[MaxGhenis]

What and why

PolicyEngine encoded Texas Child Care Services payment rates only from Board Contract Year (BCY) 2025, which starts October 1, 2024. Since it copies a parameter's earliest value back to January 2015, it applied October 2024 rates to every month from then through September 2024. We add the BCY 2024 rates (October 1, 2023 to September 30, 2024) and fix four Dallas County BCY 2025 rates. We also date the statewide copay scale from January 13, 2025, when TWC launched its TX3C case management system and began phasing the scale in.

BCY 2024 payment rates

TWC listed every Board's maximum provider payment rates for BCY 2024 in WD Letter 20-23, Attachment 1. It dated the letter August 29, 2023, with rates effective October 1, 2023, and later revised the attachment under the same effective date. TWC created the revised PDF on October 16, 2023, according to the file's metadata, and links it from its WD 20-23 page as the letter's only attachment. The file TWC serves today matches, byte for byte, the copies Wayback captured in February 2024 and July 2025 (sha256 dcb54a8d…). TWC still serves the August 29 original at the policy-letter path; we encoded it in 1541fef973 and replaced it with the revision in 1ca5db7e38.

In the revision, TWC raised 2,646 Texas Rising Star and Texas School Ready rates by $0.01 to $0.17 a day and left every regular rate unchanged. It also changed ten relative care rates, raising nine in East Texas by $3.35 to $4.55 and cutting Texoma's toddler full-time rate from $17.26 to $16.64.

TWC laid out the BCY 2024 table like its BCY 2025 four-age-group table:

• four age groups (infant, toddler, preschool and school age), which TWC's Commission defined in 40 TAC §809.20(a)(2) as 0 to 17 months, 18 to 35 months, 36 to 71 months, and 72 months and older;
• four provider types: licensed centers (LCCC), licensed homes (LCCH), registered homes (RCCH) and relatives;
• regular, Texas Rising Star Two- to Four-Star (TRS2 to TRS4) and Texas School Ready (TSR) rates, with a single unrated row for relatives;
• full-time and part-time rates, plus a blended rate for school age.

Each of the 4,032 cells in rates.yaml matches one cell of the BCY 2024 table, so we give each a 2023-10-01 value. We extracted the same 4,032 values from the revision with Poppler's pdftotext and with pdfplumber.

Under Government Code §2308.315, each Board must set graduated rates that align TWC's age groupings with the child-to-caregiver ratios and group sizes that Texas's Health and Human Services Commission sets. In S.B. 1555, which amended §2308.315 in 2021, the Legislature gave Boards until December 1, 2023 to set those rates. In 2022, TWC's Commission amended §809.20 to meet that deadline (preamble, pp. 7-8). From December 1, 2023, Boards would set rates for eight age groups, the ones that the Health and Human Services Commission's Child Care Regulation division defines (§809.20(a)(3)). TWC put that change off until TX3C (WD 06-24, Change 2) and kept the four groups in its tables through January 12, 2025 (WD 19-24, Change 1). PolicyEngine already switches to the eight groups in January 2025: uses_expanded_age_groups.yaml starts on 2025-01-01 for TWC's January 13 start. We record the question in source_conflicts/candidates.yaml as us-tx:ccs-rate-age-groups-before-tx3c.

In its BCY 2025 table, TWC printed 2,413 rates above their BCY 2024 levels, by a median of 5.6% and by as much as 59% in Northeast Texas's registered-home infant full-time care. It printed another 1,618 at their BCY 2024 levels and one below: East Texas's relative care rate for school-age blended care, $12.12 against $12.21. Under the method TWC's Commission approved on September 10, 2024, TWC kept any rate outside relative care that would have dropped at its prior-year level (WD 19-24, Change 1, pp. 1-2). TWC sets no minimum for relative care rates, and Boards may set them locally (p. 3).

Panhandle's Board table

Panhandle's Board printed its BCY 2024 rates as an attachment to its Child Care Services policy, CC 3.1, effective December 6, 2023, and cited WD 20-23, Attachment 1 as their source. We read that attachment in the packet for the December 7, 2023 meeting of the Panhandle Workforce Development Consortium's Governing Body. The Board printed all 16 rows and put dashes in the school-age cells of its three Texas School Ready rows, 9 of Panhandle's 144 cells. All 135 cells it printed match the revision, and 81 of them, all Texas Rising Star or Texas School Ready rates, exceed the August original by $0.01 to $0.10 a day.

Under 40 TAC §809.20(b)(2), Boards set Texas School Ready rates only for infants, toddlers and preschoolers, while TWC printed school-age ones equal to the Two-Star rate in its BCY 2024 and BCY 2025 tables. We follow TWC's tables and record the question in draft #10622 as us-tx:ccs-tsr-school-age-enhanced-rate in source_conflicts/candidates.yaml.

Four Dallas County BCY 2025 fixes

On main, four Dallas County cells in rates.yaml differ from TWC's BCY 2025 four-age-group table, and every other BCY 2025 cell matches it. We change the four to TWC's values:

Cell	Was	TWC
LCCC TRS4 infant part time	45.45	46.60
LCCH TRS3 school-age full time	38.65 (the TRS2 value)	39.43
RCCH regular infant part time	37.00 (the preschool full-time value)	34.80
RCCH regular school-age full time	36.50	34.00

Three of the old values break the 40 TAC §809.20 floors. Against a regular rate of $37.00 or $36.50, TWC's Texas Rising Star rates in the same column fall short of their markups. A Three-Star rate of $38.65 would equal the Two-Star rate. The fourth old value, $45.45, meets those floors but falls short of the $46.60 that TWC set for both BCY 2024 and BCY 2025.

TWC's live copy of the BCY 2025 table matches every Wayback capture since March 2025 byte for byte (sha256 33361411…), so we found no second edition of it.

Copay scale

In September 2022, TWC's Commission amended 40 TAC §809.19 so that from December 1, 2023, it would establish the parent share of cost on a statewide sliding fee scale (§809.19(b)). On November 7, 2022, it adopted a statewide method to assess each family's share. TWC put both off until TX3C, and each Board set its own scale until then:

• In WD 06-24, Change 2, TWC wrote that the statewide parent share of cost from the 2022 amendments would take effect with TX3C, and the November 2022 method with TX3C's Phase 2, Release 1.
• In WD 24-23, TWC gave Boards BCY 2024 income ranges, and each Board set its own amounts for them.
• In WD 20-24, TWC wrote that Boards set the parent share of cost locally until TX3C's launch, when TWC would put a statewide method in place.
• TWC's Commission adopted a 7% cap on the parent share of cost in 40 TAC §809.19(a)(1)(C) on July 9, 2024, effective July 29. In WD 20-24, TWC wrote that the cap would take effect with TX3C's launch too.
• In its December 17, 2024 CCDF waiver request, TWC said it would phase the statewide method in "as new families enroll ... or existing families recertify" from January 13, 2025.
• In a January 27, 2025 bulletin, TWC told providers that it had launched TX3C and that its first TX3C payments would cover January 13 to 26.

So we start first_child.yaml, additional_child.yaml and maximum.yaml on 2025-01-01 (actual date 2025-01-13) and cite those sources in each. We keep the values, and since PolicyEngine still copies the earliest value back to 2015, it computes the same copays at every date and applies this scale before TX3C too. We record the question of which scale applies from December 1, 2023 to January 12, 2025 in source_conflicts/candidates.yaml as us-tx:ccs-board-psoc-before-tx3c. We picked each Board's own scale where we have its chart, and TWC's statewide scale for other Boards and for every family from TX3C's launch. In draft #10623, we add six Boards' own scales for the months before TX3C and record the cap's timing as us-tx:ccs-psoc-7pct-cap-timing.

Impact

Take a Panhandle household with one four-year-old in registered-home regular full-time care, 21 days a month, and $12,000 of annual care expense. TWC set Panhandle's daily rate for that care at $27.33 for BCY 2024 and $30.40 from October 2024.

2024 subsidy	Before	After
No income (zero copay)	$7,660.80	$7,080.57 (9 × $573.93 + 3 × $638.40)
$24,000 earnings ($88.4495 monthly copay)	$6,599.41	$6,019.18

PolicyEngine now applies BCY 2024 rates to every month from January 2015 through September 2024. For those months, we compute a lower maximum payment than main in 2,414 cells: the 2,413 where TWC's BCY 2025 rate exceeds its BCY 2024 rate, and Dallas's registered-home regular infant part-time cell, where main had $37.00. We compute a higher one in two cells: the East Texas relative care cell, and Dallas's licensed-center Four-Star infant part-time cell, where main had $45.45. From October to December 2024, we compute a higher maximum payment in two of the four Dallas cells and a lower one in the other two. Where the maximum falls, we compute a lower subsidy for a family whose care costs more than the new maximum, unless its copay already cancels the subsidy.

Invariants

test_tx_ccs_rates_table.py reads rates.yaml and checks every cell of both years:

• Each of the 4,032 cells (28 Boards × 16 provider type and rating rows × 9 age and schedule columns) carries both a BCY 2024 and a BCY 2025 value.
• Each Two-, Three- and Four-Star rate exceeds the matching regular rate by at least 5%, 7% and 9% (40 TAC §809.20(c)), with no rounding allowance. Under §809.20(b)(2) and (c)(1)(B), Boards must set Texas School Ready rates for infants, toddlers and preschoolers at least 5% above the regular rate. The test holds every Texas School Ready rate, school age included, to that floor.
• Each star level exceeds the one below by at least 2 percentage points of the regular rate (40 TAC §809.20(d)).
• Outside relative care, each part-time rate falls below the full-time rate, and each blended rate falls strictly between the two. In its BCY 2024 table, TWC printed equal part-time and full-time rates for Concho Valley's home-based infant care at every rating, and the test requires those ten cells to stay equal. Some Boards pay relative care the same for both schedules, so for relative care the test allows equal rates. No rule requires this ordering, but TWC printed both tables this way, so we read a violation as a value in the wrong column.
• TWC printed each Texas School Ready rate equal to the Two-Star rate in both tables, school age included.
• Outside relative care, TWC kept each BCY 2025 rate at or above its BCY 2024 level, under the method its Commission approved on September 10, 2024 (WD 19-24, Change 1, pp. 1-2). The test checks every cell outside relative care.

The revision and the BCY 2025 table pass every check. The August original misses the markup in 605 cells and the star gap in 366, each by under a cent, so the test fails on its values.

We mutated rates.yaml one cell at a time to see which errors the test catches. Outside relative care, it fails on all 3,350 copies of a part-time rate into a full-time cell that change a value. It also fails when we raise any Two-Star rate by $0.50, move a Concho Valley part-time rate off its full-time rate, or load the August values. In relative care, where the test checks only schedule order, it passes 136 of 190 such copies. Outside relative care, it passes 210 of 1,680 copies of a toddler rate into the infant cell of the same row and schedule, and in relative care, 66 of 82. When we lower a regular rate by $1, it passes in 967 of 1,512 cells, and when we raise a Three- or Four-Star rate by $0.50, it passes in 105 and 1,110 of 1,512. Both parsers' output matches rates.yaml in every cell, and the YAML cases pin single cells in CI.

Test plan

• Twelve tx_ccs_payment_rate YAML cases pin rates from TWC's tables across Boards, provider types, ratings and schedules, including the four Dallas cells and three relative care cells that TWC changed in the revision. Three tx_ccs cases pin the Panhandle annual subsidy ($7,080.57) and a household with a copay across the October boundary (September $485.48, October $549.95, annual $6,019.18). Ten of the 15 fail on main at 0ea88946c6, eight of the twelve rate cases fail with the August original's values, and every case fails on at least one of the two. All 15 pass at e16026e171.
• At e16026e171, CI passes all 141 YAML cases in the Texas CCS folder, including the 51 in tx_ccs.yaml and tx_ccs_payment_rate.yaml. test_tx_ccs_rates_table.py passes 10 tests, and test_source_conflicts.py passes 103.
• We record us-tx:ccs-board-psoc-before-tx3c, with PolicyEngine's pick, and us-tx:ccs-rate-age-groups-before-tx3c, not yet reviewed, in source_conflicts/candidates.yaml. Neither has a rulespec-us register entry yet, so no file carries a # Source conflict marker.
• ruff format and ruff check on the two changed Python files
• Changelog fragment changelog.d/tx-ccs-bcy24-rates.fixed.md

axiom: TheAxiomFoundation/rulespec-us#1990 queued

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (b0a8c09) to head (da0af4e).
⚠️ Report is 189 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main    #10598   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            1         1           
  Lines           15        36   +21     
  Branches         0         1    +1     
=========================================
+ Hits            15        36   +21     

Flag	Coverage Δ
unittests	100.00% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[MaxGhenis]

Ready for hub review
VERDICT: REQUEST_CHANGES

Head reviewed: 1ca5db7e380ab49d5e9f9b9a26cfa77d47ac818a (merge base 0ea88946c6).

This review doesn't cover the rate data against TWC's tables. This review's network policy blocked every primary source host: www.twc.texas.gov, web.archive.org, archive.org, law.cornell.edu and theprpc.org. Neither curl nor a web fetch could reach them. I didn't download, hash or extract any of the ten sources. So I couldn't compare any BCY 2024 or BCY 2025 cell with a TWC PDF, or check any statement about what a TWC letter or 40 TAC §809.20 says. Every such claim below is marked unverifiable. A reviewer with access to those hosts still needs to run checks 1 to 4 (version, 4,032 cells, Panhandle, BCY 2025) before this merges. The request for changes rests on one demonstrated docstring error, P2 below, which is independent of the sources.

What I ran

• policyengine-core test policyengine_us/tests/policy/baseline/gov/states/tx/twc/ccs -c policyengine_us: 132 passed.
• pytest test_tx_ccs_rates_table.py: 8 passed.
• CI on this head: 36 check runs, all completed / success. They include every Baseline, Contrib and Python shard, Microsimulation, Household API Partners, PR validation and codecov.
• git merge-tree --write-tree origin/main 1ca5db7e38: clean, no conflicts.
• I copied the 15 new YAML cases onto origin/main: 10 failed and 5 passed. With rates.yaml from 1541fef973 (the August values), 8 of the 12 rate cases failed. The two failing sets together cover all 12 rate cases. The 3 tx_ccs cases all fail on main, so every new case fails on at least one of the two.
• Mutation runs on test_tx_ccs_rates_table.py (see P2 and P3).
• tx_ccs_copay and tx_ccs for four Panhandle households (1 to 3 children, $15,000 to $60,000 of earnings) in five months of each of 2023, 2024 and 2025, run on origin/main and on this head.
• Composition with Correct Texas CCS expanded-age rates and pay relative and blended care from January 2025 #10636 (head 311f7b04e8), in both merge orders.
• test_source_conflicts.py: 103 passed.

Findings

P2: the rate-table test's docstring claims coverage the test lacks. policyengine_us/tests/policy/baseline/gov/states/tx/twc/ccs/payment/test_tx_ccs_rates_table.py:12-13

• What's wrong: the docstring says "Each check runs over all 4,032 cells of both board contract years, so a value copied into the wrong cell fails here." That's false.

• Evidence:

  • Copy Alamo LCCC regular infant part-time (40.66) into the BCY 2024 full-time cell (44.40). All 8 tests still pass.
  • Lower any regular cell by $1. All 8 still pass, because a lower regular rate only loosens the markup and star-gap floors.
  • Copy the toddler full-time value (40.10) into infant full time instead, and one test fails.

  So the checks catch some misplaced values but not all of them. Under the review rules, a false statement in a test docstring blocks the merge.

• Fix: reword the sentence, for example "so many misplaced values fail here, though an understated regular rate or a part-time value copied into the full-time column can still pass." Alternatively, add a check that would catch those cases, such as part time strictly below full time if every TWC table satisfies that, but only after you verify it against the PDFs.

P3: the invariant checks can't catch understated regular rates or overstated star rates. Same file.

• Evidence: as above, a regular cell lowered by $1 passes every check. A BCY 2024 star cell raised by any amount up to its BCY 2025 value also passes.
• Fix: none required. This is inherent to bound checks, and the 12 pinned YAML cases cover specific cells. A full-table comparison against the extracted PDF is the only complete guard.

P3: PR body, Panhandle count. The body says Panhandle's table matches the revision in all 135 cells it prints and differs from the August original in 81. In rates.yaml, 90 of Panhandle's 144 BCY 2024 cells differ between 1541fef973 and this head: 12 rows, none of them relative care. If the 135 printed cells are every non-relative cell, the count should be 90. A count of 81 fits only if the packet prints the relative row and omits one row with 9 differences, such as LCCC TRS3, LCCH TRS2 or LCCH TSR. I couldn't open the packet. Please state which 135 cells the packet prints, or correct the count.

P3: PR body wording. "The brief's datapoint holds" refers to a document that public readers can't see. The value it cites does match rates.yaml: 27.33 in BCY 2024 and 30.40 in BCY 2025.

P3, outside this PR's scope: after either merge order with #10636, its UNCORRECTED_FOUR_AGE skip never fires, because this PR replaces all four values. #10636 can drop that block once this PR merges. I opened no work on it.

Checked and found consistent

These checks compare the PR's claims with the repository, not with TWC's PDFs.

• Completeness: 4,032 of 4,032 cells carry both a 2023-10-01 and a 2024-10-01 value (28 Boards × 16 rows × 9 columns).

• Revision vs. August original (1541fef973 against this head): 2,656 cells differ.

  • 2,646 are TRS2, TRS3, TRS4 or TSR cells, all raised, by $0.01 to $0.17.
  • 10 are relative care cells. Nine East Texas cells rise by $3.35 to $4.55, and Texoma toddler full time falls from 17.26 to 16.64.
  • No regular cell changes.

  This matches claim 1.

• BCY 2025 vs. main: exactly four cells differ, the four Dallas County cells listed (45.45→46.60, 38.65→39.43, 37.00→34.80, 36.50→34.00). Whether they match TWC's BCY 2025 table is unverifiable here.

• BCY 2024 to BCY 2025:

  • 2,413 cells rise, with a median of 5.62%. The maximum is 59.2% (Northeast Texas RCCH infant full time), with East Texas at 56.0% and Concho Valley at 54.8%.
  • 1,618 cells don't change.
  • One cell falls: East Texas relative school-age blended, 12.21 to 12.12.

  This matches claim 5.

• Invariants on the YAML data: the August values fail the markup check in 605 cells (largest shortfall $0.0097) and the star-gap check in 366 (largest shortfall $0.008). The revised BCY 2024 values and the BCY 2025 values fail neither. This matches the PR body and the docstring's "under a cent".

• YAML test arithmetic: all 12 tx_ccs_payment_rate expectations equal the rates.yaml value × days. The "August 29 original printed …" comments (46.62, 17.26, 11.37) match 1541fef973.

  • tx_ccs Case 6: 9 × 573.93 + 3 × 638.40 = 7,080.57.
  • Case 7: 94,466 × 0.68 / 12 = 5,353.07; 2,000 / 5,353.07 = 37.36%; 2% + 36.36% × 0.0493 / 0.74 = 4.42%, which gives a copay of 88.45 and subsidies of 485.48 and 549.95.
  • Case 8: 6,019.18.
  • The impact table's "before" values are 12 × 638.40 = 7,660.80 and 12 × (638.40 − 88.45) = 6,599.41.

• Copay date move: tx_ccs_copay is identical on main and this head in all 60 household-months checked, across 2023, 2024 and 2025. The move changes no copay. tx_ccs changes only through the BCY 2024 rates. One effect the body states only implicitly: periods before October 2023 now also use BCY 2024 rates instead of BCY 2025 rates.

• Composition with Correct Texas CCS expanded-age rates and pay relative and blended care from January 2025 #10636: git merge-tree gives the same clean tree in both orders. On the merged tree, both rate-table test files pass (31 passed), and the Texas CCS YAML folder passes (170 passed).

• Repo rules:

  • The changelog fragment is a top-level changelog.d/tx-ccs-bcy24-rates.fixed.md.
  • The PR touches no partner test files.
  • References use one entry per document, with #page=N only in the href.
  • The variable reference tuple has no titles.
  • test_source_conflicts.py passes.

Unverifiable here (sources blocked)

• Which version of Attachment 1 governs BCY 2024: dates, titles, which file TWC's WD 20-23 page links, and whether TWC issued any later revision or mid-year change.
• All 4,032 BCY 2024 values and all BCY 2025 values against TWC's PDFs, including the four Dallas fixes, the Alamo "Reg" relative row and the Board-name mapping.
• Panhandle's packet, beyond the count consistency noted above.
• Every statement about what WD 20-23, WD 19-24 Change 1, WD 20-24, WD 24-23, the December 17, 2024 waiver request and 40 TAC §809.20(c) and (d) say. That includes the copay YAML comments ("Each Board set its own parent share of cost scale until … TX3C … launched on January 13, 2025", "phased in … as families enrolled or recertified"), the docstring's paraphrases of §809.20 and WD 20-23, and the reference titles.
• The two-parser extraction claim and the sha256 claims.

---

Generated by Claude Code

[MaxGhenis]

3dae28f689 and 4f8545d76e respond to the review above, which ran at 1ca5db7e38.

• 3dae28f689 narrows the rate-table docstring to what holds and adds two checks. Each Texas School Ready rate must equal the Two-Star rate. Outside relative care, each part-time rate must fall strictly below the full-time rate, except in Concho Valley's BCY 2024 home-based infant cells. In mutation runs, the test now fails when someone copies a part-time rate into a full-time cell in either year. It still passes a regular rate lowered by $1 or a Four-Star rate raised by $0.50, as the docstring now says.
• 4f8545d76e rewrites the docstring, the YAML comments and the changelog so that TWC and the Boards act. It changes no parameter value or test logic.
• The PR body now explains the Panhandle count. The Board printed 135 of its 144 cells and left the nine school-age Texas School Ready cells blank, so the editions differ in 81 printed cells and 90 in all. The body also corrects its Dallas paragraph: three of the four old values break the §809.20 floors, and the fourth, $45.45, meets them.
• A second review is now comparing the data with TWC's PDFs.

[MaxGhenis]

Ready for hub review
VERDICT: REQUEST_CHANGES

Head reviewed: 4f8545d76e0f616ae821bf2614972171e4ef9d90 (merge base 0ea88946c6).

This review reached the primary sources. The rate data are right: every BCY 2024 and BCY 2025 cell in rates.yaml matches TWC's PDFs, and so does every cell Panhandle's Board printed. At this head, three statements block the merge:

• what the PR body says the test catches;
• the copay comments' account of the statewide scale;
• the PR body's Texas School Ready floor for school-age care.

The PR moved to 571a8e8841 while this review ran. That commit fixes the first and third. The copay finding still blocks there; see "The new head" at the end.

Sources

web.archive.org reset every connection, and a web fetch refused it, so I took each file from its live URL. TWC serves files to a browser user agent.

Source	Result
1. WD 20-23 Attachment 1, revised (wf/docs/20-23-att1-twc.pdf)	sha256 dcb54a8d…d234 matches. PDF title "WD 20-23 Attachment 1", created 2023-10-16.
2. WD 20-23 Attachment 1, original (wf/policy-letter/wd/20-23-att1-twc.pdf)	sha256 1b068d13…7c19 matches. PDF title "BCY'24 CC Provider Payment Rates", created 2023-08-29.
3. WD Letter 20-23	sha256 9d09c880…3b1c matches.
4. BCY 2025 four-age-group table	sha256 33361411…f1c4 matches. Footer "WD Letter 19-24, Attachment 1".
5–7. WD 19-24 Change 1, WD 20-24, WD 24-23	Not obtainable. Each file URL returns 404, and the letters' pages return "Access denied". TWC's rescinded-letters page offers copies only by email. I got WD 24-23's two attachments from TWC.
8. December 17, 2024 waiver request	sha256 8008606e…2633 matches.
9. Panhandle packet, December 7, 2023	sha256 76377b88…f78a matches. The table is on file page 112.
10. 40 TAC §809.20 (Cornell)	Read. It shows the text effective 2024-07-29. TWC's adoption documents below give the 2022 text.

I also used these TWC primary sources:

• the Chapter 809 rules TWC adopted on September 13, 2022 (ogc/docs/fr-809-ch-rev-09-22-twc.pdf) and July 9, 2024 (ogc/docs/fr-809-trs-updates-7-24-twc.pdf), and TWC's current Chapter 809 with source notes;
• the November 7, 2022 Commission minutes and that meeting's statewide parent share of cost (PSoC) methodology paper;
• the September 10, 2024 minimum-rates paper and that meeting's minutes;
• the January 28, 2025 PSoC methodology paper;
• WD 06-24, Change 2 (October 10, 2024) and its Attachment 3;
• WD 08-25;
• TWC's BCY 2025 PSoC income ranges;
• TWC's eight-age-group BCY 2025 and BCY 2026 tables;
• TWC's BCY 2023 rate table (WD 22-22 Attachment 1, sha256 afc5c264…29bf);
• the Child Care Services Guide (August 19, 2026).

Two Board sources: South Texas's BCY 2024 PSoC chart and policy CC 3.1 in Panhandle's packet.

What I ran

• Extraction:
  • Ran poppler's pdftotext -layout and pdfplumber on sources 1, 2 and 4: 4,032 cells each.
  • The two engines agree on all 12,096 values. Each page holds one Board, 16 rows and nine values per row.
  • I took the column order from the header boxes in each PDF, not from the YAML.
  • Both engines also read Panhandle's page 112: 144 positions, 135 printed.
  • A second extraction, written separately and using three text readings, agrees on every cell and found no hidden or overlapping text.
• Tests:
  • pytest test_tx_ccs_rates_table.py: 10 passed.
  • policyengine-core test policyengine_us/tests/policy/baseline/gov/states/tx/twc/ccs -c policyengine_us: 134 passed.
  • ruff format --check and ruff check on the two changed Python files: clean.
• CI on this head: 36 check runs, all completed with success. They include every Baseline, Contrib and Python shard, Microsimulation, Household API Partners, PR validation, Package and compatibility, and codecov.
• Mutations:
  • 89,742 single-error mutations of rates.yaml in 65 error classes, each run through the test's own functions.
  • On 14 sampled mutations, that method gave the same result as running pytest on a mutated copy of the file. I reran each example quoted below with pytest.
• Commits:
  • 4f8545d76e against 3dae28f689: the loaded data in all four parameter files are identical. The test file's syntax tree is identical once docstrings are removed. So 4f8545d76e changes only comments, the docstring and the changelog, as it says.
  • 3dae28f689 changes only the test file.
• Merges: git merge-tree is clean against current main (cd821899ae) and against Correct Texas CCS expanded-age rates and pay relative and blended care from January 2025 #10636's head (02f4163256).
• Re-checks: separate re-checks tried to refute each P2 finding below. None did. Where they found overreach, I narrowed the finding.

Findings

P2: the PR body says the test catches errors it lets through, and the docstring's example reads the same way. PR body, "Invariants"; policyengine_us/tests/policy/baseline/gov/states/tx/twc/ccs/payment/test_tx_ccs_rates_table.py:20-23.

• What's wrong:
  • The PR body says: "It fails if someone copies a part-time rate into a full-time cell or a toddler rate into an infant cell". The author's earlier comment on this PR says the test "now fails when someone copies a part-time rate into a full-time cell in either year".
  • The docstring says: "A test fails when a value typed into the wrong cell breaks one of these bounds, as a part-time rate copied into a full-time cell does."
  • The docstring's previous paragraph limits the ordering pattern to care outside relative care, but its example is unqualified.
  • Relative care gets the non-strict check (line 152), and no other check constrains it.
• Evidence:
  • Part-time rate copied into the full-time cell: this changes a value in 3,540 cells.
    • Outside relative care, all 3,350 such mutations fail.
    • In relative care, 136 of 190 pass all 10 tests: infant, toddler and preschool cells, 68 a year.
    • Alamo relative infant full time, BCY 2024, 18.61 → 15.19 (the part-time rate): 10 passed.
    • Alamo relative preschool full time, BCY 2025, 13.84 → 12.19: 10 passed.
  • Toddler rate copied into the infant cell: passes in 276 of 1,762 cases.
    • 66 of the 276 are relative care.
    • Most of the rest fall under gaps the docstring already names, such as a regular rate typed too low.
  • Classes the docstring's "can still pass" sentence leaves out:
    • changes to relative care rates: 3,088 of 3,528 changes of ±$0.01 to ±$1.00 pass;
    • a Three- or Four-Star rate a cent too low: 1,235 and 1,231 of 1,512 pass;
    • transposed cents digits, such as 44.40 → 44.04: 1,796 of 6,987 pass;
    • a Board's BCY 2024 values replaced by its BCY 2025 values: 27 of 28 Boards pass.
• Fix:

  • Narrow the PR body's sentence and the docstring's example to care outside relative care, and add relative care and cross-age copies to the "can still pass" list.

  • Optionally, make the relative check strict with its own tie list. TWC printed relative part time equal to full time for the same 17 Board and age pairs in both tables:

    • infant: Brazos Valley, Cameron County, Coastal Bend, Concho Valley, Deep East, East Texas, Golden Crescent, Gulf Coast, Heart of Texas and Lower Rio;
    • toddler: Coastal Bend, Deep East, East Texas and Heart of Texas;
    • preschool: Deep East and Heart of Texas;
    • school age: Deep East, where part time, blended and full time are all $8.00.

    I ran that variant. It passes both years of the current data and fails all 190 relative part-time-into-full-time copies that change a value.

P2: the copay comments say TWC set no statewide scale before TX3C, but from December 1, 2023, TWC's rule required one, and the Commission had already adopted it. policyengine_us/parameters/gov/states/tx/twc/ccs/copay/rate/first_child.yaml:28-31, additional_child.yaml:9-12, maximum.yaml:3-7; changelog.d/tx-ccs-bcy24-rates.fixed.md:1; PR body, "Copay scale".

• What's wrong:
  • The comments say "Each Board set its own parent share of cost scale until TWC launched its TX3C case management system on January 13, 2025 (WD 24-23 for BCY 2024)".
  • They also say "TWC set no statewide scale for earlier periods". maximum.yaml says "no statewide scale or cap".
  • That holds before December 1, 2023, when §809.19(a) had each Board set its own policy.
  • It is wrong for the scale from December 1, 2023 and for the cap from July 29, 2024. The comments describe practice as if it were the whole history.
• Evidence:
  • The 2022 rule: TWC's 2022 adoption of §809.19 (adopted September 13, 2022, effective October 3, 2022) splits the section.
    • "(a) The following provisions are effective prior to December 1, 2023: … A Board shall set a parent share of cost policy".
    • "(b) The following provisions are effective on December 1, 2023: … The parent share of cost shall be: … established by the Commission and determined by a sliding fee scale based on the family's size and gross monthly income".
    • The preamble says: "New §809.19(b) will be effective December 1, 2023, and establishes a statewide PSoC policy."
  • The November 7, 2022 methodology: that day the Commission approved Agenda Item 19, the "Board Contract Year 2024 Statewide Parent Share of Cost Methodology".
    • Its paper sets 2% to 7% of income for one child, with add-ons for more children up to 13%.
    • It says the methodology "would go into effect with implementation of the new CCCM system in the fall of 2023".
  • TWC's own account:
    • The July 2024 adoption preamble (page 4) says "the rules that became effective December 1, 2023, stated that the PSoC amount is established by TWC", and that "the PSoC is no longer determined or established by the Board".
    • The January 28, 2025 paper says the Commission "had previously set a statewide PSoC in Ch. 809 rule amendments adopted on September 13, 2022".
    • No rule moved December 1, 2023: TWC's Chapter 809 source notes list no 2023 amendment to §809.19.
  • The cap: since July 29, 2024, §809.19(a)(1)(C) has capped the parent share of cost at 7 percent of family income "regardless of the number of children receiving child care services". The waiver request the PR cites says the same (page 2): "On July 9, 2024, TWC’s Commission adopted rule amendments to limit all co-payment amounts".
  • The deferral: TWC tied the start to the new system, not to the rule's dates.
    • WD 06-24, Change 2 says the 2022 "statewide PSoC" changes "will become effective with the implementation of TX3C", and the November 7, 2022 methodology "will become effective when Phase 2, Release 1, is released".
    • TWC's BCY 2025 "Income Ranges for Parent Share of Cost Assessment" are "Effective October 1, 2024–December 8, 2024".
    • South Texas's own chart is "Effective 10/1/2023 - 9/30/2024".
    • Panhandle's December 2023 policy keeps a Board "PSoC fee schedule".
    • The January 28, 2025 paper and the waiver request start the statewide method with TX3C.
• Why it blocks:
  • The comments, the changelog and the PR body state the history wrongly.
  • The rule's fixed dates and TWC's start at TX3C conflict, and the PR dates the scale by TX3C without recording that. CLAUDE.md says to record every such choice. CONTRIBUTING applies that to a PR that "takes such a reading, or finds one already in the code".
  • Under the reading the code follows now, no copay changes at any date. The model copies each parameter's earliest value back to 2015, so the 2025 statewide scale already applies to December 2023 through January 2025.
• Fix:
  • Reword the three comments, the changelog and the PR body to name both the rule and the deferral. For example: "From December 1, 2023, 40 TAC §809.19(b) had the Commission set the parent share of cost statewide, and the Commission had approved a BCY 2024 statewide methodology on November 7, 2022. TWC tied both to its new case management system (WD 06-24, Change 2). Until TX3C launched on January 13, 2025, Boards kept assessing their own scales within TWC's income ranges. TWC then phased in the statewide scale as families enrolled or recertified."
  • Record the split. A # Source conflict marker has to name an entry in the pinned register. So first add a rulespec-us entry that lists this PR under proposed, or a source_conflicts/candidates.yaml item until then, and mark the 2025-01-01 lines once the entry is pinned.
  • The entry should say the code applies the 2025 statewide scale to those months. The Board-scale reading belongs to Apply Texas Workforce Board parent share of cost scales before TX3C #10623.
  • Alternatively, drop the copay changes from this PR and let Apply Texas Workforce Board parent share of cost scales before TX3C #10623 carry the dates, the comments and the entry.

P2: the PR body says every Texas School Ready rate must clear 5%, but §809.20 provides Texas School Ready rates only through preschool age. PR body, "Invariants" and "Test plan"; test_tx_ccs_rates_table.py:167-169.

• What's wrong:
  • The PR body says "TWC tiers Texas School Ready rates under the same rule (WD 20-23), so they must clear 5% too". That is right for the 504 infant, toddler and preschool TSR cells a year. It is not right for the 252 school-age ones.
  • The comment at lines 167-169 says the same: "Under §809.20(c), Boards need only set Texas School Ready rates at least 5 percent above the regular rate".
  • The body also says "Panhandle's Board table matches the revision, so no source conflict needs recording". That passes over the nine school-age TSR cells the Board didn't print.
• Evidence:
  • §809.20(b)(2) reads the same in the versions effective October 3, 2022 and July 29, 2024: "only for infant, toddler, and preschool-age children at child care providers that participate in integrated school readiness models". The 5% in (c)(1)(B) covers those rates.
  • The 2022 preamble says (b)(2) was "amended to include infants and toddlers for enhanced rates for providers participating in the Texas School Ready program for those age groups".
  • The Child Care Services Guide, B-703, says Boards must set TSR enhanced rates "for infant, toddler, and preschool-age children".
  • TWC's own BCY 2023 table (WD 22-22 Attachment 1) prints "-" in all 252 school-age TSR cells. Its BCY 2024 table, in both editions, and its BCY 2025 table print them equal to the Two-Star rate. Panhandle's Board printed "-" in its nine, perhaps keeping the earlier layout.
  • B-703 gives a Texas Rising Star provider that also takes part in TSR the star rate at every age. So the printed school-age TSR rates go beyond the rule only for a TSR provider without a star rating. Whether (b)(2)'s "only" bars a higher rate or only limits what Boards must set is a reading the PR should state.
• Fix:
  • Restate both statements, and say that the school-age TSR cells follow TWC's tables.
  • Then record that reading as a source conflict, or explain why (b)(2) doesn't bar it. Either way, the test results don't change.
  • main already had the BCY 2025 school-age TSR cells; this PR adds BCY 2024 values and a test that enforces them.

P3: commit messages. Squash merges here keep every commit message in the body; see #10585 and #10599 on main.

• 59d9721606 says "The table's own tier arithmetic confirms each one" and "the table's TRS2 rate of $36.55 is $34.80 x 1.05". But $34.80 × 1.05 = $36.54, and the floors don't tell $45.45 from $46.60.
• 1541fef973 says "so no statewide BCY 2024 copay scale exists", which the November 7, 2022 approval contradicts.
• Edit the squash message.

P3: WD 19-24, Change 1 citations. The docstring (lines 10-12), the case comment at tx_ccs_payment_rate.yaml:526 and the PR body cite WD 19-24, Change 1 for two points: the BCY 2025 no-reduction rule and Board-set relative care rates.

• I couldn't check that letter: TWC no longer serves it, and WD 08-25 rescinds it.
• Public TWC sources support both points:
  • On September 10, 2024, the Commission unanimously approved a minimum-rate method "for board contract year 2025 and subsequent years". The paper says "Ensure that no rate is reduced; … TWC will retain the current rate".
  • WD 08-25, for BCY 2026, says "TWC does not set minimum thresholds for payment rates for child care provided by relative providers. Boards may set these rates locally."
• Consider citing those as well.

P3: PR body details.

• "So first_child.yaml, additional_child.yaml and maximum.yaml now … cite those sources": they cite WD 24-23 and the waiver request, not WD 20-24.
• "by up to 59% for home-based infant full-time care in Northeast Texas, East Texas and Concho Valley": only Northeast Texas reaches 59.2%. East Texas tops out at 56.0% and Concho Valley at 54.8%.

P3: rounding in the copay comments. policyengine_us/tests/policy/baseline/gov/states/tx/twc/ccs/tx_ccs.yaml:208-209 and :246; PR body, "Impact".

• Lines 208-209 say the rate is 4.42%, "so the copay is 4.42% * $2,000 = $88.45". But 4.42% of $2,000 is $88.40. The unrounded rate, 4.4225%, gives the model's $88.4495.
• Line 246 says "9 months at $485.48 + 3 months at $549.95 (Case 7) = $6,019.18". That sum is $6,019.17. The expected $6,019.18 is the model's total with the unrounded copay, and it passes within the 0.01 margin.
• For the same reason, the Impact table's $6,599.41 and $6,019.18 would be $6,599.40 and $6,019.17 with the $88.45 copay it shows.
• The expected values are right. Write the rate as 4.4225%, and say the totals use the unrounded copay.

Checked and found correct

• Version: the revision supersedes the original for BCY 2024.
  • TWC's WD 20-23 page (status Active, published 09/20/2023, no change number) links wf/docs/20-23-att1-twc.pdf as its only attachment. That file is the revision. It was created 2023-10-16 and is still headed "Effective October 1, 2023".
  • The original was created 2023-08-29, the letter's date, and sits unlinked at the old policy-letter path.
  • Panhandle's packet prints the revision's values.
  • The revision fixes the original's 605 markup and 366 star-gap shortfalls.
  • TWC's letter index lists no change to WD 20-23 and no other BCY 2024 rate letter. I found no mid-year BCY 2024 rate change.
• BCY 2024: 4,032 of 4,032 2023-10-01 values match the revised PDF.
  • All 28 Board names map to their keys.
  • Alamo's relative row prints "Reg" in the revision and in the BCY 2025 table. Its values sit under RELATIVE: NONE and match.
• 1541fef973: 4,032 of 4,032 2023-10-01 values match the August original.
• Claim 1: the revision changes 2,656 cells.
  • 2,646 TRS2, TRS3, TRS4 or TSR cells rise by $0.01 to $0.17, across all 28 Boards.
  • Nine East Texas relative cells rise by $3.35 to $4.55.
  • Texoma's relative toddler full-time rate falls from $17.26 to $16.64.
  • No regular rate changes.
• BCY 2025: 4,032 of 4,032 2024-10-01 values match TWC's BCY 2025 table. On main, exactly four differ: the Dallas cells, with the old and new values the PR lists.
• Dallas: the PR body's account holds.
  • $37.00 is the RCCH regular preschool full-time rate, and $38.65 is the LCCH Two-Star school-age full-time rate.
  • As regular rates, $37.00 and $36.50 put every TRS and TSR rate in their rows under the markup.
  • As a Three-Star rate, $38.65 fails both the 7% markup ($39.38) and the gap.
  • $45.45 clears both floors: the regular rate is $40.60 and the Three-Star rate $43.50.
• Panhandle (claim 5):
  • The table is the BCY 2024 rates attachment to the Board's policy CC 3.1 (effective December 6, 2023). It is on file page 112 of the packet for the December 7, 2023 meeting of the Panhandle Workforce Development Consortium's Governing Body.
  • The Board printed all 16 rows. In nine cells, all TSR school-age, it printed "-" instead of a rate.
  • All 135 printed cells equal the revision and rates.yaml.
  • 81 exceed the original by $0.01 to $0.10: TRS2 24, TRS3 25, TRS4 17, TSR 15.
  • Each of those nine cells differs by $0.01 between the editions, so 90 cells differ in all.
• Test patterns on the PDF values, not the YAML:
  • TSR equals TRS2 in every cell of both tables.
  • The revision and the BCY 2025 table clear every markup and gap. The original fails 605 markups (largest shortfall $0.0097) and 366 gaps ($0.0080).
  • The 366 counts Board, provider-type and column positions with a failing step, as the test reports them. Counted per tier step, like the 605, it is 461.
  • Outside relative care, part time is strictly below full time, and blended strictly between, in every cell except the ten BCY 2024 Concho Valley LCCH and RCCH infant cells. Those are exactly the cells EQUAL_PART_AND_FULL_TIME lists. BCY 2025 has no exception.
  • No non-relative cell falls from BCY 2024 to BCY 2025.
• Mutation claims that hold:
  • every change to a Two-Star or TSR rate fails: 21,168 of 21,168;
  • every untied Concho Valley cell fails: 80 of 80;
  • the August values fail, with 605 markup and 366 gap failures;
  • every swap of two rating rows fails;
  • a Four-Star rate raised by $0.50 can pass (1,110 of 1,512), and so can a regular rate lowered by $1 (967 of 1,512).
• Rule statements:
  • §809.20(c)'s 5, 7 and 9 percent and (d)'s 2-point gap read the same in the versions effective October 3, 2022 and July 29, 2024.
  • WD 20-23 says "Texas Rising Star and Texas School Ready (TSR) maximum payment rates continue to be tiered in accordance with … §809.20".
  • The docstring states (c), (d) and WD 20-23 correctly. The school-age TSR point above concerns the PR body and the comment at lines 167-169.
  • On September 10, 2024, the Commission approved keeping each rate at or above its prior-year level for BCY 2025. That supports the docstring's no-reduction sentence.
  • The PR body's four age groups match §809.20(a)(2) as in force before December 1, 2023: infants 0 to 17 months, toddlers 18 to 35 months, preschool 36 to 71 months, and school age 72 months and older. The tables print no age ranges.
• YAML cases:
  • All 24 expected outputs in the 12 rate cases equal the PDF rate times days.
  • The three "August 29 original printed" comments ($46.62, $17.26, $11.37) match the original.
  • The Impact section's $27.33 and $30.40 match.
  • I reproduced the PR's counts:
    • at 0ea88946c6 with the head test files, 10 of the 15 new cases fail and 24 of 34 cases pass;
    • at this head with 1541fef973's rates.yaml, 8 of the 12 rate cases fail;
    • every new case fails in one of the two runs, and all 34 pass at this head.
• Copay quotes:
  • The waiver request (page 2) says the new copays "will be phased in as new families enroll in the program or existing families recertify their eligibility effective January 13, 2025". That matches the PR's quote and the comments' "phased in" sentence.
  • WD 24-23's Attachment 2 is the "Income Ranges for Parent Share of Cost Assessment, Effective October 1, 2023–September 30, 2024".
• PR body, "Outside this PR":
  • TWC's eight-age-group tables differ from bcy2026.csv in 127 of 8,512 cells and from bcy2025.csv in 1 of 6,720: Golden Crescent LCCH regular 6–13 part time, 22.20 against TWC's 22.80.
  • TWC's January 2025 table prints no relative row and no blended columns.
• Other source statements in the PR body:
  • WD 20-23 is dated August 29, 2023 and effective October 1, 2023.
  • TWC's page links the revision as the only attachment, and the revision's metadata gives October 16, 2023.
  • TWC still serves the original at the policy-letter path.
  • The live files match dcb54a8d… and 33361411….
  • The two parsers extract the same 4,032 values.
  • From BCY 2024 to BCY 2025, 2,413 rates rise (median 5.6%), 1,618 stay the same, and East Texas's relative school-age blended rate falls from 12.21 to 12.12.
  • Apply Texas Workforce Board parent share of cost scales before TX3C #10623 is a draft that applies Board scales before TX3C.
• Repo rules:
  • The changelog fragment is a top-level .fixed.md.
  • References keep one entry per document, with #page= only in the href.
  • No partner test file changes.
  • The Texas CCS entry in programs.yaml has no year field to update.

Not verified

• The text of WD 19-24 Change 1, WD 20-24 and the body of WD 24-23: not obtainable. So I couldn't check three statements:

  • "TWC sets no floor for relative care rates and leaves them to Boards (WD 19-24, Change 1, page 3)";
  • "In WD 20-24, TWC wrote that Boards set the parent share of cost locally until TX3C's launch";
  • what page 2 of WD 24-23 says.

  The TWC documents above support the substance of each.

• The Wayback captures: web.archive.org was blocked, so I couldn't check the statements that captures match the live files byte for byte.

• The Texas Register: the Secretary of State's archive returns "not found" for 2022–2024 issues. The rule texts above come from TWC's own adoption documents, which TWC says will have "no substantive changes" in the Register.

• The axiom: line: rulespec-us is outside this review's access.

• The full test suite: CI is the evidence for it.

Against the first review

• Panhandle: its guess doesn't hold. The packet omits no row: it prints all 16 rows and puts "-" in the nine TSR school-age cells. So the PR's 81 printed cells plus those 9 cells give the 90 the first review counted.
• Impact arithmetic: its check "12 × (638.40 − 88.45) = 6,599.41" is a cent off. That product is 6,599.40; 6,599.41 is the model's result with the unrounded copay (see the P3 above).
• Everything else: nothing else it reported is contradicted. I reproduced its failure counts on main and with the August values. Its counts of 132 and 8 passing tests at 1ca5db7e38 become 134 and 10 here, with the two tests 3dae28f689 adds.

The new head

While this review ran, the PR moved to 571a8e8841 (pushed 05:04Z), and the body was edited at 05:18Z. That commit changes only comments, the docstring and the changelog. Against the findings above:

• Test claims: fixed.
  • The docstring now says the part-time-into-full-time catch holds "outside relative care". It lists relative care copies and copies from another age group among the errors that can pass.
  • The body now gives counts that match mine: 3,350 of 3,350 fail outside relative care, 136 of 190 relative copies pass, and 210 of 1,680 non-relative toddler-into-infant copies pass.
• Copay scale: still open.
  • first_child.yaml and additional_child.yaml still say "TWC set no statewide scale for earlier periods". The body and the changelog still say each Board set its own scale until TX3C.
  • maximum.yaml now says the Commission adopted the 7% cap in July 2024, and that TWC said it would apply it with TX3C. That settles the cap point.
  • Nothing yet mentions §809.19(b)'s December 1, 2023 date or the November 7, 2022 methodology, and nothing records the split.
• School-age TSR: fixed. The docstring, the test comment and the body now state the §809.20(b)(2) limit. The body says the PR follows TWC's tables, and draft Add Texas CCS provider payment rates for October 2016 to September 2023 #10622's candidates.yaml adds us-tx:ccs-tsr-school-age-enhanced-rate, which cites this PR.
• P3 "cite those sources": fixed. The three copay files now cite WD 20-24. The other P3 items are unchanged.
• New statement checked: "PolicyEngine copies a parameter's earliest value back to January 2015". policyengine_us/system.py backdates every parameter to FIRST_MODELED_YEAR, which tools/parameters.py sets to 2015.
• CI on 571a8e8841: still running when I posted: 24 runs complete, all green, and 11 in progress.
• Not reviewed: the January 27, 2025 bulletin the new comments cite.

---

Generated by Claude Code

[MaxGhenis]

We correct these statements from our earlier reply, the earlier description, our code comments and two commit messages. We changed the code comments in 571a8e8841, b08c480125, 2434ac6c7f and c43a3f37e3, and the description in place.

• We wrote that the test now fails when someone copies a part-time rate into a full-time cell in either year. We left the same claim in the docstring at 3dae28f689, the commit we said had narrowed it to what holds. Outside relative care, the test fails on all 3,350 such copies that change a value; in relative care, where it checks only schedule order, it passes 136 of the 190 that do. We now say both in the docstring.
• We listed a toddler rate copied into an infant cell among the errors the test catches. Outside relative care, it passes 210 of 1,680 such copies, and in relative care, 66 of 82.
• We wrote that the test still passes a regular rate lowered by $1 or a Four-Star rate raised by $0.50. It passes 967 of 1,512 such regular rates and 1,110 of 1,512 such Four-Star rates.
• We wrote in the docstring that the test builds no tax-benefit system. Under pytest it builds one, because pytest loads policyengine_us/conftest.py and importing the package builds the system. We removed the sentence.
• We wrote that PolicyEngine applies a parameter's earliest value to every earlier period. It copies that value back to January 2015.
• In the test comments, we cited §809.20(c) for a 5% floor on every Texas School Ready rate. Under (b)(2) and (c)(1)(B), Boards set Texas School Ready rates only for infants, toddlers and preschoolers, at least 5% above the regular rate, while TWC printed school-age ones too. We record that question in draft Add Texas CCS provider payment rates for October 2016 to September 2023 #10622 as us-tx:ccs-tsr-school-age-enhanced-rate, and we removed the description's line that no source conflict needs recording.
• We wrote that TWC set no statewide copay scale before TX3C, and in maximum.yaml, no cap. In September 2022, TWC's Commission amended §809.19 so that from December 1, 2023, it would establish the parent share of cost on a statewide scale. On November 7, 2022, it adopted a statewide method, and in July 2024 it adopted the 7% cap, effective July 29. TWC put the scale and the cap off until TX3C (WD 06-24, Change 2; WD 20-24), so each Board kept setting its own scale. We record the question, with PolicyEngine's pick, as us-tx:ccs-board-psoc-before-tx3c.
• We attributed the four age ranges to TWC's rate table, but TWC's Commission defined them in §809.20(a)(2). In S.B. 1555, the Legislature gave Boards until December 1, 2023 to set the graduated rates that Government Code §2308.315 requires. To meet that deadline, the Commission set eight age groups to replace the four, and TWC put that change off until TX3C. We record the question as us-tx:ccs-rate-age-groups-before-tx3c.
• We wrote that the three copay files cite WD 20-24. They didn't until we added it to each in 571a8e8841, along with TWC's January 27, 2025 TX3C bulletin.
• We placed Panhandle's table in a December 7, 2023 Board packet and called it Attachment 1 to the Board's policy. The Board attached the table, unnumbered, to its Child Care Services policy, CC 3.1, effective December 6, 2023, and cited TWC's WD 20-23, Attachment 1 as its source. The December 7 packet went to the Panhandle Workforce Development Consortium's Governing Body. In the nine school-age Texas School Ready cells, the Board printed dashes, not blanks.
• We wrote that draft Apply Texas Workforce Board parent share of cost scales before TX3C #10623 applies each Board's own scale before TX3C. So far we've added six there: Alamo, Borderplex, Brazos Valley, Dallas County, Panhandle and South Texas.
• We wrote that BCY 2025 rates rose by up to 59% for home-based infant full-time care in Northeast Texas, East Texas and Concho Valley. Only Northeast Texas reaches 59%, in registered homes; East Texas tops out at 56.0% and Concho Valley at 54.8%.
• We wrote that before October 2024, we compute a lower maximum payment in 2,413 cells and a higher one in one. Against main, which carried the four Dallas errors back to 2015, we compute a lower one in 2,414 cells and a higher one in two. From October to December 2024, we compute a higher maximum in two of the Dallas cells and a lower one in the other two.
• In tx_ccs.yaml, we gave the copay rate as 4.42% and summed rounded monthly subsidies to $6,019.18. The rate comes to 4.422476% and the copay to $88.4495. The rounded monthly amounts sum to $6,019.17, and we compute $6,019.18 from the unrounded copay.
• In 59d9721606's commit message, we gave $34.80 × 1.05 as $36.55, TWC's Two-Star rate; it comes to $36.54. In 1541fef973's, we wrote that no statewide BCY 2024 copay scale exists. None took effect, but TWC's Commission had amended its rules in 2022 to set one from December 1, 2023, and TWC put it off until TX3C. We'll leave both claims out of the squash commit message.

We changed no parameter value or test logic in 571a8e8841, b08c480125, 2434ac6c7f or c43a3f37e3. At c43a3f37e3, the rate-table, source-conflict and Texas CCS YAML tests pass.

[MaxGhenis]

Hand-off to the US + core hub. Head e16026e171d6d1c4f3dd629dd065b856abdc7bfc, which merges main at d161c9715b into c43a3f37e3.

• Review. A Claude Opus 5.5 agent reviewed each head from 4f8545d76e on, and ~/reviews/tx-ccs-bcy24/reviews/ holds each verdict. At this head, it agreed with the code, our correction comment and the description, once we fixed two statements in the description that Correct Texas CCS expanded-age rates and pay relative and blended care from January 2025 #10636's merge had made stale. It found that the merge of main leaves this PR's own changes as they were.
• Earlier reviews. Both cloud reviews in this thread asked for changes. The first couldn't reach TWC's site and rested its request on one error, in the test's docstring, which we fixed by 571a8e8841. The second, at 4f8545d76e, matched every rate cell to TWC's PDFs, and the Opus agent found each of its findings resolved at c43a3f37e3.
• Passed. On the merged tree, test_source_conflicts.py, test_tx_ccs_rates_table.py and Correct Texas CCS expanded-age rates and pay relative and blended care from January 2025 #10636's test_tx_ccs_expanded_rates_table.py pass 136 tests. CI ran the Texas CCS YAML folder on this head with Correct Texas CCS expanded-age rates and pay relative and blended care from January 2025 #10636's code, and all 141 cases passed, including the 51 in tx_ccs.yaml and tx_ccs_payment_rate.yaml.
• Pending. At 10:02Z, 28 checks had passed on this head and none had failed. Microsimulation and eight Contrib shards were still running.
• Merge. Squash it with the subject "Add Texas CCS Board Contract Year 2024 provider payment rates (Add Texas CCS Board Contract Year 2024 provider payment rates #10598)" and the body in ~/reviews/tx-ccs-bcy24/drafts/squash_body_v2.md. GitHub's default body would repeat claims from earlier commit messages that we corrected in this thread. No hold, decision or other PR blocks the merge. Drafts Add Texas CCS provider payment rates for October 2016 to September 2023 #10622 and Apply Texas Workforce Board parent share of cost scales before TX3C #10623 build on this branch, and GitHub retargets them to main when it deletes the branch.
• Follow-ups. Two, outside this PR:
  • On main, the YAML loader cuts four us: notes in source_conflicts/candidates.yaml at " Follow the statute where IRS forms differ: refundable CTC housing deduction, joint filing test, EITC trust royalties, former 32(c)(1)(F) #10297", since it reads " #" as the start of a comment.
  • Texas CCS reads hhs_smi, which PolicyEngine computes for a whole calendar year from each state's figure on January 1. So from October to December 2024, PolicyEngine applies HHS's $94,466 four-person figure for Texas for the year from October 1, 2023, not the $103,213 for the year from October 1, 2024.

[MaxGhenis]

US + core hub merge audit, us#10598 at da0af4eaeda12ba19ec3876f30cd78d87901df41 (squash): Add Texas Child Care Services payment rates for Board Contract Year 2024, correct four Dallas County rates for 2025, and date the statewide parent share of cost scale from January 2025.

• Review: a cloud review at 4f8545d76e checked every BCY 2024 and BCY 2025 cell against TWC's PDFs and asked for changes (review). An Opus 5.5 agent then reviewed each head through e16026e171, found each of that review's findings resolved by c43a3f37e3, and agreed with the code and the description at e16026e171 (hand-off). A GitHub branch update merged main 8afdcb55e9 into e16026e171, giving da0af4eaed. Every file's changes match.
• Source conflict: for us-tx:ccs-board-psoc-before-tx3c, we follow the reading we took for draft Apply Texas Workforce Board parent share of cost scales before TX3C #10623 on November 2: each Board's own scale where we have its chart, and TWC's statewide scale for other Boards and from TX3C's launch on January 13, 2025. Until Apply Texas Workforce Board parent share of cost scales before TX3C #10623 adds the Board scales, the model has only TWC's statewide scale. For us-tx:ccs-rate-age-groups-before-tx3c, we apply TWC's four-group tables through January 12, 2025, the only rates TWC published for those months. Under Government Code §2308.315 and S.B. 1555, each Board had until December 1, 2023 to set rates that align TWC's age groupings with HHSC's child-to-caregiver ratios, and the Legislature set no amounts. TWC kept rates on the four age ranges until TX3C launched (WD 19-24, Change 1, p. 3).
• Gates the hub checked live at merge:
  • gh pr checks exits 0;
  • the latest Pull request run at the head concluded success;
  • MERGEABLE;
  • not a draft;
  • no reviewer's latest review requests changes;
  • the semantic-overlap guard;
  • head pinned.
• Impact: no output moves in 2026 or 2030. In each month of both years, every value in the four parameter files this PR changes matches main's, except four Dallas County cells in rates.yaml. tx_ccs_payment_rate, the only variable that reads that file, switches to the eight-age-group tables from January 2025, and this PR leaves those tables alone.
• Axiom: us-tx:policies/ccdf/rate-schedules/board-max-provider-payment-rates-bcy2024, us-tx:policies/ccdf/rate-schedules/board-max-provider-payment-rates-bcy2025 queued (TheAxiomFoundation/rulespec-us#1990)

🤖 Generated with Claude Code