Syam Adusumilli
Chief Evangelist, GroundGame.Health
The Shift That Never Starts#
Darnell Williams clocks in at 6:47 AM at the Wendy’s on Martin Luther King Boulevard, thirteen minutes before his shift officially begins because that’s when the morning manager needs help prepping the breakfast station. He’ll work until 2:00 PM, then walk three blocks to the Burger King on Commerce Street, where he picks up another five hours most days, sometimes six when someone calls in sick. Between the two jobs, he averages 35 to 40 hours per week. Sometimes more.
Neither restaurant gives him a schedule more than a week in advance. Neither provides pay stubs automatically. At Wendy’s, his hours get tracked through a system called Kronos that he’s never seen and doesn’t have access to. At Burger King, the assistant manager writes hours in a spiral notebook that lives in the back office. When Darnell asked for a printout of his hours last month, the assistant manager looked at him like he’d asked for the nuclear codes.
“We don’t do that,” she said. “Corporate handles payroll.”
Corporate, it turns out, is a franchisee in another state who contracts with a payroll processing company whose customer service number routes to a recorded message explaining that employment verification requests must be submitted in writing with a seven-to-ten business day turnaround. Darnell doesn’t have seven to ten business days. His Medicaid work verification is due in four.
At Wendy’s, the situation is marginally better. The general manager is willing to provide a letter confirming Darnell’s employment, but she’s only there Monday through Wednesday, and Darnell works Thursday through Sunday at that location. When he finally catches her on a Monday, she writes a letter on lined notebook paper stating that “Darnell Williams works here” with her signature at the bottom. No hours. No dates. No letterhead. No contact information.
The state portal rejects the letter. The error message says verification must include “total hours worked per month” and “employer contact information” and must be submitted on “official letterhead or company documentation.” Darnell doesn’t know what “official letterhead” means, exactly, and when he tries to explain the situation to the Medicaid call center, the wait time is ninety-three minutes. He hangs up after forty because he has to get to his afternoon shift.
Two weeks later, his Medicaid terminates. The notice says he failed to verify work activity. Darnell stares at the letter in disbelief. He’s working forty hours a week. He’s been working forty hours a week for eight months. He has the pay stubs in his bank account, the muscle memory of the grill, the grease burns on his forearms.
He just can’t prove it.
The Arkansas Lesson#
When Arkansas became the first state to implement Medicaid work requirements in June 2018, it offered the country’s first real-world test of whether such policies could function as intended. The requirements seemed straightforward enough: adults aged 30 to 49 needed to work 20 hours per week, participate in qualifying activities like job training or community service, or obtain an exemption. Those who complied could keep their healthcare coverage. Those who didn’t would lose it.
Within ten months, more than 18,000 people lost their coverage. By the time a federal judge halted the program in April 2019, nearly one in four people subject to the requirements had been disenrolled from Medicaid for non-compliance.
The policy’s supporters had predicted that work requirements would promote employment, encourage personal responsibility, and transition people from public assistance to private coverage. What happened instead became a case study in the gap between policy design and administrative reality.
Researchers from Harvard examined what actually occurred. Their findings, published in the New England Journal of Medicine and Health Affairs, revealed something that should reshape how we think about work requirements entirely: the vast majority of coverage losses occurred among people who were already working or who qualified for exemptions. They didn’t fail the work test. They failed the verification test.
The Harvard research team found that 97 percent of the affected population was already meeting the requirements through work, disability, caregiving, or other qualifying activities. Only 3 to 4 percent were genuinely not working and not exempt. Yet 25 percent of the target population lost coverage. The arithmetic tells the story: coverage loss far exceeded actual non-compliance.
More than 70 percent of Arkansans subject to the requirements were unaware the policy was in effect or unsure whether it applied to them. Among those who did know about it, confusion about how to comply was pervasive. The state required monthly online reporting through a web portal, but many affected residents lacked reliable internet access. Others couldn’t navigate the portal interface. Some didn’t understand that failure to report, even while working sufficient hours, would result in coverage termination.
The policy produced zero measurable increase in employment. It produced significant increases in medical debt, delayed care, and forgone medications among those who lost coverage. It produced administrative chaos and legal challenges that ultimately led to its suspension. What it did not produce was any evidence that people who weren’t working started working as a result of the requirement.
This outcome was not an implementation failure in the sense of a good policy poorly executed. It revealed something more fundamental: the premise that coverage loss reflects behavioral non-compliance is flawed. Most coverage loss reflected administrative non-compliance, meaning people who couldn’t prove what they were already doing rather than people who weren’t doing what they were supposed to do.
The Verification Architecture Problem#
Work requirement verification systems are designed around an employer that increasingly doesn’t exist: stable, cooperative, documented, and equipped to provide standardized proof of employment on demand. The labor market that Medicaid expansion adults actually navigate looks nothing like this.
Consider the structural mismatch. Work requirements typically demand monthly hour-counting, usually 80 hours per month, with documentation submitted within narrow windows after each month concludes. This framework assumes workers have single employers who track hours systematically, provide accessible records, and respond promptly to verification requests. It assumes schedules are predictable enough that workers know in advance whether they’ll meet the threshold. It assumes the administrative infrastructure exists to document compliance.
The Medicaid expansion population, however, works in a labor market characterized by precisely the opposite conditions. Kaiser Family Foundation analysis shows that Medicaid adults are concentrated in industries with volatile scheduling, limited documentation, and minimal human resources infrastructure. Retail, food service, agriculture, construction, and domestic work together employ the majority of working Medicaid beneficiaries. These sectors feature seasonal fluctuations, variable hours, multiple part-time positions, cash payments, and informal employment relationships.
The gig economy compounds these challenges. Ride-share drivers, delivery workers, and platform-based service providers are classified as independent contractors with no employer to verify their hours at all. They may work substantial hours across multiple platforms with no centralized record of their activity. Bank statements showing deposits can’t distinguish between Uber earnings and birthday money from a grandmother. App-generated reports may not match the documentation format states require.
Multiple part-time jobs create their own verification nightmare. A worker with 20 hours at one restaurant, 15 hours at another, and 10 hours of informal childcare needs three separate verification sources to document a compliant 45-hour month. Each source operates on different timelines, uses different documentation systems, and may have different willingness or capacity to cooperate. Missing any single component renders the entire verification incomplete.
Seasonal and irregular work patterns fit poorly with monthly verification requirements. Agricultural workers who log 60-hour weeks during harvest season may have little documented work during winter months. Retail workers see hours surge before holidays and collapse afterward. Construction schedules depend on weather, permitting, and project timelines. The requirement to verify 80 hours each and every month penalizes the reality of how low-wage work actually operates.
The cash economy remains largely invisible to formal verification systems. Day laborers, house cleaners, yard workers, and informal caregivers may work substantial hours without any documentation trail. Asking an elderly neighbor for a signed letter confirming that you helped with grocery shopping and meal preparation raises practical and social barriers that formal employment verification does not. The work is real; the documentation is nonexistent.
Small employers, who employ a disproportionate share of low-wage workers, often lack the human resources infrastructure to respond to verification requests. The franchise owner of a single fast-food location may have no dedicated HR function. The landscaping company with eight employees may track hours on paper without any system for generating compliant documentation. Nearly half of working Medicaid beneficiaries are employed by companies with fewer than 50 employees, firms that are not required to provide health insurance and often lack basic administrative capacity.
The Exemption Documentation Challenge#
Work requirements include exemptions for those who cannot work or whose circumstances make work requirements inappropriate. Medical conditions, caregiving responsibilities, pregnancy, enrollment in education or training, and other situations can qualify someone for an exemption from the hour requirements. The theory is straightforward: people who genuinely cannot work shouldn’t lose healthcare coverage for not working.
The practice is considerably more complicated. Exemptions require documentation, and documentation requires navigating systems that may be even more burdensome than work verification itself.
Medical exemptions illustrate the challenge. A person with a chronic condition that limits work capacity needs a healthcare provider to complete exemption documentation. But many Medicaid beneficiaries lack established relationships with providers. They may use emergency departments for episodic care rather than maintaining primary care relationships. The provider who treated their back pain three months ago may not be available or willing to complete disability paperwork. New patients seeking exemption documentation face wait times for appointments that may exceed their verification deadlines.
Mental health conditions present particular documentation difficulties. Depression, anxiety, and other conditions that substantially impair work capacity may not be formally diagnosed. The symptoms that make employment difficult also make navigating bureaucratic systems difficult. Executive function impairment affects deadline management, form completion, and follow-through on documentation requirements. The very conditions that justify exemptions also make obtaining exemptions harder.
Caregiving exemptions assume formal arrangements that may not exist. A woman caring for her elderly mother doesn’t have an employer to verify her caregiving hours. She may not have formal documentation of her mother’s care needs. The work is invisible to administrative systems because it occurs within families rather than formal care settings. Verification may require medical documentation of the care recipient’s condition, affidavits from multiple parties, or other evidence that transforms unpaid family labor into bureaucratic paperwork.
Perhaps most troublingly, many people who qualify for exemptions don’t know they qualify. The exemption categories are numerous and their boundaries unclear. Does chronic back pain that prevents standing for extended periods qualify as a medical exemption? Does caring for a grandchild while a parent works count as caregiving? Does taking two community college classes satisfy the education exemption if you’re not enrolled full-time? The answers depend on state-specific definitions, documentation requirements, and administrative discretion that are opaque to most beneficiaries.
The administrative sophistication required to successfully claim an exemption may exceed the capacity of those most likely to qualify. Understanding which exemption category applies, gathering the necessary documentation, submitting it in the correct format through the correct channel by the correct deadline, and following up if something goes wrong requires a level of bureaucratic navigation that correlates inversely with many exemption-qualifying conditions.
The Profile of the Compliant But Terminated#
If compliance failures were randomly distributed across the affected population, they would represent implementation friction requiring process improvement. But the evidence suggests that compliance failures concentrate among specific populations whose characteristics systematically disadvantage them in documentation systems. Understanding who falls through helps explain why the gap between compliance and verification exists.
Educational attainment shapes administrative navigation capacity. Those with lower educational attainment are more likely to struggle with complex forms, written instructions, and online portals. They may have difficulty understanding verification requirements, identifying which documents satisfy those requirements, and composing written explanations when documentation doesn’t match standard formats. Administrative literacy, meaning the ability to successfully navigate bureaucratic systems, correlates with formal educational achievement.
Language barriers compound documentation challenges. Verification systems typically operate in English, with limited translation availability and variable quality of translated materials. A Spanish-speaking worker may understand the work requirement itself but struggle to comprehend portal instructions, error messages, or correspondence about missing documentation. Limited English proficiency affects not just direct system navigation but also the ability to seek help from call centers, navigate appeal processes, or advocate effectively when problems arise.
Digital access remains unevenly distributed despite the assumption of universal connectivity. Rural areas may lack reliable broadband internet. Low-income households may have mobile devices without unlimited data plans suitable for uploading documents or completing lengthy online forms. Public library computer access requires transportation, time, and digital literacy that are not universally available. Arkansas’s portal-only verification design effectively excluded populations without consistent internet access.
Social capital, meaning the network of relationships that can provide assistance, information, and advocacy, varies dramatically across the affected population. Someone whose sister works in healthcare administration can get help understanding verification requirements. Someone whose neighbor speaks English fluently can get help interpreting notices. Someone with no such connections must navigate alone. The documentation gap is partly a social capital gap.
Mental health conditions affect administrative capacity in ways that may not qualify for formal exemptions. Executive function challenges associated with depression, anxiety, ADHD, and other conditions make deadline management difficult. Procrastination, avoidance of stressful tasks, difficulty with organization, and impaired decision-making can all prevent timely verification even when the substantive requirements are met. The Medicaid population has elevated rates of mental health conditions compared to the general population.
Housing instability creates address-based failure modes. Verification notices mailed to outdated addresses never reach their intended recipients. Portal accounts tied to email addresses that change with phone plans become inaccessible. The documentation that proves work activity, including pay stubs, bank statements, and employer letters, gets lost in moves between temporary housing situations. People experiencing housing instability are overrepresented in the Medicaid expansion population.
The intersection of these factors creates compound disadvantage. A Spanish-speaking worker with depression, limited internet access, low educational attainment, and unstable housing faces documentation barriers at every turn. Each disadvantage multiplies the difficulty created by others. These are not rare edge cases; they describe substantial portions of the population work requirements are designed to reach.
Designing Systems That Work#
If the documentation gap is fundamentally a system design problem rather than a behavioral compliance problem, then solutions must focus on system redesign rather than stronger enforcement. The question shifts from “how do we make people comply” to “how do we verify what people are already doing.”
Automated verification should be the primary pathway for the vast majority of compliant beneficiaries. State agencies already have access to wage data through unemployment insurance systems, which track quarterly earnings for most formal employment. Matching Medicaid enrollment against wage records can automatically verify work hours for W-2 employees without requiring any individual action. Ohio’s approach to work requirements emphasizes this data matching strategy, attempting to verify compliance through existing administrative data before requiring individual documentation.
The limitation of automated verification is that it captures formal employment more reliably than the gig economy, informal work, or qualifying non-work activities. Data matching works well for a Medicaid beneficiary employed 40 hours per week at Walmart. It works poorly for someone who drives for Uber, babysits for neighbors, and takes care of an elderly parent. The system design must accommodate both.
Self-attestation with strategic audit represents an alternative to universal documentation. Rather than requiring every beneficiary to prove compliance, systems can allow beneficiaries to attest to their work status and apply verification requirements selectively to a sample of attestations. This approach, borrowed from tax administration, reduces burden on the compliant majority while maintaining program integrity through targeted review. The key is calibrating audit rates and consequences to maintain accurate self-reporting without creating universal documentation burden.
Presumptive compliance for populations with high demonstrated compliance rates offers another design option. If 97 percent of a population is already meeting requirements, treating that population as presumptively compliant and focusing verification resources on genuine non-compliance may be more efficient than universal documentation. The administrative cost of verifying the compliant 97 percent may exceed any savings from identifying the non-compliant 3 percent.
Outreach before termination represents a minimal intervention that could significantly reduce inappropriate coverage loss. Arkansas terminated coverage for members who failed to report, without meaningful outreach to determine whether non-reporting reflected non-compliance or merely administrative failure. Systems that contact members before termination, through multiple channels and with genuine assistance in completing verification, would distinguish between those who cannot comply and those who simply haven’t yet.
Multiple verification channels prevent single-point failures. Arkansas’s portal-only design guaranteed that anyone unable to access the portal would fail verification regardless of their work status. Systems offering phone verification, mail submission, in-person assistance, and mobile applications provide redundancy that accommodates diverse circumstances. The goal is ensuring that no one loses coverage solely because one specific verification method was inaccessible to them.
Distributed submission authority shifts verification burden from individuals to institutions. Rather than requiring workers to obtain documentation from employers and submit it to the state, employers can submit verification directly to state systems. Educational institutions can verify enrollment. Healthcare providers can verify medical exemptions. Community organizations can verify volunteer activities. This approach recognizes that institutions have documentation capacity that individuals often lack.
Georgia’s current approach to work requirements attempts to incorporate some of these design principles. The state has moved toward annual rather than monthly reporting, reduced portal dependence, and expanded verification channels. Early results suggest lower administrative failure rates than Arkansas experienced, though enrollment has been far below projections for reasons that include factors beyond verification design.
Reframing the Policy Debate#
The dominant framing of Medicaid work requirements assumes that coverage loss reflects behavioral non-compliance. Under this framing, losing coverage is a consequence people bring upon themselves by failing to meet obligations they could have met. The appropriate policy response is enforcement: clearer rules, stronger consequences, more rigorous verification.
The documentation gap reframes the problem. Most coverage loss reflects administrative non-compliance, not behavioral non-compliance. People lose coverage because they cannot prove what they are already doing, not because they refuse to do what they are required to do. The appropriate policy response is system redesign: better data matching, reduced documentation burden, more accessible verification channels, presumptive compliance where appropriate.
This reframe has profound implications for how we measure success. If coverage loss indicates system failure rather than behavioral change, then high coverage loss rates represent poor outcomes rather than policy working as intended. A work requirement that results in 25 percent coverage loss while producing zero employment gains has failed, not succeeded. It has burdened the working poor with administrative requirements they cannot meet while doing nothing to promote work among those who are not working.
The reframe also has implications for program integrity. Anti-fraud measures that produce more harm to compliant beneficiaries than they prevent in fraud represent net costs rather than net benefits. If documentation requirements cause 1,000 people to lose coverage for administrative failure for every one person caught committing fraud, the requirements are causing more harm than they prevent. Program integrity requires balancing fraud prevention against the administrative burden that prevents legitimate access.
The concept of administrative burden, developed by public policy scholars Pamela Herd and Donald Moynihan, provides theoretical grounding for this reframe. Administrative burden refers to the costs citizens bear when interacting with government programs: learning costs to understand requirements, compliance costs to document eligibility, and psychological costs of navigating bureaucratic systems. These burdens are not neutral features of program administration. They are policy choices that determine who successfully accesses benefits and who does not.
Herd and Moynihan argue that administrative burdens often function as policymaking by other means. When legislators cannot or will not directly restrict program eligibility, they can achieve similar effects by imposing administrative requirements that eligible populations cannot meet. The burdens appear neutral but operate selectively, excluding those with the least capacity to navigate complex systems.
Work requirements, viewed through this lens, may function less as genuine work promotion and more as coverage restriction through administrative burden. If most people are already working or exempt, and if work requirements don’t increase employment, then the primary effect of work requirements is coverage loss among people who cannot navigate verification systems. The question is whether that outcome reflects policy success or policy failure.
Different stakeholders will answer that question differently depending on their underlying views about Medicaid’s purpose, the deserving poor, and the appropriate role of administrative gatekeeping in safety-net programs. What the documentation gap analysis provides is not an answer but a more accurate framing of the question. The debate is not between promoting work and enabling dependency. It is between designing systems that verify compliance and designing systems that create barriers to coverage.
What Would Have Helped Darnell#
Return to Darnell Williams, working his two fast-food jobs, unable to obtain documentation that satisfies verification requirements. What would a well-designed system do differently for him?
Automated data matching would catch him first. His earnings from both restaurants flow through payroll processors that report wage data to the state unemployment insurance system. A system designed to match Medicaid enrollment against existing wage records would verify his work hours without requiring him to obtain employer letters at all. He meets the requirements. The data proves it. No documentation burden necessary.
If data matching failed because one employer paid in cash or operated outside normal payroll systems, alternative verification channels would provide backup. Darnell could photograph his bank deposits showing regular income and submit them through a mobile app. He could call a verification assistance line where a caseworker could help him document his employment through whatever evidence was available. He could visit a community organization authorized to submit verification on his behalf.
If his verification remained incomplete as the deadline approached, the system would reach out rather than terminate. A text message would alert him to the missing documentation. A phone call would offer assistance completing verification. A navigator would help him identify what was needed and how to obtain it. Coverage would continue while verification was pending, with termination occurring only after genuine, informed non-compliance was established.
The system would recognize that Darnell’s challenge is not unwillingness to work but inability to document. It would design around that reality rather than punishing it. It would treat verification as a system problem to solve rather than an individual obligation to enforce.
Darnell would keep his healthcare coverage. He would continue managing his hypertension and monitoring his borderline diabetes. He would show up for his shifts at both restaurants, taking orders and working the grill and earning the wages that were never in question. The only thing that would change is that a documentation gap would no longer stand between his work and his coverage.
The 18.5 million expansion adults who will face work requirements beginning in December 2026 include millions of people like Darnell. They are working. They will continue working. The question is whether the systems built to verify their work will function as neutral measurement or as barriers that transform working people into coverage casualties.
The documentation gap is not inevitable. It is a design choice. States can choose differently.