policyengine 6.2.2__tar.gz → 6.2.3__tar.gz

{policyengine-6.2.2 → policyengine-6.2.3}/.github/workflows/draft-pdf.yml

@@ -1,5 +1,8 @@
 on:
+  # Feature branches get the pull_request run below; building the draft again
+  # on every branch push duplicated it.
   push:
+    branches: [main]
     paths:
       - paper.md
       - paper.bib
@@ -18,6 +21,7 @@ jobs:
   paper:
     runs-on: ubuntu-latest
     name: Draft PDF
+    timeout-minutes: 15
     steps:
       - uses: actions/checkout@v4
       - uses: openjournals/openjournals-draft-action@master

{policyengine-6.2.2 → policyengine-6.2.3}/.github/workflows/pr_code_changes.yaml

@@ -22,6 +22,7 @@ jobs:
   check-changelog:
     name: Check changelog fragment
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     steps:
       - uses: actions/checkout@v6
         with:
@@ -32,6 +33,7 @@ jobs:
         run: .github/check-changelog.sh
   Lint:
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     steps:
         - uses: actions/checkout@v6
         - name: Install ruff
@@ -44,6 +46,7 @@ jobs:
           run: ruff check .
   Mypy:
     runs-on: ubuntu-latest
+    timeout-minutes: 20
     steps:
         - uses: actions/checkout@v6
         - name: Install uv
@@ -61,6 +64,7 @@ jobs:
   Python-Compat:
     name: Install + smoke-import (py${{ matrix.python-version }})
     runs-on: ubuntu-latest
+    timeout-minutes: 15
     strategy:
       fail-fast: false
       matrix:
@@ -85,6 +89,7 @@ jobs:
     # Private manifest credentials are available only to trusted repository PRs.
     if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
     runs-on: ubuntu-latest
+    timeout-minutes: 30
     env:
       POLICYENGINE_SKIP_COUNTRY_IMPORTS: "1"
     steps:
@@ -136,6 +141,9 @@ jobs:
         run: policyengine bundle verify --country us --country uk --packages-only --json
   Test:
     runs-on: ubuntu-latest
+    # Slowest observed 2026-10-06..08: 28 minutes. Fail a hung job instead of
+    # holding a shared org runner for the 6-hour default.
+    timeout-minutes: 60
     strategy:
       fail-fast: false
       matrix:

{policyengine-6.2.2 → policyengine-6.2.3}/.github/workflows/pr_docs_changes.yaml

@@ -19,6 +19,7 @@ jobs:
   Test:
     runs-on: ubuntu-latest
     name: Test documentation builds
+    timeout-minutes: 30
     steps:
       - name: Checkout repo
         uses: actions/checkout@v6

{policyengine-6.2.2 → policyengine-6.2.3}/.github/workflows/push.yaml

@@ -28,6 +28,7 @@ jobs:
   Lint:
     if: github.event.head_commit.message != 'Update package version'
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     steps:
       - uses: actions/checkout@v6
       - name: Install ruff
@@ -40,6 +41,8 @@ jobs:
   Test:
     if: github.event.head_commit.message != 'Update package version'
     runs-on: ubuntu-latest
+    # Slowest observed 2026-10-06..08: 26 minutes.
+    timeout-minutes: 60
     strategy:
       fail-fast: false
       matrix:

{policyengine-6.2.2 → policyengine-6.2.3}/AGENTS.md

@@ -33,3 +33,16 @@ round-trip is a documented property of the public surface.
 
 Do not cache arbitrary Python objects in public result structures. The
 `core.Simulation` output must stay serialisable.
+
+## UK country filtering
+
+Select Scotland, Wales, and Northern Ireland through the stored household
+`region` column: `SCOTLAND`, `WALES`, and `NORTHERN_IRELAND`, respectively.
+Never require a `country` column when filtering raw UK household inputs;
+`country` is derived by the UK model after loading the data. The public selector
+`country/scotland` describes the requested area, not an input column name.
+
+England must select the nine English `region` values, not `ENGLAND` and not
+the complement of the other countries. Preserve these rules in new filters,
+refactors, and tests. Read the UK country filtering section in
+`docs/engineering/skills/repository-guidance.md` before changing this behavior.

{policyengine-6.2.2 → policyengine-6.2.3}/CHANGELOG.md

@@ -1,3 +1,14 @@
+## [6.2.3] - 2026-10-09
+
+### Changed
+
+- Give every lint, test, docs and paper CI job a timeout, so a hung job frees its shared GitHub Actions runner instead of holding it for six hours.
+
+### Fixed
+
+- Filter UK country simulations using stored household regions, selecting all nine English regions for England and preserving entity relationships and weights without requiring a derived country column.
+
+
 ## [6.2.2] - 2026-10-07
 
 ### Changed

{policyengine-6.2.2/src/policyengine.egg-info → policyengine-6.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: policyengine
-Version: 6.2.2
+Version: 6.2.3
 Summary: A package to conduct policy analysis using PolicyEngine tax-benefit models.
 Author-email: PolicyEngine <hello@policyengine.org>
 License:                     GNU AFFERO GENERAL PUBLIC LICENSE

{policyengine-6.2.2 → policyengine-6.2.3}/docs/engineering/skills/repository-guidance.md

@@ -52,6 +52,35 @@ representative-data runs unless the change specifically needs that coverage.
 When a test needs country package data, make the dependency explicit and skip
 cleanly if credentials or local artifacts are unavailable.
 
+## UK country filtering
+
+Raw UK household data stores `region`. The UK model derives `country` from
+that input during calculation. Filtering precedes calculation, so agents and
+developers must use `region` for country selection even though the registry's
+public identifiers start with `country/`.
+
+| Public selector | Required input filter |
+| --- | --- |
+| `country/scotland` | `region == "SCOTLAND"` |
+| `country/wales` | `region == "WALES"` |
+| `country/northern_ireland` | `region == "NORTHERN_IRELAND"` |
+| `country/england` | `region` is one of the nine English values below |
+
+The English values are `NORTH_EAST`, `NORTH_WEST`, `YORKSHIRE`, `EAST_MIDLANDS`,
+`WEST_MIDLANDS`, `EAST_OF_ENGLAND`, `LONDON`, `SOUTH_EAST`, and `SOUTH_WEST`.
+There is no stored `ENGLAND` region value. Use explicit membership so unknown
+or missing regions do not become English households.
+
+Keep all four country entries as `RowFilterStrategy` so the simulation API can
+combine them with `RegionGroupStrategy`. Do not add a dataset `country` column
+or run a preliminary model calculation to work around an incorrect filter.
+Preserve household membership, related people/benefit units, and weights.
+
+Regression tests must exercise Scotland, Wales, Northern Ireland, and England
+against household inputs with `region` and **no `country` column**, including
+the string and byte representations accepted by the dataset loader. See
+`tests/test_uk_regions.py` and the public explanation in `docs/regions.md`.
+
 ## Anti-Patterns
 
 - Do not bypass the wrapper layer without a clear reason.

{policyengine-6.2.2 → policyengine-6.2.3}/docs/regions.md

@@ -54,6 +54,38 @@ for row in impacts.district_results:
 at-large districts use `00`). Congressional district regions load the same
 ACS-local file and scope its rows by `congressional_district_geoid`.
 
+## UK countries
+
+UK country simulations filter the national dataset's stored household `region`
+column. The UK model derives `country` from `region`; the input dataset does not
+need a separate `country` column.
+
+| Country selector | Stored household input selection |
+| --- | --- |
+| `country/scotland` | `region == "SCOTLAND"` |
+| `country/wales` | `region == "WALES"` |
+| `country/northern_ireland` | `region == "NORTHERN_IRELAND"` |
+
+The `country/` prefix identifies a public geographic selector; it does not name
+a dataset column. Always filter these countries through `region` before running
+the simulation.
+
+Each country uses `RowFilterStrategy`. England matches a list of nine ITL1 regions:
+`NORTH_EAST`, `NORTH_WEST`, `YORKSHIRE`, `EAST_MIDLANDS`, `WEST_MIDLANDS`,
+`EAST_OF_ENGLAND`, `LONDON`, `SOUTH_EAST`, and `SOUTH_WEST`. Scotland, Wales and
+Northern Ireland each match their corresponding stored value: `SCOTLAND`,
+`WALES`, or `NORTHERN_IRELAND`. Unknown regions are excluded
+from all four countries. These filters preserve the selected households'
+people, benefit units and weights.
+
+```python
+scotland = pe.uk.model.get_region("country/scotland")
+england = pe.uk.model.get_region("country/england")
+```
+
+Pass the registry entry's `scoping_strategy` to `Simulation` when using a
+national UK dataset.
+
 ## UK parliamentary constituencies
 
 Constituency-level impacts group household output rows by the longwise

{policyengine-6.2.2 → policyengine-6.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "policyengine"
-version = "6.2.2"
+version = "6.2.3"
 description = "A package to conduct policy analysis using PolicyEngine tax-benefit models."
 readme = "README.md"
 authors = [

{policyengine-6.2.2 → policyengine-6.2.3}/src/policyengine/core/scoping_strategy.py

@@ -24,6 +24,7 @@ from microdf import MicroDataFrame
 from pydantic import BaseModel, Discriminator, Field
 
 from policyengine.utils.entity_utils import (
+    HouseholdFilterValue,
     filter_dataset_by_household_ids,
     filter_dataset_by_household_variable,
     matching_household_ids,
@@ -69,12 +70,13 @@ class RowFilterStrategy(RegionScopingStrategy):
     """Scoping strategy that filters dataset rows by a household variable.
 
     Used for regions where we want to keep only households matching a
-    specific variable value (e.g., US states or congressional districts).
+    specific variable value or any value in a list (e.g., the nine English
+    regions). Additional filters must also match.
     """
 
     strategy_type: Literal["row_filter"] = "row_filter"
     variable_name: str
-    variable_value: Union[str, int, float]
+    variable_value: HouseholdFilterValue
     additional_filters: dict[str, Union[str, int, float]] = Field(default_factory=dict)
 
     def apply(

{policyengine-6.2.2 → policyengine-6.2.3}/src/policyengine/countries/uk/regions.py

@@ -39,6 +39,21 @@ UK_COUNTRIES = {
     "northern_ireland": "Northern Ireland",
 }
 
+# Stored names from policyengine_uk's Region enum. England spans nine ITL1
+# regions; each other UK country is itself one ITL1 region.
+# https://www.ons.gov.uk/methodology/geography/ukgeographies/eurostat
+ENGLISH_REGIONS = (
+    "NORTH_EAST",
+    "NORTH_WEST",
+    "YORKSHIRE",
+    "EAST_MIDLANDS",
+    "WEST_MIDLANDS",
+    "EAST_OF_ENGLAND",
+    "LONDON",
+    "SOUTH_EAST",
+    "SOUTH_WEST",
+)
+
 
 def _load_constituencies_from_csv() -> list[dict]:
     """Load UK constituency data from CSV.
@@ -135,18 +150,19 @@ def build_uk_region_registry(
         )
     )
 
-    # 2. Country regions (filter from national by 'country' variable)
+    # 2. Countries scope the stored region input; country is a derived variable.
     for code, name in UK_COUNTRIES.items():
+        scoping_strategy = RowFilterStrategy(
+            variable_name="region",
+            variable_value=list(ENGLISH_REGIONS) if code == "england" else code.upper(),
+        )
         regions.append(
             Region(
                 code=f"country/{code}",
                 label=name,
                 region_type="country",
                 parent_code="uk",
-                scoping_strategy=RowFilterStrategy(
-                    variable_name="country",
-                    variable_value=code.upper(),
-                ),
+                scoping_strategy=scoping_strategy,
             )
         )
 

{policyengine-6.2.2 → policyengine-6.2.3}/src/policyengine/data/bundle/manifest.json

@@ -1,10 +1,10 @@
 {
-  "bundle_version": "6.2.2",
+  "bundle_version": "6.2.3",
   "citation": {
     "publisher": "PolicyEngine",
-    "title": "PolicyEngine bundle 6.2.2",
+    "title": "PolicyEngine bundle 6.2.3",
     "type": "software-bundle",
-    "version": "6.2.2"
+    "version": "6.2.3"
   },
   "countries": {
     "uk": {
@@ -17,7 +17,7 @@
   "data_releases": {
     "uk": {
       "build_id": "policyengine-uk-data-1.56.16",
-      "bundle_id": "uk-6.2.2",
+      "bundle_id": "uk-6.2.3",
       "certification": {
         "built_with_model_git_sha": "12a1e028afeef08d8b2d74ee03fd9de3a78b2dd3",
         "built_with_model_version": "2.89.2",
@@ -99,7 +99,7 @@
         "version": "2.102.3",
         "wheel_url": "https://files.pythonhosted.org/packages/37/e6/e8d9069d498f1510a1380f24a78e957902d01f00b10248c146d6710d5b37/policyengine_uk-2.102.3-py3-none-any.whl"
       },
-      "policyengine_version": "6.2.2",
+      "policyengine_version": "6.2.3",
       "region_datasets": {
         "national": {
           "path_template": "enhanced_frs_2024_25.h5"
@@ -112,7 +112,7 @@
     },
     "us": {
       "build_id": "populace-us-2024-spm-20260915",
-      "bundle_id": "us-6.2.2",
+      "bundle_id": "us-6.2.3",
       "certification": {
         "built_with_model_version": "2.2.1",
         "certified_by": "policyengine.py bundle certification",
@@ -217,7 +217,7 @@
         "version": "2.2.1",
         "wheel_url": "https://files.pythonhosted.org/packages/5e/09/86ac3968d2a18e28fce3ba1186fc21728e9f299b96a806d6f204f1e0aaed/policyengine_us-2.2.1-py3-none-any.whl"
       },
-      "policyengine_version": "6.2.2",
+      "policyengine_version": "6.2.3",
       "region_datasets": {
         "congressional_district": {
           "path_template": "populace_us_2024_acs_local.h5"
@@ -295,10 +295,10 @@
   "packages": {
     "policyengine": {
       "import_name": "policyengine",
-      "install_requirement": "policyengine==6.2.2",
+      "install_requirement": "policyengine==6.2.3",
       "name": "policyengine",
       "role": "bundle_carrier",
-      "version": "6.2.2"
+      "version": "6.2.3"
     },
     "policyengine-core": {
       "import_name": "policyengine_core",
@@ -334,6 +334,6 @@
       "wheel_url": "https://files.pythonhosted.org/packages/19/a0/c484f69a0ebf88a9b9fd0ac28176f8df66550f46714f1b0c5ec0b360824e/spm_calculator-1.0.0-py3-none-any.whl"
     }
   },
-  "policyengine_version": "6.2.2",
+  "policyengine_version": "6.2.3",
   "schema_version": 2
 }

{policyengine-6.2.2 → policyengine-6.2.3}/src/policyengine/data/bundle/uk.trace.tro.jsonld

@@ -18,18 +18,18 @@
         "schema:url": "https://policyengine.org"
       },
       "schema:dateCreated": "2026-08-18T14:20:32.100961Z",
-      "schema:description": "TRACE TRO for certified runtime bundle uk-6.2.2 covering the bundle manifest, the certified dataset artifact, the country model wheel, and the country data release manifest when it is available.",
+      "schema:description": "TRACE TRO for certified runtime bundle uk-6.2.3 covering the bundle manifest, the certified dataset artifact, the country model wheel, and the country data release manifest when it is available.",
       "schema:name": "policyengine uk certified bundle TRO",
       "trov:createdWith": {
         "@type": "schema:SoftwareApplication",
         "schema:name": "policyengine",
-        "schema:softwareVersion": "6.2.2"
+        "schema:softwareVersion": "6.2.3"
       },
       "trov:hasArrangement": [
         {
           "@id": "arrangement/1",
           "@type": "trov:ArtifactArrangement",
-          "rdfs:comment": "Certified arrangement for bundle uk-6.2.2.",
+          "rdfs:comment": "Certified arrangement for bundle uk-6.2.3.",
           "trov:hasArtifactLocation": [
             {
               "@id": "arrangement/1/location/bundle_manifest",
@@ -75,7 +75,7 @@
             "@type": "trov:ResearchArtifact",
             "schema:name": "policyengine.py bundle manifest for uk",
             "trov:mimeType": "application/json",
-            "trov:sha256": "f7563d813c99ae69da849631d57e498703ce338490aac1a17f127e17c15d6f5f"
+            "trov:sha256": "4f4d1c5802ff917e555ce9f5520150fa4e7e1584853782afc38c03818101c9c6"
           },
           {
             "@id": "composition/1/artifact/data_release_manifest",
@@ -102,7 +102,7 @@
         "trov:hasFingerprint": {
           "@id": "composition/1/fingerprint",
           "@type": "trov:CompositionFingerprint",
-          "trov:sha256": "ac5ce54dbdff4fc4aeeb67f89a3ec0455b8bda26c87a5744adf34d5a77fe6c97"
+          "trov:sha256": "23c38fee2f5fe5e6dd89137ba46b18b0add70f1a906cccbd28025f13c9061f64"
         }
       },
       "trov:hasPerformance": {

{policyengine-6.2.2 → policyengine-6.2.3}/src/policyengine/data/bundle/us.trace.tro.jsonld

@@ -17,18 +17,18 @@
         "schema:name": "PolicyEngine",
         "schema:url": "https://policyengine.org"
       },
-      "schema:description": "TRACE TRO for certified runtime bundle us-6.2.2 covering the bundle manifest, the certified dataset artifact, the country model wheel, and the country data release manifest when it is available.",
+      "schema:description": "TRACE TRO for certified runtime bundle us-6.2.3 covering the bundle manifest, the certified dataset artifact, the country model wheel, and the country data release manifest when it is available.",
       "schema:name": "policyengine us certified bundle TRO",
       "trov:createdWith": {
         "@type": "schema:SoftwareApplication",
         "schema:name": "policyengine",
-        "schema:softwareVersion": "6.2.2"
+        "schema:softwareVersion": "6.2.3"
       },
       "trov:hasArrangement": [
         {
           "@id": "arrangement/1",
           "@type": "trov:ArtifactArrangement",
-          "rdfs:comment": "Certified arrangement for bundle us-6.2.2.",
+          "rdfs:comment": "Certified arrangement for bundle us-6.2.3.",
           "trov:hasArtifactLocation": [
             {
               "@id": "arrangement/1/location/bundle_manifest",
@@ -74,7 +74,7 @@
             "@type": "trov:ResearchArtifact",
             "schema:name": "policyengine.py bundle manifest for us",
             "trov:mimeType": "application/json",
-            "trov:sha256": "f7563d813c99ae69da849631d57e498703ce338490aac1a17f127e17c15d6f5f"
+            "trov:sha256": "4f4d1c5802ff917e555ce9f5520150fa4e7e1584853782afc38c03818101c9c6"
           },
           {
             "@id": "composition/1/artifact/data_release_manifest",
@@ -101,7 +101,7 @@
         "trov:hasFingerprint": {
           "@id": "composition/1/fingerprint",
           "@type": "trov:CompositionFingerprint",
-          "trov:sha256": "f9c4770614de85fadde53b9f18d3f05f7d2e10962ee0b67ca99d659938bdee2e"
+          "trov:sha256": "f601acd275cdd7a90632c491593474090dac2c4478f372fe65472e23352e1194"
         }
       },
       "trov:hasPerformance": {

{policyengine-6.2.2 → policyengine-6.2.3}/src/policyengine/utils/entity_utils.py

@@ -8,6 +8,8 @@ from microdf import MicroDataFrame
 
 logger = logging.getLogger(__name__)
 
+HouseholdFilterValue = Union[str, int, float, list[Union[str, int, float]]]
+
 
 def _resolve_id_column(person_data: pd.DataFrame, entity_name: str) -> str:
     """Resolve the ID column name for a group entity in person data.
@@ -55,10 +57,10 @@ def build_entity_relationships(
 def _household_mask(
     household_data: pd.DataFrame,
     variable_name: str,
-    variable_value: Union[str, int, float],
+    variable_value: HouseholdFilterValue,
     additional_filters: Optional[dict[str, Union[str, int, float]]] = None,
 ):
-    """Boolean mask over the household table for a single-value filter.
+    """Boolean mask matching one value or any value in a list.
 
     Local intermediate only — never crosses a function boundary — so no
     positional-alignment invariant leaks out (callers key on household_id).
@@ -85,7 +87,7 @@ def _household_mask(
 def matching_household_ids(
     entity_data: dict[str, MicroDataFrame],
     variable_name: str,
-    variable_value: Union[str, int, float],
+    variable_value: HouseholdFilterValue,
     additional_filters: Optional[dict[str, Union[str, int, float]]] = None,
 ) -> set:
     """Return the set of ``household_id`` values matching a household filter.
@@ -163,7 +165,7 @@ def filter_dataset_by_household_variable(
     entity_data: dict[str, MicroDataFrame],
     group_entities: list[str],
     variable_name: str,
-    variable_value: Union[str, int, float],
+    variable_value: HouseholdFilterValue,
     additional_filters: Optional[dict[str, Union[str, int, float]]] = None,
 ) -> dict[str, MicroDataFrame]:
     """Filter dataset entities to only include households matching variables.
@@ -177,7 +179,8 @@ def filter_dataset_by_household_variable(
                      (from YearData.entity_data).
         group_entities: List of group entity names for this country.
         variable_name: The household-level variable to filter on.
-        variable_value: The value to match. Handles both str and bytes encoding.
+        variable_value: One value or a list of allowed values. String values
+                        match both str and bytes encoding.
         additional_filters: Optional household-level filters that must also
                             match, keyed by variable name.
 
@@ -201,7 +204,13 @@ def filter_dataset_by_household_variable(
     )
 
 
-def _values_match(values, expected: Union[str, int, float]):
+def _values_match(values, expected: HouseholdFilterValue):
+    if isinstance(expected, list):
+        allowed = [
+            *expected,
+            *(value.encode() for value in expected if isinstance(value, str)),
+        ]
+        return pd.Series(values).isin(allowed).to_numpy(copy=True)
     if isinstance(expected, str):
         return (values == expected) | (values == expected.encode())
     return values == expected

{policyengine-6.2.2 → policyengine-6.2.3/src/policyengine.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: policyengine
-Version: 6.2.2
+Version: 6.2.3
 Summary: A package to conduct policy analysis using PolicyEngine tax-benefit models.
 Author-email: PolicyEngine <hello@policyengine.org>
 License:                     GNU AFFERO GENERAL PUBLIC LICENSE

{policyengine-6.2.2 → policyengine-6.2.3}/tests/test_scoping_strategy.py

@@ -15,6 +15,63 @@ from policyengine.core.scoping_strategy import (
 from policyengine.core.simulation import Simulation
 
 
+@pytest.mark.parametrize(
+    "values,selected,expected",
+    [
+        (["LONDON", b"NORTH_WEST", "SCOTLAND"], ["LONDON", "NORTH_WEST"], [0, 1]),
+        ([6, 34, 36], [6, 36], [0, 2]),
+        ([6, "6", b"6"], [6], [0]),
+        ([6, "6", b"6"], ["6"], [1, 2]),
+        (["LONDON", "SCOTLAND", "WALES"], [], []),
+    ],
+)
+def test_row_filter_matches_any_list_value(values, selected, expected):
+    household = MicroDataFrame(
+        pd.DataFrame(
+            {
+                "household_id": [0, 1, 2],
+                "household_weight": [10.0, 20.0, 30.0],
+                "geography": values,
+            }
+        ),
+        weights="household_weight",
+    )
+    person = MicroDataFrame(
+        pd.DataFrame({"person_id": [0, 1, 2], "person_household_id": [0, 1, 2]})
+    )
+    strategy = RowFilterStrategy(variable_name="geography", variable_value=selected)
+    data = {"household": household, "person": person}
+
+    if not expected:
+        with pytest.raises(ValueError, match="No households found"):
+            strategy.apply(data, ["household"], 2026)
+    else:
+        result = strategy.apply(data, ["household"], 2026)
+        assert result["household"]["household_id"].tolist() == expected
+        assert result["person"]["person_id"].tolist() == expected
+
+
+def test_row_filter_list_values_respect_additional_filters(us_entity_data):
+    strategy = RowFilterStrategy(
+        variable_name="state_fips",
+        variable_value=[6, 34],
+        additional_filters={"congressional_district_geoid": 601},
+    )
+
+    result = strategy.apply(
+        us_entity_data,
+        ["household", "tax_unit", "spm_unit", "family", "marital_unit"],
+        2026,
+    )
+
+    assert result["household"]["household_id"].tolist() == [1]
+
+
+def test_row_filter_scalar_cache_key_is_unchanged():
+    strategy = RowFilterStrategy(variable_name="region", variable_value="SCOTLAND")
+    assert strategy.cache_key == "row_filter:region=SCOTLAND"
+
+
 class TestRowFilterStrategy:
     """Tests for RowFilterStrategy."""