Elder care: does workforce or funding actually cap home care?

The baseline: Medicaid pays for this, not Medicare — and only after the savings are gone

More than half of Americans turning 65 will develop a severe long-term-care need before death — 56% under HHS's current microsimulation model. The 70% figure still in circulation is also HHS's, from an older survey-based study with a different definition; cite the 56% and say which model it came from. Medicare covers acute and narrow post-acute care; it is not the long-term-care payer most people assume. Medicaid, reached only after a household spends down its assets, covers 61% of the $415B this country spends annually on long-term care; within Medicaid specifically, home- and community-based spending overtook institutional spending in 2013. That 61% is KFF's long-term services and supports construct, which excludes Medicare post-acute care by definition — and we should have said so, because the other standard construct answers differently. In CMS's National Health Expenditure Accounts for 2024, Medicare is the largest single payer of home health care in the country ($55.7B of $169.4B, 32.9%, ahead of Medicaid's 22.6%) and the third-largest payer of nursing-facility care ($47.3B of $219.9B, 21.5%). Across both settings Medicare pays 26.5 cents of every dollar to Medicaid's 30.1. No single canonical "LTSS spending" figure exists — federal accounting categories mix short-stay Medicare rehab with long-stay Medicaid custodial care, and we previously named that problem only in the direction that understates Medicaid, never the one that understates Medicare.

The worker problem: wages rose, vacancies didn't fall, and nobody's proven which caused what

The best natural experiment available is a 2021 funding surge — $37.1B in state and federal money that states planned to spend, 71% of it on workforce recruitment and retention: provider rate increases, wage add-ons and hiring bonuses, not wages alone. Wages moved substantially where states spent it that way — Colorado's direct-care wage rose from $12.41 to roughly $18/hour — but vacancy and turnover did not improve to match, and the national HCBS waiting list did not fall: it has sat at roughly 0.7 million in most years since 2016, and stood above 710,000 in 2024.

That pattern fits the workforce theory. It equally fits economy-wide, sector-agnostic wage inflation touching nearly every low-wage service job in 2021–2024 — a confound nobody has ruled out. The program carried no evaluation requirement, and the federal assessment that could settle this is still pending. The honest verdict is open, not proven.

Two things we should have said the first time. Only 5% of that money ($1.7B) was aimed at waiting lists at all, so a flat waiting list is weaker evidence against the funding theory than we made it sound. And this is Medicaid home-care data, which is mostly not elder-care data: 63% of Medicaid home-care users are under 65, and most people on these waiting lists have intellectual or developmental disabilities. Both facts are in our own sources; neither was on this page.

Money and insurance: the last federal program died before it opened; the new one has no track record yet

The CLASS Act, a federal long-term-care insurance program enacted in 2010, collapsed in 2013 without ever enrolling a person, opening a claim, or collecting a premium — HHS's own 19-month review found voluntary signup let only the sickest people buy in, and no benefit design could price around it. Washington State's WA Cares removes exactly that channel with a mandatory 0.58% payroll tax, but its $36,500 lifetime cap versus CLASS's proposed open-ended benefit means the two programs are not a clean contrast — a design that fails on adverse selection and one that fails on an inadequate benefit could look identical from outside. WA Cares benefits first became payable on July 1, 2026, weeks before this report; there is no outcome data yet.

A softer fix was tried and didn't pencil out: Washington modeled a state reinsurance backstop for private insurers instead of a public benefit, and its 2015 feasibility study found "little potential to generate savings" — suggestive evidence against that specific design, from one state, reported secondhand, not a verdict on every possible reinsurance mechanism. The standalone individual LTC insurance market has shrunk from 125 carriers (2000) to roughly 7–10 today.

Who's waiting, and why: nursing homes are a legal right, home care is a waiting list

Nursing home care is a mandatory Medicaid entitlement; home- and community-based care runs through optional, capped waivers with waiting lists in most states — backwards from the "home care is cheaper, so of course it's prioritized" intuition. The national waitlist was undersold in this project's own starting assumptions: roughly 710,000 people by 2024, about 40% above the seeded prior. Indiana matches the theory: a waitlist that grew from 12,800 to over 17,000 in nineteen months, atop a spending mix sending two-thirds to three-quarters of Medicaid long-term-care dollars to institutional care. But it is one state, and the state most commonly cited for this argument — Texas — is the one state where the relevant spending data is unknowable, excluded from the two national trackers that would answer the question.

Two things we left out, and the second one is ours to own. First, the cap has a reason: Congress made home care available by waiver on the statutory condition that it cost no more per person than the institutional care it replaces (42 U.S.C. §1396n(c)(2)(D)). Limiting enrollment is one of the two levers a state has to meet that test — so a waiting list is partly what a budget-neutral optional benefit looks like, not only what a preference for institutions looks like. Second, Congress has already partly removed the cap, in the very statute we cite two sections up for killing the staffing rule. Pub. L. 119-21 §71121 — in a subchapter titled "Expanding Access to Care" — lets the Secretary approve, from July 1, 2028, standalone waivers covering home care for people who do not meet the institutional level-of-care test, on state-set needs-based criteria, with $150 million appropriated to build the systems. We scored HCBS de-capping as though it were untried policy. It isn't. The enacted version also keeps the cost cap (the state must attest per-person spending won't exceed institutional per-person spending), requires that the new waiver not materially lengthen the wait for people already in the queue — Congress wrote queue competition into the law — and bars the money from paying for direct-care workers' health insurance or training where Medicaid is that class of practitioners' primary source of revenue.

Institutional quality: the federal staffing rule is dead, twice over

The 2024 federal nursing-home minimum-staffing rule was not merely under litigation risk — it is dead. Two federal courts vacated it, Congress barred CMS from implementing it until September 30, 2034, and CMS formally repealed it effective February 2026. State law is the surviving lever, but it is an uneven one: before the federal rule existed there was no numeric federal staffing floor at all, and 29 states still require under 3.5 hours per resident day, 15 under 2.5. No state staffing law has been struck down — but New York shows statutory survival isn't enforcement: roughly 400 violating facilities identified with almost no penalties for years.

Separately, on much stronger evidence: private-equity ownership raised short-stay mortality by 11% for the patients steered into those homes, in a 4.2-million-patient study designed specifically to correct for which patients end up where. No direct econometric rebuttal has been found — and the authors themselves report the harm falls on a subset of patients, with small benefits for others.

Assisted living's blind spot: still the smaller setting, and still uncounted

This inquiry assumed assisted living had overtaken nursing homes in resident count. The data says otherwise — roughly 1.24 million nursing-home residents versus 1.016 million in assisted living — but the comparison is three years stale on one side, and since nursing-home census is flat-to-declining while assisted living keeps growing, the actual current gap could be smaller or closed. What holds: the federal government still can't see this setting. GAO found in 2018 that 26 of the 48 states it studied couldn't report how many critical incidents happened in Medicaid-funded assisted living; GAO found in 2026 that Washington can't even count what it spends there — "at least $12 billion" in 2024, which GAO itself calls likely an undercount. Two different blind spots, eight years apart. The incident-reporting fix won't take effect until 2027 at the earliest.

Family caregivers: no honest dollar figure, one genuine trial

This inquiry's own kill condition fired: unpaid family caregiving's economic value cannot be reconciled to a single defensible figure. Independent, credible estimates run from $234 billion to $1.01 trillion — and part of that spread is simply that they measure different years, from 2011 data to 2024 data. Strip the calendar out and two estimates for the same era still differ by more than 2× ($234B and $522B), on population scope, wage-rate assumption, and hours source: three named, defensible methodological choices, not chaos. There is no "the" number.

What is solid: 63 million Americans provide unpaid care, over a quarter reduced work hours because of it, and CDC's own national health survey shows caregivers reporting worse mental health than non-caregivers, with the gap wider at the end of the last decade than the start — though that endpoint sits inside the pandemic, so read the level as solid and the trend as suggestive. The one architecture in this entire report anchored by an actual randomized controlled trial — Cash and Counseling, which lets Medicaid recipients hire their own caregivers, including family — measurably improved satisfaction and health. It did not save money; it delivered more of the authorized care rather than the same amount more cheaply.

What other countries did: three systems, all mid-crisis, none a finished blueprint

Germany, Japan, and the Netherlands are the usual comparators for LTC financing, and all three show real, current strain rather than a settled model to import. Germany's contribution rate rose again for 2025, and reporting says federal loans were needed to hold the 2026 rate steady — a claim we have not been able to source to a German government document, so treat it as reported rather than verified. Japan projects a 570,000-worker shortfall by 2040 and is already proposing copay increases toward 30%. The Netherlands' 2015 reform is documented as working against its own cost-containment goal, with a projected 266,000-worker shortfall by 2035. Score all three as donors of design fragments — partial coverage, age-40 pre-funding, severity-tiered financing splits — not whole-system precedents.

The scorecard: the best-evidenced fix is already law in every state, and nobody has measured whether scaling it worked

Nine candidate architectures, scored 1–5 against five objectives. Only Cash and Counseling is anchored by an actual randomized trial rather than a projection, a single-state study, or plausibility. It wins decisively on caregiver relief regardless of weighting — but its own trial data show a real cost increase, so an honest scorecard lets that pull its cost-containment rank down too. It does not win "no matter what," and our first-pass scorecard overstated that it did.

Then a second, harder re-score moved 25 of these 40 cells, and the most important thing it found was how little this board actually knows. Our first re-score was told not to look at the scorecard; this one was structurally prevented from reaching it. It caught a defect in our own scoring scale: an architecture nobody had researched got marked down, because "no evidence found" was a low score while a high score required evidence. Twenty cells sitting at 1 or 2 turned out to rest on no evidence at all — in either direction. Corrected, 30 of the 40 cells sit at dead neutral, and the differences between most of these instruments disappear. We are withdrawing the finding we stated most confidently: that private long-term-care insurance reform "ranks last under all three weightings, with no exceptions — the clearest, most stable negative finding in the whole scorecard." Four of that row's five cells were empty. It was the least stable finding on the board, not the most. The wage floor also no longer leads on workforce, where it now ties the do-nothing cash comparator — which means our own sequencing argument gets no support from our own scorecard, and we say so rather than let the two look like they agree.

Then a steelman pass went after the three things this filing says loudest, and two of them broke. We had headlined this section "the best-evidenced fix isn't the one anyone's proposing." That is withdrawn: it is false. Congress enacted the Cash and Counseling design as a permanent state-plan option in 2005 (42 U.S.C. §1396n(j)) and again in the ACA as Community First Choice (§1396n(k)), which pays states a six-percentage-point federal match bonus to adopt it. Self-direction now runs in all fifty states and DC and about 1.5 million people use it, up 23% since 2019. Our own research file said so; this page said the opposite. Worse for us: the one study that evaluated the statutory version found nothing — the same paper we cite for this board's strongest positive cell reports no significant workforce effect for Community First Choice (1.51%, CI −12.77% to 15.79%) alongside the Balancing Incentive Program's +13.24%, and we published only the half that helped. The Cash and Counseling trial still stands, and it is still the only randomized evidence on this board. What falls is the idea that this is an untried fix waiting to be proposed. It has been law for twenty years, it has been scaled, and nobody checked.

One more, on the row we just moved to last place. Federal LTC social insurance ranks last under all three weightings on the corrected board, and we said it earned that "on evidence." One cell carries it: a workforce score marked down because Japan projects a 570,000-worker shortfall and the Netherlands 266,000. Those are facts about ageing countries, not measurements of what the insurance does — and the country without such a program, this one, faces 9.7 million direct-care openings over the same decade by our own anchor table. Score that cell at neutral, as our corrected scale says an unevidenced cell should be, and the row ties for last under all three weightings instead of holding it alone. We are publishing that as a sensitivity rather than changing the cell, but the honest summary is that our replacement headline rests on the same kind of thin cell as the headline it replaced.