Add UT CCAP Table 4 copays and income limits effective October 1, 2026

[hua7450]

Summary

Adds Utah's already-published next CCAP copay/income schedule: the DWS Table 4 "Child Care Income Eligibility and Co-Payment" effective October 1, 2026 (table1026.pdf), as a 2026-10-01 layer in ut/dwf/ccap/copay/. Copays rise ~3–4% (e.g., household of 2: group 3 $42 → $43, group 5 $109 → $113, group 16 $352 → $366; household of 4 group 16 $518/$538/$558 → $539/$559/$580 by child count) and every income-group bound is uprated (household of 2 group 16: $5,668 → $5,896; 12+: $12,502 → $13,006).

Found by the #9365 copay sweep. The existing 2020-01-01 layers (the April 2026 edition at the backdating floor, per the file comments) are untouched.

Changes

File	Change
amount.yaml	528 cells (11 household sizes × 16 income groups × 3 child-count columns) at 2026-10-01
income_limits.yaml	176 group upper bounds at 2026-10-01
Tests	5 new ut_ccap_copay cases and 2 new ut_ccap_income_group cases at period 2027-01 (inside the new schedule): an uprated interior cell, both sides of the raised group-2 boundary (inclusive bound $2,140 → $18 vs. $2,141 → $43), the above-group-16 clamp ($366), and a size-4 group-16 three-child cell ($580)

Verification: transcription ran through a generation script with built-in assertions (all nine sweep-identified sample deltas, group-1 zeros, within-column monotonicity, contiguous bounds), then 22 cells re-checked independently against the PDF, plus a further independent 15-cell check of the household-size-7 block against the source extract.

Transcription notes (documented in file comments)

• This edition prints 0 (not "n/a") in impossible child-count cells (e.g., household of 2, "2 children" column); the established repeat-padding convention for TWO/THREE_PLUS is retained so those cells stay non-binding, consistent with the April layer.
• Household 10 group 16 upper bound prints $12,846, breaking size-monotonicity (HH9 $12,225, HH11 $12,746, HH12+ $13,006; the uniform step suggests $12,486 was intended). Encoded as published and flagged as an apparent typo — behaviorally inert since the group-16 bound only separates group 15 from 16.
• Household 6 group 16 prints its range without the usual .01 offset; only upper bounds are encoded, so no effect.

Out of scope (deliberately)

The October 2026 table introduces a two-tier eligibility structure — Entry Limit at income Group 6 (applications), Exit Limit at Group 10 (reviews), Open Child Care at Group 16 — versus the single 85% SMI ceiling currently in ut_ccap_income_eligible. That is an eligibility-variable restructure left for a follow-up; this PR only notes it in the income_limits.yaml comment.

Part of #9365.

Test plan

• CI passes (tests intentionally not run locally)

🤖 Generated with Claude Code

https://claude.ai/code/session_01ASXyC4DNpinDieKCdkNgL8

[DTrim99]

PR #9367 Review — Utah CCAP Table 4 (copays + income limits, effective 2026-10-01)

Program Review

PR #9367 adds the October 1, 2026 edition of Utah DWS Table 4 (Child Care Income
Eligibility and Co-Payment) to the Utah Child Care Assistance Program (CCAP) parameters.
It is a pure parameter/table update — no .py files change. A second dated layer
(2026-10-01) is added on top of the existing layer (2020-01-01, holding the April 2026
edition at the backdating floor) in two files:

• policyengine_us/parameters/gov/states/ut/dwf/ccap/copay/amount.yaml — copay amounts by
  household size (2–12) × income group (1–16) × child-count group (ONE / TWO / THREE_PLUS).
• policyengine_us/parameters/gov/states/ut/dwf/ccap/copay/income_limits.yaml — inclusive
  upper income bound by household size (2–12) × income group (1–16).

The addition is strictly additive: every prior 2020-01-01 value is untouched (diff shows
only + insertions of new-date lines, no modification/deletion of any prior-edition line).
The existing formulas (ut_ccap_copay, ut_ccap_income_group) read the breakdown generically,
so the new dated layer is picked up automatically for periods on/after 2026-10-01 with no code
change. Three test YAMLs are updated (ut_ccap_copay, ut_ccap_income_group,
ut_child_care_subsidies) with cases pinned to period: 2027-01 (which resolves the new layer)
plus a mid-year year-blend integration case.

Source Documents

• New (October 2026): Utah DWS Table 4 — Child Care Income Eligibility and Co-Payment,
  effective October 1, 2026 — https://jobs.utah.gov/occ/provider/table1026.pdf#page=1
  (HTTP 200, valid PDF, date-stamped durable filename). Cited in both files.
• Retained (April 2026): Utah DWS Table 4, effective April 1, 2026 (Infosource .htm).
  Covers the 2020-01-01 layer. Retained; see Should Address for a durability note.
• Regulatory: Utah Admin. Code R986-700-707(1) (copay) and R986-700-710(6)(d) (income
  limits), retained in the respective files as authority backup.

Critical

None.

Should Address

1. New-block test coverage has real gaps (tests). The new-block cases correctly pin
   representative cells across sizes 2–4, groups 2/3/4/16, and all three child-count columns at
   period: 2027-01, plus a true at-bound/above-bound income-group pair (size 2: 2,140 / 2,141),
   an above-ceiling clamp, and a mid-year year-blend integration case (9×791 + 3×826 = 9,597).
   Gaps that would pass CI silently if the corresponding new-block cells regressed:

   • No 2027-01 analogue of the old Case 11 for the size-2 (>2 children) n/a-column padded cell.
     The PR's own comment describes the October edition's changed rendering of these cells, yet
     no new-block test confirms the padded value (18, not 0) under the October layer.
   • No large-household new-block case (e.g. size 12, group 16 → 870) to pin the far corner of
     the new table; large-size values are asserted only on the old block.
   • Only one household size (size 2) gets a new-block income-group threshold edge; a second size
     (e.g. size 4 at the 8,288 / 8,289 grp-15/16 edge) is not exercised on the new limits.
     No incorrect expected values were found — every asserted new-block number matches the parameter
     diff. These are additive test requests, not corrections.

2. Retained April-2026 citation is not edition-durable (references). The retained April block
   cites the Infosource "current Table 4" .htm page, which Utah overwrites in place each time
   Table 4 is reissued; once DWS publishes the next edition this link will no longer show the
   April 2026 values. The new October citation, by contrast, uses a date-stamped PDF
   (table1026.pdf) and is durable. Suggest replacing the .htm with a date-stamped April PDF if
   one exists (e.g. a table0426-style filename). Non-blocking: the April block is corroborated by
   the R986-700 regulatory citation and internal documentation.

3. Oct-2026 entry/exit eligibility thresholds not modeled (regulatory — documented follow-up).
   The October 2026 edition introduces application/review/interim eligibility gates
   (Entry = Group 6, Exit = Group 10, Open/Special-Needs = Group 16). The income_limits.yaml
   comment explicitly discloses that this structure "is not yet modeled in
   ut_ccap_income_eligible," which applies a single 85%-SMI ceiling. This gap is pre-existing
   (the April layer also lacked it), orthogonal to this PR's stated scope (loading the new
   copay/income-limit numbers), and does not touch the new block (ut_ccap_income_eligible uses
   hhs_smi, not these table bounds). Recommend a tracking issue for follow-up; do not block
   this PR on it.

Suggestions

1. Tighten the n/a-cell header comment (code / copay audit). The amount.yaml header comment
   reads that the impossible child-count columns are what "the October 2026 edition prints as 0
   instead; in both layers these cells repeat the largest published column." The two clauses read
   as contradictory. The encoding is correct — impossible cells repeat the largest published
   column (e.g. size 2 group 6 = 126/126/126), never 0 — so the wording should be tightened to
   avoid implying the values are stored as 0. Documentation only; no behavioral impact. (This is
   also the minor cross-report doc inconsistency: one source described these cells as "printing 0,"
   the copay audit found the repo "repeats the top column" — both describe the same intentional
   convention.)

2. Optionally mirror the PDF's literal 0 in the n/a cells (copay audit). For strict
   source-fidelity, the 45 structurally-unreachable n/a cells (size 2 with TWO/THREE_PLUS; size 3
   with THREE_PLUS) could be encoded as 0 to match what the October PDF literally prints. The repo
   instead repeats the largest published column so the three-key lookup is always defined and
   degrades safely. These cells can never bind (a 2-person household cannot have 2+ children in
   care; a 3-person household cannot have >2), so this is a convention choice, not a defect.

3. Optionally pin the size-10 group-16 anomaly with a test (code / income audit). The
   published size-10 group-16 upper bound is 12,846, which breaks cross-size monotonicity (size 9
   = 12,225; size 11 = 12,746; size 12 = 13,006) and is likely a Utah DWS typo for 12,486. The PR
   faithfully encodes the published 12,846 and documents it. Since the group-16 bound only
   separates group 15 from group 16 (group 16 being the maximum band), there is no behavioral
   effect. A one-line YAML test pinning size-10 group-15/16 classification would guard the
   intentional discrepancy against a future well-meaning "fix."

Value Audit Summary

Both tables were audited cell-by-cell against the October 1, 2026 PDF (table1026.pdf).

Table	Scope audited	Match	Mismatch	Notes
Income limits (income_limits.yaml)	176 size×group upper bounds	176 / 176	0	100% match. size-10 grp-16 = 12,846 is published as-is by Utah (likely typo for 12,486); faithfully encoded; no behavioral effect (grp-16 = 85%-SMI band).
Copay amounts (amount.yaml)	483 published copay cells (size×group×child-count)	483 / 483	0	Every published value transcribed exactly.
Copay — n/a cells	45 structurally-unreachable cells	n/a	0 binding	PDF prints literal 0; repo repeats the largest published column so the lookup stays defined. Never selectable, never binds. Suggestion only.

Zero value errors. The only source–repo divergences are (a) the deliberate n/a-column repeat
convention (45 unreachable cells) and (b) one faithfully-mirrored Utah source anomaly (12,846),
both documented and behaviorally inert.

Validation Summary

• CI: 31/31 green.
• Value audit: income limits 176/176 exact; copay 483/483 published cells exact; 0 mismatches.
• Structure: breakdowns preserved (range(2,13) × range(1,17) × child-count); additive dated
  layer; prior edition untouched.
• Effective-date layering: 2026-10-01 correct; resolved by period: 2027-01 tests and by
  mid-year period arithmetic in the year-blend case.
• References: both files cite the correct October 2026 Table 4 (durable, date-stamped) plus
  admin-code authority; jurisdiction (Utah) correct.
• Changelog: top-level fragment changelog.d/ut-ccap-copay-table4-oct-2026.changed.md; .changed
  is the correct type (patch bump) for a rate/value update.
• Code patterns: 0 critical / 0 should / 2 suggestions; no hard-coded logic, no .py changes,
  formatting clean.

Review Severity

APPROVE. All values are verified against the authoritative source (176/176 income limits,
483/483 published copays, 0 mismatches). There are no criticals and no should-address items that
affect any household outcome. The should-address items are additive test coverage, a citation-
durability improvement on the retained (pre-existing) April block, and a documented, pre-existing
eligibility-modeling follow-up — none of which block this parameter/table update.

Next Steps

Run /fix-pr 9367 to optionally apply the non-blocking improvements: add the three requested
new-block test cases (n/a padded cell, large-household corner, second-size boundary), tighten the
n/a-cell header comment, and (optionally) swap the retained April .htm citation for a date-stamped
PDF. Consider filing a tracking issue for the Group 6 entry / Group 10 exit eligibility thresholds
in ut_ccap_income_eligible.

🤖 Generated with Claude Code /review-program

[DTrim99]

Fix minor nits, then ready

[hua7450]

Addressed the review in 20bbfe9:

Should address

1. New-block test coverage — added four October-layer cases: the Case 11 analogue pinning the padded n/a cell at 2027-01 (ut_ccap_copay Case 17: size 2, forced TWO column → 18); the far-corner large-household case (Case 18: size 12, group 16, 3+ children → 870); and a second size's boundary pair on the new limits (ut_ccap_income_group Cases 15–16: size 4 at 8,288 → group 15 / 8,289 → group 16).
2. April citation durability — Utah publishes no date-stamped April PDF (probed table0426-style names on jobs.utah.gov), so I archived the live Infosource page to the Wayback Machine and verified the capture shows the April edition ("effective April 1, 2026", size-2 group-2 bound 2,082). Both parameter files now cite the archived capture instead of the overwritable live page.
3. Entry/exit eligibility tiers — tracking issue filed: UT CCAP: model the October 2026 entry/exit income eligibility tiers (Table 4 groups 6/10/16) #9372.

Suggestions

1. n/a-cell header comment — reworded in amount.yaml: the two editions' renderings (n/a vs 0) and the repeat-largest-column encoding are now stated as separate facts, with an explicit "neither the n/a nor the 0 is stored."
2. Literal 0 in n/a cells — declined, keeping the repeat convention: it keeps the three-key lookup defined and degrading safely, and the cells can never bind (now pinned on both layers by Cases 11 and 17).
3. Size-10 group-15/16 guard — added as ut_ccap_income_group Cases 17–18 (11,935 → group 15, 11,936 → group 16) with a comment noting the as-published 12,846 bound; since income above any group-16 bound clamps to group 16, classification is identical under either reading of the bound, so the boundary pair plus the documentation is the strongest guard available.