The model denies Medicaid to expansion-group adults in Virginia, Maine, Idaho and Utah in 2019 or 2020, when those states covered them

policyengine_us/parameters/gov/hhs/medicaid/eligibility/categories/adult/income_limit.yamlから始め、引用されているCMSの承認文書と各州の日付を照合してください。policyengine_us/variables/gov/hhs/medicaid/eligibility/categories/adult/is_adult_for_medicaid_fc.pyにある1月1日の検索動作を読み、その後、issueに記載された再現手順を実行してください。YAMLの日付と上限額が文書に記載された適用開始日に一致し、2019～2021年の結果がissueの修正済みの行と一致したら完了です。

The model opens the Medicaid adult group in Virginia, Maine, Idaho and Utah on January 1, 2021. Virginia and Maine covered that group throughout 2019 and 2020, and Idaho and Utah throughout 2020, so an adult who qualifies only through it gets no Medicaid from the model in those years.

For example, a 30-year-old with no children and $15,000 of wages gets a medicaid_category of NONE in each of those states and years, and is_aca_ptc_eligible returns true. Those wages came to 120% of the federal poverty guideline for one person in 2019 ($12,490) and 118% in 2020 ($12,760). The same adult in Kentucky gets the ADULT category in both years.

Any calculation for 2019 or 2020 finds the adult group closed in these states, whether it covers one household or a microsimulation. Calculations for 2021 onward read the 138% limit in all four states.

What the model reads

The adult group's income_limit.yaml holds -.inf from 2018-01-01 and 1.38 from 2021-01-01 for Idaho, Maine, Nebraska, and Utah and Virginia.

The file cites a KFF table filtered to January 2018, January 2021 and January 2025 as its first reference. Each of these five states expanded between the first two dates and starts at 2021-01-01 in the file.

is_adult_for_medicaid_fc reads the limit at the start of the year, so each state's adult group opens in 2021.

When each state began covering the adult group

• CMS approved Maine's amendment in April 2019 with an effective date of July 2, 2018, and said federal funds would match payments for services those adults received from that date (page 4).
• We already date Missouri's adult group from July 1, 2021, the effective date of the amendment CMS approved that December. Dating Maine the same way gives July 2, 2018.
• In its approval for Utah, CMS describes the 95% standard of an earlier demonstration group as effectively 100% with the five-point income disregard (pages 6 and 25). That group's members moved into the adult group on April 1, 2019 (page 26), so we read the April 2019 limit as 1.00, the way the file stores 133% as 1.38.
• Nebraska's October start changes no calendar-year result, because the model reads the limit on January 1.
• The Oklahoma, South Dakota and North Carolina rows match the dates in CMS's approval for Oklahoma, CMS's approval of South Dakota's matching-rate amendment and North Carolina's announcement.

Reproduction

from policyengine_core.reforms import Reform
from policyengine_us import Simulation

STATES = ["KY", "VA", "ME", "ID", "UT", "NE"]
YEARS = [2019, 2020, 2021]
GROUPS = ["tax_units", "spm_units", "families", "marital_units", "households"]


def situation():
    s = {"people": {}, **{group: {} for group in GROUPS}}
    for st in STATES:
        s["people"][st] = {
            "age": {y: 30 for y in YEARS},
            "employment_income": {y: 15_000 for y in YEARS},
        }
        for group in GROUPS:
            s[group][f"{group}_{st}"] = {"members": [st]}
        s["households"][f"households_{st}"]["state_code"] = {y: st for y in YEARS}
    return s


def show(label, sim):
    for year in YEARS:
        for variable in ["medicaid_category", "is_aca_ptc_eligible"]:
            values = sim.calculate(variable, year)
            if variable == "medicaid_category":
                values = values.decode_to_str()
            cells = [f"{st}={v}" for st, v in zip(STATES, values)]
            print(label, year, f"{variable:19}", *cells)


show("main ", Simulation(situation=situation()))

path = "gov.hhs.medicaid.eligibility.categories.adult.income_limit."
dated = Reform.from_dict(
    {
        path + "VA": {"2019-01-01.2020-12-31": 1.38},
        path + "ME": {"2018-07-02.2020-12-31": 1.38},
        path + "ID": {"2020-01-01.2020-12-31": 1.38},
        path + "UT": {"2019-04-01.2019-12-31": 1.0, "2020-01-01.2020-12-31": 1.38},
        path + "NE": {"2020-10-01.2020-12-31": 1.38},
    },
    country_id="us",
)
show("dated", Simulation(situation=situation(), reform=dated))

On main at 06c0dcb (2.116.19), this prints:

main  2019 medicaid_category   KY=ADULT VA=NONE ME=NONE ID=NONE UT=NONE NE=NONE
main  2019 is_aca_ptc_eligible KY=False VA=True ME=True ID=True UT=True NE=True
main  2020 medicaid_category   KY=ADULT VA=NONE ME=NONE ID=NONE UT=NONE NE=NONE
main  2020 is_aca_ptc_eligible KY=False VA=True ME=True ID=True UT=True NE=True
main  2021 medicaid_category   KY=ADULT VA=ADULT ME=ADULT ID=ADULT UT=ADULT NE=ADULT
main  2021 is_aca_ptc_eligible KY=False VA=False ME=False ID=False UT=False NE=False
dated 2019 medicaid_category   KY=ADULT VA=ADULT ME=ADULT ID=NONE UT=NONE NE=NONE
dated 2019 is_aca_ptc_eligible KY=False VA=False ME=False ID=True UT=True NE=True
dated 2020 medicaid_category   KY=ADULT VA=ADULT ME=ADULT ID=ADULT UT=ADULT NE=NONE
dated 2020 is_aca_ptc_eligible KY=False VA=False ME=False ID=False UT=False NE=True
dated 2021 medicaid_category   KY=ADULT VA=ADULT ME=ADULT ID=ADULT UT=ADULT NE=ADULT
dated 2021 is_aca_ptc_eligible KY=False VA=False ME=False ID=False UT=False NE=False

The main rows use the file as it stands. The dated rows apply the table's dates in memory.

Fix

Set the start dates in income_limit.yaml to Virginia 2019-01-01, Maine 2018-07-02, Idaho 2020-01-01, Utah 2019-04-01 (1.00) and 2020-01-01 (1.38), and Nebraska 2020-10-01, and cite the CMS documents above. With those dates, the adult qualifies in Virginia and Maine in 2019 and 2020, and in Idaho and Utah in 2020 (the dated rows).