Populate rent (ACS donor) and childcare (CPS ASEC) to close Gate-2 zero-targets (#180)

Always-enable the ACS donor (rent / real_estate_taxes imputation) and read childcare from CPS ASEC (SPM_CHILDCAREXPNS -> spm_unit_pre_subsidy_childcare_expenses), closing the two largest national Gate-2 residual targets (rent $0->~$735B, childcare $0->~$348B). Removes the now-dead include_acs flag. Independently reviewed via /cycle.

Author: MaxGhenis

src/microplex_us/data_sources/cps.py

@@ -117,6 +117,9 @@
     "WICYN": "_receives_wic",
     "SPM_CAPHOUSESUB": "_spm_capped_housing_subsidy",
     "SPM_ENGVAL": "spm_unit_energy_subsidy",
+    # Capped work childcare expenses are a PolicyEngine-computed variable
+    # (derived from this pre-subsidy input), so only the input is exported.
+    "SPM_CHILDCAREXPNS": "spm_unit_pre_subsidy_childcare_expenses",
     # Person relationship-to-householder code (eCPS cps.py:190-195, :1219).
     # Codes 43/44/46/47 mark an unmarried partner of the household head.
     "PERRP": "_person_relationship_to_householder",
@@ -200,6 +203,7 @@
     "social_security_survivors",
     "social_security_dependents",
     "spm_unit_energy_subsidy",
+    "spm_unit_pre_subsidy_childcare_expenses",
 )
 
 PERSON_ZERO_DEFAULT_VALUE_COLUMNS = (
@@ -216,6 +220,7 @@
     "social_security_survivors",
     "social_security_dependents",
     "spm_unit_energy_subsidy",
+    "spm_unit_pre_subsidy_childcare_expenses",
 )
 
 PERSON_CACHE_REQUIRED_COLUMNS = (

src/microplex_us/pipelines/pe_us_data_rebuild.py

@@ -105,7 +105,6 @@ def default_policyengine_us_data_rebuild_source_providers(
     puf_demographics_path: str | Path | None = None,
     puf_expand_persons: bool = True,
     include_donor_surveys: bool = True,
-    include_acs: bool | None = None,
     include_sipp: bool | None = None,
     include_scf: bool | None = None,
     acs_year: int = 2022,
@@ -155,19 +154,21 @@ def default_policyengine_us_data_rebuild_source_providers(
             social_security_split_strategy=SOCIAL_SECURITY_SPLIT_STRATEGY_PE_QRF,
         ),
     ]
-    resolved_include_acs = include_donor_surveys if include_acs is None else include_acs
     resolved_include_sipp = (
         include_donor_surveys if include_sipp is None else include_sipp
     )
     resolved_include_scf = include_donor_surveys if include_scf is None else include_scf
-    if resolved_include_acs:
-        providers.append(
-            ACSSourceProvider(
-                year=int(acs_year),
-                policyengine_us_data_repo=policyengine_us_data_repo,
-                policyengine_us_data_python=policyengine_us_data_python,
-            )
+    # The ACS donor is always enabled. It supplies the rent and real_estate_taxes
+    # source imputation that eCPS also draws from ACS, so omitting it leaves those
+    # variables at zero. ACS as a population spine ("multispine") is a separate,
+    # independently controlled feature that is not enabled here.
+    providers.append(
+        ACSSourceProvider(
+            year=int(acs_year),
+            policyengine_us_data_repo=policyengine_us_data_repo,
+            policyengine_us_data_python=policyengine_us_data_python,
         )
+    )
     if resolved_include_sipp:
         providers.extend(
             [

src/microplex_us/pipelines/pe_us_data_rebuild_checkpoint.py

@@ -1979,7 +1979,6 @@ def run_policyengine_us_data_rebuild_checkpoint(
     puf_demographics_path: str | Path | None = None,
     puf_expand_persons: bool = True,
     include_donor_surveys: bool = True,
-    include_acs: bool | None = None,
     include_sipp: bool | None = None,
     include_scf: bool | None = None,
     acs_year: int = 2022,
@@ -2066,7 +2065,6 @@ def run_policyengine_us_data_rebuild_checkpoint(
                 puf_demographics_path=puf_demographics_path,
                 puf_expand_persons=puf_expand_persons,
                 include_donor_surveys=include_donor_surveys,
-                include_acs=include_acs,
                 include_sipp=include_sipp,
                 include_scf=include_scf,
                 acs_year=acs_year,
@@ -2289,15 +2287,6 @@ def main(argv: list[str] | None = None) -> None:
         action=argparse.BooleanOptionalAction,
         default=True,
     )
-    parser.add_argument(
-        "--include-acs",
-        action=argparse.BooleanOptionalAction,
-        default=None,
-        help=(
-            "Include the ACS donor provider. Defaults to --include-donor-surveys; "
-            "use --no-include-acs for an eCPS-shaped run that keeps SIPP/SCF."
-        ),
-    )
     parser.add_argument(
         "--include-sipp",
         action=argparse.BooleanOptionalAction,
@@ -2488,7 +2477,6 @@ def main(argv: list[str] | None = None) -> None:
         puf_demographics_path=args.puf_demographics_path,
         puf_expand_persons=not args.no_puf_expand_persons,
         include_donor_surveys=args.include_donor_surveys,
-        include_acs=args.include_acs,
         include_sipp=args.include_sipp,
         include_scf=args.include_scf,
         acs_year=args.acs_year,

src/microplex_us/pipelines/us.py

@@ -9891,6 +9891,7 @@ def _attach_spm_unit_source_columns(
             "takes_up_snap_if_eligible": "max",
             "takes_up_tanf_if_eligible": "max",
             "spm_unit_energy_subsidy": "first",
+            "spm_unit_pre_subsidy_childcare_expenses": "first",
         }
         aggregations = {
             column: aggregation

tests/pipelines/test_pe_us_data_rebuild.py

@@ -145,17 +145,21 @@ def test_default_policyengine_us_data_rebuild_source_providers_use_pe_style_bund
     assert isinstance(providers[5], SCFSourceProvider)
 
 
-def test_default_policyengine_us_data_rebuild_source_providers_can_disable_donor_surveys() -> (
+def test_default_policyengine_us_data_rebuild_source_providers_keeps_acs_when_donor_surveys_disabled() -> (
     None
 ):
+    # include_donor_surveys=False disables the SIPP/SCF donors, but the ACS donor is
+    # always enabled (it supplies the rent / real_estate_taxes imputation), so it
+    # remains alongside the CPS spine and PUF.
     providers = default_policyengine_us_data_rebuild_source_providers(
         include_donor_surveys=False,
         cps_download=False,
     )
 
-    assert len(providers) == 2
+    assert len(providers) == 3
     assert isinstance(providers[0], CPSASECSourceProvider)
     assert isinstance(providers[1], PUFSourceProvider)
+    assert isinstance(providers[2], ACSSourceProvider)
 
 
 def test_default_policyengine_us_data_rebuild_source_providers_can_include_donor_surveys() -> (
@@ -177,26 +181,6 @@ def test_default_policyengine_us_data_rebuild_source_providers_can_include_donor
     assert isinstance(providers[5], SCFSourceProvider)
 
 
-def test_default_policyengine_us_data_rebuild_source_providers_can_disable_only_acs() -> (
-    None
-):
-    providers = default_policyengine_us_data_rebuild_source_providers(
-        include_donor_surveys=True,
-        include_acs=False,
-        cps_download=False,
-    )
-
-    assert len(providers) == 5
-    assert isinstance(providers[0], CPSASECSourceProvider)
-    assert isinstance(providers[1], PUFSourceProvider)
-    assert isinstance(providers[2], SIPPSourceProvider)
-    assert providers[2].block == "tips"
-    assert isinstance(providers[3], SIPPSourceProvider)
-    assert providers[3].block == "assets"
-    assert isinstance(providers[4], SCFSourceProvider)
-    assert not any(isinstance(provider, ACSSourceProvider) for provider in providers)
-
-
 def test_build_policyengine_us_data_rebuild_pipeline_returns_configured_pipeline() -> (
     None
 ):

tests/pipelines/test_us.py

@@ -1014,6 +1014,7 @@ def test_build_policyengine_entity_tables_preserves_spm_source_inputs(
                 "takes_up_housing_assistance_if_eligible": [False, True, False],
                 "takes_up_snap_if_eligible": [False, True, False],
                 "spm_unit_energy_subsidy": [90.0, 90.0, 0.0],
+                "spm_unit_pre_subsidy_childcare_expenses": [1500.0, 1500.0, 0.0],
             }
         )
 
@@ -1028,6 +1029,10 @@ def test_build_policyengine_entity_tables_preserves_spm_source_inputs(
         ]
         assert spm_units["takes_up_snap_if_eligible"].tolist() == [True, False]
         assert spm_units["spm_unit_energy_subsidy"].tolist() == [90.0, 0.0]
+        assert spm_units["spm_unit_pre_subsidy_childcare_expenses"].tolist() == [
+            1500.0,
+            0.0,
+        ]
 
     def test_build_policyengine_entity_tables_adds_deterministic_snap_takeup(
         self,

tests/policyengine/test_us.py

@@ -2189,6 +2189,7 @@ def __init__(self, entity):
         )
         spm_unit_contract_inputs = (
             "receives_housing_assistance",
+            "spm_unit_pre_subsidy_childcare_expenses",
             "spm_unit_tenure_type",
         )
         legacy_spm_unit_contract_inputs = (
@@ -2246,6 +2247,7 @@ class FakeSystem:
                     "spm_unit_id": [1000],
                     "household_id": [10],
                     "receives_housing_assistance": [True],
+                    "spm_unit_pre_subsidy_childcare_expenses": [1500.0],
                     "spm_unit_tenure_type": ["RENTER"],
                     **{name: [1.0] for name in legacy_spm_unit_contract_inputs},
                 }

tests/test_cps_source_provider.py

@@ -447,6 +447,7 @@ def test_load_cps_asec_derives_policyengine_value_inputs(tmp_path):
             "WICYN": [1, 2],
             "SPM_CAPHOUSESUB": [700, 0],
             "SPM_ENGVAL": [90, -1],
+            "SPM_CHILDCAREXPNS": [1500, -1],
             "PHIP_VAL": [900, -1],
             "POTC_VAL": [120, -1],
             "PMED_VAL": [450, -1],
@@ -474,6 +475,7 @@ def test_load_cps_asec_derives_policyengine_value_inputs(tmp_path):
     assert persons["receives_housing_assistance"].tolist() == [True, False]
     assert persons["takes_up_housing_assistance_if_eligible"].tolist() == [True, False]
     assert persons["spm_unit_energy_subsidy"].tolist() == [90, 0]
+    assert persons["spm_unit_pre_subsidy_childcare_expenses"].tolist() == [1500, 0]
     assert persons["health_insurance_premiums_without_medicare_part_b"].tolist() == [900, 0]
     assert persons["over_the_counter_health_expenses"].tolist() == [120, 0]
     assert persons["other_medical_expenses"].tolist() == [450, 0]