Follow the adopted California Premium Subsidy design at the 150% FPL boundary

changelog.d/statute-first-ca-premium-subsidy.fixed.md

@@ -0,0 +1 @@
+Select California Premium Subsidy bands from exact nonnegative ACA MAGI divided by prior-year FPL under the adopted [2026 Program Design, (d)(4)](https://board.coveredca.com/meetings/2025/July%2028,%202025/CoveredCA_2026_Premium_Subsidy_Program_Design_Final.pdf#page=2): exactly 150% changes from a 3.19% contribution to 0%, while income above 150% enters the 3.19% band. Retain the truncated fraction for the rate within that band, following [FTB 3849 Table 1-2 (2020)](https://www.ftb.ca.gov/forms/2020/2020-3849-instructions.pdf#page=6) and [2021](https://www.ftb.ca.gov/forms/2021/2021-3849-instructions.pdf#page=6); [2025 Form 8962 Worksheet 2 line 4](https://www.irs.gov/pub/irs-prior/i8962--2025.pdf#page=9) likewise tests "more than" on the exact amount before truncating.

policyengine_us/parameters/gov/states/ca/hbex/premium_subsidy/applicable_percentage/threshold.yaml

@@ -1,9 +1,14 @@
 description: California sets the federal poverty line thresholds for the applicable percentage brackets under the California Premium Subsidy program.
 values:
-  # 150% is untiered in the Program Design prose but sits in the 3.19% band:
-  # the Explainer labels the band "150% - At or Below 165%", and under
-  # 26 CFR 1.36B-3(g)(2) FPL-percentage truncation to 2 decimals collapses
-  # [150%, 151%) to 1.50, so 150% takes the 150-165% bracket rate.
+  # The adopted Program Design runs its bands "100% up to 150%" (0%) and
+  # "Above 150% up to 165%", so an exact ratio equal to a threshold takes the
+  # lower band: 150% pays 0%. The Policy Explainer labels the upper band
+  # "150% - At or Below 165%"; the model follows the adopted design.
+  # Band selection uses exact nonnegative ACA MAGI divided by prior-year FPG;
+  # only the rate lookup uses truncated aca_magi_fraction. The Designs leave
+  # rounding unspecified. FTB 3849 Table 1-2 (2020/2021) tests exact income
+  # for "above" boundaries before truncating for lookup; Form 8962 Worksheet 2
+  # line 4 likewise checks "more than" before truncation.
   2026-01-01:
     - 1
     - 1.5

policyengine_us/tests/code_health/test_source_conflicts.py

@@ -152,6 +152,7 @@
     "policyengine_us/variables/gov/local/ca/cc/ga/eligibility/ca_cc_general_assistance_eligible.py": 1,
     "policyengine_us/variables/gov/local/tax/sales/local_sales_tax.py": 1,
     "policyengine_us/variables/gov/states/ar/ade/oec/sra/rates/ar_sra_daily_copay.py": 1,
+    "policyengine_us/variables/gov/states/ca/hbex/premium_subsidy/ca_premium_subsidy_applicable_percentage.py": 1,
     "policyengine_us/variables/gov/states/ct/tax/income/rebate/ct_child_tax_rebate.py": 1,
     "policyengine_us/variables/gov/states/fl/commerce/liheap/eligibility/fl_liheap_eligible.py": 1,
     "policyengine_us/variables/gov/states/fl/commerce/liheap/eligibility/fl_liheap_income_limit.py": 1,

policyengine_us/tests/policy/baseline/gov/states/ca/hbex/premium_subsidy/ca_premium_subsidy.yaml

@@ -9,6 +9,9 @@
 # figure (spread across 12 months and re-summed by add()), matching the NM
 # convention. aca_magi, aca_magi_fraction, slcsp, aca_ptc and is_aca_ptc_eligible
 # are supplied directly so each expectation is exactly hand-verifiable.
+# Prior-year tax_unit_fpg is pinned so supplied MAGI and FPL fractions agree.
+# The negative-MAGI clamp case supplies eligibility and a positive rate to
+# isolate the amount formula; its income fraction is correctly zero.
 #   CA schedule: threshold [1.0, 1.5, 1.65], initial [0, 0.0319], final [0, 0.0391]
 
 # ---- 100-150% FPL band (applicable percentage 0%) ----
@@ -27,6 +30,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 21_000
         aca_magi_fraction: 1.40
         slcsp: 6_000
@@ -50,6 +55,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 21_000
         aca_magi_fraction: 1.40
         slcsp: 6_000
@@ -77,6 +84,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_873.015873015873
         aca_magi: 25_000
         aca_magi_fraction: 1.575
         slcsp: 6_000
@@ -102,6 +111,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_757.575757575758
         aca_magi: 26_000
         aca_magi_fraction: 1.65
         slcsp: 7_000
@@ -127,6 +138,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 24_000
         aca_magi_fraction: 1.60
         slcsp: 3_000
@@ -141,7 +154,11 @@
 # ---- Negative-MAGI clamp ----
 
 - name: Case 6, negative ACA MAGI is clamped to zero (does not inflate subsidy).
-  # magi_frac 1.575 -> pct 0.0355; aca_magi -10_000 clamps to 0 so the -pct*income
+  # Supply eligibility and pct 0.0355 to isolate the amount clamp:
+  # exact negative MAGI would otherwise be ineligible and select the 0% band,
+  # hiding a missing amount clamp. The income fraction is zero because
+  # aca_magi_fraction floors negative MAGI at zero before dividing by FPG.
+  # aca_magi -10_000 clamps to 0 so the -pct*income
   # term is 0 rather than +355. amount = max(0, 6_000 - 2_000 - 0.0355*0) = 4_000.
   # Without the clamp it would be 6_000 - 2_000 - 0.0355*(-10_000) = 4_355.
   period: 2026
@@ -154,8 +171,12 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 10_000
         aca_magi: -10_000
-        aca_magi_fraction: 1.575
+        aca_magi_fraction: 0
+        ca_premium_subsidy_eligible: true
+        ca_premium_subsidy_applicable_percentage: 0.0355
         slcsp: 6_000
         aca_ptc: 2_000
         tax_unit_is_filer: true
@@ -180,6 +201,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_873.015873015873
         aca_magi: 25_000
         aca_magi_fraction: 1.575
         slcsp: 6_000
@@ -204,6 +227,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2024: 15_873.015873015873
         aca_magi: 25_000
         aca_magi_fraction: 1.575
         slcsp: 6_000
@@ -228,6 +253,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 16_265.060240963856
         aca_magi: 27_000
         aca_magi_fraction: 1.66
         slcsp: 6_000
@@ -252,6 +279,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 21_000
         aca_magi_fraction: 1.40
         slcsp: 0
@@ -280,6 +309,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 15_000
         aca_magi_fraction: 1.00
         slcsp: 6_000
@@ -310,6 +341,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 24_000
         aca_magi_fraction: 1.60
         slcsp: 6_000
@@ -338,6 +371,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2027: 15_873.015873015873
         aca_magi: 25_000
         aca_magi_fraction: 1.575
         slcsp: 6_000
@@ -368,6 +403,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 16_265.060240963856
         aca_magi: 27_000
         aca_magi_fraction: 1.66
         slcsp: 6_000
@@ -396,6 +433,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_000
         aca_magi: 14_850
         aca_magi_fraction: 0.99
         slcsp: 6_000
@@ -427,6 +466,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2026: 15_873.015873015873
         aca_magi: 25_000
         aca_magi_fraction: 1.575
         slcsp: 6_000
@@ -458,6 +499,8 @@
     tax_units:
       tax_unit:
         members: [person1]
+        tax_unit_fpg:
+          2025: 15_873.015873015873
         aca_magi: 25_000
         aca_magi_fraction: 1.575
         slcsp: 6_000
@@ -476,7 +519,8 @@
   # add(is_aca_ptc_eligible) > 0, which fires on any eligible member, so the tax
   # unit is eligible. The amount uses tax-unit slcsp/aca_ptc (not per-member), so
   # the asymmetry does NOT halve or zero the payout. magi_frac 1.575 -> pct 0.0355.
-  # income 30_000. amount = max(0, 8_000 - 3_000 - 0.0355*30_000) = 5_000 - 1_065 = 3_935.
+  # Pin prior-year FPG to 20_000: income 31_500 / 20_000 = 1.575.
+  # amount = max(0, 8_000 - 3_000 - 0.0355*31_500) = 5_000 - 1_118.25 = 3_881.75.
   # Confirms the add()>0 gate + defined_for + person->tax_unit aggregation produce
   # the full subsidy rather than a per-eligible-member proration.
   period: 2026
@@ -492,7 +536,9 @@
     tax_units:
       tax_unit:
         members: [person1, person2]
-        aca_magi: 30_000
+        tax_unit_fpg:
+          2025: 20_000
+        aca_magi: 31_500
         aca_magi_fraction: 1.575
         slcsp: 8_000
         aca_ptc: 3_000
@@ -503,4 +549,5 @@
         state_code: CA
   output:
     ca_premium_subsidy_eligible: true
-    ca_premium_subsidy: 3_935
+    ca_premium_subsidy_applicable_percentage: 0.0355
+    ca_premium_subsidy: 3_881.75