Michigan is the only state that has both attempted work requirements with genuine investment in implementing them well and concluded, from its own experience, that they cannot be implemented without significant coverage losses. Robert Gordon, director of the Michigan Department of Health and Human Services under Governor Gretchen Whitmer, spent more than $30 million and a year building what he believed was the best possible Medicaid work requirement system. His team reprogrammed eligibility systems, designed plain-language communications tested with actual enrollees, built phone and online reporting channels, trained navigators, and established automatic deemed compliance for people already meeting work requirements through SNAP or TANF. When work requirements took effect on January 1, 2020, Michigan was as ready as any state had ever been. Gordon’s own analysis showed that more than 100,000 Michiganders were on track to lose coverage within the year before a federal judge struck down the waiver in March 2020.
That institutional memory makes Michigan’s second attempt the most informed in the country. Whether information translates to better outcomes when mandate comes from Congress rather than state legislature remains the central question. Michigan now faces federal mandate affecting its Healthy Michigan Plan’s approximately 711,000 to 750,000 adults, with MDHHS estimating between 100,000 and 290,000 beneficiaries could lose coverage in first year. State officials estimate administrative costs of approximately $75 million, more than double the $30 million spent on 2020, reflecting expanded scope and more complex federal requirements.
Michigan’s 2020 implementation, though brief, generated findings no other state possesses from direct experience. The state discovered that most enrollees were already meeting the law’s requirements. Roughly 60 percent of Healthy Michigan Plan enrollees were already working, enrolled in school, or serving as homemakers. Many others qualified for exemptions based on medical frailty, caregiving, disability, or age. The population that actually needed to change its behavior to comply was considerably smaller than the population that needed to navigate verification systems to prove existing compliance. This distinction, between the population that does not meet requirements and the population that cannot document meeting them, proved to be the defining operational challenge.
Michigan invested in human-centered design for member communications, testing all notices with actual enrollees before deployment. The state built multiple reporting channels including online portal, telephone hotline, mail, and in-person options to avoid the digital-only trap that contributed to Arkansas’s catastrophic outcomes. Navigator organizations participated in system design reviews. Cross-program deemed compliance automatically recognized SNAP and TANF work activity without requiring separate Medicaid documentation. Even with all of this, Gordon’s team projected that more than 100,000 people, many of them working or qualifying for exemptions, would lose coverage through verification failures rather than genuine non-compliance. Communication design could not overcome fundamental policy complexity.
The cost data is equally instructive. Michigan spent more than $30 million on implementation and would have spent more than twice that had the program continued. Reprogramming eligibility systems, training staff across 83 counties, conducting member outreach, and operating verification infrastructure consumed resources that could have been directed toward healthcare delivery. The state now estimates federal mandate will cost approximately $75 million in administrative expenses, in a program MDHHS describes as already “lean” with “less room to cut.”
Michigan’s geography imposes implementation constraints that compound administrative complexity. The state spans 83 counties across two peninsulas. Detroit and surrounding counties contain approximately 40 percent of expansion population with dense service infrastructure. The Upper Peninsula presents entirely different environment: limited broadband, no public transit, MDHHS offices requiring long drives, and seasonal tourism, mining, and forestry economies where hours fluctuate unpredictably.
Michigan’s political landscape shifted dramatically between 2020 and current federal mandate. Democrats gained full control after 2022 elections, repealed work requirement law in January 2025, but Republicans reclaimed House in November 2024. This divided government means Whitmer administration controls implementation through MDHHS but requires legislative cooperation for appropriations.
The federal mandate adds complexity that did not exist in 2020. Semi-annual redetermination cycles double verification frequency. The marketplace exclusion provision eliminates coverage fallback. The SNAP work requirements rolling out simultaneously create cross-program coordination demands the 2020 system did not face.
Michigan’s approach will emphasize protective implementation through the Whitmer administration’s control of MDHHS policy design. The state will likely pursue maximum exemptions, interpreting federal allowances expansively across disability, caregiving, pregnancy, medical frailty, substance use disorder treatment, and education categories. Cross-program deemed compliance will automatically recognize SNAP and TANF participation. Verification will follow data-first approaches using wage records and unemployment insurance data before requesting member documentation. The MDHHS presentation on H.R.1 prepared for state legislators included framework capturing the state’s analytical position: “Administrative Burden equals Coverage Loss. Many enrollees meet requirements, but may lose coverage due to inability to navigate complex reporting systems.”
Michigan is the experiment within the experiment. Every other state implementing work requirements for first time is doing so without operational experience. Michigan alone has design templates, cost data, navigator training materials, system specifications, and outcome projections from actual implementation. The state knows what $30 million buys and what it does not. It knows that human-centered design improves but does not solve communication challenge. It knows that deemed compliance reduces but does not eliminate verification burden. It knows that even under most favorable conditions, six-figure coverage losses are projected.
Whether Michigan’s institutional memory translates to meaningfully better outcomes depends on variables state does not fully control. Federal guidance may permit or restrict exemption interpretations and deemed compliance provisions that formed backbone of Michigan’s protective design. Divided government may support or starve administrative investment that implementation requires. Michigan’s answer will carry more weight than any other state’s, precisely because Michigan has already answered the preliminary question that every other state is still asking: can this be done well? Michigan’s experience suggests it can be done better or worse, but that gap between “as well as possible” and “without significant harm” may not be closeable.