Release 2.2.2: calculate 2036 on the 2.2 line with spm-calculator 1.1

.github/release-lock.md

@@ -75,11 +75,11 @@ The opt-in probe creates an isolated, minimal project with a standard PyPI
 dependency and exercises actual uv locking. CI enables it. These commands do not
 install or import the country model.
 
-For the SPM integration, `spm-calculator` 1.0.0 reached PyPI on 2026-09-11, and
+For the SPM integration, `spm-calculator` 1.1.0 reached PyPI on 2026-11-02, and
 the committed country lock resolves it from `https://pypi.org/simple` with sdist
 and wheel SHA-256 hashes, so `--committed` passes on the committed files.
-`pyproject.toml` requires `>=1.0.0,<=1.0.0.post1` and the lock pins 1.0.0;
-moving that pin to the post-release is an ordinary reviewed lock change. This
+`pyproject.toml` requires `>=1.1.0,<1.2` and the lock pins 1.1.0; moving to a
+later 1.1.x release is an ordinary reviewed lock change. This
 guard validates dependency sources, artifact URLs and hashes, never a version
 string, so the calculator requirement is held by `pyproject.toml` and review
 rather than by `release_lock.py`. A stale root version or an older calculator

.github/workflows/pr.yaml

@@ -1,7 +1,7 @@
 name: Pull request
 on:
   pull_request:
-    branches: [main]
+    branches: [main, release/2.2.x]
 # Concurrency: auto-cancels previous runs when new commits are pushed
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}

.github/workflows/push.yaml

@@ -1,7 +1,7 @@
 name: Push
 on:
   push:
-    branches: [main]
+    branches: [main, release/2.2.x]
 
 jobs:
   ReleaseLock:
@@ -333,6 +333,13 @@ jobs:
       (github.repository == 'PolicyEngine/policyengine-us')
       && (github.event.head_commit.message == 'Update PolicyEngine US')
     needs: [ReleaseLock, ReleaseLockTests, Baseline, HouseholdAPIPartners, Contrib, Rest, Microsimulation]
+    # PyPI trusted publishing, as on main: PyPI accepts uploads only from this
+    # workflow in the pypi environment, whose deployment branches must include
+    # release/2.2.x. No stored PyPI token.
+    environment: pypi
+    permissions:
+      contents: read
+      id-token: write
     steps:
       - name: Checkout repo
         uses: actions/checkout@v6
@@ -355,6 +362,4 @@ jobs:
       - name: Publish a Python distribution to PyPI
         uses: pypa/gh-action-pypi-publish@release/v1
         with:
-          user: __token__
-          password: ${{ secrets.PYPI }}
           skip-existing: true

changelog.d/cbo-cpi-2036.changed.md

@@ -0,0 +1 @@
+- Add CBO's February 2026 calendar-2036 CPI-U and chained CPI-U points and keep their authored 2035 points instead of extrapolating over them, so single-year datasets extend through 2036 by default and IRS uprating projects through tax year 2036.

changelog.d/spm-calculator-1-1.changed.md

@@ -0,0 +1 @@
+- Require spm-calculator 1.1, whose canonical forecast covers 2022–2036, so SPM thresholds and SPM resources calculate for 2036.

docs/spm.md

@@ -3,7 +3,7 @@
 The country model uses the installed `spm-calculator` artifact for all five SPM
 measurement amounts. It does not extrapolate thresholds with country CPI or income
 parameters. Each supported year and scenario comes from the verified artifact.
-The installed 1.0.0 artifact covers 2022 through 2035; unavailable years fail.
+The installed 1.1.0 artifact covers 2022 through 2036; unavailable years fail.
 Taxes, benefits, resources and the housing-assistance cap
 remain country-model formulas. General household benefits and CBO transfer
 aggregates count actual `housing_assistance`; only SPM resources count

policyengine_us/parameters/gov/bls/cpi/c_cpi_u.yaml

@@ -332,7 +332,8 @@ values:
   2033-02-01: 210.8 # 2034 value.
   2034-02-01: 214.9 # 2035 value.
   2035-02-01: 219.2 # 2036 value.
-  # Values for 2036-2100 are set programmatically in uprating_extensions.py
+  2036-02-01: 223.5 # 2037 value.
+  # Values for 2037-2100 are set programmatically in uprating_extensions.py
 metadata:
   economy: false
   unit: currency-USD

policyengine_us/parameters/gov/bls/cpi/cpi_u.yaml

@@ -1373,7 +1373,8 @@ values:
   2033-02-01: 389.6 # 2034 value.
   2034-02-01: 398.4 # 2035 value.
   2035-02-01: 407.3 # 2036 value.
-  # Values for 2036-2100 are set programmatically in uprating_extensions.py
+  2036-02-01: 416.5 # 2037 value.
+  # Values for 2037-2100 are set programmatically in uprating_extensions.py
 metadata:
   unit: currency-USD
   label: CPI-U

policyengine_us/parameters/uprating_extensions.py

@@ -373,7 +373,9 @@ def set_all_uprating_parameters(parameters: ParameterNode) -> ParameterNode:
 
     # IRS uprating - special case that computes from CPI
     IRS_UPRATING_START_YEAR = 2010
-    IRS_LAST_PROJECTED_YEAR = 2035
+    # 2036 averages chained CPI-U from August 2034 to July 2035, the 2034 and
+    # 2035 February instants, which hold CBO's February 2026 calendar points.
+    IRS_LAST_PROJECTED_YEAR = 2036
 
     uprating_index = parameters.gov.irs.uprating
 
@@ -446,19 +448,22 @@ def set_all_uprating_parameters(parameters: ParameterNode) -> ParameterNode:
         period_day=1,
     )
 
-    # CPI-U (February values, last projection year 2034)
+    # CPI-U (February values, last projection year 2036). Each February
+    # instant holds CBO's February 2026 calendar-year average for that year
+    # (the "# YYYY value" comments name the tax year it feeds), so CBO's last
+    # point, calendar 2036, sits at the 2036 instant.
     extend_parameter_values(
         parameters.gov.bls.cpi.cpi_u,
-        last_projected_year=2034,
+        last_projected_year=2036,
         end_year=END_YEAR,
         period_month=2,
         period_day=1,
     )
 
-    # Chained CPI-U (February values, last projection year 2034)
+    # Chained CPI-U (February values, last projection year 2036), as for CPI-U.
     extend_parameter_values(
         parameters.gov.bls.cpi.c_cpi_u,
-        last_projected_year=2034,
+        last_projected_year=2036,
         end_year=END_YEAR,
         period_month=2,
         period_day=1,

policyengine_us/tests/microsimulation/data/test_extend_single_year_dataset.py

@@ -1103,14 +1103,14 @@ class FakeParam:
 
     def test_with_real_cpi_u_parameter(self):
         """Verify get_parameter_last_year works on the actual CPI-U
-        parameter YAML, returning 2035 (not 2100 from uprating_extensions)."""
+        parameter YAML, returning 2036 (not 2100 from uprating_extensions)."""
         from policyengine_us.system import system
 
         cpi_u = _resolve_parameter(system.parameters, "gov.bls.cpi.cpi_u")
         last_year = get_parameter_last_year(cpi_u)
-        # CPI-U YAML currently extends to 2035; if the YAML is updated,
+        # CPI-U YAML currently extends to 2036; if the YAML is updated,
         # this assertion should be updated to match.
-        assert last_year == 2035
+        assert last_year == 2036
 
 
 # ---------------------------------------------------------------------------

policyengine_us/tests/policy/baseline/household/income/spm_unit/test_spm_unit_spm_threshold.py

@@ -74,7 +74,7 @@ def canonical(forecast, year, tenure, *, adults=2, children=2, county=None):
 )
 def test_every_amount_uses_canonical_final_value_for_every_supported_year(tenure):
     forecast = load_forecast()
-    assert list(forecast.years) == list(range(2022, 2036))
+    assert list(forecast.years) == list(range(2022, 2037))
     # Explicit period inputs cover the artifact, including its historical years.
     situation = household(county="06037", tenure=tenure)
     for person in situation["people"].values():
@@ -164,10 +164,11 @@ def test_unavailable_selection_or_consumer_extrapolation_rejected(config):
         create_spm_provider(config)
 
 
-def test_unknown_year_has_no_country_cpi_extrapolation():
+@pytest.mark.parametrize("year", [2037, 2099])
+def test_unknown_year_has_no_country_cpi_extrapolation(year):
     simulation = Simulation(situation=household(county="06037"))
     with pytest.raises(ValueError, match="no entry"):
-        simulation.calculate("spm_unit_spm_threshold", 2099)
+        simulation.calculate("spm_unit_spm_threshold", year)
 
 
 @pytest.mark.parametrize(

pyproject.toml

@@ -24,7 +24,7 @@ dependencies = [
     # extension and is available on every supported Python version.
     "pandas>=3.0",
     "policyengine-core>=3.30.1",
-    "spm-calculator>=1.0.0,<=1.0.0.post1",
+    "spm-calculator>=1.1.0,<1.2",
     "tables>=3.9",
     "tqdm>=4.67.1",
 ]