Use CMS Medicare spending per enrollee for the Medicare per capita cost

changelog.d/medicare-per-enrollee-cms.fixed.md

@@ -0,0 +1 @@
+Value Medicare enrollees at CMS's Medicare spending per enrollee (National Health Expenditure projections, 2013-2034, extended to 2100) instead of a $14,500 estimate held flat from 2025.

policyengine_us/parameters/calibration/gov/hhs/medicare/per_capita_cost.yaml

@@ -1,12 +1,64 @@
-description: Average Medicare spending per enrolled beneficiary (national average). Represents the annual value of Medicare coverage including Part A and Part B.
-values:
-  2021-01-01: 11_080  # Traditional Medicare (Part A and/or Part B) per KFF
-  2025-01-01: 14_500  # Estimated - update with official Centers for Medicare & Medicaid Services (CMS) data when available
+description: The Centers for Medicare and Medicaid Services (CMS) reports and projects this Medicare spending per enrollee, which is all Medicare spending, including payments to Medicare Advantage plans, Part D and program administration, divided by average enrollment in Part A, Part B or both.
 metadata:
-  label: Medicare per capita cost
   unit: currency-USD
   period: year
+  label: CMS Medicare spending per enrollee
   reference:
-    - title: KFF Medicare Spending Per Beneficiary (2021)
-      href: https://www.kff.org/medicare/state-indicator/per-beneficiary/
-    - title: Note - 2024 value is estimated and should be updated with official Centers for Medicare & Medicaid Services (CMS) data
+  # Table17 Health Insurance Enrollment and Enrollment Growth Rates,
+  # "Spending per Enrollee ($)", "Medicare" row: Medicare spending (the same
+  # row as Table03's Medicare source of funds) divided by Medicare
+  # enrollment. Historical for 2013-2024, projected for 2025-2034. ZIP
+  # Last-Modified 2026-09-03, sha256
+  # 264f822618a5e1829138aea57cc6c8bf45bbaf072a72898fef5547f7ece6cd70,
+  # retrieved 2026-10-11 from
+  # https://www.cms.gov/data-research/statistics-trends-and-reports/national-health-expenditure-data/projected
+  - title: CMS National Health Expenditure Projections 2025-2034, Table 17
+    href: https://www.cms.gov/files/zip/nhe-projections-tables.zip
+  # Medicare spending covers Parts A, B, C (Medicare Advantage) and D (file
+  # page 15) and, as program spending, federal administration and the
+  # non-medical expenditures of Part C and Part D private plans (file page
+  # 23). Premiums finance part of that spending rather than reduce it: the
+  # sponsor analysis counts SMI premiums as household spending on Medicare
+  # (file page 30). Medicare enrollment is Part A and/or B enrollment by "a
+  # person-year methodology, which sums the number of months each
+  # beneficiary is enrolled during the year and divides by 12", with U.S.
+  # territories removed (file page 34). sha256
+  # 9c5755666847a78103016f04da937326744071a9a6736f4e77e91ce9e7af8629.
+  - title: "CMS National Health Expenditure Accounts: Methodology Paper, 2024 #page 15,23,30,34"
+    href: https://www.cms.gov/files/document/definitions-sources-methods.pdf#page=15
+  # "Projections of Medicare spending and enrollment used in the NHE
+  # projections were generated for the most recent Trustees Report" (file
+  # page 7); file page 6 names the 2026 report for the economic assumptions.
+  # Last Updated June 24, 2026, sha256
+  # ce62fe2c236cc96ce287027784b141584ddf39e62c5fd8ddc3e771a014944817.
+  - title: "Projections of National Health Expenditures and Enrollment: Methodology and Model Specification #page 6-7"
+    href: https://www.cms.gov/research-statistics-data-and-systems/statistics-trends-and-reports/nationalhealthexpenddata/downloads/projectionsmethodology.pdf#page=6
+values:
+  # Historical.
+  2013-01-01: 11_483
+  2014-01-01: 11_683
+  2015-01-01: 11_929
+  2016-01-01: 12_119
+  2017-01-01: 12_319
+  2018-01-01: 12_795
+  2019-01-01: 13_366
+  2020-01-01: 13_547
+  2021-01-01: 14_293
+  2022-01-01: 14_921
+  2023-01-01: 15_915
+  2024-01-01: 16_779
+  # Projections.
+  2025-01-01: 17_751
+  2026-01-01: 18_983
+  2027-01-01: 20_096
+  2028-01-01: 21_310
+  2029-01-01: 22_543
+  2030-01-01: 23_789
+  2031-01-01: 25_185
+  2032-01-01: 26_552
+  2033-01-01: 28_277
+  2034-01-01: 30_129
+  # 2035 to 2100 grow at the 2031-2032 rate (26_552 / 25_185, 5.4% a year),
+  # not the 2033-2034 rate, which carries the end of Medicare sequestration
+  # on August 31, 2033; later years hold the 2100 value. See
+  # set_all_uprating_parameters in uprating_extensions.py.

policyengine_us/parameters/uprating_extensions.py

@@ -880,6 +880,23 @@ def set_all_uprating_parameters(parameters: ParameterNode) -> ParameterNode:
         period_day=1,
     )
 
+    # CMS Medicare spending per enrollee (January values, last projection
+    # year 2034), which Medicare cost variables read directly. Sequestration
+    # cuts Medicare expenditures 2 percent through August 31, 2033 (2026
+    # Medicare Trustees Report, file page 209); its end lifts 2033 and 2034
+    # growth, and the Trustees' per-beneficiary growth falls from 6.6
+    # percent in 2034 to 5.4 percent in 2035 (Table V.D1, file page 210).
+    # So each year from 2035 to 2100 grows at the 2031-2032 rate, and later
+    # years hold the 2100 value.
+    extend_parameter_values(
+        parameters.calibration.gov.hhs.medicare.per_capita_cost,
+        last_projected_year=2034,
+        end_year=END_YEAR,
+        period_month=1,
+        period_day=1,
+        growth_years=(2031, 2032),
+    )
+
     # CBO income-by-source aggregates are used directly and as anchors for
     # SOI-based income upraters. Extending them in the baseline path keeps
     # long-run nominal data aging independent of scenario-specific reforms.

policyengine_us/tests/core/test_medicare_per_capita_cost.py

@@ -0,0 +1,247 @@
+"""The CMS Medicare spending per enrollee series that values Medicare coverage.
+
+``calibration.gov.hhs.medicare.per_capita_cost`` holds the "Medicare" row of
+"Spending per Enrollee ($)" in CMS's National Health Expenditure projections,
+Table 17 (2025-2034 release): Medicare spending over Medicare enrollment.
+Spending covers every Part, including payments to Medicare Advantage plans and
+program administration; enrollment counts Part A and/or Part B enrollees in
+person-years; both exclude the territories (NHEA methodology paper, file pages
+15, 23 and 34). 2013-2024 are historical and 2025-2034 projected, from the
+projections prepared for the 2026 Medicare Trustees Report. 2035 to 2100 grow
+at the 2031-2032 rate, because the end of sequestration on August 31, 2033
+lifts 2033 and 2034 growth.
+
+``medicare_gross_cost`` returns the series for each enrolled person,
+``medicare_cost`` subtracts the person's Part A and gross Part B premiums, and
+``qmb_cost_sharing`` pays QMB enrollees a share of it each month.
+
+Invariants:
+
+1. Transcription: every value is Table 17's published dollar figure, and is
+   consistent with the same table's Medicare spending over Medicare enrollment
+   within rounding.
+2. Cross-source level: the Trustees Report's per beneficiary cost (Table V.D1)
+   runs 5% to 8% above the series in every year from 2015 to 2034. V.D1 sums
+   per-Part averages (its footnote 2), while Table II.B1 counts 63.4 million
+   Part B and 56.8 million Part D enrollees among 69.3 million in 2025.
+3. Bounds and order: the series is positive and strictly increasing from 2013
+   to 2100, and holds the 2100 value afterwards.
+4. Extension rule: each year from 2035 to 2100 equals the 2034 value times
+   (26,552 / 25,185) ** (year - 2034), the closed form of the iterative
+   extension; the 2035 value is within 0.1% of the 2034 value grown at the
+   Trustees' 2034-2035 per beneficiary growth.
+5. Calendar years: every date within a year reads the January 1 value.
+6. Consumers: for an enrolled person, ``medicare_gross_cost`` equals the
+   series; ``medicare_cost`` equals the series less the Part A premium and the
+   gross Part B premium, floored at zero, so never exceeds
+   ``medicare_gross_cost``; a QMB enrollee's monthly ``qmb_cost_sharing`` is
+   the series over 12 times the QMB rate. A person not enrolled gets zero.
+"""
+
+import pytest
+from hypothesis import given, settings
+from hypothesis import strategies as st
+from policyengine_core.parameters.operations.get_parameter import get_parameter
+
+from policyengine_us import Simulation
+from policyengine_us.system import system
+
+SERIES = "calibration.gov.hhs.medicare.per_capita_cost"
+FIRST_YEAR, LAST_PROJECTED_YEAR, END_YEAR = 2013, 2034, 2100
+YEARS = range(FIRST_YEAR, LAST_PROJECTED_YEAR + 1)
+GROWTH_YEARS = (2031, 2032)
+
+# CMS NHE projections tables, Table 17, "Medicare" rows
+# (https://www.cms.gov/files/zip/nhe-projections-tables.zip, Last-Modified
+# 2026-09-03, sha256 264f8226...cd70).
+MEDICARE_SPENDING_PER_ENROLLEE = dict(
+    zip(
+        YEARS,
+        (
+            *(11_483, 11_683, 11_929, 12_119, 12_319, 12_795, 13_366, 13_547),
+            *(14_293, 14_921, 15_915, 16_779, 17_751, 18_983, 20_096, 21_310),
+            *(22_543, 23_789, 25_185, 26_552, 28_277, 30_129),
+        ),
+    )
+)
+MEDICARE_SPENDING_BILLIONS = dict(
+    zip(
+        YEARS,
+        (
+            *(588.9, 617.3, 647.7, 675.7, 704.9, 751.6, 805.2, 833.8, 894.6),
+            *(951.7, 1037.3, 1118.0, 1204.0, 1314.6, 1422.7, 1544.2, 1664.6),
+            *(1783.7, 1911.1, 2035.8, 2188.3, 2354.3),
+        ),
+    )
+)
+MEDICARE_ENROLLMENT_MILLIONS = dict(
+    zip(
+        YEARS,
+        (
+            *(51.3, 52.8, 54.3, 55.8, 57.2, 58.7, 60.2, 61.6, 62.6, 63.8),
+            *(65.2, 66.6, 67.8, 69.2, 70.8, 72.5, 73.8, 75.0, 75.9, 76.7),
+            *(77.4, 78.1),
+        ),
+    )
+)
+# 2026 Medicare Trustees Report, Table V.D1 (file page 210), "Total" average
+# per beneficiary cost: historical through 2025, intermediate estimates after.
+TRUSTEES_PER_BENEFICIARY_COST = dict(
+    zip(
+        range(2015, 2036),
+        (
+            *(12_734, 12_922, 13_127, 13_604, 14_224, 14_401, 15_170, 15_820),
+            *(16_809, 17_837, 18_894, 20_430, 21_497, 22_783, 24_076, 25_401),
+            *(26_826, 28_169, 29_989, 31_967, 33_687),
+        ),
+    )
+)
+
+years = st.integers(FIRST_YEAR, END_YEAR)
+
+
+def _series(date):
+    return get_parameter(system.parameters, SERIES)(date)
+
+
+def _value(year):
+    return _series(f"{year}-01-01")
+
+
+@pytest.mark.parametrize("year", YEARS)
+def test_series_is_table_17_medicare_spending_per_enrollee(year):
+    assert _value(year) == MEDICARE_SPENDING_PER_ENROLLEE[year]
+
+
+@pytest.mark.parametrize("year", YEARS)
+def test_published_values_match_spending_over_enrollment(year):
+    """Checks the transcribed rows against each other, so one mistyped
+    spending, enrollment or per-enrollee figure fails."""
+    spending = MEDICARE_SPENDING_BILLIONS[year]
+    enrollment = MEDICARE_ENROLLMENT_MILLIONS[year]
+    # Both components are rounded to one decimal and the published value to
+    # the dollar.
+    low = 1_000 * (spending - 0.05) / (enrollment + 0.05) - 0.5
+    high = 1_000 * (spending + 0.05) / (enrollment - 0.05) + 0.5
+    assert low <= MEDICARE_SPENDING_PER_ENROLLEE[year] <= high
+
+
+@pytest.mark.parametrize("year", range(2015, LAST_PROJECTED_YEAR + 1))
+def test_trustees_per_beneficiary_cost_runs_5_to_8_percent_above(year):
+    ratio = TRUSTEES_PER_BENEFICIARY_COST[year] / _value(year)
+    assert 1.05 <= ratio <= 1.08
+
+
+def test_2035_extension_follows_the_trustees_2035_growth():
+    trustees_growth = (
+        TRUSTEES_PER_BENEFICIARY_COST[2035] / TRUSTEES_PER_BENEFICIARY_COST[2034]
+    )
+    assert _value(2035) == pytest.approx(
+        _value(LAST_PROJECTED_YEAR) * trustees_growth, rel=0.001
+    )
+
+
+# The year domains below are finite and small, so they are checked
+# exhaustively rather than sampled.
+@pytest.mark.parametrize("year", range(FIRST_YEAR, END_YEAR))
+def test_series_is_positive_and_strictly_increasing(year):
+    """Adjacent years, so strict order holds for every pair of years."""
+    assert 0 < _value(year) < _value(year + 1)
+
+
+@pytest.mark.parametrize("year", range(LAST_PROJECTED_YEAR + 1, END_YEAR + 1))
+def test_extension_is_the_closed_form_of_the_2031_2032_growth(year):
+    earlier, later = GROWTH_YEARS
+    growth = (
+        MEDICARE_SPENDING_PER_ENROLLEE[later]
+        / (MEDICARE_SPENDING_PER_ENROLLEE[earlier])
+    )
+    expected = MEDICARE_SPENDING_PER_ENROLLEE[LAST_PROJECTED_YEAR] * growth ** (
+        year - LAST_PROJECTED_YEAR
+    )
+    assert _value(year) == pytest.approx(expected, rel=1e-9)
+    assert _value(year) / _value(year - 1) == pytest.approx(growth, rel=1e-12)
+
+
+@settings(deadline=None)
+@given(st.integers(END_YEAR, END_YEAR + 1_000))
+def test_series_holds_the_2100_value_afterwards(year):
+    assert _value(year) == _value(END_YEAR)
+
+
+@settings(deadline=None)
+@given(years, st.integers(1, 12), st.integers(1, 28))
+def test_every_date_in_a_year_reads_the_january_value(year, month, day):
+    assert _series(f"{year}-{month:02d}-{day:02d}") == _value(year)
+
+
+def _person_simulation(year, person):
+    members = ["person"]
+    return Simulation(
+        situation={
+            "people": {"person": person},
+            "tax_units": {"tax_unit": {"members": members}},
+            "spm_units": {"spm_unit": {"members": members}},
+            "families": {"family": {"members": members}},
+            "marital_units": {"marital_unit": {"members": members}},
+            "households": {
+                "household": {"members": members, "state_code": {year: "CA"}}
+            },
+        }
+    )
+
+
+@settings(max_examples=20, deadline=None)
+@given(
+    st.integers(2015, END_YEAR),
+    st.floats(0, 12_000, allow_nan=False),
+    st.floats(0, 40_000, allow_nan=False),
+)
+def test_medicare_cost_nets_premiums_from_the_series(
+    year, part_a_premium, part_b_premium
+):
+    simulation = _person_simulation(
+        year,
+        {
+            "age": {year: 70},
+            "is_medicare_eligible": {year: True},
+            "medicare_enrolled": {year: True},
+            "base_part_a_premium": {year: part_a_premium},
+            "gross_medicare_part_b_premium": {year: part_b_premium},
+        },
+    )
+    gross = simulation.calculate("medicare_gross_cost", year)[0]
+    net = simulation.calculate("medicare_cost", year)[0]
+    # Variables are float32, so compare to single precision.
+    assert gross == pytest.approx(_value(year), rel=1e-6)
+    expected = max(_value(year) - part_a_premium - part_b_premium, 0)
+    assert net == pytest.approx(expected, rel=1e-6, abs=0.01)
+    assert net <= gross
+
+
+@settings(max_examples=10, deadline=None)
+@given(st.integers(2015, END_YEAR))
+def test_person_not_enrolled_has_no_medicare_cost(year):
+    simulation = _person_simulation(
+        year,
+        {
+            "age": {year: 70},
+            "is_medicare_eligible": {year: True},
+            "medicare_enrolled": {year: False},
+        },
+    )
+    assert simulation.calculate("medicare_gross_cost", year)[0] == 0
+    assert simulation.calculate("medicare_cost", year)[0] == 0
+
+
+@settings(max_examples=10, deadline=None)
+@given(st.integers(2015, END_YEAR), st.integers(1, 12))
+def test_qmb_cost_sharing_is_the_qmb_rate_of_monthly_spending(year, month):
+    period = f"{year}-{month:02d}"
+    rate = system.parameters(
+        period
+    ).gov.hhs.medicare.savings_programs.qmb.cost_sharing_rate
+    simulation = _person_simulation(year, {"msp_category": {period: "QMB"}})
+    assert simulation.calculate("qmb_cost_sharing", period)[0] == pytest.approx(
+        _value(year) / 12 * rate, rel=1e-6
+    )

policyengine_us/tests/policy/baseline/gov/hhs/medicare/costs/medicare_cost.yaml

@@ -9,7 +9,7 @@
     tax_exempt_interest_income:
       2023: 0
   output:
-    medicare_cost: 12_280  # $14,500 spending - $2,220 Part B premium (2025)
+    medicare_cost: 15_531  # $17,751 spending - $2,220 Part B premium (2025)
 
 - name: unit test 2 - enrolled, pays full Part A premium
   period: 2025
@@ -22,7 +22,7 @@
     tax_exempt_interest_income:
       2023: 0
   output:
-    medicare_cost: 6_064  # $14,500 - $6,216 Part A - $2,220 Part B (2025)
+    medicare_cost: 9_315  # $17,751 - $6,216 Part A - $2,220 Part B (2025)
 
 - name: unit test 3 - enrolled with IRMAA
   period: 2025
@@ -36,7 +36,7 @@
       2023: 0
     filing_status: SINGLE
   output:
-    medicare_cost: 6_953.2  # $14,500 - $7,546.80 Part B with IRMAA (2025)
+    medicare_cost: 10_204.2  # $17,751 - $7,546.80 Part B with IRMAA (2025)
 
 - name: unit test 4 - eligible but not enrolled
   period: 2025
@@ -58,7 +58,7 @@
     tax_exempt_interest_income:
       2019: 0
   output:
-    medicare_cost: 9_298  # $11,080 - $1,782 Part B (2021: $148.50*12)
+    medicare_cost: 12_511  # $14,293 - $1,782 Part B (2021: $148.50*12)
 
 - name: unit test 6 - direct Part B premium input
   period: 2025
@@ -68,4 +68,4 @@
     medicare_quarters_of_coverage: 40
     gross_medicare_part_b_premium: 3_000
   output:
-    medicare_cost: 11_500  # $14,500 spending - $3,000 Part B premium
+    medicare_cost: 14_751  # $17,751 spending - $3,000 Part B premium