Use modeled Medicare Part B inputs and targets (#714)

* Use modeled Medicare Part B inputs and targets

* Fix Medicare Part B changelog fragment

* Fix Medicare Part B data compatibility

* Format Part B data compatibility changes

* Populate Medicare enrollment independently of reported premiums

Author: MaxGhenis

changelog.d/713.fixed

@@ -0,0 +1 @@
+Modeled Medicare Part B premiums from enrollment and premium schedules, netted a cycle-free MSP standard-premium offset, and documented the national Part B calibration target as an approximate beneficiary-paid out-of-pocket benchmark rather than gross CMS premium income.

policyengine_us_data/datasets/cps/cps.py

@@ -37,6 +37,10 @@
     assign_takeup_with_reported_anchors,
     reported_subsidized_marketplace_by_tax_unit,
 )
+from policyengine_us_data.utils.policyengine import (
+    supports_medicare_enrollment_input,
+    supports_modeled_medicare_part_b_inputs,
+)
 
 
 CURRENT_HEALTH_COVERAGE_REPORTED_VAR_MAP = {
@@ -820,7 +824,12 @@ def add_personal_income_variables(cps: h5py.File, person: DataFrame, year: int):
     cps["health_insurance_premiums_without_medicare_part_b"] = person.PHIP_VAL
     cps["over_the_counter_health_expenses"] = person.POTC_VAL
     cps["other_medical_expenses"] = person.PMED_VAL
-    cps["medicare_part_b_premiums"] = person.PEMCPREM
+    if supports_medicare_enrollment_input():
+        cps["medicare_enrolled"] = person.MCARE == 1
+    if supports_modeled_medicare_part_b_inputs():
+        cps["medicare_part_b_premiums_reported"] = person.PEMCPREM
+    else:
+        cps["medicare_part_b_premiums"] = person.PEMCPREM
 
     # Get QBI simulation parameters ---
     yamlfilename = (

policyengine_us_data/datasets/cps/extended_cps.py

@@ -19,6 +19,9 @@
     impute_tax_unit_mortgage_balance_hints,
 )
 from policyengine_us_data.utils.policyengine import has_policyengine_us_variables
+from policyengine_us_data.utils.policyengine import (
+    supports_modeled_medicare_part_b_inputs,
+)
 from policyengine_us_data.utils.retirement_limits import (
     get_retirement_limits,
     get_se_pension_limits,
@@ -150,7 +153,6 @@ def _supports_structural_mortgage_inputs() -> bool:
     "health_insurance_premiums_without_medicare_part_b",
     "over_the_counter_health_expenses",
     "other_medical_expenses",
-    "medicare_part_b_premiums",
     "child_support_expense",
     # Hours/employment
     "weekly_hours_worked",
@@ -164,6 +166,9 @@ def _supports_structural_mortgage_inputs() -> bool:
     "self_employment_income_last_year",
 ]
 
+if not supports_modeled_medicare_part_b_inputs():
+    CPS_ONLY_IMPUTED_VARIABLES.append("medicare_part_b_premiums")
+
 # Set for O(1) lookup in the splice loop.
 _CPS_ONLY_SET = set(CPS_ONLY_IMPUTED_VARIABLES)
 

policyengine_us_data/datasets/puf/puf.py

@@ -802,11 +802,13 @@ class PUF_2024(PUF):
     url = "release://policyengine/irs-soi-puf/1.8.0/puf_2024.h5"
 
 
+# Leave Medicare Part B out of the generic PUF medical-expense split:
+# the baseline model now derives Part B premiums separately.
 MEDICAL_EXPENSE_CATEGORY_BREAKDOWNS = {
     "health_insurance_premiums_without_medicare_part_b": 0.453,
     "other_medical_expenses": 0.325,
-    "medicare_part_b_premiums": 0.137,
     "over_the_counter_health_expenses": 0.085,
+    "medicare_part_b_premiums": 0.137,
 }
 
 if __name__ == "__main__":

policyengine_us_data/db/etl_national_targets.py

@@ -12,6 +12,11 @@
 from policyengine_us_data.storage.calibration_targets.soi_metadata import (
     RETIREMENT_CONTRIBUTION_TARGETS,
 )
+from policyengine_us_data.utils.cms_medicare import (
+    get_beneficiary_paid_medicare_part_b_premiums_notes,
+    get_beneficiary_paid_medicare_part_b_premiums_source,
+    get_beneficiary_paid_medicare_part_b_premiums_target,
+)
 from policyengine_us_data.utils.db import (
     DEFAULT_YEAR,
     etl_argparser,
@@ -152,9 +157,15 @@ def extract_national_targets(year: int = DEFAULT_YEAR):
         },
         {
             "variable": "medicare_part_b_premiums",
-            "value": 112e9,
-            "source": "CMS Medicare data",
-            "notes": "Medicare Part B premium payments",
+            "value": get_beneficiary_paid_medicare_part_b_premiums_target(
+                HARDCODED_YEAR
+            ),
+            "source": get_beneficiary_paid_medicare_part_b_premiums_source(
+                HARDCODED_YEAR
+            ),
+            "notes": get_beneficiary_paid_medicare_part_b_premiums_notes(
+                HARDCODED_YEAR
+            ),
             "year": HARDCODED_YEAR,
         },
         {

policyengine_us_data/utils/cms_medicare.py

@@ -0,0 +1,41 @@
+MEDICARE_PART_B_GROSS_PREMIUM_INCOME = {
+    2024: 139.837e9,
+}
+
+
+MEDICARE_STATE_BUY_IN_MINIMUM_BENEFICIARIES = {
+    2024: 10_000_000,
+}
+
+
+BENEFICIARY_PAID_MEDICARE_PART_B_PREMIUM_TARGETS = {
+    2024: 112e9,
+}
+
+
+def get_beneficiary_paid_medicare_part_b_premiums_target(year: int) -> float:
+    try:
+        return BENEFICIARY_PAID_MEDICARE_PART_B_PREMIUM_TARGETS[year]
+    except KeyError as exc:
+        raise ValueError(
+            f"No beneficiary-paid Medicare Part B premium target sourced for {year}."
+        ) from exc
+
+
+def get_beneficiary_paid_medicare_part_b_premiums_source(year: int) -> str:
+    gross_income = MEDICARE_PART_B_GROSS_PREMIUM_INCOME[year] / 1e9
+    minimum_buy_in = MEDICARE_STATE_BUY_IN_MINIMUM_BENEFICIARIES[year]
+    return (
+        "CMS 2025 Medicare Trustees Report Table III.C3 actual 2024 Part B "
+        f"premium income (${gross_income:.3f}B), plus CMS State Buy-In FAQ "
+        f"noting states paid Part B premiums for over {minimum_buy_in:,} people"
+    )
+
+
+def get_beneficiary_paid_medicare_part_b_premiums_notes(year: int) -> str:
+    return (
+        "Approximate beneficiary-paid Medicare Part B out-of-pocket premiums "
+        "for SPM/MOOP calibration. This intentionally does not target gross "
+        "trust-fund premium income because Medicaid and other MSP pathways pay "
+        "premiums on behalf of some enrollees."
+    )

policyengine_us_data/utils/loss.py

@@ -12,6 +12,9 @@
 from policyengine_us_data.storage.calibration_targets.soi_metadata import (
     RETIREMENT_CONTRIBUTION_TARGETS,
 )
+from policyengine_us_data.utils.cms_medicare import (
+    get_beneficiary_paid_medicare_part_b_premiums_target,
+)
 from policyengine_us_data.db.etl_irs_soi import get_national_geography_soi_target
 from policyengine_core.reforms import Reform
 from policyengine_us_data.utils.soi import pe_to_soi, get_soi
@@ -26,7 +29,9 @@
 HARD_CODED_TOTALS = {
     "health_insurance_premiums_without_medicare_part_b": 385e9,
     "other_medical_expenses": 278e9,
-    "medicare_part_b_premiums": 112e9,
+    "medicare_part_b_premiums": get_beneficiary_paid_medicare_part_b_premiums_target(
+        2024
+    ),
     "over_the_counter_health_expenses": 72e9,
     "spm_unit_spm_threshold": 3_945e9,
     "child_support_expense": 33e9,

policyengine_us_data/utils/policyengine.py

@@ -134,3 +134,13 @@ def has_policyengine_us_variables(*variables: str) -> bool:
         return False
 
     return set(variables).issubset(available_variables)
+
+
+def supports_medicare_enrollment_input() -> bool:
+    return has_policyengine_us_variables("medicare_enrolled")
+
+
+def supports_modeled_medicare_part_b_inputs() -> bool:
+    return has_policyengine_us_variables(
+        "medicare_part_b_premiums_reported",
+    )

tests/unit/test_cms_medicare_targets.py

@@ -0,0 +1,25 @@
+import pytest
+
+from policyengine_us_data.utils.cms_medicare import (
+    get_beneficiary_paid_medicare_part_b_premiums_notes,
+    get_beneficiary_paid_medicare_part_b_premiums_source,
+    get_beneficiary_paid_medicare_part_b_premiums_target,
+)
+
+
+def test_beneficiary_paid_medicare_part_b_target_2024_is_sourced():
+    assert get_beneficiary_paid_medicare_part_b_premiums_target(2024) == pytest.approx(
+        112e9
+    )
+
+
+def test_beneficiary_paid_medicare_part_b_source_mentions_primary_sources():
+    source = get_beneficiary_paid_medicare_part_b_premiums_source(2024)
+    assert "2025 Medicare Trustees Report" in source
+    assert "State Buy-In FAQ" in source
+
+
+def test_beneficiary_paid_medicare_part_b_notes_describe_out_of_pocket_semantics():
+    notes = get_beneficiary_paid_medicare_part_b_premiums_notes(2024)
+    assert "out-of-pocket" in notes
+    assert "gross trust-fund premium income" in notes

tests/unit/test_medicare_part_b_inputs.py

@@ -0,0 +1,27 @@
+from policyengine_us_data.datasets.cps.extended_cps import (
+    CPS_ONLY_IMPUTED_VARIABLES,
+    supports_modeled_medicare_part_b_inputs,
+)
+from policyengine_us_data.datasets.puf.puf import MEDICAL_EXPENSE_CATEGORY_BREAKDOWNS
+from policyengine_us_data.utils import policyengine as policyengine_utils
+
+
+def test_medicare_part_b_clone_imputation_matches_installed_model_support():
+    assert ("medicare_part_b_premiums" in set(CPS_ONLY_IMPUTED_VARIABLES)) is (
+        not supports_modeled_medicare_part_b_inputs()
+    )
+
+
+def test_puf_medical_breakdown_still_sums_to_one():
+    assert sum(MEDICAL_EXPENSE_CATEGORY_BREAKDOWNS.values()) == 1.0
+
+
+def test_supports_medicare_enrollment_input_allows_partial_support(monkeypatch):
+    monkeypatch.setattr(
+        policyengine_utils,
+        "has_policyengine_us_variables",
+        lambda *variables: variables == ("medicare_enrolled",),
+    )
+
+    assert policyengine_utils.supports_medicare_enrollment_input() is True
+    assert policyengine_utils.supports_modeled_medicare_part_b_inputs() is False