Panel Paper: Estimated Impacts for the Full SSDI Caseload (Stage 1)

Saturday, November 10, 2018
8228 - Lobby Level (Marriott Wardman Park)

*Names in bold indicate Presenter

David Wittenburg1, David Stapleton1, Stephen Bell2, Daniel Gubits3, Denise Hoffman1, Sarah Croake1, David Mann1, Judy Geyer3 and Michelle Wood3, (1)Mathematica Policy Research, (2)Westat, (3)Abt Associates, Inc.


The first presentation will summarize findings from Stage 1 of BOND, which includes an analysis of nearly one million SSDI beneficiaries who were assigned to a treatment group or a control group (i.e., a non-volunteer sample). A major advantage of this design is that it provides impact estimates for a nationally representative sample that the evaluation team could use to generalize outcomes under a national policy that is administered in a manner comparable to demonstration implementation. The evaluation tracked employment, earnings, and benefit outcomes and present impact estimates using regression-adjusted comparisons of outcomes between treatment and control subjects.

The findings indicate that the Stage 1 benefit offset intervention indicate no impacts on earnings in any of the five years after random assignment and an increase in average SSDI benefit amounts in all years. The increases in SSDI benefit amounts were small, a little over one percent of the control group average.

While no impact of the benefit offset on mean earnings was observed in any year, small impacts at several earnings thresholds were observed in the fourth and fifth years after random assignment. The benefit offset increased the share of treatment group members who were employed during these years and the share earning above the cliff amount. However, mean earnings remained essentially unchanged in these years because of reductions in earnings among higher-earning beneficiaries.

The presentation will discuss multiple possible explanations of why this large change in earnings rules did not induce more beneficiaries to earn above the benefit cliff.