Date the Texas SSI supplement's PNA history from primary sources

changelog.d/tx-ssi-supplement-history.fixed.md

@@ -0,0 +1 @@
+Backfill the Texas SSI supplement's personal needs allowance at $30 from July 1988 (no supplement before September 1999), so Texas simulations for months from July 1988 through August 1999 no longer fail, and document the supplement's sources from 1999 to 2026.

policyengine_us/parameters/gov/states/tx/hhsc/ssi_state_supplement/personal_needs_allowance.yaml

@@ -1,6 +1,11 @@
 description: Texas provides this amount as the personal needs allowance for Supplemental Security Income recipients in Medicaid-funded long-term care facilities under the Optional State Supplement program.
 
 values:
+  # H-1500 and the 2002 DHS handbook give $30 before September 1999 without a
+  # start date. The modeled backfill starts July 1, 1988, when the $30 federal
+  # minimum in 42 USC 1396a(q)(2) took effect; Texas's own adoption date is
+  # not in these sources.
+  1988-07-01: 30
   1999-09-01: 45
   2001-09-01: 60
   2003-09-01: 45
@@ -16,3 +21,17 @@ metadata:
       href: https://statutes.capitol.texas.gov/Docs/HR/htm/HR.32.htm
     - title: HHSC MEPD Handbook Section H-1500 - Personal Needs Allowance
       href: https://www.hhs.texas.gov/handbooks/medicaid-elderly-people-disabilities-handbook/h-1500-personal-needs-allowance
+    - title: 1 Texas Administrative Code Section 358.381(c)
+      href: https://www.law.cornell.edu/regulations/texas/1-Tex-Admin-Code-SS-358-381
+    - title: 42 USC 1396a(q)(2) - Minimum monthly personal needs allowance
+      href: https://www.law.cornell.edu/uscode/text/42/1396a#q_2
+    - title: Texas DHS Medicaid Eligibility Handbook Revision Notice 02-1 (December 21, 2001), Personal Needs Allowance - Effective September 1, 2001
+      href: https://web.archive.org/web/20020504200936/http://www.dhs.state.tx.us:80/handbooks/archives/meh/02-1/memo.htm
+    - title: H.B. 2292, 78th Texas Legislature (2003), Section 2.207
+      href: https://capitol.texas.gov/tlodocs/78R/billtext/html/HB02292F.htm
+    - title: Texas DHS Medicaid Eligibility Handbook Item 3231, Personal Needs Allowance (Revision 03-7, effective September 1, 2003)
+      href: https://web.archive.org/web/20030910100506/http://www.dhs.state.tx.us:80/handbooks/meh/3000/3231.htm
+    - title: HHSC MEPD Handbook Section H-6000 (Revision 09-4, effective December 1, 2009)
+      href: https://web.archive.org/web/20100126020411/http://www.dads.state.tx.us/handbooks/mepd/H/H-6000.htm
+    - title: H.B. 54, 88th Texas Legislature (2023)
+      href: https://capitol.texas.gov/tlodocs/88R/billtext/html/HB00054F.htm

policyengine_us/tests/core/test_tx_ssi_state_supplement_invariants.py

@@ -0,0 +1,150 @@
+"""Income invariance of the Texas SSI supplement for Medicaid-facility residents.
+
+1 TAC 358.381(c)(2): "For a person receiving the reduced SSI federal benefit
+rate, HHSC issues a supplement to give the person a PNA at the minimum level
+set by the HHSC executive commissioner." HHSC MEPD Handbook H-6000 states it as
+a fixed amount: "HHSC supplements the reduced SSI payment standard by $45 per
+month". No Texas source subtracts the recipient's own income from it.
+
+Checked for single residents of a Medicaid facility who are paid SSI on the
+reduced standard (a positive federal payment), with generated pensions,
+earnings and ages, in June 2011 and June 2024:
+
+1. Income invariance: each such eligible resident gets exactly PNA - $30 a
+   month, whatever their income.
+2. Federal identity: federal SSI plus countable income equals the $30 reduced
+   standard.
+3. Differential: PNA - SSI - countable income, the reading of SSA's
+   description that subtracts the recipient's income, equals the supplement.
+4. Bounds: 0 <= supplement <= PNA - $30 for every resident.
+
+Residents with no federal payment (countable income at or above $30) are left
+to the eligibility tests: the sources do not settle whether Texas pays one
+whose countable income is exactly $30.
+
+Why Python: YAML checks fixed expected outputs for listed households. These
+properties compare three computed variables against each other for generated
+residents, which YAML cannot express. The dated amounts are in
+tx_ssi_state_supplement.yaml.
+"""
+
+import numpy as np
+from hypothesis import HealthCheck, given, settings
+from hypothesis import strategies as st
+
+from policyengine_us import Simulation
+
+PERIODS = ("2011-06", "2024-06")
+TOLERANCE = 0.01
+FACILITY_INPUTS = dict(
+    is_in_medicaid_facility=True,
+    ssi_lives_in_medical_treatment_facility=True,
+    ssi_medicaid_pays_majority_of_care=True,
+)
+
+
+def _situation(residents, years):
+    situation = {
+        group: {}
+        for group in (
+            "people",
+            "tax_units",
+            "marital_units",
+            "spm_units",
+            "families",
+            "households",
+        )
+    }
+    for index, (age, pension, earnings) in enumerate(residents):
+        name = f"resident{index}"
+        person = {
+            "age": {year: age for year in years},
+            "pension_income": {year: 12 * pension for year in years},
+            "employment_income": {year: 12 * earnings for year in years},
+        }
+        for variable, value in FACILITY_INPUTS.items():
+            person[variable] = {year: value for year in years}
+        situation["people"][name] = person
+        for group in ("tax_units", "marital_units", "spm_units", "families"):
+            situation[group][f"{group}_{index}"] = {"members": [name]}
+        situation["households"][f"household_{index}"] = {
+            "members": [name],
+            "state_code": {year: "TX" for year in years},
+        }
+    return situation
+
+
+def _check(residents):
+    years = sorted({int(period[:4]) for period in PERIODS})
+    simulation = Simulation(situation=_situation(residents, years))
+    paid = []
+    for period in PERIODS:
+        parameters = simulation.tax_benefit_system.parameters(period)
+        pna = (
+            parameters.gov.states.tx.hhsc.ssi_state_supplement.personal_needs_allowance
+        )
+        reduced = parameters.gov.ssa.ssi.amount.institutional.individual
+        eligible = simulation.calculate("tx_ssi_state_supplement_eligible", period)
+        supplement = simulation.calculate("tx_ssi_state_supplement", period)
+        ssi = simulation.calculate("ssi", period)
+        countable = simulation.calculate("ssi_countable_income", period)
+        recipient = eligible & (ssi > 0)
+        paid.append(recipient)
+
+        # 4. Bounds.
+        assert (supplement >= 0).all()
+        assert (supplement <= pna - reduced + TOLERANCE).all()
+        # 1. Income invariance.
+        np.testing.assert_allclose(supplement[recipient], pna - reduced, atol=TOLERANCE)
+        # 2. Federal SSI plus countable income is the reduced standard.
+        np.testing.assert_allclose(
+            (ssi + countable)[recipient], reduced, atol=TOLERANCE
+        )
+        # 3. Agreement with "allowance less federal SSI less countable income".
+        np.testing.assert_allclose(
+            supplement[recipient],
+            (pna - ssi - countable)[recipient],
+            atol=TOLERANCE,
+        )
+    return np.any(paid, axis=0)
+
+
+residents = st.tuples(
+    st.integers(min_value=65, max_value=95),
+    # Monthly pension: $0-$150, so countable unearned income spans the $30
+    # reduced standard.
+    st.integers(min_value=0, max_value=150),
+    # Monthly earnings: $0-$250.
+    st.integers(min_value=0, max_value=250),
+)
+
+# A batch's cost is mostly per-variable overhead, so each example is a large
+# batch and there are few examples.
+SETTINGS = dict(
+    max_examples=5,
+    deadline=None,
+    derandomize=True,
+    suppress_health_check=[HealthCheck.too_slow, HealthCheck.data_too_large],
+)
+
+
+@settings(**SETTINGS)
+@given(st.lists(residents, min_size=10, max_size=30))
+def test_generated_residents(population):
+    _check(population)
+
+
+def test_seeded_residents():
+    # Pensions below, at and above the $50 that puts countable income at the
+    # $30 standard, earnings with and without a pension, and no income.
+    population = [
+        (age, pension, earnings)
+        for age in (66, 85)
+        for pension in (0, 20, 40, 49, 50, 51, 120)
+        for earnings in (0, 100, 145, 200)
+    ]
+    paid = _check(population)
+    # The grid reaches residents paid SSI with and without other income, and
+    # residents with no federal payment.
+    assert paid.sum() >= 10
+    assert (~paid).any()