use logistic regression analyses to estimate the effect of the explanatory factors on the probability of an adverse determination. In addition to a pooled regression with yearly fixed-effects, separate regressions are run on each calendar-year redetermination cohort to determine if the effects of the explanatory factors change over time. This model can be expressed as—

(1)  $\ln \frac{P(Y_i=1)}{1-P(Y_i=1)}=\alpha +{\displaystyle \sum _{k=1}^K{\beta }_k{X}_{ik}+{\epsilon }_i}$ .

Here, Y_i = 1 indicates an adverse determination for individual i; the X_i are characteristics identified shortly; the β_k are estimated coefficients; and ε_i is an error term. We express the effects as odds ratios (exponentiated coefficients). Additionally, we use a similar model to estimate the effect of the explanatory factors on postredetermination SSI participation: successfully appealing an initial adverse determination or successfully reapplying for SSI.

The control variables (X_i) include several programmatic and demographic variables available from administrative records. We identify the following variables reflecting the age-18 redetermination:²¹ primary disability, whether or not a CE was requested by SSA, and whether or not the individual had a prior CDR.²² We also identify information on the youth's age at his or her initial SSI entitlement and the adjudicative level of that award decision. Additionally, we control for sex and the year in which the redetermination occurred. We also include a variable identifying individuals who had annual reported earnings greater than or equal to $250 at age 17. We use this as a proxy for having had a serious work effort. This is roughly the 25^th percentile of earnings for sample members with earnings.²³ In models of the appeal of the initial determination or reapplication to SSI, we also include reason for the initial adverse determination as an explanatory covariate.

We identify the following periods of SSI entry (defined by age at eligibility), which may be of interest: before 1991; from 1991 through 1996 (under new childhood mental disorders listings, other revised listings, and under Zebley, but before the PRWORA); from 1997 through 1999 (under the interim PRWORA rules); and after 1999 (under the final PRWORA rules). Other factors that we do not measure, such as the economy, may have also affected participation and may have changed in these time periods as well, confounding any pure policy effect. The yearly fixed-effects will capture some of these effects at the national level, but local effects are not captured. There were also other changes to the listings and regulations that may not be reflected in the estimates for these time frames. We leave further analysis of those changes to future work.

There are several ways to examine a youth's participation in SSI after the initial redetermination decision. The simplest method takes a look at either a successful appeal of the decision or reapplication to the SSI program. This "global" approach is the broadest in the sense that it includes whether or not an initial cessation determination is overturned. We also estimate a multinomial logistic regression model where the possible paths—appeal and reapplication—are separated from each other. Each path is also estimated using separate logistic regression models. These specifications describe the odds of successful appeal or reapplication to SSI after an adverse initial redetermination.

As previous research has shown, appeals of adverse initial redeterminations are common. During this time, individuals may never stop receiving SSI payments.²⁴ To examine the return to SSI of youth who actually leave the program, we also specify a postappeal reapplication model based solely on those who have not successfully appealed an initial cessation determination and do not have an open appeal. We discuss these models in greater detail later.

Sample Selection

We placed several restrictions on the data, which lead to there being a different number of age-18 redeterminations than the number reported in SSA's Office of the Actuary's Annual Report of the Supplemental Security Income Program.²⁵ These restrictions are listed in Table A-1. We exclude 133 individuals from the analysis who died before their redeterminations were completed. We also limit the sample to individuals whose age-18 redeterminations occurred between the day of and 3 years after their 18^th birthday.²⁶ This excludes 6,314 individuals (1.5 percent of the remaining redeterminations). Finally, 212 individuals were found to have first received SSI payments outside the age range from birth to age 17. Excluding these individuals from the data resulted in a final sample population of 402,601 youth who had an age-18 redetermination occurring from 1998 through 2005. Of these individuals, 170,376 had an initial decision of not disabled (42.32 percent).

The two postinitial determination options for continued program participation—successful appeal and successful reapplication—require different amounts of time to complete. The appeals process can take many years to complete, and most individuals wait for completion of this process before reapplying for SSI, although some attempt both means of return simultaneously. Because of this, later cohorts most likely have not had enough time to experience the full range of postinitial determination options. To eliminate this censoring issue, our postinitial determination analyses are limited to individuals with 4 years of observed follow-up time—those whose redeterminations occurred during the 1998–2001 period (N = 81,458).²⁷ We then consider only successful appeals or reapplications within a rolling 4-year period (beginning at the date of the initial redetermination decision for each individual). This method allows enough time for an individual to go through both the appeal and reapplication processes. Additionally, the method we use incorporates most ages commonly suggested as alternative definitions of the child/adult age boundary.²⁸

As mentioned earlier, we separately consider postappeal reapplications. These individuals have not successfully appealed and are no longer eligible to appeal. Because individuals have 60 days to appeal their decision at each level of the appeal process, individuals for whom 60 days have not passed after their final appeal are removed from the sample population. This leaves 62,953 individuals in the reapplication-only analysis.

Descriptive Characteristics of the Age-18 Redetermination Population

We first present the characteristics of the age-18 redetermination population, the outcomes of the determination, and trends of the outcomes. This provides comparability with other studies and will allow for an analysis of trends in the population over time.

Table 1 presents summary statistics of the age-18 redetermination population.²⁹ A relatively constant proportion of individuals are initially either continued or determined not disabled as a result of these redeterminations (about 57 percent and 43 percent, respectively). This is very similar to the numbers produced by the Office of the Actuary; the difference is due to the selection differences described earlier. The majority of adverse determinations are due to the recipient not meeting the adult criteria for disability—although a sizable number, remaining relatively stable at around 8 percent, were also due to FTC. This is about 3 percentage-points less than what was estimated by Rogowski and others (2002). Although that study used a different data source than the current analysis, the higher number may also reflect differences in the implementation of the redetermination policy.

The largest share of redeterminations was for individuals with mental retardation as their primary disability. However, this share has been declining, as in the general child SSI population. Part of this is likely the result of SSA policy changes and training in the classification of mental retardation and "other" mental disorders. By 2005, individuals with mental disorders other than mental retardation were slightly more common in the age-18 redetermination population than those with mental retardation (37 percent versus 36 percent).³⁰ Combined, individuals with "other" mental disorders and mental retardation make up over two-thirds of the redetermination population.³¹

The proportion of age-18 redeterminations that first became eligible for SSI during the time period each policy was in effect has shifted as expected. In 1998, the majority of redeterminations (55 percent) were for children who first became eligible from 1991 through 1996 (under Zebley rules). This is still the most common time period when these children entered SSI overall, but those who first became eligible after 1996 have become increasingly more common (46 percent, total, in 2005). The proportion of cases that became eligible before 1990 has lessened by about one-third. This movement is natural as individuals who entered SSI in an earlier period age into adulthood.

Well over half of redeterminations are conducted for male SSI participants, similar to the fraction of those in the child SSI population. The share of age-18 redeterminations of youth who became eligible for SSI payments before age 5 has increased over time. The share of redeterminations of recipients who first became eligible at ages 5–12 increased from 1998 through 2002, but has since declined; the share who first became eligible as a teenager declined from a high of 49 percent in 1998 to 32 percent in the early 2000s. However, this proportion has since risen to 37 percent in 2005. Among the many reasons the average age of first eligibility is dropping may be an increased awareness of disabilities at younger ages and a greater acceptance of mental disorders in the general population.

Over three-quarters of youth undergoing an age-18 redetermination had reported earnings of less than $250 in the year they turned age 17. This proportion has increased from 78 percent in 1998 to 84 percent in 2005. We do not know the reason for this increase, but we do note that there has been a general shift in the age at which youth first achieve significant earnings (Compson 2008). It may also reflect a behavioral response to economic cycles or the result of individuals attempting to ensure a favorable redetermination.

The majority of the redetermination population (80 percent) was originally entitled to SSI at the initial application level. Only 4 percent were allowed at the reconsideration level and 2.5 percent at higher levels, although the share of both of these groups has grown over time. A large minority (13 percent) have an unknown adjudication level.³² The proportion with no prior CDRs almost halved from 1999 through 2005, from 83 percent to 43 percent. A relatively steady proportion of youth (61 percent) required a CE for their redetermination.

Initial Redetermination Decision

We now turn our attention to the initial redetermination decision. We focus on adverse determinations—those where the youth was found not to have a disability under the adult definition—because these decisions set the stage for later work on postinitial determination participation in SSI.

Descriptive Characteristics

The percentage of age-18 redeterminations initially determined not to be disabled by year and characteristic is shown in Table 2. All disability types, with the exception of other/uncodable disabilities, saw a decrease in the proportion initially receiving an adverse determination over time, although there has been a slight upward movement in 2004 and 2005. Youth with mental retardation and those with sensory impairments were initially determined not to be disabled under SSA's definition at a relatively low rate of about 20 percent. Some of these individuals may have initially been misclassified (particularly youth with mental retardation); however, the low percentage receiving an adverse determination generally reflects the similar definition of disability between adults and children as well as the small expected changes in the severity of these disabilities. Similarly, only 15 percent of youth with schizophrenia, psychoses, or other neuroses had an adverse initial determination.

Over half of individuals with major affective disorders and over two-thirds of those with "other" mental disorders were initially determined not disabled under the adult definition. Youth with other/uncodable diagnostic codes were the most likely to receive an initial cessation determination, with over 90 percent receiving an adverse decision. The proportion of youth with musculoskeletal and physical disabilities receiving an initial cessation determination was over two-thirds; this share declined about 8 percentage points from 1998 through 2004, but has since risen slightly.

We find that over half of youth originally entitled under Zebley (from 1991 through 1996) initially received an adverse determination from 1998 through 2001. These youth included many who may have been unprepared for the changes in the PRWORA. Although this share has decreased somewhat, 45 percent of this group was initially determined not disabled in 2005. The share determined not disabled of those initially allowed under PRWORA (for both cohorts) has increased over time. In 1998, only 18 percent of the 1997–1999 cohort, under the initial PRWORA rules, had an adverse determination; in 2005, over 50 percent of that cohort was found not disabled. The post-1999 cohort, also, initially had a low percentage of youth receiving an adverse determination (8 percent in 2000, the first year they would have been eligible for an age-18 redetermination), but this proportion grew to almost half by 2005. This may reflect the shorter time between initial eligibility and the age-18 redetermination in the early cohorts; these youth have had less time for their disability to improve. Additionally, many of the functional equivalence rules for older children allow for an easier transition to the adult disability rules.

A smaller fraction of female SSI participants had an adverse determination than their male counterparts almost every year, by about 5 percentage points (39 percent versus 44 percent). Only 20 percent of female redeterminations who first became eligible for SSI before age 5 are determined not disabled under the adult definition compared with almost 50 percent of those who first became eligible at ages 5–17. However, in estimates not reported, children first receiving SSI as a teenager are less likely to have had a previous CDR, which would have removed some older children from the program rolls before the age-18 redetermination.

A higher proportion of individuals with no prior CDR initially received an adverse determination than those with a prior CDR. This was more of an issue in 1998 when the difference was 42 percentage points. The difference shrunk to only about 5 percentage points in 2005. Because individuals in the early cohorts were less likely to have had a prior CDR (Table 1), the percentage initially receiving an adverse determination decreased over time among those who did not have a prior CDR. Among youth who had a prior CDR, it is unclear why the percentage initially receiving an adverse determination increased from 1998 through 2000 and then decreased to around 40 percent thereafter; it is possible that earlier CDRs used somewhat different criteria than more recent ones.

Working youth have demonstrated a capacity for employment, which may signal an ability to perform SGA (the adult definition of disability). Those with a recent work history are, in fact, more likely to receive an adverse determination than those who did not work in the year they turned age 17 (56 percent versus 39 percent).

There are large differences in the percentages initially receiving an adverse determination by level of initial adjudication. The majority of individuals entitled at the initial application level are continued as a result of the initial age-18 redetermination; over 70 percent of those initially entitled at the Office of Disability Adjudication and Review (ODAR) level (ALJ and higher) are initially found not to have a disability under the adult rules. This is consistent with the hypothesis that individuals who have had a more difficult time proving that their disability meets SSA criteria are less likely to continue receiving SSI after age 18. We also find that cases requiring CEs, indicating that the disability does not obviously meet SSA criteria or lacked medical evidence, are more likely to receive an adverse determination than those not requiring a CE (49 percent versus 31 percent).

Odds of an Initial Cessation Determination

The results from a pooled logistic regression model controlling for all of these factors, expressed as odds ratios, are presented in the first column of Table 3. Most effects are significant at the 5 percent or 1 percent level. The results are largely consistent with the statistics from Table 2.

Relative to youth with physical disabilities, those with mental retardation and schizophrenia, psychoses, or other neuroses and those with sensory disabilities are significantly less likely to receive an adverse determination; those with all other disorders are significantly more likely to receive one. Among the larger effects, the odds of initial cessation for youth with schizophrenia, psychoses, and other neuroses are almost 80 percent lower than for those with physical disabilities; for youth with mental retardation, the odds are 74 percent lower.

Relative to those who first entered SSI from 1991 through 1996 (under Zebley), all other cohorts are less likely to have an initial cessation determination. Youth who first entered before 1991 have 29 percent lower odds, all else equal; youth who first entered under the interim PRWORA rules have 30 percent lower odds; and youth who first entered under the final PRWORA rules have 44 percent lower odds of having an initial cessation determination. Part of this may be due to individuals allowed after 1996 having very recently demonstrated their disability, whereas those in previous cohorts had more time for their disability to improve. Directly relating this result to the changes in legislation may be confounded by selective attrition and, particularly among earlier cohorts, a higher reliance on the program that has grown with the length of participation. However, the strong effect of the 1991–1996 cohort does suggest there may be some factors that should be explored more carefully.

Female child SSI participants are slightly less likely to have an adverse determination, all else equal; the estimated odds ratio, relative to their male counterparts, is 0.95. The odds of an adverse determination for children first eligible for SSI before age 5 are about half that of those who became eligible at ages 5–12. Those who became eligible as a teenager have only 6 percent lower odds of an adverse determination than those who became eligible at ages 5–12.

The likelihood of receiving an adverse determination for those who have reported earnings greater than $250 is large and significant. Controlling for other characteristics, the odds of an adverse determination for an individual who earned at least $250 at age 17 are 73 percent higher than the odds for someone who did not work.

Individuals who became eligible for payments at successively higher adjudication decision levels are more likely to have an initial adverse determination during an age-18 redetermination than those who first became eligible for payments at the initial application level (as shown in Table 2). Youth who had a prior CDR are also still less likely to have an adverse determination. All else equal, when a CE is requested, indicating a difficult evaluation or a lack of available medical information, the odds of an adverse determination more than double, relative to when a CE is not requested.

Robustness

These results are largely robust to calendar-year-specific regressions (Table 3, columns 2–9). One important trend to note is that youth who first entered SSI from 1991 through 1996 were more likely to have an adverse determination in the earlier years. In 2004, the odds ratio for the 1997–1999 cohort is not significantly different from that of the 1991–1996 cohort; in 2005, the 1997–1999 cohort has 13 percent greater odds of an adverse determination. For the post-1999 cohort, the odds ratio in 2005 is not significantly different from that of the 1991–1996 cohort.

The significant effect of prior CDRs on reducing the likelihood of an adverse determination is largely driven by redeterminations that occurred in 1998 (odds ratio = 0.13), which includes a large proportion of individuals with no prior CDR and a very low adverse determination rate among those with a prior CDR (see Tables 1 and 2). Most other years have an odds ratio that is either insignificant or greater than 0.9.

Because the policy changes we identify altered the regulations with respect to certain disabilities and not others, we also estimated policy/cohort-specific regressions (not reported, but available upon request). If there are lasting effects from Zebley or other changes, we would expect to find stronger results among youth with disabilities primarily affected by these policies. Instead, the results for each policy cohort are largely similar to the combined results. This suggests that the policy/cohort effect is driven by population changes, such as the general selection issue raised earlier, and not policy-specific changes.

Appeals and Reapplications

We next turn our attention to the postredetermination participation of individuals who initially received an adverse decision and have had at least 4 years to either have that decision overturned on appeal or reapply. For this reason, the remaining results reported in this article are conditional on having an initial cessation determination from 1998 to 2001. Because the time frame is limited to 4 years for postinitial decisions for all cohorts, only the pooled results are presented. Additionally, because of death in our sample population, we removed 536 individuals for ease of computation and to maintain a comparable comparison group.

Descriptive Characteristics

Descriptive statistics for youth who received an initial cessation determination by their postinitial determination outcomes are shown in Table 4.³³ The first column shows the composition of all youth receiving an adverse determination and largely reflects the findings in Tables 2 and 3. Youth with "other" mental disorders make up one-third of the population that initially received an adverse determination. Individuals with mental retardation and those with other/uncodable disabilities each make up an additional 20 percent of this population. About 2 percent of youth who initially received an adverse determination have either sensory disabilities and musculoskeletal disabilities. Youth with physical disabilities; schizophrenia, psychoses, or other neuroses; and major affective disorders make up 13 percent, less than 1 percent, and 8 percent of the population, respectively.

Consistent with the high adverse determination rate in the early cohorts, most of the postinitial decision population (66 percent) initially became eligible for SSI under Zebley rules (from 1991 through 1996). However, the proportion of the total age-18 redetermination population initially allowed in that time period has been decreasing over time, and increasingly fewer individuals from that cohort have initially received an adverse determination at the age-18 redetermination.

Almost two-thirds of the population is comprised of male participants, and half became eligible for SSI at ages 5–12. Most (69 percent) did not have earnings greater than or equal to $250 in the year they turned age 17. The vast majority received an adverse determination because of their failure to meet the adult disability criteria (73 percent), although a sizable minority (19 percent) failed to cooperate during the redetermination. Most of our sample population were first entitled to SSI as a child at the initial application level (83 percent). The majority also did not have any prior CDRs (76 percent) and required CEs during the redetermination process (72 percent).

Summary characteristics of youth who either successfully appealed their initial cessation determination or successfully reapplied for payments are shown in Table 4, column 2. Compared with the population in column 1, a larger proportion of youth who successfully appealed the decision or reapplied for SSI has schizophrenia, psychoses, and other neuroses; mental retardation; and physical disabilities. A smaller proportion has "other" mental disorders. The gap between male participants and their female counterparts decreases, with the male group who successfully appealed or reapplied at 56 percent as opposed to 63 percent of the full initial cessation determination group. The initial cessation population also has a larger proportion with earnings of at least $250 at age 17 (31 percent versus 25 percent), which is not surprising because those with a work history have demonstrated an ability to work. Youth who successfully appeal or reapply are also less likely to have been initially allowed from 1991 through 1996, but in any other cohort, are more likely to be allowed.

There are only minor differences in the characteristics of individuals who successfully appeal and those who successfully reapply (columns 3 and 4). However, differences by level of appeal could be masked by this taxonomy. Individuals for whom we do not observe either a successful appeal or a successful reapplication (that is, who are censored after 4 years) are very different from those who did successfully appeal or reapply within 4 years. Comparing columns 5 and 2, a larger proportion of the censored group has "other" mental disorders (36 percent versus 28 percent), and a smaller proportion has mental retardation (17 percent versus 23 percent). The censored group is also composed of a larger proportion of male participants and more frequently were employed at age 17 (33 percent versus 25 percent).

The proportion of each characteristic group in the initially ceased population in each of the postredetermination events is shown in Table 5. We find that a large proportion (27 percent) of these early cohorts successfully appealed their determination or successfully reapplied for SSI. Certain groups are much more likely to return, however, such as individuals with schizophrenia, psychoses, and other neuroses (48 percent). Other groups are less likely to return, such as those who were initially entitled in the 1991–1996 period or who worked at age 17 (24 percent and 22 percent, respectively). Table 5 also shows how prevalent the appeals process is, with almost 20 percent of initial decisions overturned upon appeal. Comparatively, less than 8 percent successfully reapply for SSI within 4 years.

Odds of Successful Appeal or Reapplication

The results from logistic and multinomial logistic regressions of the likelihood of having a successful appeal or successful reapplication, controlling for all of the other individual characteristics, are presented in Table 6. The specifications are similar to those for the regressions in Table 3, with the inclusion of the reason for the initial cessation decision as an additional explanatory variable. Recall that this population is limited to those individuals who we could follow for 4 years after the initial cessation determination. The specification in column 1 (model 1) does not differentiate between a successful appeal and a successful reapplication. The results are similar to the patterns observed in the descriptive statistics in Tables 4 and 5.

Relative to youth with physical disabilities, those with schizophrenia, psychoses, or other neuroses; mental retardation; and sensory disabilities are significantly more likely to return to SSI within 4 years of their initial cessation determination. These groups are also less likely to receive an initial cessation determination. Youth with other/uncodable disabilities and "other" mental disabilities are less likely to successfully appeal or reapply for SSI than those with physical disabilities.

Controlling for the year of the redetermination and age at first SSI receipt, youth first allowed under PRWORA regulations (the post-1996 cohorts) are most likely to successfully appeal or reapply. The odds of successfully appealing or reapplying are 60 percent higher for the 1997–1999 cohort than for the 1991–1996 cohort; those allowed under the pre-1991 policies have a 43 percent higher odds ratio of successfully appealing or reapplying. The very large odds ratio for the post-1999 cohort should be taken with caution; the relatively small sample size (429) may be driving this result.³⁴

Female participants are much more likely to successfully appeal or successfully reapply than their male counterparts. Individuals who first became eligible before age 5 have 14 percent higher odds than those who became eligible at ages 5–12. Those who had at least $250 of earnings at age 17 have 32 percent lower odds of returning to the program than those who did not.

The odds of successfully appealing or reapplying if individuals fail to cooperate during the redetermination are 19 percent higher than if they did not meet the adult eligibility criteria. Individuals with an unknown initial level of adjudication are more likely to successfully appeal or reapply relative to those entitled at the initial application level. We also find that youth who had a prior CDR are more likely to successfully appeal or reapply for SSI after an initial cessation determination, and youth with more difficult cases—who required a CE—are less likely to successfully appeal or reapply than those who did not, all else equal.

Odds of Successful Appeal or Reapplication: Alternative Specifications

As indicated, a successful appeal is not differentiated between a successful reapplication in the specification in column 1 of Table 6. The specifications in columns 2–6 each demonstrate the difference between the two pathways under various econometric and operational assumptions. The specification in column 2 (model 2) models successful appeal against successful reapplication, conditional on returning. A logit regression was run on the returning sample, and the dependent variable indicates whether or not an individual successfully appealed. The specifications in columns 3 and 4 (models 3a and 3b) model each path separately using a logit regression for each pathway, relative to not using that path. If the individual was successful using the other path (that is, he or she successfully reapplied in the appeal model or successfully appealed in the reapplication model), that person is treated as not using the path in that specification—but instead, is treated identical to those in the censored group. Under somewhat stronger restrictions, the specification in columns 5 and 6 (model 4) estimates a multinomial logit regression of the two paths with individuals who do not successfully appeal or reapply (the censored group) as the reference.

Focusing on model 2 (Table 6), individuals with schizophrenia, psychoses, and other neuroses; and musculoskeletal disorders are more likely to be successful via the appeal route over reapplication, conditional on successfully appealing or reapplying for SSI within 4 years, relative to those with physical disabilities. Turning to the year-of-entry effects, the estimates do not indicate any difference between either of the post-1996 cohorts and the 1991–1996 cohort. Youth who entered the program before 1991, however, are more likely to be successful appealing than reapplying. Among the other effects, female youth who had a prior CDR, and youth first entitled to payments after the initial level of adjudication or with an unknown level of adjudication are more likely to be successful through the appeal route. Youth who had earnings greater than or equal to $250 and those who failed to cooperate during the redetermination are more likely to return through the reapplication route. Additionally, youth in later redetermination cohorts are less likely to return via the appeal route.

Models 3a/3b and 4 largely support the findings in models 1 and 2 (Table 6). The odds ratios are generally similar to those in model 1. When model 2 indicated that a successful appeal was more likely, the odds ratio for the "appeal" portion of the model is greater than that for the "reapplication" portion (and vice versa). For most groups, the effect on successfully regaining SSI payments is driven by the appeals process. For example, the odds of successfully appealing or reapplying among those who first became eligible before age 5 are 14 percent higher than for those who first became eligible at ages 5–12 (model 1). However, we found no difference in path conditional on returning (model 2), but models 3a/3b and 4 suggest that this effect is only significant with respect to the appeals choice.

Odds of Successful Postappeal Reapplication

The appeals process can be thought of as part of the redetermination process itself, as described earlier. The majority of youth who receive an initial cessation determination appeal the decision with a high level of success (SSA 2007b). In this section, we focus on youth who regain SSI eligibility through a postappeal reapplication. The population is limited to those who have not successfully appealed, do not have an open appeal, and whose 60-day appeal window has closed (technically), that is, the initial cessation determination became final.³⁵ The results of a logistic regression of this population successfully appealing (model 5) is shown in Table 6, column 7. These results are qualitatively quite similar to the estimates in models 1, 3, and 4, which is not surprising because almost all attempted reapplications occur after the appeal window has closed.

Even after the appeals process, youth with schizophrenia, psychoses, and other neuroses; mental retardation; and sensory disabilities are more likely to successfully reapply than those with physical disabilities. The lower likelihood of successful appeal or reapplication in model 1 that was found for youth with "other" mental disorders and other/uncodable disabilities mostly disappears. Additionally, youth with musculoskeletal disabilities are less likely to return to SSI through a postappeal reapplication, relative to youth with other physical disabilities.³⁶

Female youth are more likely to successfully reapply after the appeals process than their male counterparts. Individuals who had earnings greater than or equal to $250 at age 17 are less likely to have a successful postappeal reapplication than those who did not. Other youth more likely to have a successful postappeal reapplication include those who received an initial cessation determination for FTC, youth with prior CDRs, and those in later redetermination-year cohorts. Youth initially entitled at the ALJ or higher level of appeal and those who required a CE are less likely to have a successful postappeal.

Concluding Remarks

In this article, we present the characteristics and initial outcomes of youth with disabilities in the SSI program who have undergone the age-18 redetermination process as well as the likelihood of successfully appealing or reapplying for SSI. The age-18 redetermination is a major event in the lives of youth receiving SSI, with potentially long-lasting effects. Our results are largely consistent with previous research. We find that the characteristics of the redetermination population and the percentage with an initial cessation determination have remained stable over the 8-year period under study.

The analysis reveals that there are large differences in the probability of an initial cessation determination by demographic characteristics and program background. One important finding is that in recent years fewer youth are working before their redetermination than previously. Whether this decrease is due to a conscious effort to try to remain on SSI, because fewer job opportunities are available, or for other reasons is not identified. Additionally, youth with a history of work are less likely to successfully appeal or reapply for SSI payments after an adverse age-18 determination. This suggests that efforts to employ youth, such as SSA's Youth Transition Demonstration projects (Fraker and Rangarajan 2009), may help reduce long-term dependence on SSI.

Even controlling for several observable characteristics identifiable in administrative records, there are still differences in the risk of successfully overturning an initial cessation determination or successfully reapplying for SSI payments. Those youth with a higher likelihood of initially having an adverse redetermination are not necessarily those who have the highest likelihood of a successful appeal or reapplication. This suggests that the age-18 redetermination is being implemented in a manner consistent with the criteria of the decision process. For example, youth with "other" mental disorders are the most likely to be initially ceased, but are less likely to successfully appeal or successfully reapply relative to youth with physical disabilities. The strongly significant odds ratios of return to the program for individuals with schizophrenia, psychoses, and other neuroses; sensory disabilities; and mental retardation suggest that closer attention to these cases may be warranted during the initial redetermination decision. Targeting redeterminations to youth likely to have their payments ceased may allow SSA or the DDS to reallocate resources to the sizable backlog of other decisions.

The results also imply that the policy in effect at the time of initial entry may have lasting effects on SSI participation, significantly affecting the probability of initially receiving an adverse age-18 redetermination and of appealing that determination or reapplying for SSI. In particular, we find that youth who were originally allowed from 1991–1996, when Zebley policies were in effect, are much more likely than other cohorts to initially receive an adverse determination during their age-18 redetermination and are less likely to successfully appeal or reapply afterward. The source of this difference, whether this is due to selection issues, policies, or a variety of other factors, is not clear from these results and warrants further analysis.

Although we find that over one-quarter of the population that initially received an adverse determination successfully appealed or successfully reapplied, important questions remain. Do individuals who have been determined not to have a qualifying disability seek employment before attempting to appeal or reapply? When do they return to SSI? Is it after a few years of trying to become self-sufficient or immediately after payments officially cease? Why do they return to SSI? Do individuals who file new applications do so on the basis of the same impairments or new ones? Questions about whether certain groups are more likely to return before others also remain.

There are likely several factors not captured in this study, such as current employment and education, which address reasons individuals return or how they become reeligible. These factors will likely play a large role in any policy concerning the age-18 redetermination. Employment opportunities and education quite likely have large roles in this process. The results here and in Loprest and Wittenburg (2007) and Hemmeter, Kauff, and Wittenburg (2009) suggest that there are likely to be large differences in return to SSI by nonprogrammatic individual characteristics. Such results warrant further study.

Appendix