A Medi-Cal Plan Just Decided Assisted Living Costs Too Much — And 3,500 Californians Become the Experiment on December 31

An Insurer Put a Number on the Thesis, and the Number Was No

Health Net, one of the larger Medi-Cal managed care plans in California, has notified the state that it is ending its assisted living benefit — the CalAIM Community Supports offering that pays for 24-hour services in board-and-care homes, memory care facilities and similar residences. The benefit ends December 31, 2026, and roughly 3,500 low-income seniors are enrolled (Skilled Nursing News, August 24, 2026, reporting on CalMatters). CalViva Health and the Community Health Plan of Imperial Valley, which contracted through Health Net, are affected as well.

The termination itself is a California story. The reason is not.

In a termination notice filed with the California Department of Health Care Services, Health Net said the decision was driven partly by a rise in patients moving into assisted living directly from their own homes rather than from nursing facilities — a pattern the company says increases cost rather than generating savings (CalMatters, August 2026).

Read that again with an underwriter's eye. The entire policy case for Medicaid-funded assisted living rests on diversion: it is cheaper to serve a frail person in a board-and-care home than in a skilled nursing facility, so the state saves money by moving people down the acuity ladder. Health Net is not disputing the price gap. It is disputing who walked through the door. If the enrollee was never going to a nursing home in the first place, the plan did not avoid a $10,000 month — it added a $6,000 one.

The Arithmetic Everyone Agrees On

Nobody in this dispute disagrees about the unit economics. Monthly assisted living costs generally run $5,000 to $7,000, while nursing home care can exceed $10,000 (CalMatters, August 2026). CalAIM — California's broader Medi-Cal overhaul — was explicitly designed to cut repeat emergency room visits and expensive hospital stays by funding lower-cost settings (California Health Care Foundation).

So the spread is real and roughly 40 to 50 percent. The savings are only real if the counterfactual is a nursing home. That conditional is doing all the work, and it is the part nobody was measuring carefully enough to defend when a plan decided to stop paying.

This is a distinction worth internalizing well outside California. Any argument that a lower-acuity setting saves a payer money is an argument about substitution, not about price. A program that grows by attracting people who would otherwise have stayed home is, from the payer's ledger, a program that grows cost — no matter how good the care is or how much the residents prefer it.

What Happens to the 3,500

The population is not marginal. Many residents receiving the benefit have dementia or similarly high care needs (Skilled Nursing News, August 24, 2026). Pauline Shatara, deputy director of California Advocates for Nursing Home Reform, told CalMatters the fallout is "going to be a disaster."

The mechanics are messier than the headline date suggests. Some provider contracts expire October 7, 2026, even though the benefit itself is supposed to run through December 31 — so a subset of families face an earlier cliff, or at minimum a period of not knowing which applies to them (Skilled Nursing News, August 24, 2026). New enrollees, meanwhile, may struggle to get approved for costly long-term care at all during the wind-down.

Where do they go? The sector's assumption is nursing homes and hospitals — which is precisely the outcome CalAIM was built to prevent, and precisely the reversal that makes this worth an investor's attention rather than only an advocate's.

The Receiving End Is Not Ready

Here is the part that turns a benefits story into a capacity story.

Between 2019 and 2024, the number of licensed SNF beds declined only modestly — but operating capacity declined by 5%, according to a study published in the National Library of Medicine earlier this year. The researchers found capacity declines were larger once staffing shortages were factored in, and that nursing home capacity for new admissions varies wildly by geography.

Licensed beds are not staffed beds. A building can hold a license for 120 and admit against 90.

Layer on payer behavior: operators are still limiting admissions and reducing services because of reimbursement inadequacy under Medicare Advantage, per a Skilled Nursing News survey conducted in June 2026. A nursing home that is already declining admissions on margin grounds does not become an eager taker of displaced Medi-Cal assisted living residents with dementia — arguably the least attractive admission on a rate-per-hour-of-care basis.

So the displaced population meets a receiving sector that has less usable capacity than its bed count implies and an active financial reason to be selective.

What We Would Actually Underwrite Differently

Three concrete adjustments, none of them dramatic:

1. Treat Medicaid-funded assisted living revenue as a policy position, not a contract. This benefit was not cut by the legislature or by CMS. It was cut by a managed care plan filing a notice with a state agency. That is a much shorter decision chain and a much faster clock than most rent rolls assume. If a community's census depends on a Community Supports line item, the question for diligence is not "is the program funded?" but "which plan pays it, and what is that plan's stated view of whether it saves them money?"

2. Ask where the residents came from. Health Net's rationale means referral origin is now an underwriting variable. A Medicaid-heavy assisted living community whose admissions come predominantly from the community rather than from institutional discharge is, by the payer's own logic, the profile most exposed to the next termination notice. That is a data point most operators can produce and few are asked for.

3. Stop treating licensed beds as capacity in any market-absorption model. The 2019–2024 evidence is that the gap between licensed and operable is material and geographically uneven. Where we are testing whether a submarket can absorb displaced acuity, staffed capacity is the only number worth modeling.

The Honest Uncertainty

We do not yet know how many of the 3,500 land in nursing homes, how many return to family care, and how many churn through hospitals first — and we will not have clean data on that for several quarters. We are also reporting a single plan's decision, not a statewide repeal: other Medi-Cal plans continue to offer the benefit today, and Health Net has said members will continue receiving services through the stated period.

What is not uncertain is the precedent. A payer looked at the diversion thesis behind Medicaid-funded assisted living, decided the substitution was not happening, and exited. Anyone whose model assumes that thesis holds should know it has now been tested in public — and, in at least one plan's judgment, failed.

---

*Crawford Commercial Group Research. Figures carry their sources and as-of dates inline. Where the public record conflicts or is incomplete, we say so rather than fill the gap.*