Medicaid Can Now Pay for Housing Help. Be Precise About What That Means.

By Common Ladder · June 18, 2026 · 8 min read

For most of Medicaid's history, the program drew a hard line: it paid for medical care, not for housing. A doctor could treat the pneumonia a person caught sleeping outside, but Medicaid would not help that person find a room to recover in — even when everyone involved knew the room would do more for their health than the antibiotics. That line is now blurring, and the shift is one of the more consequential things happening in homelessness policy that almost no one outside the field is talking about.

The change is real, and it matters. But it is also easy to overstate, and overstating it sets people up for disappointment at the eligibility desk. So this brief tries to do two things at once: take the development seriously, and be precise about its limits.

What Medicaid will — and won't — pay for

The mechanism is a tool called a Section 1115 demonstration waiver, which lets a state test approaches that fall outside Medicaid's normal rules. Beginning in 2022, the federal government started approving waivers that let states spend Medicaid dollars on what it calls "health-related social needs" — the non-medical conditions, like housing and food, that drive so much medical cost. By early 2025 roughly 18 states had approval to do some version of this.¹

What that buys, in practice, is housing services more than housing itself. The covered list typically includes housing transition and navigation services — someone to help a person search for a unit, gather documents, and get through an application — plus tenancy-sustaining services like tenant-rights education, landlord mediation, and eviction prevention. It also covers one-time setup costs: a security deposit, first-month essentials, and home modifications for accessibility or safety.⁴ In a smaller number of states, including Arizona, a waiver can also cover short-term rent or temporary housing for up to six months while a person transitions out of an institution, a shelter, or the child welfare system.²

Here is the line worth underlining: Medicaid is not becoming a rent-subsidy program. It will not pay your rent indefinitely, and it is not a substitute for a Housing Choice Voucher or a public-housing unit. What it funds is the connective tissue around housing — the help that gets someone into a unit and keeps them there — and, at most, a bridge of temporary rent during a defined transition. Treating it as a voucher will only produce frustration.

Does it work? The early evidence says yes — modestly

The strongest data so far comes from North Carolina, whose Healthy Opportunities Pilots have been paying for housing, food, and transportation supports since 2022 and building an evaluation in alongside the program. An independent assessment by the Cecil G. Sheps Center for Health Services Research at the University of North Carolina, reported in 2026, found that participants cost Medicaid about $164 less per month than they otherwise would have — on the order of roughly $1,000 a year per person — while using fewer emergency-department visits and hospital stays and more lower-cost primary care.³ An earlier interim look had estimated savings closer to $85 a month, so the later, larger analysis pointed in the same direction, more strongly.

This fits a longer line of research on what happens when you stabilize where someone sleeps. Studies of medical respite care — short-term recuperative housing for people discharged from the hospital while homeless — have found that patients sent to respite used fewer hospital days than similar patients who were not, suggesting that a stable place to heal prevents the bounce-back admissions that drive cost.⁴ None of this is a silver bullet, and the cleanest savings estimates come from observational and quasi-experimental work rather than randomized trials, so they should be read as strong signals rather than final numbers. But the direction is consistent: meeting a housing need can lower medical spending, not just improve lives.

Arizona's version: the H2O program

Arizona is one of the states putting this into practice. AHCCCS — the state's Medicaid agency — runs the Housing and Health Opportunities demonstration, known as H2O, which began enrolling members on October 1, 2024. Its stated goals are exactly the two threads above: stabilize members' health and reduce the cost of care by cutting crisis, emergency-department, and inpatient use.²

It also illustrates how narrow the front door is. AHCCCS is starting with its most acute population, and the criteria stack up: to be flagged as potentially eligible, a member generally must be experiencing homelessness (or housing instability), and carry a serious mental illness (SMI) designation, and have a qualifying chronic health condition — or be transitioning into or out of a correctional facility within 90 days.² Referrals are generated through state data and routed to a program administrator, Solari, which confirms eligibility against HUD's definition of homelessness before connecting the member to a provider. As of this writing, the program is not taking open external referrals. In other words, this is targeted, clinically gated help for people with the highest combined health-and-housing risk — not a benefit the average person experiencing homelessness can simply sign up for.

The federal door narrowed in 2025

Any precise account has to include what changed in Washington. On March 4, 2025, CMS rescinded the Biden-era informational bulletins and framework that had spelled out — and encouraged — how states could use Medicaid to address health-related social needs, housing among them.⁵

What the rescission did not do is cancel programs already approved. Arizona's H2O and the other states' housing-services waivers remain in effect and continue to operate. What it changed is the road ahead: CMS signaled it will weigh new health-related-social-needs requests case by case rather than under a standing invitation, which makes launching a brand-new state housing-services program slower and less certain than it looked a year ago. For anyone tracking this field, the honest summary is that the tool still exists and is still being used where it was already approved, but the federal posture toward expanding it has cooled.

What this means for practice

For a case manager or a person navigating the system, the practical takeaways are concrete. First, if you or a client has both a serious health condition and a housing crisis, it is worth asking the health plan specifically about Medicaid housing services — this is real coverage that many eligible people don't know exists. Second, set expectations honestly: this is navigation, tenancy support, and at most short-term transitional rent, not a permanent subsidy, so it works best paired with HUD programs rather than in place of them. Third, because eligibility is data-driven in Arizona, keeping a member's record current — the housing Z-codes, the SMI designation, the chronic-condition diagnosis — is what actually gets them flagged.

The bigger picture is that Medicaid quietly became a homelessness-response tool, backed by early evidence that meeting housing needs can lower medical costs. That is genuinely good news. It is also a narrow, clinically targeted, and now federally uncertain tool — not the rent program the headlines can imply. Both halves of that sentence are true, and saying only the first one does no one a favor.

Have a correction or an updated figure? Email corrections.commonladder@gmail.com.