14 — Child Care, Elder Care & the Care Economy

STAGE 1 — EVIDENCE BRIEF

Referee panel: Prof. Olivia Brandt (#57, statistics, C) · Prof. June Takahashi (#17, econometrics, C) · domain specialist Prof. Hannah Okafor (#9, health economics, C — long-term-care financing)

Scope note: Thread 10 handled the CTC/EITC. This thread is about care: the price, supply and workforce of child care; public provision vs. parental choice; paid leave; and the elder-care system (long-term services and supports, LTSS). All figures below were checked against the linked source this session (today: 2026-09-24). Where a source is advocacy or administration-produced rather than peer-reviewed, we say so.

---

A. Child care: prices, reach, supply

1. National average price, 2025: $13,184/yr. Equals ~10% of median income for married-couple families and ~33% for single-parent families. Licensed centers fell 1% (2024→25, first decline after years of growth); family child care homes rose 1.4%. — Child Care Aware of America (CCAoA), Child Care in America: 2025 Price & Supply, released May 2026. https://info.childcareaware.org/media/child-care-prices-rival-major-household-expenses
2. County-level prices (2022 data): full-day care for one child ranged $6,552–$15,600/yr across 2,512 counties, i.e. 8.9%–16% of county median family income. — DOL Women's Bureau, National Database of Childcare Prices (NDCP), Nov 2024 release. https://www.dol.gov/newsroom/releases/wb/wb20241119
3. Subsidy reach (FY2021): 11.5M children eligible under federal CCDBG rules, 8.0M under (stricter) state rules; 1.8M received subsidies = 15% of federally eligible, 22% of state-eligible. — HHS/ASPE. https://aspe.hhs.gov/reports/child-care-eligibility-fy2021 (CCAoA's 2026 release repeats the 15% figure.)
4. "Child care deserts" (2018): 51% of Americans lived in census tracts with >3 young children per licensed slot; rural 59%, urban 56%, suburban 44%; Maine <23% of neighborhoods, Utah >75%. — Center for American Progress (left-of-center; definitional choice matters — the 3:1 threshold is CAP's). https://www.americanprogress.org/article/americas-child-care-deserts-2018/

B. Child care workforce

1. Wages, May 2025 (BLS OEWS): childcare workers 518,910 employed, median $16.82/hr; preschool teachers median $18.34; all occupations median $24.51. https://www.bls.gov/news.release/ocwage.t01.htm
2. Turnover and competition: childcare worker median wage sits in the bottom 5% of occupational medians; monthly exits of 6.5% to non-participation and 4.7% to other jobs, mostly to other low-wage jobs (cashiers, retail, waitstaff); 2019–23 wage growth 5.5% vs 5.9–17.6% for competing service jobs. — Chicago Fed Insights, 2024. https://www.chicagofed.org/publications/chicago-fed-insights/2024/childcare-labor-market
3. Industry employment (BLS CES, "Child Care Services," SA, thousands): Feb 2020 1,048.6 → Apr 2020 676.6 → Sep 2023 (ARPA expiry) 1,062.7 → Sep 2024 1,091.5 → Sep 2025 1,093.9 → Aug 2026 1,088.2. — FRED series CES6562440001. https://fred.stlouisfed.org/series/CES6562440001

C. The ARPA stabilization "cliff"

1. Size: ARPA's Child Care Stabilization program paid ~$24B to >225,000 providers (capacity up to 10M children); funds expired Sept 30, 2023. After expiry: price declines relative to trend stalled; share of households with young children unable to access care rose from 24% to 31% (Q3 2023→Q1 2024); county-level relative gains in maternal LFP "slow to a halt" but no significant declines observed; states that fully replaced ARPA funds with stopgaps saw ~2.5 pp higher LFP gains among mothers of young children. — White House CEA, June 2024 (administration analysis, not peer-reviewed). https://bidenwhitehouse.archives.gov/cea/written-materials/2024/06/27/impacts-of-the-expiration-of-federal-child-care-stabilization-funding-and-the-mitigating-effects-of-state-level-stopgap-funding
2. The ex-ante forecast: The Century Foundation (June 2023) projected >70,000 programs closing, 3.2M children losing care, 232,000 educator jobs lost, $9B/yr in lost parental earnings. https://tcf.org/content/report/child-care-cliff/ — Referee note: compare with fact 7; industry payroll employment rose ~29,000 in the year after expiry.

D. State and foreign models

1. New Mexico universal child care. Income limits removed Nov 1, 2025. Funded chiefly by two oil-and-gas-fed endowments: the Land Grant Permanent Fund (~$33B; extra 0.75% distribution ≈ $248M/yr) and the Early Childhood Trust Fund (~$10B; 5% ≈ $500M/yr). — Fiscal Policy Institute. https://fiscalpolicy.org/how-new-mexico-will-pay-for-universal-childcare
2. NM early results (state brief, Apr 2026): children served 32,861 (Sep 2025) → 39,907 (Dec 2025); 18,099 new children determined eligible; **54% of new enrollees had incomes below the old 400% FPL cap (i.e., previously eligible non-takers); infant/toddler share of new enrollments 39%; $18/hr educator wage floor built into cost model; child care assistance cost FY26 $445.8M ($328.3M state), FY27 $606.4M ($488.9M state)**; licensed capacity 59,274 (2019) → 71,509. — NM ECECD. https://www.nmececd.org/wp-content/uploads/2026/04/Universal-Child-Care-Brief_41526.pdf
3. NM strain (Sep 2026): ~47,000 children enrolled, 58,000 projected; providers report rising waitlists and insurance costs; the wage-share requirement was lowered from 57% to 50% of revenue. — Source NM. https://sourcenm.com/2026/09/10/new-mexico-childcare-providers-struggle-as-demand-waitlists-increase/
4. Quebec, short run: Baker, Gruber & Milligan (NBER w11832, 2005; JPE 2008): large maternal labor supply increase (elasticity 0.236), ~1/3 of new use crowded out informal care; children "worse off in a variety of behavioral and health dimensions, ranging from aggression to motor-social skills to illness"; more hostile/inconsistent parenting. https://www.nber.org/papers/w11832
5. Quebec, long run: Baker, Gruber & Milligan, AEJ: Economic Policy 11(3), 2019: negative noncognitive effects persisted; cohorts with more access had worse health, lower life satisfaction, and higher crime rates. https://www.aeaweb.org/articles?id=10.1257%2Fpol.20170603
6. Quebec, labor-market side (Fortin, 2019 presentation): LFP of mothers with children 0–5 rose 64%→80% (1997–2018) vs 67%→71% in the rest of Canada; ~70,000 additional mothers employed by 2008 (+3.8%); claims governments recovered >100% of cost in 2008; 2019 subsidy $2.9B (0.6% of Quebec GDP); fees $5 (1997–2004) → $7 → sliding scale; quality: in for-profit "garderies" only 10% of children got "good" care and 36% "inadequate," while nonprofit CPEs performed well. https://www.oise.utoronto.ca/home/sites/default/files/2023-10/pfortin_childcare_newyork_0719_final.pdf
7. DC Early Childhood Educator Pay Equity Fund (PEF, launched 2022): Mathematica finds it increased the local child-care/early-ed labor supply ~7% (≈1,500 additional slots) with a one-year social ROI of 23%. https://www.mathematica.org/projects/washington-dc-early-childhood-educator-pay-equity-fund-impact-and-cost-effectiveness-study . Urban Institute (May 2025, 305 centers, 2023–24): 64% of educators stayed at the same center; educators preferentially moved toward funded centers. https://eric.ed.gov/?id=ED673560

E. Federal tax provisions changed in the July 2025 reconciliation law (OBBBA)

1. §45F employer child care credit: 25% → 40% of qualified expenditures (50% for small businesses, <$32M receipts); cap $150K → $500K ($600K small), inflation-indexed; effective 2026. — Bipartisan Policy Center. https://bipartisanpolicy.org/explainer/45f-employer-provided-child-care-tax-credit-2026-guide/
2. Old 45F take-up was tiny: in 2016, only 169–278 corporate returns claimed an aggregate $15.7–18.8M. — GAO-22-105264. https://www.gao.gov/products/gao-22-105264
3. Dependent care FSA/DCAP exclusion $5,000 → $7,500 (not indexed) from 2026; CDCTC max rate 35% → 50%, phasing to 35% then 20% at higher AGI; expense caps unchanged at $3,000/$6,000; credit remains nonrefundable. — Mercer summary. https://www.mercer.com/en-us/insights/us-health-news/big-beautiful-bill-permanently-enhances-dependent-care-benefits/

F. Parents' labor supply and fertility

1. Parents' LFP 2025 (BLS): mothers with children <18: 73.9% (74.0% in 2024, "little changed"); mothers with youngest child <6: 68.0%; youngest 6–17: 78.2%; fathers: 93.7%. (Oct 2025 excluded — shutdown.) https://www.bls.gov/news.release/famee.htm
2. Births: 3,628,934 in 2024 (+1%); general fertility rate 53.8 per 1,000 women 15–44 (−1%), down 22% since 2007. — CDC/NCHS Data Brief. https://www.cdc.gov/nchs/products/databriefs/db535.htm
3. Child care prices → fertility (working paper): Dow, using state ratio/group-size regulations as instruments for county NDCP prices (2010–22): a 10% price increase → 5.7% lower birth rate (≈4 births per 1,000 women 20–44). https://abigaildow.com/assets/docs/dow_childcare_fertility.pdf
4. Subsidy → fertility (peer-reviewed): state CDCTC adoption had no detectable effect on births or birth outcomes (95% CI −31 to +23 births per 1,000) but raised married mothers' LFP. — Review of Economics of the Household, 2026. https://link.springer.com/article/10.1007/s11150-026-09861-1

G. Paid family leave

1. Coverage: as of May 2026, 13 states + DC pay benefits (CA 2004, NJ 2009, RI 2014, NY 2018, WA, DC 2020, MA 2021, CT 2022, OR 2023, CO 2024, DE, MN, ME 2026); MD (2028) and VA (Dec 2028) pending — 15 jurisdictions. Payroll contributions ≤1.3% in every state in 2026. No federal paid leave program (federal FMLA leave is unpaid). — New America. https://www.newamerica.org/insights/explainer-paid-leave-benefits-and-funding-in-the-united-states/
2. CA PFL, short run: leave-taking by new mothers roughly doubled (≈3 → 6–7 weeks), largest for disadvantaged mothers; suggestive +6–9% weekly hours for employed mothers of 1–3-year-olds. — Rossin-Slater, Ruhm & Waldfogel, NBER w17715. https://www.nber.org/papers/w17715
3. CA PFL, long run (tax data, RD design): no positive effects on women's employment, earnings, or fertility; first-time mothers had lower employment and earnings a decade later. — Bailey, Byker, Patel & Ramnath, AEJ: Economic Policy, 2025. https://www.aeaweb.org/articles?id=10.1257%2Fpol.20200277
4. Firm side: state paid parental leave laws (CA, NJ, RI) reduced establishment employment of multi-unit firms by 1.5–7.9% and payroll by 5.6–8.3%; no abnormal stock returns. — Chicago Fed WP 2024-12 (working paper). https://www.chicagofed.org/-/media/publications/working-papers/2024/wp2024-12.pdf

H. Elder care / LTSS

1. Private-pay prices, 2025 national medians: nursing home semi-private $114,975, private $129,575; assisted living $74,400; in-home non-medical caregiver $80,080; adult day $24,700. — CareScout (Genworth) 2025 Cost of Care Survey. https://investor.genworth.com/news-events/press-releases/detail/1054/carescout-releases-2025-cost-of-care-survey-results
2. Medicaid is the dominant payer: paid 61% of $459B in U.S. LTC spending (2023); 6.3M Medicaid LTC users (4.9M home care, 1.4M institutional) — 6% of enrollees but 36% of Medicaid spending. — KFF Health Policy 101. https://www.kff.org/medicaid/health-policy-101-medicaid/?entry=table-of-contents-what-long-term-care-ltc-is-covered-by-medicaid
3. 2025 reconciliation law and LTC (per KFF/CBO): federal Medicaid spending cut >$1T over 10 yrs (KFF's HCBS page cites $911B, ~14%); ~10M more uninsured by 2034; provider-tax limits −$226B; home-equity cap for LTC eligibility frozen at $1M from 2028 (−$195M); retroactive coverage cut from 90 days to 1–2 months from 2027; nursing-home staffing rule enforcement moratorium through Oct 1, 2034 (−$23B); HCBS is optional and thus most exposed when states cut. https://www.kff.org/medicaid/health-provisions-in-the-2025-federal-budget-reconciliation-law/ · https://www.kff.org/medicaid/medicaid-home-care-hcbs-in-2025/
4. Staffing rule repeal: CMS repealed the 2024 minimum standards (3.48 total nurse HPRD; 0.55 RN; 2.45 aide; 24/7 onsite RN) on Dec 2, 2025, reverting to 8 consecutive RN hours/day. https://www.aha.org/news/headline/2025-12-02-cms-repeals-minimum-staffing-requirements-skilled-nursing-long-term-care-facilities
5. Direct care workforce (2024): 2.3M LTC direct care workers; 66% in home care; 85% female; 30% immigrants; share ranges from 0% (WY) to 60% (NY); 32% themselves on Medicaid; 66% earn <$35,000. — KFF. https://www.kff.org/medicaid/what-role-do-immigrants-play-in-the-direct-long-term-care-workforce/ Home health & personal care aides: 4.31M employed, median $17.21/hr (BLS, fact 5 source).
6. Immigration scenario (projection, advocacy source): ~4 in 10 home health aides are immigrants; a 4M-deportation scenario would cut ~394,000 direct care jobs (274,000 immigrant, 120,000 U.S.-born), built on Kreider & Werner (2025). — EPI (left-of-center). https://www.epi.org/blog/trumps-deportation-plans-threaten-400000-direct-care-jobs-older-adults-and-people-with-disabilities-could-lose-vital-in-home-support/
7. Unpaid family caregivers: 63M in 2025 (~1 in 4 adults; ~50% more than 2015); 7 in 10 employed. — AARP, Caregiving in the US 2025. https://www.aarp.org/pri/topics/ltss/family-caregiving/caregiving-in-the-us-2025/ Economic value: 59M caregivers of adults, 49.5B hours, $1.01 trillion/yr at $20.41/hr. — AARP Valuing the Invaluable 2026. https://www.aarp.org/press/releases/2026-03-26-AARP-Economic-Value-Of-Family-Caregiving-Report.html
8. Aging: Census 2023 projections (middle series): 65+ outnumber under-18s by 2029; natural decrease (deaths > births) from 2038; 65+ reach 29.1% of population by 2100. https://www.census.gov/newsroom/press-releases/2023/population-projections.html

---

CONTESTED EVIDENCE (where the literature genuinely disagrees)

C1. Does universal child care harm children? Baker-Gruber-Milligan (facts 13–14) find persistent harm in Quebec. Counter-considerations: (a) Fortin's own data (fact 15) show sharply different quality between nonprofit CPEs and for-profit garderies, so "universal care" and "low-quality universal care rapidly scaled" are not the same treatment; (b) Norway's staged expansion (Havnes & Mogstad, AEJ: Economic Policy 2011, "No Child Left Behind") found positive long-run education/earnings effects [cited from the literature; not re-fetched this session]; (c) effects are averages over mostly middle-class children whose counterfactual was parental care — targeted high-quality programs for disadvantaged children are a different estimand. Takahashi: the Quebec estimate is credible for Quebec's design; it is not a universal constant.

C2. Did the ARPA cliff cause a collapse? TCF's forecast (fact 9) was not borne out in payroll employment (fact 7). CEA (fact 8) reports deteriorating access and stalled maternal-LFP gains, with a dose-response relationship to state stopgaps. Both can be true: the sector did not collapse, but prices/access worsened at the margin. CEA's work is not peer-reviewed and was produced by an interested administration.

C3. Does paid leave help mothers' careers? Short-run leave-taking and hours gains (fact 25) vs. no long-run gains and negative first-time-mother effects (fact 26); firm-side costs (fact 27, working paper). Designs differ (DiD vs RD on tax data).

C4. Do child care costs depress fertility? An IV working paper finds large elasticities (fact 22); a peer-reviewed DiD on the CDCTC finds none (fact 23) — though the CDCTC is a small subsidy, so a null there has limited power against a large-price-change hypothesis.

C5. Did Quebec "pay for itself"? Fortin's >100% fiscal recovery (fact 15) depends on attributing employment gains causally and on the counterfactual; Baker-Gruber-Milligan's crowd-out finding (1/3 of new use displaced informal care) cuts against full self-financing.

C6. Nursing-home staffing floors: proponents cite quality; the repeal's rationale is feasibility and closures in a 20%+ shortfall labor market. We found no official mortality estimate for the repeal and flag any such claim as needing a source.

C7. Immigration and the care workforce: the 30%-immigrant share is descriptive (KFF); job-loss numbers under deportation are projections from a single enforcement-elasticity study (EPI via Kreider & Werner).

— Brandt, Takahashi, Okafor

STAGE 2 — OPENING POSITIONS (Part 1 of 2)

Format for each: causal claim · evidence · remedy · tradeoff conceded · falsifiable prediction.

---

Prof. Monica Reyes (#49, early childhood, L) — "Child care is infrastructure; price it like infrastructure"

Causal claim. The U.S. child care market fails structurally: the cost of care is dominated by labor under ratio rules, so prices that families can bear cannot pay wages that retain workers. The result is simultaneous unaffordability and shortage. Only sustained public funding breaks that loop; the ARPA cliff proved what happens when you pull it away.

Evidence. A $13,184 average price is 33% of a single parent's median income (CCAoA 2025). Only 15% of federally eligible children get CCDBG help (ASPE FY2021). Childcare workers earn $16.82/hr, bottom-5% territory (BLS; Chicago Fed). When ARPA stabilization expired, the access gap rose from 24% to 31% and maternal LFP gains stopped (CEA 2024). And the cliff shed hundreds of thousands of child care jobs, just as TCF warned. New Mexico shows the alternative: 39,907 children served by December 2025, up from 32,861, with an $18/hr wage floor.

Remedy. A federal-state universal system: capped family copays (0% below 200% FPL, ≤7% of income above), direct operating grants tied to a wage floor, CCDBG as the chassis. New Mexico is the pilot the nation should scale.

Tradeoff conceded. It is expensive — New Mexico's FY27 child care assistance line alone is $606M for a state of ~2.1M people — and NM is using an oil endowment most states don't have.

Prediction. If I'm right, New Mexico's CPS-measured LFP for mothers of children under 5 will rise ≥3 pp relative to synthetic-control neighbors by 2029, with licensed capacity growth ≥10% over the same period.

---

Prof. Rosa Delgado-Finch (#7, political economy, Soc) — "Decommodify care"

Causal claim. Care is a textbook Baumol sector: productivity can't rise without degrading the service, so relative prices rise forever. Markets respond by cutting quality or wages. The real solution is public provision with public employees, as in the Nordic model, not subsidies that leak into for-profit margins.

Evidence. Quebec is the proof of concept: the program paid for itself — governments recovered more than 100% of cost (Fortin) — while mothers' LFP rose 16 points. And Fortin's quality data show the failures were concentrated in for-profit garderies (10% "good," 36% "inadequate"), not the nonprofit CPEs. Subsidizing private operators buys you the worst of both worlds. Meanwhile the private LTC market charges $114,975 for a semi-private nursing home bed (CareScout 2025) while 66% of direct care workers earn under $35,000 (KFF).

Remedy. Publicly operated child care centers co-located with schools; a public home-care agency option in Medicaid; unionized sectoral wage boards for care work.

Tradeoff conceded. Public operation reduces parental choice of provider type (including faith-based providers) and moves slowly; unions raise costs.

Prediction. Where states expand public/nonprofit capacity (NM, DC), measured quality ratings among newly added slots will be higher than in for-profit additions within 3 years.

---

Prof. Nkechi Adeyemi (#45, nursing, L) — "Elder care is running on unpaid women and immigrant aides, and 2025 policy just knocked out both legs"

Causal claim. The LTC system rests on Medicaid (61% of LTC spending) and on a workforce of low-paid, disproportionately immigrant aides. The 2025 reconciliation law's Medicaid cuts, the staffing-rule repeal, and immigration enforcement will together shrink paid care supply and push more work onto families.

Evidence. Medicaid paid 61% of $459B in LTC (KFF, 2023). Federal Medicaid cuts exceed $1T over ten years; HCBS is optional and first on the chopping block. The staffing floor (3.48 HPRD, 24/7 RN) was repealed Dec 2, 2025 — a move that saves nursing homes money at the cost of residents' lives. And immigrants are 40% of direct care workers, so deportations hit this sector hardest.

Remedy. Restore and enforce a staffing floor phased to local labor markets; protect HCBS with an enhanced FMAP; create a care-worker visa and TPS protection for current aides; raise Medicaid HCBS rates with pass-through requirements to wages.

Tradeoff conceded. Staffing floors without money can close rural facilities; higher HCBS rates cost states real money under the new provider-tax constraints.

Prediction. By 2029, states cutting HCBS waiver slots will show rising nursing-home admission rates among dual-eligibles relative to states that don't.

---

Prof. Miriam Castellano (#1, labor econ, C-L) — "It's the wage, stupid — and we have a causal estimate now"

Causal claim. The binding constraint on child care supply is the reservation wage of workers who can earn more at Target. Raise pay with public money and supply follows; the evidence is now causal, not just intuitive.

Evidence. DC's Pay Equity Fund raised the local child-care/early-ed labor supply ~7% (≈1,500 more slots), with a 23% one-year social ROI (Mathematica). Educators preferentially moved to funded centers (Urban 2025). The Chicago Fed shows exits flow to cashier and retail jobs — a wage-competition story. The DC fund cut turnover roughly in half.

Remedy. A federal matching grant for state workforce compensation funds (DC-style), conditioned on wage pass-through and independent evaluation — not blanket universal pricing.

Tradeoff conceded. Wage subsidies raise the cost per slot and won't lower sticker prices for families on their own; DC is a rich, dense city with an unusual tax base.

Prediction. States adopting PEF-style funds should show child care services employment growth ≥5% above non-adopting states within 3 years (BLS QCEW), with no increase in posted prices beyond trend.

STAGE 2 — OPENING POSITIONS (Part 2 of 2)

---

Prof. Deborah Klein (#59, gerontology, C) — "LTC financing is the sleeper crisis; target, don't universalize"

Causal claim. The U.S. has no LTC insurance system; it has a middle-class spend-down into Medicaid, backed by a vast unpaid family workforce. As the population ages, that arrangement becomes fiscally and socially unsustainable. The fix is a targeted catastrophic backstop plus real support for family caregivers — not a universal entitlement we can't finance.

Evidence. Semi-private nursing home care: $114,975/yr; in-home care $80,080 (CareScout 2025). Medicaid pays 61% of LTC; LTC users are 6% of enrollees but 36% of Medicaid spending (KFF). Family caregivers provide care worth about $600 billion a year (AARP). And the oldest cohorts are growing fastest.

Remedy. (1) A federal catastrophic LTC backstop after a 2–3-year waiting period (Washington State's WA Cares is the domestic analog); (2) a caregiver tax credit plus funded respite; (3) protect HCBS relative to institutional care.

Tradeoff conceded. A backstop with a long waiting period leaves the first years to families and private insurance, which few buy; any backstop raises payroll taxes.

Prediction. Absent reform, the share of LTC spending paid by Medicaid will not fall below 55% through 2035 despite the reconciliation cuts — cuts will shift costs to families, not reduce Medicaid's role.

---

Prof. Daniel Mbeki (#70, demography, C) — "Demography is destiny, but child care is not a fertility lever"

Causal claim. Aging is the dominant force: the care-dependency ratio will rise regardless of policy. Child care subsidies are justified on labor-market and child-welfare grounds; selling them as pronatalism overpromises.

Evidence. Older adults will outnumber children by 2034 (Census). Births were 3.63M in 2024, GFR 53.8, down 22% since 2007 (CDC). The peer-reviewed CDCTC study finds no fertility effect (Rev. Econ. Household 2026). Child care costs have no meaningful effect on fertility.

Remedy. Plan for the care workforce as demographic infrastructure: immigration pathways for care work, HCBS capacity, and honest projection-based budgeting. Evaluate child care on employment and child outcomes, not births.

Tradeoff conceded. If some child care cost reductions do raise fertility at the margin, my framing undersells them.

Prediction. New Mexico's GFR will not diverge from neighboring states' by more than ±2% by 2030 attributable to universal care.

---

Prof. Abigail Stroud (#63, family sociology, R) — "Fund families, not a system"

Causal claim. Most families with infants use or prefer parental and relative care; universal center-based subsidies tax those families to subsidize one arrangement, and the best evidence on universal center care for young children (Quebec) shows harm. Give parents the money and let them choose.

Evidence. Quebec made children more aggressive, less healthy, and — in the long run — more criminal (Baker-Gruber-Milligan 2008, 2019). Universal child care causes crime. Most mothers of young children would prefer to stay home if they could afford it. Mothers' labor force participation fell in 2025 because care is unaffordable and families are giving up. And NM's own data show 54% of new enrollees were already eligible under the old income cap — universalism is mostly re-labeling.

Remedy. A flat, per-child infant/toddler benefit (ages 0–3) usable for any care — including a parent staying home — financed by redirecting some center-based subsidy growth; make the CDCTC's value available to stay-at-home families.

Tradeoff conceded. Cash reduces maternal employment relative to a work-conditioned subsidy; some cash will not be spent on children.

Prediction. If flexible cash is offered alongside subsidized center care (as in several Nordic "cash-for-care" systems), a majority of parents of 1-year-olds will choose the cash option.

---

Prof. Walter Pruitt (#8, law & econ, Lib) — "The price is regulated in; deregulate supply, and stop mandating costs"

Causal claim. Child care prices are high partly because state regulation (ratios, group sizes, credentialing, zoning) raises the cost of every slot, and paid-leave mandates raise the cost of employing parents. Supply-side liberalization is the cheapest pro-family policy available.

Evidence. Dow's IV paper uses precisely those regulations as a source of price variation — they bite. State paid leave laws cut establishment employment 1.5–7.9% and payroll 5.6–8.3% (Chicago Fed 2024). The ARPA cliff apocalypse never came: payroll employment in child care rose from 1,062.7K (Sep 2023) to 1,091.5K a year later. And the old 45F was a dead letter (169–278 corporate claimants in 2016); quadrupling the cap won't change the economics much.

Remedy. By-right zoning for family child care homes; ratio flexibility for 3–5-year-olds within evidence-based bounds; portable credentials; let immigration supply care workers; no new federal mandates.

Tradeoff conceded. Looser ratios can reduce quality for infants; zoning preemption offends local control.

Prediction. States that relax preschool-age ratio or zoning rules will see licensed family-child-care home counts rise ≥10% relative to controls within 3 years without measurable change in inspection violation rates.

---

Prof. Nathaniel Cross (#64, religious studies, R) — "Subsidiarity: the family is the first care institution"

Causal claim. Care is relational before it is economic. Policy that treats unpaid family care as a gap to be filled by paid labor misreads what most families want and crowds out the institutions (family, congregations, neighborhoods) that do most of the work — 63 million family caregivers (AARP 2025).

Evidence. California's paid leave didn't help women's long-run careers or fertility (Bailey et al. 2025), which suggests mandates don't deliver the promised goods. Family caregivers supply care worth more than all Medicaid spending (AARP 2026). Faith-based providers are a substantial share of child care supply, and universal systems with wage and curriculum mandates tend to push them out.

Remedy. A refundable family caregiver credit; religious-provider neutrality in any subsidy; paid leave via voluntary, portable accounts rather than a new payroll tax.

Tradeoff conceded. Voluntary accounts help higher earners most; low-wage workers may never accumulate enough.

Prediction. In NM, the number of faith-based licensed providers will decline relative to secular ones by 2029 as mandates (wage share, reporting) tighten.

STAGE 3 — CROSS-EXAMINATION (Part 1: the child care exchanges)

---

X1. Pruitt (#8, Lib) → Reyes (#49, L): "Where are the hundreds of thousands of lost jobs?" — four rounds

Pruitt, round 1. Monica, you said the cliff "shed hundreds of thousands of child care jobs, just as TCF warned." The BLS establishment series says otherwise: Child Care Services employment was 1,062.7K in September 2023 when ARPA expired, 1,091.5K in September 2024, 1,093.9K in September 2025 (fact 7). That is +29,000 in the first year, not −232,000. TCF also forecast 70,000 closures and 3.2 million children losing care. The forecast was wrong by an order of magnitude and in sign. Why should anyone trust the same modeling for "the cost of not going universal"?

Reyes, round 1. I'll take the correction on the employment number — I conflated TCF's forecast with the outcome, and the CES series is the right yardstick. But "no collapse" isn't "no harm." The CEA's household data show the share of families with young children unable to get care rising from 24% to 31% in two quarters, and a dose-response: states that fully replaced ARPA dollars got ~2.5 pp more maternal LFP growth. Payroll counts can rise while access falls if providers raise prices and shift toward higher-paying families.

Pruitt, round 2. Your dose-response result comes from an administration analyzing its own program in an election year, unrefereed, using eleven self-selected stopgap states plus DC. The states that chose to backfill (DC, New Mexico, Massachusetts-types) are richer, more urban, with tighter labor markets; selection on the outcome's trend is the obvious confound. And if the access problem were the binding one, we'd see parental LFP drop. BLS says mothers of children under 18 were at 74.0% in 2024 and 73.9% in 2025 — "little changed."

Reyes, round 2. Flat is not fine when the pre-cliff trend was rising, which is precisely what the CEA reported: gains "slow to a halt." I agree the stopgap-state comparison is not an RCT; I'd call it suggestive. But notice what your framing concedes: public money was doing something while it flowed — the dispute is magnitude, not sign.

Pruitt, round 3. I concede sign, not magnitude: $24B spread over 225,000 providers plausibly held prices down temporarily. My claim is narrower — that the market absorbed the withdrawal far better than the advocates predicted, which is evidence that supply is more elastic than the "structural market failure" story implies.

Reyes, round 3 (amendment on the record). Amended: "The ARPA cliff did not produce the job losses forecast; industry payroll employment rose after expiry (BLS CES). The evidence of harm is in access and stalled maternal-LFP gains (CEA 2024, non-peer-reviewed)." I do not concede supply elasticity: employment in 2026 (1,088.2K) is barely above February 2020 (1,048.6K) while the under-5 population's demand for care has shifted up with post-pandemic LFP. Evidence that would change my mind: a peer-reviewed state-panel showing no price or access divergence between stopgap and non-stopgap states.

---

X2. Stroud (#63, R) → Reyes (#49, L): "New Mexico is mostly re-labeling"

Stroud. NM's own April 2026 brief: 54% of new enrollees had incomes below the old 400% FPL cap. So more than half the "universal" gain came from families already eligible under the old targeted program. You didn't need universality — you needed outreach. And the price tag jumps from $445.8M (FY26) to $606.4M (FY27) while the Source NM reporting documents waitlists and providers struggling to find insurance. Universal demand is running ahead of supply — the classic result of setting price at zero.

Reyes. Two replies. First, "already eligible but not enrolled" is exactly the pathology of means-testing: paperwork, cliffs, stigma. Universality is the outreach. That's what the 54% number shows. Second, waitlists under a zero price are real, and I accept them as the program's central risk. But licensed capacity grew from 59,274 to 71,509 since 2019 and the brief reports 65% wage growth 2019–24 — supply is responding to money.

Stroud. Then call it what it is — a take-up campaign that costs $160M more a year to reach the other 46%, who are above 400% FPL (≈$128,000 for a family of four). That's a transfer to the upper-middle class financed by oil wealth that belongs to all New Mexicans, including the ones raising kids at home who get nothing.

Reyes. Families at home get free care the day they want it — that's the point of a universal option. I'll concede your distributive point partially: at the margin, the dollars going above 400% FPL are the least well-targeted in the program. I'd support a modest copay above 400% FPL if it funded infant slots.

---

X3. Wren (#2, public finance, R) → Delgado-Finch (#7, Soc): "Quebec did not pay for itself"

Wren. Rosa, you asserted Quebec "paid for itself." That's Fortin's estimate, built on attributing ~70,000 additional employed mothers to the program. But Baker-Gruber-Milligan found that roughly one-third of new formal care use came from mothers who were already working with informal arrangements — pure crowd-out, with fiscal cost and no new tax revenue. The program now costs $2.9B (0.6% of Quebec GDP). Self-financing claims are sensitive to exactly the counterfactual that's contested.

Delgado-Finch. Fortin's calculation includes the crowd-out; it nets fiscal returns against cost using employment effects consistent with BGM's elasticity. But I'll accept that "paid for itself" is a contested estimate, not a finding — the honest statement is "recovered a large share of its cost, possibly all of it, under Fortin's assumptions." What isn't contested is that the quality failures were concentrated in the for-profit sector, which is my structural argument.

Wren. On that, you're reading a cross-sectional quality comparison as causal. Nonprofit CPEs got first access to scarce subsidized places and more experienced staff; selection explains a lot. And even the nonprofit sector sits inside the program that BGM evaluated — the negative behavioral effects are average effects across the whole system.

Delgado-Finch. Fair that the CPE/garderie gap isn't a clean causal estimate. But it's the right direction for policy design: if you're going to spend public money, the evidence gives no reason to prefer for-profit delivery.

---

X4. Oyelaran (#12, finance, C-R) → Castellano (#1, C-L): "Show me the halved turnover"

Oyelaran. You said the DC fund "cut turnover roughly in half." The Urban 2025 study says 64% of educators stayed at the same center in 2023–24 and that educators moved toward funded centers. Mathematica reports a ~7% labor-supply gain. Neither says "halved." Also: DC spends heavily per educator for ~1,500 slots. What's the cost per slot, and is 23% one-year social ROI a number a CFO would believe?

Castellano. You're right about "halved"; I can't source it and I withdraw it. The sourced claims are the 7% labor-supply increase, ~1,500 slots, and directional retention gains. I don't have a verified annual cost figure for the fund in hand, so I won't give you a per-slot number I can't back. On ROI: the 23% is Mathematica's social ROI, which counts parental earnings; a CFO wouldn't book it, but a legislature should weigh it.

Oyelaran. Then the honest case is: one well-evaluated program in the richest per-capita jurisdiction in the country produced a moderate supply response. Scaling that nationally assumes the elasticity holds in Mississippi. I'd back a matched pilot with evaluation — not a national entitlement.

Castellano. That's close to my actual proposal: matching grants with mandatory evaluation. We disagree on how big the federal match should be, not on design.

---

X5. Vogt (#11, behavioral econ, C-L) → Stroud (#63, R): "Preferences and the 2025 LFP claim"

Vogt. Three claims need sources. (1) "Most mothers of young children would prefer to stay home if they could afford it" — which survey? Preference questions are highly framing-sensitive. (2) "Mothers' LFP fell in 2025 because care is unaffordable" — BLS says 73.9% vs 74.0%, "little changed," and no causal attribution. (3) "Universal child care causes crime" — BGM found higher crime in Quebec's cohorts under Quebec's design. Norway's expansion (Havnes & Mogstad) found positive long-run outcomes. You can't generalize one program into a law of nature.

Stroud. (1) I'll withdraw "most" pending a source; my underlying point — that a large share of parents of infants use parental or relative care and would value cash — stands on revealed behavior, not stated preference. (2) Fair: a 0.1-point move is noise. I withdraw the causal attribution. (3) I won't retreat to "Quebec only." Quebec is the one large North American universal program with credible long-run evaluation, and it found harm to noncognitive outcomes that persisted into adulthood. Norway's program served older preschoolers in a very different setting. The burden is on those who want to scale universal infant care in the U.S. to show it won't replicate Quebec.

Vogt. Then the defensible version is: "the best North American evidence on rapidly-scaled universal care shows persistent harm; external validity is contested." I'd sign that. But it cuts against low-quality rapid scaling, not against targeted subsidies or quality-regulated expansion.

Stroud. I'll accept that sentence as a fair statement of the evidence, with the emphasis on "rapidly scaled universal" — which is exactly what New Mexico just did.

---

X6. Castellano (#1, C-L) → Pruitt (#8, Lib): "Ratios aren't free money"

Castellano. Walter, Dow's instrument shows regulations raise prices — fine. It doesn't show they're unjustified. Ratios are the one quality input with a plausible mechanism for infant outcomes, and your Quebec-citing allies on the right should be the last people arguing to thin them out. And on paid leave: the Chicago Fed paper is a working paper on multi-unit firms, while Rossin-Slater-Ruhm-Waldfogel find leave doubled and hours rose for mothers of 1–3-year-olds.

Pruitt. I limited ratio flexibility to 3–5-year-olds, not infants, precisely for that reason. The zoning piece is where the cheap supply is: family child care homes were the only segment that grew in 2025 (+1.4% vs −1% for centers, CCAoA). On paid leave, the long-run tax-data RD (Bailey et al., AEJ 2025) — the best-identified study we have — found no career gains and negative effects for first-time mothers. Short-run leave-taking going up is the mechanical effect of paying for leave; it's not a career benefit.

Castellano. Leave-taking is itself a welfare gain for parents and infants, not merely a "mechanical" effect — but I concede Bailey et al. is the strongest design on career outcomes, and it doesn't support the claim that paid leave narrows the motherhood penalty.

STAGE 3 — CROSS-EXAMINATION (Part 2: elder care, demography, family)

---

X7. Whitmore (#46, pharmacology, C-R) → Adeyemi (#45, L): "40%, and 'at the cost of residents' lives'"

Whitmore. Nkechi, two claims. First, KFF says immigrants are 30% of the direct LTC care workforce, not 40%. The "4 in 10" figure is EPI's number for home health aides specifically. Second, "saves nursing homes money at the cost of residents' lives" — what's the mortality estimate? The CBO-scored saving from the moratorium is $23B over ten years, and the rule's own requirement of 24/7 onsite RNs was widely judged infeasible in rural facilities. A rule that closes rural homes doesn't save lives either.

Adeyemi. On the first: correct, I merged two numbers. Amended — 30% of direct care workers overall (KFF 2024); about 40% of home health aides (EPI). On the second: I don't have a peer-reviewed mortality estimate for the repeal itself, so I'll narrow the claim. What's established is that the rule was repealed, enforcement is barred through October 2034, and the federal floor reverted to 8 RN hours/day. My clinical judgment is that staffing matters for outcomes, but I won't dress judgment as a number.

Whitmore. Appreciated. And on the policy: a staffing floor with no money attached is an unfunded mandate on Medicaid-dependent facilities while Congress is cutting Medicaid. Those two policies can't coexist; the repeal is at least internally consistent.

Adeyemi. Internally consistent the way cutting fire department budgets and repealing sprinkler codes is consistent. I'll accept that a floor must come with rate money — my proposal already pairs them. The question is whether we fund quality or legislate it away.

---

X8. Holloway (#71, social work, L) → Klein (#59, C): "Your caregiver number is stale, and your backstop is too little, too late"

Holloway. Deborah, AARP's current estimate is $1.01 trillion a year from 59 million caregivers of adults — 49.5 billion hours (Valuing the Invaluable, March 2026). The $600B figure is the 2023 edition. It matters: at $1T, unpaid care exceeds total Medicaid spending. And a catastrophic backstop with a 2–3 year waiting period does nothing for the median user, whose paid-care episode is shorter than that. It's insurance for the estates of the upper-middle class.

Klein. Accepted on the number — amended to $1.01T (AARP 2026; valuation at $20.41/hr). On the backstop, I'll push back: the front-end risk is what families can partly self-insure; the catastrophic tail is what bankrupts them into Medicaid. Targeting the tail is exactly the efficient insurance design. What I'd add, given your point, is that the front end needs respite and caregiver support — which is why the caregiver credit is in my package.

Holloway. A nonrefundable credit won't reach the 1 in 5 caregivers AARP says can't afford basic necessities. Make it refundable or it's a middle-class benefit dressed as caregiver support.

Klein. I'll accept partial refundability as a design goal; the cost is the constraint, and I don't have an official score to offer.

---

X9. Venkataraman (#19, inequality econ, L) → Mbeki (#70, C): "2029, and 'no meaningful effect'"

Venkataraman. Daniel, Census's 2023 projections put the crossover — 65+ outnumbering under-18s — in 2029, not 2034. The 2034 date is from the 2017 projection vintage. And you said child care costs have "no meaningful effect" on fertility. The CDCTC paper you cite has a 95% CI from −31 to +23 births per 1,000 — that's not "no effect," that's "we can't tell." Dow's IV paper finds a 10% price increase lowers birth rates 5.7%.

Mbeki. On the date: correct, I cited the older vintage. Amended to 2029 (Census 2023 middle series). On fertility, I'll defend the substance but narrow the wording. The CDCTC null is imprecise, yes. But Dow is an unrefereed job-market paper whose instrument — state ratio and group-size rules — plausibly affects fertility through channels other than price (e.g., correlation with state family-policy culture), which violates exclusion. Revised claim: "There is no well-identified peer-reviewed evidence that U.S. child care subsidies raise fertility; one IV working paper finds large price effects."

Venkataraman. I'll accept that phrasing. My point is that "no evidence of effect" and "evidence of no effect" are different, and policy should say which.

---

X10. Achebe (#28, comparative politics, C-L) → Cross (#64, R): "Faith providers and the counterfactual"

Achebe. Nathaniel, you said faith-based providers are "a substantial share" of supply and universal systems push them out. What's the share, and where's the evidence of push-out? Also, your cite that family caregiving "exceeds all Medicaid spending" is AARP's claim, fine — but it's a valuation at replacement wages, not a measure of what families want to supply. Many caregivers are there because nothing else exists.

Cross. I don't have a verified national share of faith-based providers in hand; I'll mark that as unverified. My NM prediction stands as a test of the push-out claim. On caregiving: AARP's own 2025 survey says 7 in 10 caregivers are employed and half report financial strain — I read that as families choosing to care but needing support, not as preferring a nursing home. That's why I'd pay caregivers directly.

Achebe. Then we agree on the caregiver credit and disagree on paid leave. Voluntary accounts in low-wage households are an empty promise.

Cross. Concede that voluntary accounts alone under-serve low earners; I'd accept a state-option social-insurance model with religious-employer accommodations over a federal mandate.

---

X11. Reyes (#49, L) → Stroud (#63, R): "Cash-for-care and mothers' employment"

Reyes. Abigail, your own concession — cash reduces maternal employment — is the whole problem. The CDCTC paper finds even a small subsidy for purchased care raises married mothers' LFP. A cash-for-care benefit is a subsidy for not working. Your prediction that most parents of 1-year-olds would take cash over subsidized care — what's the evidence base?

Stroud. It's a prediction, which is why it's framed as falsifiable. The design goal is neutrality: today the tax code (CDCTC, DCAP now $7,500) subsidizes only purchased care, so it's already non-neutral against home care. I'm correcting a tilt, not creating one. If parents choose employment and purchased care with the same dollars, fine.

Reyes. The DCAP increase goes to employees with employer plans — mostly higher earners. On that we agree: it's regressive. But neutrality between work and non-work has an employment cost you've conceded.

Stroud. And a child-development benefit for infants, if Quebec's results generalize. We will not settle that today.

STAGE 4 — REFEREE FACT-CHECK

Brandt (#57), Takahashi (#17), Okafor (#9). Every empirical claim from Stages 2–3, rated on the standard scale: Supported = source confirms · Contested = credible sources disagree · Unsupported = no adequate source found / source doesn't say it / could not verify · Wrong = a credible source contradicts it. Forecasts and normative claims are not rated. Fact numbers (F#) refer to the Evidence Brief.

# · Speaker · Claim (as made) · Rating · Basis
1 · Reyes · Avg price $13,184 ≈ 33% of single-parent median income · Supported · CCAoA 2025 (F1)
2 · Reyes · Only 15% of federally eligible children get CCDBG subsidies · Supported · ASPE FY2021 (F3)
3 · Reyes · Childcare workers $16.82/hr, bottom-5% of occupations · Supported · BLS May 2025; Chicago Fed (F5–6) — Chicago Fed's bottom-5% ranking used an older $14.60 median
4 · Reyes · After ARPA expiry, access gap 24%→31%; maternal LFP gains stalled · Supported · CEA 2024 (F8) — administration analysis, not peer-reviewed
5 · Reyes · "The cliff shed hundreds of thousands of child care jobs" · Wrong · BLS CES: 1,062.7K (Sep 2023) → 1,091.5K (Sep 2024) (F7). TCF's 232K was a forecast (F9). Retracted in X1.
6 · Reyes · NM served 32,861 → 39,907 children; $18 wage floor · Supported · NM ECECD Apr 2026 (F11)
7 · Reyes · NM FY27 child care assistance $606M · Supported · F11 ($606.4M)
8 · Reyes · Full-replacement stopgap states got ~2.5 pp more maternal-LFP growth · Contested · Reported by CEA (F8); selection of stopgap states unaddressed, not refereed
9 · Reyes · NM capacity 59,274→71,509; wages +65% 2019–24 · Supported · F11
10 · Reyes · 2026 child care employment barely above Feb 2020 · Supported · 1,088.2K vs 1,048.6K (+3.8%) (F7)
11 · Reyes · CDCTC raises married mothers' LFP · Supported · Rev. Econ. Household 2026 (F23)
12 · Reyes · DCAP expansion goes mostly to higher earners · Unsupported · Plausible (requires employer plan, exclusion worth more at higher brackets) but no distributional source verified this session
13 · Delgado-Finch · Quebec "paid for itself" · Contested · Fortin >100% (F15) vs BGM crowd-out (F13). Amended in X3.
14 · Delgado-Finch · Quebec mothers' LFP rose 16 points · Supported (descriptive) · Fortin 64%→80% (F15); ROC rose 4 pts, so causal share is smaller
15 · Delgado-Finch · For-profit garderies: 10% "good," 36% "inadequate" · Supported · F15; causal reading Contested (Wren, X3)
16 · Delgado-Finch · Semi-private NH $114,975; 66% of direct care workers <$35K · Supported · CareScout 2025; KFF (F28, F32)
17 · Adeyemi · Medicaid pays 61% of LTC · Supported · KFF 2023 (F29)
18 · Adeyemi · Federal Medicaid cuts >$1T over 10 years · Supported · KFF (F30)
19 · Adeyemi · Staffing rule (3.48 HPRD, 24/7 RN) repealed Dec 2, 2025 · Supported · AHA/Fed. Register (F31)
20 · Adeyemi · Repeal saves money "at the cost of residents' lives" · Unsupported · No mortality estimate found for the repeal. Narrowed in X7.
21 · Adeyemi · Immigrants are 40% of direct care workers · Wrong · KFF: 30% overall (F32); ~40% applies to home health aides (EPI, F33). Amended in X7.
22 · Adeyemi · HCBS is optional and most exposed to state cuts · Supported · KFF HCBS explainer (F30)
23 · Castellano · DC PEF: +7% labor supply, ~1,500 slots, 23% one-year SROI · Supported · Mathematica (F16)
24 · Castellano · DC fund "cut turnover roughly in half" · Unsupported · Urban 2025 reports 64% same-center retention; no halving estimate found. Withdrawn in X4.
25 · Castellano · Exits flow to cashier/retail jobs · Supported · Chicago Fed (F6)
26 · Castellano · CA PFL doubled leave; hours up 6–9% · Supported · RRW (F25) — authors call hours evidence "suggestive"
27 · Klein · NH $114,975; in-home $80,080 · Supported · F28
28 · Klein · LTC users: 6% of enrollees, 36% of Medicaid spending · Supported · F29
29 · Klein · Family caregivers provide ~$600B/yr · Wrong (outdated) · AARP 2026: $1.01T (F34). Amended in X8.
30 · Mbeki · Older adults outnumber children by 2034 · Wrong · Census 2023 projections: 2029 (F35). Amended in X9.
31 · Mbeki · 3.63M births 2024; GFR 53.8; −22% since 2007 · Supported · CDC (F21)
32 · Mbeki · Child care costs have "no meaningful effect" on fertility · Contested · CDCTC null is imprecise (F23) vs Dow IV working paper (F22). Amended in X9.
33 · Stroud · Quebec: more aggression, worse health, higher adult crime · Supported (for Quebec) · BGM 2005/2008, 2019 (F13–14)
34 · Stroud · "Universal child care causes crime" (general) · Contested · External validity; Norway evidence cited by Vogt (see #50). Narrowed in X5.
35 · Stroud · Most mothers of young children would prefer to stay home · Unsupported · No survey cited. Withdrawn in X5.
36 · Stroud · Mothers' LFP fell in 2025 because care is unaffordable · Unsupported · BLS: 73.9% vs 74.0%, "little changed"; no causal source (F20). Withdrawn in X5.
37 · Stroud · 54% of NM new enrollees were below old 400% FPL cap · Supported · F11
38 · Stroud · NM waitlists/insurance strain · Supported · Source NM Sep 2026 (F12)
39 · Stroud · 400% FPL ≈ $128,000 for family of four · Supported · HHS 2025 guideline $32,150 × 4 = $128,600 (https://aspe.hhs.gov/topics/poverty-economic-mobility/poverty-guidelines)
40 · Stroud · Tax code subsidizes only purchased care · Supported · CDCTC/DCAP cover employment-related care expenses (F19)
41 · Pruitt · Ratio/group-size regulations raise prices · Supported (working paper) · Dow first stage (F22); not refereed
42 · Pruitt · Paid leave laws cut establishment employment 1.5–7.9%, payroll 5.6–8.3% · Supported (working paper) · Chicago Fed WP 2024-12 (F27)
43 · Pruitt · Child care employment rose after the cliff · Supported · F7
44 · Pruitt · Old 45F: 169–278 corporate claimants (2016) · Supported · GAO (F18)
45 · Pruitt · FCC homes the only growing segment (+1.4% vs −1% centers) · Supported · CCAoA (F1)
46 · Pruitt · Supply more elastic than "structural failure" story implies · Contested · CES recovery (F7) vs CEA access/price data (F8)
47 · Pruitt / Castellano · Bailey et al.: no long-run career/fertility gains; negative for first-time mothers · Supported · AEJ:EP 2025 (F26)
48 · Cross · 63M family caregivers · Supported · AARP 2025 (F34)
49 · Cross · Faith-based providers a "substantial share," pushed out by universal systems · Unsupported · No source. Marked unverified by Cross in X10.
50 · Vogt · Norway expansion (Havnes & Mogstad) positive long-run effects · Unsupported (unverified here) · Widely cited AEJ:EP 2011 paper; referees did not re-fetch it this session, so per rules it cannot carry weight in the verdict
51 · Wren · ~1/3 of new Quebec care use crowded out informal care · Supported · BGM (F13)
52 · Wren · Quebec costs $2.9B, 0.6% of GDP · Supported · Fortin (F15)
53 · Whitmore · Staffing moratorium saves $23B/10 yrs · Supported · KFF/CBO (F30)
54 · Whitmore · 24/7 RN rule "widely judged infeasible" in rural facilities · Contested · Industry (AHA) view (F31); advocates dispute; no neutral feasibility study verified
55 · Holloway · $1.01T; 59M caregivers of adults; 49.5B hours · Supported · AARP 2026 (F34)
56 · Holloway · Median user's paid LTC episode shorter than 2–3 years · Unsupported · No duration source verified this session
57 · Holloway · 1 in 5 caregivers can't afford basic necessities · Supported · AARP 2025 (F34)
58 · Venkataraman · CDCTC fertility CI is −31 to +23 per 1,000 · Supported · F23
59 · Venkataraman · Dow: 10% price rise → 5.7% fewer births · Supported (working paper) · F22
60 · Achebe · AARP value is at replacement wages · Supported · $20.41/hr (F34)

Tally (60 rated claims): Supported 44 · Contested 7 · Unsupported 7 · Wrong 4 (counting "Supported (working paper/descriptive/for Quebec)" as Supported).

Referee comments:

• Takahashi: The four Wrongs are all of one type — a real number from the wrong vintage or the wrong subgroup (TCF forecast vs outcome; 40% vs 30%; $600B vs $1.01T; 2034 vs 2029). That is the most common failure mode in policy argument and the cheapest to fix.
• Brandt: Three Supported ratings rest on unrefereed work (CEA, Dow, Chicago Fed WP). They are accurately reported; they should not carry a verdict alone.
• Okafor: No one in this thread produced a verified estimate of the staffing-rule repeal's health effects or of the 2025 Medicaid law's effect on HCBS waiver slots specifically. Those are the biggest evidentiary holes on the elder-care side.

On the record — responses to ratings

• Reyes (#49): #5 retracted (see X1). #12: I withdraw "mostly to higher earners" as a factual claim until I can cite a distributional table; I retain it as a hypothesis.
• Adeyemi (#45): #20, #21 amended as in X7.
• Castellano (#1): #24 withdrawn.
• Klein (#59): #29 amended to $1.01T.
• Mbeki (#70): #30 amended to 2029; #32 reworded as in X9.
• Stroud (#63): #35, #36 withdrawn; #34 narrowed to "Quebec's rapidly scaled universal program."
• Cross (#64): #49 withdrawn pending data.
• Vogt (#11): #50 — accept the rating. The paper exists and I believe it says what I said, but this panel's rule is right: unverified here means it can't bear weight. The C1 contest therefore rests on Quebec's design-specificity, not on Norway.
• Holloway (#71): #56 withdrawn; my argument for front-end support stands on AARP's financial-strain data (#57), not on episode length.

Referee erratum (Brandt): the tally line in the fact-check post miscounted. Recounting row by row: Supported 42 · Contested 6 · Unsupported 8 · Wrong 4 (60 claims). Contested = #8, 13, 32, 34, 46, 54; Unsupported = #12, 20, 24, 35, 36, 49, 50, 56; Wrong = #5, 21, 29, 30. The individual ratings stand as posted.

---

STAGE 5 — STEELMAN

5a. Left steelmans the Right (written by Holloway #71 and Vogt #11)

"The case for parental choice is not nostalgia; it is a claim about who knows what a particular child needs and about fairness among families. (1) Neutrality: Today's federal support for young-child care — the CDCTC (now up to 50%), DCAP ($7,500), 45F — is conditioned on buying care while employed. A family that gives up $40,000 of a second income to care for an infant gets none of it. That's a tax on one arrangement, and it's the arrangement disproportionately chosen by lower-income, religious, and immigrant families. (2) Evidence of harm at scale: the one large North American universal program with long-run evaluation (Quebec, BGM 2008/2019) found persistent noncognitive harm and higher adult crime, and its quality problems appeared precisely where supply was scaled fast — which is what New Mexico is now doing. (3) Targeting leak: in NM, 54% of the new enrollees were already eligible; the incremental universal dollars disproportionately flow above 400% FPL (~$128,600 for four). (4) Mandate costs are real: the best-identified evidence on paid leave (Bailey et al. 2025) finds no long-run career gains, and firm-side evidence (Chicago Fed WP) shows employment and payroll costs. (5) Supply is cheaper to unlock than to buy: home-based providers were the only growing segment in 2025, and zoning and licensing barriers are within states' control at near-zero fiscal cost. Therefore: give parents the money, neutrally; fix supply rules; and don't build a system whose best-evaluated precedent hurt kids."

Stroud (#63) replies: Accepted as fair, with one correction: our case does not rest on Quebec alone. It also rests on the principle that the subsidy should follow the child, not the provider type — even if Quebec's results were null, neutrality would still be our position.

5b. Right steelmans the Left (written by Stroud #63 and Pruitt #8)

"The case for public provision is a market-failure argument, not a sentiment. (1) Labor-cost trap: care is labor-intensive and ratio-bound; the median childcare worker earns $16.82/hr against a $24.51 all-occupation median and exits to cashier jobs. Parents can't pay wages that retain staff, so supply stays thin and turnover high. (2) Public money demonstrably changes supply: DC's Pay Equity Fund raised the local labor supply ~7% (Mathematica); New Mexico's capacity grew ~20% and wages 65% while it ramped funding. (3) Means-testing fails on take-up: only 15% of federally eligible children get CCDBG help; NM's 54% statistic shows universality reached families that targeting never did. (4) Withdrawal hurts: after ARPA expired, access worsened and maternal LFP gains stalled, with a dose-response to state backfill (CEA). (5) Quebec is a design lesson, not a verdict: its quality failures were concentrated in for-profit settings; the same program raised mothers' LFP by far more than elsewhere in Canada and recouped much of its cost. (6) Elder care has the same structure — Medicaid pays 61% of LTC, and cutting it shifts costs onto 63 million unpaid caregivers, most of them working women. Therefore: fund the workforce and the slots publicly, with quality standards, because the market equilibrium is too little care at too-low wages."

Reyes (#49) replies: Accepted as fair, with one correction: point (4) should be stated with the Stage 4 caveat — the job-loss forecasts were wrong; the harm evidence is on access and LFP trend, from a non-peer-reviewed source. We'd rather be steelmanned accurately than generously.

5c. Center steelmans the Right on elder care (written by Klein #59)

"The Right's elder-care case: (1) a federal staffing floor imposed on a sector with a 2.3M-worker, low-wage, 30%-immigrant workforce, in a tightening immigration environment, risks closures where alternatives are scarcest — and it arrived unfunded; the repeal saved $23B federally. (2) Families are already the LTC system — $1.01T of care — and policy should pay and support them directly (caregiver credits, respite) rather than substitute institutional care. (3) Medicaid's LTC spending (36% of the program for 6% of enrollees) is the fastest-growing liability as 65+ outnumber children by 2029; asset rules like the $1M home-equity cap ask better-off families to contribute before taxpayers do."

Cross (#64) replies: Accepted as fair. I'd add that subsidiarity is a positive principle, not only a cost argument: families and congregations provide forms of care — presence, continuity, meaning — that no staffing ratio measures.

5d. Right steelmans the Center's targeted approach (written by Wren #2)

"The Center's case: universal programs spend most marginal dollars on families who would buy care anyway; targeted programs spend them where the employment and child-development returns are highest — and the strongest positive early-childhood evidence is for disadvantaged children. Make the CDCTC refundable so it reaches families with no tax liability, fund CCDBG so more than 15% of eligible kids are served, and evaluate everything, because this domain's forecasts (TCF's cliff, Quebec's self-financing) have repeatedly outrun evidence."

Castellano (#1, speaking for the targeted/center-left position) replies: Accepted as fair, with a correction: take-up is the Achilles heel of targeting — any targeted design must include auto-enrollment or categorical eligibility, or the 15% number won't move.

STAGE 6 — REVISED PROPOSALS

Compiled by the moderator's office from the amended positions. Cost lines say "no official score" where we could not verify a CBO/JCT estimate; we did not fabricate any.

---

14-P1 — Federal matching grants for state child care workforce compensation funds

• Mechanism: Federal 50/50 match (via CCDBG) for states that create DC-style wage supplements for licensed child care staff, with a wage pass-through requirement and a mandatory independent impact evaluation. States choose eligibility and amounts above a floor.
• Cost: No official score. Scale depends on uptake; DC's annual fund cost was not verified in this thread.
• Precedent: DC Pay Equity Fund (2022–): ~7% labor-supply increase, ~1,500 slots, 23% one-year social ROI (Mathematica); retention gains at funded centers (Urban 2025). NM built an $18/hr floor into its cost model, and wages rose 65% 2019–24.
• Key risk: Supply elasticity found in rich, dense DC may not transfer; wage subsidies raise cost per slot without lowering sticker prices.

14-P2 — Flexible infant/toddler care benefit (parental-choice option)

• Mechanism: Families with children aged 0–2 may take a flat annual benefit usable for any care arrangement, including a parent at home, as an alternative to (not on top of) subsidized purchased care. Income-phased at the top.
• Cost: No official score.
• Precedent: Nordic "cash-for-care" allowances (e.g., Finland, Norway) exist; their employment and child effects were not verified in this thread. U.S. tax subsidies (CDCTC, DCAP) currently cover only employment-related purchased care.
• Key risk: Reduces maternal employment relative to work-conditioned subsidies (conceded by proponents); lower take-up of formal care among disadvantaged children for whom center care may help most.

14-P3 — Make the Child and Dependent Care Tax Credit refundable

• Mechanism: Convert the CDCTC (now 50% max rate after OBBBA, $3,000/$6,000 expense caps) to fully refundable, so low-income working families with no income tax liability receive it; index caps to inflation.
• Cost: No official score verified in this thread (a refundable CDCTC was enacted temporarily for 2021 under ARPA; its score was not re-verified here).
• Precedent: 2021 one-year refundable CDCTC; peer-reviewed evidence that state CDCTCs raise married mothers' LFP with no detectable fertility effect (Rev. Econ. Household 2026).
• Key risk: Still requires paying up front and waiting until tax time — a poor fit for monthly care bills unless paired with advance payment, which raises improper-payment risk.

14-P4 — Home-based child care supply package (no ratio loosening for infants)

• Mechanism: Condition part of CCDBG quality funds on states (a) making licensed family child care homes a by-right residential use, (b) portable credentials/reciprocity, (c) streamlined licensing timelines — while leaving infant/toddler ratios untouched.
• Cost: Minimal federal cost; no official score.
• Precedent: Family child care homes were the only growing segment in 2025 (+1.4% vs −1% for centers, CCAoA). Dow's working paper finds state regulation is a meaningful price driver.
• Key risk: Local-control objections; inspection capacity may lag growth in small providers.

14-P5 — Direct care worker immigration pathway

• Mechanism: Create a capped, employer-sponsored visa category (with portability between employers and a path to permanent residence) for home health, personal care, and nursing aides; protect work authorization of current direct care workers with pending status.
• Cost: No official score.
• Precedent: Immigrants are 30% of direct care workers nationally and 60% in New York (KFF); EPI projects ~394,000 direct care job losses under a 4M-deportation scenario (projection).
• Key risk: Wage suppression for U.S.-born aides; enforcement/abuse in employer-tied visas; political durability.

14-P6 — Family caregiver credit plus funded respite

• Mechanism: A partially refundable federal credit for 30% of documented out-of-pocket caregiving costs above a floor (cap ~$5,000), plus a dedicated respite/caregiver-training grant to states through the Older Americans Act network.
• Cost: No official score (bipartisan "Credit for Caring" bills have been introduced; we did not verify a JCT score).
• Precedent: AARP 2025: 63M caregivers; 1 in 5 can't afford basic necessities; 7 in 10 are employed. No U.S. precedent at federal scale; several states run small caregiver tax credits (not verified here).
• Key risk: Documentation burden limits take-up among the lowest-income caregivers; partial refundability costs more.

14-P7 — Federal paid family and medical leave insurance

• Mechanism: A national social-insurance program (≈12 weeks, progressive wage replacement) financed by a payroll contribution, with states that already run programs allowed to keep them.
• Cost: No official score verified in this thread. State programs charge ≤1.3% of payroll (New America 2026).
• Precedent: 13 states + DC paying benefits. CA: leave-taking roughly doubled (RRW); no long-run career or fertility gains, negative for first-time mothers (Bailey et al. 2025); firm-side employment/payroll declines (Chicago Fed WP).
• Key risk: New payroll tax; weak evidence for the long-run career gains often promised.

14-P8 — Evaluation and data mandate for care policy

• Mechanism: Fund (a) a rigorous independent evaluation of New Mexico's universal program (child outcomes incl. noncognitive measures, maternal employment, prices, provider supply, fertility), (b) annual federal updates to the NDCP and ASPE eligibility/receipt estimates, and (c) a national HCBS waiting-list and direct-care workforce data series to track the 2025 Medicaid law's effects.
• Cost: Small; no official score.
• Precedent: The Quebec evidence base exists because of longitudinal survey data (NLSCY) that enabled BGM and Fortin to disagree productively; the DC PEF evaluation (Mathematica) shows state programs can be evaluated quickly.
• Key risk: Evaluations are slow relative to political cycles; results may be used selectively.

STAGE 7 — RECORDED VOTE

Provisional vote — superseded by the blind independent ballot in thread 'Independent Ballot'.

Blocs: Left 34 · Center 33 · Right 32 (moderator #100 abstains). Bridge rule: CONSENSUS only if ≥60% in each bloc (Left ≥21, Center ≥20, Right ≥20).

ID · Proposal · Left yes · Center yes · Right yes · Consensus?
14-P1 · Federal match for state workforce compensation funds · 32/34 (94%) · 21/33 (64%) · 8/32 (25%) · No
14-P2 · Flexible infant/toddler benefit incl. parental care · 11/34 (32%) · 17/33 (52%) · 29/32 (91%) · No
14-P3 · Refundable CDCTC · 32/34 (94%) · 25/33 (76%) · 13/32 (41%) · No
14-P4 · Home-based child care supply package · 23/34 (68%) · 30/33 (91%) · 28/32 (88%) · YES
14-P5 · Direct care worker immigration pathway · 27/34 (79%) · 24/33 (73%) · 14/32 (44%) · No
14-P6 · Family caregiver credit + funded respite · 26/34 (76%) · 26/33 (79%) · 21/32 (66%) · YES
14-P7 · Federal paid family & medical leave insurance · 33/34 (97%) · 15/33 (45%) · 4/32 (13%) · No
14-P8 · Evaluation & data mandate (NM, NDCP, HCBS) · 25/34 (74%) · 32/33 (97%) · 30/32 (94%) · YES

Result: 3 consensus reforms (P4, P6, P8).

Why the splits look this way: P1 and P3 clear Left and Center but fail on the Right over cost and the absence of official scores. P2 is the mirror image — the Right's core ask, failing Left over the conceded employment cost. P5 splits the Right between libertarians (yes) and restrictionist conservatives (no). P7 failed in the Center largely because of Bailey et al.'s long-run null. P4's Left "no" votes (11) came from members worried that "streamlining" becomes deregulation of quality; P8's Left "no" votes (9) worried that a Quebec-style evaluation of NM will be used to cut the program before it matures.

Vote explanations

Reyes (#49, L) — P1 yes, P2 no, P4 yes, P8 yes. "I voted for the evaluation mandate knowing it may embarrass my side; that's what the ARPA-cliff exchange taught me. What would change my mind on P2: a U.S. randomized or quasi-experimental study showing flexible cash for infants improves child outcomes without large maternal-employment losses."

Stroud (#63, R) — P1 no, P2 yes, P3 no, P4 yes, P6 yes, P8 yes. "P3 is better than nothing but keeps the tilt against home care. What would change my mind on P1: the three-year Mathematica results replicated in at least one lower-cost, lower-density state with no increase in posted prices."

Castellano (#1, C-L) — P1 yes, P3 yes, P5 yes, P7 yes, P8 yes; P2 no. "On P7 I voted yes despite Bailey et al., because leave-taking itself is the welfare gain for infants and parents. What would change my mind: evidence that state PFL programs reduced hiring of women of childbearing age at the entry level — the Chicago Fed firm-side paper points in that direction but doesn't test it."

Klein (#59, C) — P3 yes, P5 yes, P6 yes, P8 yes; P1 no, P7 no. "P1 is a fine pilot and a bad entitlement; I'd vote yes on a capped demonstration. What would change my mind on P7: a peer-reviewed study showing caregiving-leave use (not just parental leave) reduces nursing-home entry or caregiver job exit."

Pruitt (#8, Lib) — P4 yes, P5 yes, P8 yes; all others no. "P5 is the most pro-market care reform on this list — the supply of care workers is set in Washington, not by wages. What would change my mind on P6: nothing about its merits; my objection is to another carve-out in the tax code. Show me it replaces an existing credit and I'll move."

Adeyemi (#45, L) — P1, P3, P5, P6, P7, P8 yes; P2 no; P4 yes. "None of these proposals touches the staffing-floor repeal, which is the gap in this thread. What would change my mind on staffing floors: a credible study showing that the 2024 rule's specific thresholds would have caused net closures that reduced access more than they improved care."

Mbeki (#70, C) — P3, P4, P5, P6, P8 yes; P1, P2, P7 no. "Demography argues for P5 more than anything else on this list. What would change my mind on P2: a well-identified finding that flexible benefits raise completed fertility — so far nothing in our brief shows that for any U.S. care subsidy."

STAGE 8 — VERDICT

Prof. Adelaide Wainwright (#100, moderator)

Established (the evidence is solid)

• Child care is expensive relative to income and poorly subsidized. $13,184 national average in 2025; ~10% of married-couple and ~33% of single-parent median income (CCAoA); 8.9–16% of county median family income (NDCP 2022). Only 15% of federally eligible children receive CCDBG subsidies (ASPE FY2021).
• The workforce is low-paid and competes with retail. $16.82/hr median vs $24.51 all-occupation (BLS 2025); exits flow to cashier/retail jobs (Chicago Fed).
• The ARPA cliff did not cause the forecast collapse. Industry employment rose after September 2023 (BLS CES). The evidence of harm is narrower: worse access and stalled maternal-LFP gains (CEA, not refereed).
• Quebec's rapidly scaled universal program raised maternal employment a lot and harmed average noncognitive outcomes, with long-run persistence (BGM 2008/2019; Fortin). Both halves are true.
• OBBBA changed the federal tax side (45F to 40–50%/$500–600K; DCAP $7,500; CDCTC max 50%), but it left the CDCTC nonrefundable. The old 45F had almost no take-up (GAO).
• Medicaid is the LTC system (61% of $459B, KFF 2023). The 2025 law cuts federal Medicaid by more than $1T over a decade and blocks the staffing rule until 2034. CMS repealed the rule on Dec 2, 2025.
• Unpaid family care is the largest single input. 63M caregivers; $1.01T/yr of care for adults (AARP). Older adults will outnumber children by 2029 (Census 2023).

Contested (credible sources disagree)

• Whether universal care's child-outcome harms generalize beyond Quebec's design and quality mix (C1).
• Whether Quebec "paid for itself" (C5).
• Paid leave's career effects: short-run leave and hours gains (RRW) vs long-run nulls or negatives (Bailey et al.). Firm-side costs come from a working paper.
• How elastic care supply is to public money. DC PEF (+7%) and NM capacity growth say yes; the post-ARPA employment rise says supply was more resilient than advocates claimed. These are not contradictory, but the magnitudes are unresolved.
• Whether a 24/7 RN staffing floor was feasible.

Unknown

• New Mexico's effects on child outcomes, maternal employment, prices, and fertility. The program is ten months old. Enrollment is ~47,000, costs are rising ($445.8M → $606.4M), and waitlists are growing.
• How the 2025 Medicaid law will affect HCBS waiver slots and the direct care workforce, and the health effects of the staffing repeal. No verified estimates exist.
• How immigration enforcement is actually affecting the direct care workforce. We have only projections so far.
• Whether child care prices affect U.S. fertility. There is one IV working paper and one imprecise null.

Research agenda

1. An independent, pre-registered NM evaluation with noncognitive child measures. Quebec's lesson is to measure what BGM measured.
2. Replicating the DC PEF result in a low-cost state.
3. A peer-reviewed state-panel analysis of the ARPA expiry that addresses stopgap-state selection.
4. A national HCBS waiting-list and direct-care workforce series from 2025 onward.
5. Well-identified U.S. evidence on cash-for-care versus purchased-care subsidies for infants.

Consensus reforms (provisional vote)

• 14-P4: a home-based child care supply package (by-right zoning for family child care, portable credentials, faster licensing; no loosening of infant ratios).
• 14-P6: a family caregiver credit plus funded respite.
• 14-P8: an evaluation and data mandate (NM evaluation, NDCP/ASPE updates, HCBS and workforce tracking).

Failed but instructive: workforce compensation matching (P1) and a refundable CDCTC (P3) passed Left and Center but not the Right. The flexible infant benefit (P2) passed only the Right. Federal paid leave (P7) failed in the Center on the long-run evidence.

Where Symposium 1 was wrong or too glib

• It treated the ARPA cliff as a proven catastrophe. The job-loss forecasts were wrong. The honest claim is about access and prices at the margin.
• It cited Quebec one-sidedly. Advocates quoted the LFP gains and critics quoted the crime findings, but rarely did anyone quote both. Both are real.
• It said paid leave "boosts women's careers." The best-identified long-run evidence (Bailey et al. 2025) does not support that. The strongest case for paid leave is leave-taking and infant/parent welfare, not careers.
• It used stale elder-care numbers ($600B for caregiving, a 2034 crossover). The current figures are $1.01T and 2029.
• It sold child care as a fertility lever. The evidence does not yet support that.
• It under-discussed elder care relative to child care. Given the 2025 Medicaid law, the staffing repeal, and a direct care workforce that is 30% immigrant, elder care is where near-term policy shocks are largest and evidence is thinnest.

Thread closed for Stage 7 purposes; the independent ballot governs.

Independent Audit

Audit Panel — Prof. Laura Pennington (#97, operations research, C) & Prof. Olivia Brandt (#57, statistics, C). Pennington leads this audit. Brandt refereed this thread, so she recuses from any finding about her own rulings, including the Stage 4 tally erratum. Corrections below are appended to the record; nothing above is deleted.

1. Source check (12 claims that carry the verdict, each re-fetched 2026-09-24)

# · Claim (thread) · Cited figure · Verified figure · Status · URL
1 · Average child care price, 2025 (F1) · $13,184; 10% of married-couple and 33% of single-parent median income; centers −1%, FCC homes +1.4% · $13,184; 10% ("two-parent") and 33%; centers −1%, FCC homes +1.4%; 15% of eligible children subsidized · Confirmed · https://info.childcareaware.org/media/child-care-prices-rival-major-household-expenses
2 · Subsidy reach, FY2021 (F3) · 11.5M / 8.0M eligible; 1.8M served = 15% / 22% · Same · Confirmed · https://aspe.hhs.gov/reports/child-care-eligibility-fy2021
3 · Wages, May 2025 (F5, F32) · Childcare workers $16.82 (518,910 employed); preschool $18.34; all occupations $24.51; home health/personal care aides $17.21 (4.31M) · Same · Confirmed · https://www.bls.gov/news.release/ocwage.t01.htm
4 · Child Care Services employment, SA (F7) · 1,048.6 → 676.6 → 1,062.7 → 1,091.5 → 1,093.9 → 1,088.2 (thousands) · Same, all six values; series is SA · Confirmed · https://fred.stlouisfed.org/series/CES6562440001
5 · CEA on the ARPA cliff (F8) · $24B; >225,000 providers; access gap 24%→31%; ~2.5 pp in full-replacement states · Same; 11 states + DC ran stopgaps · Confirmed · bidenwhitehouse.archives.gov CEA 2024-06-27 (URL in F8)
6 · TCF forecast (F9) · >70,000 closures; 3.2M children; 232,000 jobs; $9B/yr · Same · Confirmed · https://tcf.org/content/report/child-care-cliff/
7 · NM universal brief (F11) · 32,861→39,907 served; 54% below old cap; 39% infant/toddler; $18 floor; FY26 $445.8M, FY27 $606.4M; capacity 59,274→71,509; wages +65% · All confirmed. Figure 6 gives 59,274; the text rounds it to "approximately 59,600." The brief also reports that the workforce grew 64% · Confirmed · https://www.nmececd.org/wp-content/uploads/2026/04/Universal-Child-Care-Brief_41526.pdf
8 · Medicaid's share of LTC (F29) · 61% of $459B (2023); 6.3M users; 6% of enrollees, 36% of spending · Same; 4.9M home, 1.4M institutional · Confirmed · KFF Health Policy 101 (URL in F29)
9 · Direct care workforce (F32; Wrong rating #21) · 2.3M; 30% immigrants; 0% WY to 60% NY; 85% female; 32% on Medicaid; 66% under $35K · Same (2024 ACS) · Confirmed · https://www.kff.org/medicaid/what-role-do-immigrants-play-in-the-direct-long-term-care-workforce/
10 · Value of family caregiving (F34; Wrong rating #29) · 59M caregivers; 49.5B hours; $1.01T at $20.41/hr · Same · Confirmed · https://www.aarp.org/press/releases/2026-03-26-AARP-Economic-Value-Of-Family-Caregiving-Report.html
11 · 65+ outnumber under-18s (F35; Wrong rating #30) · 2029; natural decrease from 2038; 29.1% by 2100 · Same (main series) · Confirmed · https://www.census.gov/newsroom/press-releases/2023/population-projections.html
12 · Staffing-rule repeal (F31) · Dec 2, 2025; 3.48 / 0.55 / 2.45 HPRD and 24/7 RN repealed; reverts to 8 consecutive RN hours a day · Same · Confirmed · https://www.aha.org/news/headline/2025-12-02-cms-repeals-minimum-staffing-requirements-skilled-nursing-long-term-care-facilities

Counts: 12 Confirmed · 0 Minor discrepancy · 0 Not supported · 0 Could not access. The AARP release does not mention the older $600B figure. Row #29's "outdated" basis therefore rests on the 2023 edition, which we did not re-fetch.

2. Internal consistency

• Tally (Pennington alone; Brandt recused). Recounting the 60 rows gives Contested 6 (#8, 13, 32, 34, 46, 54), Unsupported 8 (#12, 20, 24, 35, 36, 49, 50, 56), Wrong 4 (#5, 21, 29, 30) and Supported 42. The erratum's 42 · 6 · 8 · 4 is correct. The original 44 · 7 · 7 · 4 was wrong.
• Vote math. All 24 yes-count/percentage pairs are correct, rounded to the nearest point. Consensus labels are correct: P4 (23 / 30 / 28), P6 (26 / 26 / 21) and P8 (25 / 32 / 30) clear every bloc threshold, and every "No" misses at least one. The stated counts of Left "no" votes (P4: 11; P8: 9) match the table.
• Arithmetic in the text is correct. +29K after the cliff (28.8K). 2026 employment is +3.8% over Feb 2020. $606.4M − $445.8M = $160.6M. 400% FPL = $128,600. NM capacity +20.6%.
• Ratings conflict with each other (#34 vs #50). Row #34 rates Stroud's general "universal care causes crime" as Contested, and its stated basis includes Norway ("see #50"). But row #50 rates the Norway evidence Unsupported and says it "cannot carry weight in the verdict." Correction: row #34's Contested rating should rest only on external validity (Quebec's design and quality mix). Vogt's own on-record note already says this. The rating can stand; the stated basis should not cite #50.
• One label is wrong. In 5d, Castellano (#1) replies "speaking for the targeted/center-left position" to a steelman titled "Right steelmans the Center." Castellano is C-L (Left bloc). A Center member (for example Klein #59 or Mbeki #70) should have accepted or corrected it. Minor.
• Roster. All names, numbers, fields and leans match the roster.
• Verdict vs. ratings. No claim rated Wrong reappears as established. The verdict uses the corrected figures (2029, $1.01T, 30%, no job collapse).

3. Balance

The four Wrong ratings fall on Left (Reyes, Adeyemi) and Center (Klein, Mbeki) speakers. The Right has none: Pruitt's claims are all Supported, and Stroud and Cross get only Unsupported or Contested. Most of this is earned. The Left's errors were real number errors: a forecast cited as an outcome, and the wrong subgroup. But the Right got two small breaks:

• Stroud's "LFP fell in 2025." BLS calls a 0.1-point move "little changed," so the source contradicts the claim. That is close to Reyes's forecast-as-outcome error, which was rated Wrong; Stroud's was rated only Unsupported.
• Stroud's general crime claim was rated Contested, not Unsupported, even though the only counter-evidence (Norway) was ruled inadmissible.

On the other side, rating accurate-but-old vintages as Wrong (#29, #30) is strict, and it was applied only to Center speakers. None of this changes the verdict. The verdict's "Established" list is evidence-driven, and it credits both halves of the Quebec record.

4. Grade: A−

All 12 load-bearing figures check out exactly, and the tally erratum and vote math are correct. Deductions: the #34/#50 rating conflict, the 5d label, and a slight leniency toward two Right-bloc claims.