Skip to content

Correct California foster parent payments and apply them statewide - #10836

Merged
MaxGhenis merged 10 commits into
mainfrom
ca-foster-payments-statewide
Nov 17, 2026
Merged

MaxGhenis merged 10 commits into
mainfrom
ca-foster-payments-statewide

Conversation

@MaxGhenis

@MaxGhenis MaxGhenis commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor

We apply California's infant supplement and expectant parent payment in every California county. We also correct the parent test, child count, pregnancy window, historical amounts and July 2026 rates.

For the infant supplement, we count each eligible foster parent's own children and apply that parent's placement rate. A foster youth with no own children receives $0, and unrelated household members cannot increase the child count or placement rate. The California Department of Social Services (CDSS) describes the parent-child requirement in ACL 21-123, PDF pages 2–3.

For the expectant parent payment, we calculate monthly entitlements for pregnancy months 7–9 and return $0 before January 2022. CDSS starts payments in January 2022 and begins eligibility in pregnancy month 7 under ACL 21-123, PDF pages 2–4.

CDSS publishes the infant supplement amounts in ACL 15-58, page 6, ACL 16-57, page 6 and ACL 26-44, PDF page 5. From January 2022, CDSS directs counties to pay the expectant payment at the home-based amount in every placement under ACL 21-123, PDF page 3. From July 2026, CDSS lists both payments in one table in ACL 26-44. It lists $932 for foster family homes, foster family agencies, transitional housing placements for nonminor dependents (THP-NMD), whole family foster homes, resource families and supervised independent living placements, and $1,429 for short-term residential therapeutic programs (STRTPs). We apply the residential amount when a caller sets is_in_foster_care_group_home to true. We define that input as true for a foster youth placed in a qualifying group home or an STRTP, and as false for a foster youth in one of the listed base-rate placements. We use these monthly amounts:

Period Infant base Infant residential Expectant base Expectant residential
January 2015–June 2016 $411 $890 $0 $0
July 2016–December 2021 $900 $1,379 $0 $0
January 2022–June 2026 $900 $1,379 $900 $900
From July 2026 $932 $1,429 $932 $1,429

Invariants

We check these invariants with regression and property tests:

  • We pay equivalent households the same amounts in all 58 California counties. We pay $0 outside California, and the tests check Nevada.
  • Foster youth with no own children receive no infant supplement. Each eligible parent's own child count determines their supplement, and unrelated children do not increase it.
  • We use each eligible parent's placement to determine their infant rate. Unrelated household members do not change that rate.
  • We pay the expectant payment for qualifying foster care and pregnancy months 7–9. We pay $0 when callers omit the pregnancy month.
  • We pay the same expectant amount in every placement through June 2026. From July 2026, we pay the residential amount in a group home or short-term residential therapeutic program.
  • We pay no expectant payment before January 2022 and apply each rate from its effective date. Cases in the month before and the month of a change pin each home-based step and the July 2026 group home expectant step. For the other three group home steps, cases fall before and after the change but not in both adjoining months.

Methodology choices

  • We retain the four public ca_la_* variable names and the registry IDs la_infant_supplement and la_expectant_parent for existing callers.
  • We move parameter prefixes from gov.local.ca.la.dss.* to gov.states.ca.cdss.*. Saved reforms must use the new paths. We split the expectant payment amount into amount.base and amount.group_home.
  • We retain the input name is_in_foster_care_group_home and its default of false. We label it for group homes and STRTPs. In its documentation, we name both sides of the placement split and cite ACL 15-58, ACL 16-57 and ACL 26-44.
  • We use the annual own_children_in_household input for each parent's child count. The input has no formula, so a caller who omits it receives no infant supplement. We cannot identify adult children or children reported by both parents from that count.
  • We retain is_in_foster_care and mark both registry entries partial. This boolean cannot distinguish every eligible placement and funding category. CDSS describes the expectant-payment exclusions in CFL 21/22-74, PDF page 3.
  • We pay the $900 expectant payment in every placement from January 2022 through June 2026, as CDSS directs in ACL 21-123. Under WIC 11465(e)(1), a pregnant youth's rate includes the amount the section would pay for the child "if born". Under subdivision (d)(2), that amount equals the group home infant supplement rate in a group home and the STRTP rate in an STRTP. Under that reading, a pregnant youth in a group home would receive $1,379 a month before July 2026, $479 more than we pay. Under subdivision (e)(3), counties pay a lump sum before automation of three times the subdivision (a) amount. The entry us-ca:expectant-parent-payment-residential-rate in source_conflicts/candidates.yaml records this question as open, and nobody has ruled on it.
  • From July 2026, we pay $1,429 for both payments in group homes and STRTPs. CDSS lists that amount for STRTPs in ACL 26-44 and lists no amount for group homes. We apply it to group homes under our interpretation of subdivisions (d)(2) and (e)(1): the statute names a group home rate and an STRTP rate, and we treat the two as equal. CDSS listed one $1,379 infant supplement rate for group homes and STRTPs in ACL 18-89, PDF page 7 and ACL 19-70, PDF page 7. The same candidate entry notes this choice for the expectant payment, and nobody has ruled on it.
  • We calculate monthly entitlements. We do not calculate lump-sum issuance or catch-up payments after late enrollment.

Validation

Pull request run 43986871229 at 1710a00ced passed all 37 jobs on GitHub's merge with main f8fce90f1b.

We use Python to compare each generated household across California counties and Nevada. Hypothesis generates four examples for 2024; each example also includes explicit boundary cases and combines all counties in one simulation. The existing Rest Python core group runs this file. Locally at de714c68 with Core 3.36.0, the invariant test took 66.68 seconds, including setup, and peaked at 1,064.80 MiB. The file has not changed since that commit. We did not measure its peak memory on Linux.

51 YAML cases pass in four files: 13 expectant payment, 11 expectant eligibility, 23 infant supplement and 4 infant eligibility cases. The invariant test passes, and test_programs_registry.py passes 912 tests and skips nine.

We pin the placement split with three group home cases in ca_la_expectant_parent_payment.yaml: $0 in December 2021, $900 × 1 month = $900 in June 2026 and $1,429 × 1 month = $1,429 in July 2026. At 72607617 with Core 3.33.2, all ten targeted mutants triggered regression assertion failures. No mutant has run against the group home expectant split, which we added after that commit.

Ruff formatting and lint pass, and our changes pass whitespace checks against main.

Microsimulation impact

We estimate that this branch changes none of the outputs below in 2026 on the default dataset. We pay both payments only to a foster youth under 21, and the dataset holds no is_in_foster_care column. That input has no formula and defaults to false, so no record qualifies, and both payments total $0 on main and on this branch. When a caller supplies the foster care inputs, this branch pays both payments in every California county and starts the expectant parent payment in pregnancy month 7. From July 2026, it raises each monthly amount to $932, or to $1,429 in a group home or STRTP. We compared main at fd0a748698 with this branch at 1710a00ced.

Output Main total Change Records changed (weighted)
ca_la_infant_supplement $0 $0 0 (0)
ca_la_expectant_parent_payment $0 $0 0 (0)
household_state_benefits $8.57bn $0 0 (0)
spm_unit_benefits $2,056.6bn $0 0 (0)
state_income_tax $576.6bn $0 0 (0)
income_tax $2,594.9bn $0 0 (0)
household_benefits $2,060.1bn $0 0 (0)
household_net_income $17,246.4bn $0 0 (0)

We estimate that this branch changes none of the outputs below in 2027 on the default dataset. Both payments again total $0 on main and on this branch. We compared the same two commits.

Output Main total Change Records changed (weighted)
ca_la_infant_supplement $0 $0 0 (0)
ca_la_expectant_parent_payment $0 $0 0 (0)
household_state_benefits $8.19bn $0 0 (0)
spm_unit_benefits $2,172.4bn $0 0 (0)
state_income_tax $597.2bn $0 0 (0)
income_tax $2,692.5bn $0 0 (0)
household_benefits $2,175.6bn $0 0 (0)
household_net_income $17,898.7bn $0 0 (0)

axiom: TheAxiomFoundation/rulespec-us#2297 queued (Welfare and Institutions Code § 11465 and CDSS ACL 15-58, 16-57, 21-123 and 26-44; the pinned corpus release lacks § 11465 and the corpus lacks the letters, so the signed encoder can't run yet)

🤖 Generated with Claude Code

MaxGhenis and others added 3 commits November 12, 2026 00:57
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MaxGhenis and others added 3 commits November 15, 2026 01:34
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Preserve every source-conflict candidate and add the open California rate question.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@MaxGhenis MaxGhenis changed the title Apply California foster parent payments statewide Correct California foster parent payments and apply them statewide Nov 15, 2026
MaxGhenis and others added 2 commits November 15, 2026 22:51
Split the expectant parent payment amount into base and group home
parameters. ACL 26-44 lists $932 for home-based placements and $1,429
for short-term residential therapeutic programs from July 2026, and
Welfare and Institutions Code 11465(e)(1) ties the payment to the
infant supplement a placement would pay. Before July 2026 every
placement keeps the $900 that ACL 21-123 sets. Add the statute to the
open source-conflict candidate and pin group home cases for December
2021, June 2026 and July 2026.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@MaxGhenis
MaxGhenis marked this pull request as ready for review November 16, 2026 03:53
MaxGhenis and others added 2 commits November 16, 2026 10:46
…erapeutic programs

`is_in_foster_care_group_home` selects the residential infant supplement and
expectant parent payment amounts, and its label named only group homes. Label
and document it as true for a foster youth placed in a qualifying group home or
a short-term residential therapeutic program, and false for the placements
CDSS lists at the base amount. Cite ACL 15-58, ACL 16-57 and ACL 26-44.

Say in the open residential-rate candidate that ACL 26-44 lists $1,429 for
short-term residential therapeutic programs and no amount for group homes.

No formula, parameter or test expectation changes.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
@MaxGhenis

Copy link
Copy Markdown
Contributor Author

US + core hub merge audit, us#10836 at 1710a00ced4592119700a294471ccddffda318a6 (squash): Correct California foster parent payments and apply them statewide.

  • Reviews:
    • The PR has no GitHub reviews. Three independent review rounds ran outside GitHub, each followed by a fix round.
    • The first round, at c766e2412c, read ACL 21-123, ACL 16-57, ACL 26-44 and County Fiscal Letter 21/22-74. It ran 12 household cases on the head and on the merge base and checked the Fresno and San Diego amounts by hand. It ran the 26 YAML cases then in the four files, the property test and a mutant that moved the expectant payment's first month to 7.
    • It found two P1s. Any foster youth under 21 triggered the infant supplement, parent or not, at $900 for each household member under 18. The expectant payment started in pregnancy month 6, where CDSS starts it in month 7 in ACL 21-123, and the new tests asserted both results. It also found that the amounts ignored the July 2026 increase and that the statewide formula paid the expectant payment before January 2022. The infant amounts carried no pre-July 2016 values, and the description gave no impact estimate.
    • The fix round that followed, at de714c686f, required own children in the household, counted each parent's children at that parent's placement rate and moved the first month to 7. It set $0 before January 2022, added the $411 and $890 amounts and the July 2026 amounts, and recorded the residential expectant rate as an open source-conflict candidate.
    • The second round, at de714c686f, found no P1. It recomputed all 48 YAML cases by hand and checked the rate arithmetic: $411 + $489 = $900, $890 + $489 = $1,379, and each July 2016 amount × 1.0359. It read the records of ten mutants, each of which failed an assertion. It found one P2 in the model: the expectant payment paid $932 in every placement from July 2026, where CDSS lists $1,429 for short-term residential therapeutic programs (STRTPs) in ACL 26-44. It also asked for a passing CI run at the head.
    • The next fix round, at df72e7a15b, split the expectant amount into amount.base and amount.group_home, paid $1,429 in a group home or STRTP from July 2026 and added three group home cases. CI run 43800875356 passed all 38 jobs at that head.
    • The third round covered the change since the second, at df72e7a15b, and found no P0 or P1. It fetched ACL 15-58, ACL 16-57, ACL 21-123 and ACL 26-44, hand-computed the three new cases and confirmed that the 48 earlier expectations stood. It checked that the merge of main kept all 111 registry entries and all 483 source-conflict candidates, and it read the CI logs for the four YAML files, the property test and the registry test. It found one P2: the label and documentation of is_in_foster_care_group_home named group homes alone, although the residential amounts also cover STRTPs. It listed two corrections to the description.
    • The last fix round, at 1710a00ced, broadened that input's label, documented which placements set it true or false, and cited ACL 15-58, ACL 16-57 and ACL 26-44. It corrected one sentence of the candidate's note and added one changelog sentence. It changed no formula, parameter or test.
    • No review round followed that fix. A separate check at 1710a00ced read the fix against the letters and found nothing to fix.
    • The reviews listed corrections to the description, and I made each before merge. The description now names the placements in each row of the ACL 26-44 table, says which value of the placement input selects the residential amount and calls the group home amount an interpretation. It reports test counts at the head and names the tests that have not run.
  • Methodology choices for maintainers:
    • CDSS lists $1,429 for STRTPs in ACL 26-44 and has no group home row. The model pays $1,429 for a youth in a group home from July 2026, for the infant supplement and the expectant payment. The description calls that an interpretation: Welfare and Institutions Code section 11465(d)(2) names a group home rate and an STRTP rate, and the model treats the two as equal. CDSS listed one $1,379 infant supplement rate for group homes and STRTPs in ACL 18-89 and ACL 19-70. The open entry us-ca:expectant-parent-payment-residential-rate in source_conflicts/candidates.yaml records the choice for the expectant payment, and nobody has ruled on it.
    • The same entry holds a second open question. The model pays the $900 expectant amount in every placement from January 2022 through June 2026, as CDSS directs in ACL 21-123. Under section 11465(e)(1) read with (d)(2), a pregnant youth in a group home or STRTP would receive $1,379, $479 more a month. A ruling for that reading would move the San Diego January 2024 case, the June 2026 group home case and the property test's reference amount from $900 to $1,379.
    • One boolean, is_in_foster_care_group_home, selects the residential amount for both payments. It holds true for a group home or an STRTP and can't tell them apart.
    • The infant supplement reads own_children_in_household, a yearly input with no formula. A caller who leaves it out now receives $0, where main counted every household member under 18. The count can't exclude an adult child or a child both parents report.
    • is_in_foster_care can't tell AFDC-FC, Approved Relative Caregiver and Kin-GAP placements apart, although CDSS excludes Kin-GAP from the expectant payment in County Fiscal Letter 21/22-74. Both registry entries read partial.
    • The model pays monthly entitlements from January 2022. It leaves out the $2,700 lump sum that counties issue before automation and the back payments for a pregnancy found after month 7.
    • The four ca_la_* variable names and both registry ids stay. Parameters move from gov.local.ca.la.dss.* to gov.states.ca.cdss.*, so a saved reform on an old path stops working.
    • No review or fix round read section 11465 on leginfo.legislature.ca.gov, which answers scripted requests with a bot check. The statute sentences in the description and in the candidate rest on an Internet Archive copy of that page from June 18, 2026.
  • Gates checked live at merge:
    • gh pr checks exits 0;
    • the latest Pull request run at the head, 43986871229, passed and tested this head merged into main at f8fce90f1b;
    • MERGEABLE;
    • not a draft;
    • no reviewer's latest review requests changes;
    • the semantic-overlap guard: since f8fce90f1b, main has changed no area this PR touches. That base includes Add premium tax credit rules for 2014–2017 #10846, which changed policyengine_us/programs.yaml and policyengine_us/tests/test_programs_registry.py, two files this PR also changes, so the latest run tested this head with Add premium tax credit rules for 2014–2017 #10846's changes to both files. Between the PR's merge base fd0a748698 and main at 8d8327bf88, main changed 179 files in 19 commits, and this PR touches three of those files. In programs.yaml, Add premium tax credit rules for 2014–2017 #10846 added a notes: line to the aca_subsidies entry after line 436, and this PR moves the two California entries, adding 20 lines after the merge base's line 2,269 and removing its lines 2,583 to 2,602. In test_programs_registry.py, Add premium tax credit rules for 2014–2017 #10846 added a test of the aca_subsidies entry after line 148, and this PR deletes line 33, the la_expectant_parent entry in KNOWN_UNRESOLVED. In source_conflicts/candidates.yaml, main edited and added entries for other jurisdictions from line 8,517 on, and this PR adds its California entry at lines 14 to 43. Main also changed four files of California's clean vehicle rebate under gov/states/ca/calepa/carb/cvrp in Correct countable income for school meals #10873, and this PR touches none of them. Main changed nothing under gov/states/ca/cdss, and no line it changed names either payment, is_in_foster_care, own_children_in_household, current_pregnancy_month, household_state_benefits or spm_unit_benefits. The branch merges into main at 8d8327bf88 without conflict. The merged programs.yaml and test_programs_registry.py hold both PRs' changes, and the merged candidates.yaml holds 488 unique ids, main's 487 and this PR's one;
    • the merge pins the head.
  • Impact: we estimate that this branch changes none of the eight measured outputs in 2026 or 2027 on the default dataset, which holds no is_in_foster_care column. That input defaults to false, so both payments total $0 on main and on this branch. Main fd0a748698 against 1710a00ced.
  • Next:
    • An issue should find a CDSS source for the group home infant supplement and expectant rates after July 2026. Among the letters read in these rounds, ACL 20-78, of July 2020, lists the last group home amount, $1,379.
    • An issue should replace the placement boolean with a placement type input that names each placement the letters list, so each payment reads the placement directly.
    • Four mutants of the group home expectant split have not run against ca_la_expectant_parent_payment.yaml. By hand, three of them fail case 13 or cases 3 and 12. A $900 group home amount from 2015 fails case 11, since the model applies the month 7 to 9 test in December 2021.
    • No case pins January 2022 as the first month of the group home expectant amount. A group home case for January 2022 ($900 × 1 month = $900) would catch a start moved to February 2022, which passes all 13 cases now.
    • The model omits the $200 monthly supplement of section 11465(d)(3) for a whole family foster home with a shared responsibility plan and for a supervised independent living placement with a parenting support plan.
    • The two ACL captures sit at source_conflicts/sources/ca-acl-*.txt, where main now files captures in state folders, and the candidate's statute source has no capture.
  • Axiom: Encode California's foster care infant supplement and expectant parent payment (Welf. & Inst. Code § 11465; CDSS ACL 15-58, 16-57, 21-123 and 26-44) (PE-US #10836) TheAxiomFoundation/rulespec-us#2297 queued, a new pe-parity issue for Welfare and Institutions Code section 11465 and CDSS ACL 15-58, 16-57, 21-123 and 26-44, with companion rows from this PR's YAML. The signed encoder can't run yet: the corpus release that rulespec-us pins doesn't carry section 11465, and the corpus holds none of the letters.

🤖 Generated with Claude Code

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant