Measuring the health effects of Supplemental Security Income (SSI)

• In Event: Time after time: Lessons from three behavioral science research portfolios in government

Abstract

The Social Security Administration (SSA) is committed to improving the quality of service delivery and streamlining operations to deliver accurate, secure, and timely support to all Americans. This evaluation builds evidence on the impact of outreach to older adults by analyzing the effect of Supplemental Security Income (SSI) enrollment on mortality. The prior evaluation (https://oes.gsa.gov/projects/increasing-ssi-uptake/) encouraged enrollment in SSI for eligible older adults with low-cost, informational letters. In 2017, 400,000 (out of a sample of approximately 4 million) older adults who were potentially eligible for SSI were randomly assigned to be mailed letters describing their potential eligibility and information about the application process. The letters led to an estimated 6,960 additional enrollments in SSI, an increase in SSI enrollment of 1.8 percentage points.

SSI, administered by the SSA, provides monthly payments to people with disabilities and older adults who have limited or no income or resources. At the time of the earlier evaluation (2017), an eligible individual could receive a maximum monthly payment of $735 if they met income and asset limitations and other eligibility requirements. In most states, enrollment in SSI makes it easier to enroll in other health-protective programs, including Medicaid and SNAP. 

SSI participation could reduce mortality via two related mechanisms: by increasing disposable income through monthly benefit payments and by facilitating access to additional health-promoting programs (such as Medicaid or SNAP). Random assignment to the informational letter group led to some people applying and enrolling in SSI only because they were randomly selected to be sent a letter. We can use this randomly induced enrollment to estimate the effects of an SSI award on recipients’ mortality.

Our analysis used two models to estimate the effects of award receipt on mortality: a Cox Proportional Hazards Model estimated through a control function approach, and a two-stage least squares regression. Data from the prior evaluation and from the SSA Master Beneficiary Record (MBR) were used to observe mortality across the letter and no letter groups between September 15, 2017 and January 1, 2022. The MBR data includes whether an individual enrolled in SSI and whether they died during the observation period (and if so, when). The data also includes individual characteristics such as sex, date of birth, address, WEP/GPO indicator, and an estimate of the potential SSI payment amount the person might receive if enrolled.

The two pre-registered models we estimated both indicate a decrease in mortality, but the size and statistical significance of the reduction differed between the two models. The Cox proportional hazard model produces a hazard ratio of 0.61 (p = .001, [0.513–0.708]). This estimate implies a decrease in mortality of 4.8 percentage points during the study period. The two-stage least square regression also indicates a decrease in mortality of 3.4 percentage points, but this estimate is less precise and falls short of statistical significance (p = 0.31, 95% CI [-0.100, 0.032]). The mortality rate during the study period among potentially eligible older adults who were not sent the letter was 12.25%.

Author

• Elana Safran, Office of Evaluation Sciences / U.S. General Services Administration
  Presenting Author