Poster Paper: Enrolling Denied Disability Applicants with Reported Mental Disorders in Supported Employment: New Evidence from the Social Security Administration (SSA) Supported Employment Demonstration (SED)

Saturday, November 9, 2019
Plaza Building: Concourse Level, Plaza Exhibits (Sheraton Denver Downtown)

*Names in bold indicate Presenter

David Salkever1,2, Jeffrey Taylor3, William Frey3 and Jarnee K. Riley3, (1)University of Maryland, Baltimore County, (2)Johns Hopkins University, (3)Westat

A substantial majority of applicants for SSDI/SSI disability who allege a mental illness receive a denial from SSA at the initial adjudication level. SSA has identified this large group of denied applicants as persons for whom immediate post-denial intervention, with evidence-based supported employment services, could foster improved outcomes (in employment, earnings, quality of life, and self-sufficiency), and thereby delay or eliminate future need for disability benefits.

SSA is now implementing a randomized trial of a new Supported Employment Demonstration (SED) that tests such an intervention for these denied applicants. The SED provides supported employment services and other supports intended to help these individuals return to the labor market as quickly as possible. It employs the “place then support and/or train” model of individual–placement-and-support supported employment (IPS-SE). Between December 2017 and March 2019, 3,000 denied applicants from 30 communities across the country were randomized into one of two intervention groups or a control group. The study will follow each enrollee for 3 years, assessing impacts of SED services and supports on employment, earnings, benefit allowance, and clinical recovery.

Policy implications of the SED depend both on impact assessments of IPS-SE services on study enrollees, and on generalizability to broader applicant populations. Previous SSA employment-services demonstrations for disabled beneficiaries showed relatively low take-up rates. No similar studies exist of take-up by applicants denied benefits or other non-beneficiary groups. Thus, the match between the study population and the larger population of similar denied applicants is critical for assessing SED generalizability.

With SED enrollment completed, we are now undertaking descriptive analyses and multiple regression analyses to understand key enrollee-vs.-non-enrollee differences. This paper will present results of the descriptive analyses and initial results of our regression analyses.

SED recruitment included initial mail contacts, follow-up phone contacts, brief screening interviews, and in-person meetings to fully explain the study. Recruitment statistics show attempted mail contact for 21,126 denied applicants, follow-up phone contact for 19,371, positive initial screening for 15,128, follow-up informational meeting attendance for 3,454, and final enrollment of 3,000.

Including small numbers of exclusions for other administrative ineligibility criteria (at contact, screener, meeting, or enrollment phases), the final enrollment rate was nearly 20%. This clearly exceeds rates in prior demonstrations for persons on disability benefits.

Results of our ongoing analyses, presented at the conference, will include detailed comparisons of enrollees vs. non-enrollees at the time of recruitment on:

  • Individual characteristics (e.g., demographics, socioeconomics, health, work history);
  • County of residence characteristics (e.g., economic and labor market conditions); and
  • State of residence characteristics (e.g., availability of cash and in-kind benefits, Medicaid expansion versus non-expansion).

We will also report/describe non-enrollees’ reasons for non-enrollment.

Regression analyses will incorporate enrollee, county, and state characteristics into reduced-form regressions modeling overall recruitment outcome, and structural models of intermediate outcomes (speaking to applicant, passing screener, attending meeting).

To guide future replication or implementation efforts with the SED model, regression results will also be used to identify categories of denied applicants whose predicted probabilities of enrolling appear highest based on the variables in our regressions.