Few of the service choices states write into Medicaid waivers changed what home care aides and nurses earn or how often they were out of work, a study released Oct. 20, 2026 finds: 12 of 234 policy-outcome pairs showed an effect. The issue brief is from Jean-Pierre Aubry, Gal Wettstein and Nilufer Gok of the Center for Retirement Research at Boston College.
The team coded about 750 1915(c) waiver applications for older and disabled adults in 45 states and D.C. from 2006 to 2024, then compared American Community Survey data on aides and nurses in states that changed a policy with states that didn't. "Aides" covers home health aides, personal care aides, nursing assistants and orderlies. The brief describes the waivers studied as those for older and disabled adults; it does not report results for I/DD waivers separately.
What moved
All four effects found for aides were negative, as were five of the eight for nurses. Selected results:
| Waiver policy | Effect found |
|---|---|
| Self-directed home-delivered meals | Aides' annual earnings down $8,666 |
| Family-provided equipment, technology and modifications | Aides' annual earnings down $2,047 |
| Self-directed caregiver support | Aides' unemployment up 2 points |
| Self-directed or family-provided transportation, day and home-based services | Nurses' unemployment up about 1 to 4 points |
The authors say this isn't surprising: waivers aren't designed to change worker pay, and Medicaid generally pays below market rates. They call for more research on how self-direction and paid family care affect paid workers. The full brief (PDF) also finds aides' real earnings grew roughly in line with the national average from 2006 to 2024, leaving them at just under half the average for all workers.
Why this matters for providers and DSPs
Changing a waiver's service menu did little for direct care worker pay in this study, and the authors point to below-market Medicaid rates as one reason. For agencies, that is a data point to bring to rate talks. For DSPs, it's a reminder that the I/DD workforce is still missing from this kind of research. See our guide to self-direction and FMS for how those models work.