Treat PE-US aggregate outputs as formula-owned

packages/populace-build/pyproject.toml

@@ -19,7 +19,7 @@ dependencies = [
 # The US extra adds the rules engine for formula-owned export checks. Source
 # content is declared in packaged manifests and interpreted by shared Populace
 # runtimes; country source loaders must not depend on incumbent data packages.
-us = ["policyengine-us>=1.729,<2", "h5py>=3", "microunit>=0.1.0", "tables>=3"]
+us = ["policyengine-us>=1.745.0,<2", "h5py>=3", "microunit>=0.1.0", "tables>=3"]
 # The UK extra adds the rules engine for local metric generation from a
 # Populace UK H5. Target tables remain explicit inputs, and the base package
 # still does not import policyengine-uk at import time.

packages/populace-build/src/populace/build/source_manifest.py

@@ -41,9 +41,11 @@
         "assign_by_plan_type",
         "assign_binary_from_rate",
         "calibrate_binary_assignment",
+        "calibrate_binary_assignment_joint_targets",
         "convert_interest_to_structural_mortgage_inputs",
         "compute_ratio",
         "derive",
+        "derive_puf_policyengine_variables",
         "derive_mortgage_balance_hints",
         "disaggregate_aggregate_records",
         "fit_labor_market_models",

packages/populace-build/src/populace/build/us/puf_aggregate_record_disaggregation.json

@@ -9,13 +9,13 @@
     "E00200",
     "P23250",
     "P22250",
-    "E00650",
+    "qualified_dividend_income",
     "E00300",
     "E26270",
     "E00900",
     "E02100",
     "E00400",
-    "E00600"
+    "non_qualified_dividend_income"
   ],
   "buckets": {
     "999996": {

packages/populace-build/src/populace/build/us/source_stages.json

@@ -22,6 +22,13 @@
           "table": "puf_tax_unit",
           "weight": "s006"
         },
+        {
+          "kind": "derive_puf_policyengine_variables",
+          "ordinary_dividend_source": "E00600",
+          "qualified_dividend_source": "E00650",
+          "qualified_dividend_output": "qualified_dividend_income",
+          "non_qualified_dividend_output": "non_qualified_dividend_income"
+        },
         {
           "kind": "disaggregate_aggregate_records",
           "method": "donor_template_calibration",
@@ -42,7 +49,8 @@
             "employment_income",
             "self_employment_income",
             "taxable_interest_income",
-            "ordinary_dividends",
+            "qualified_dividend_income",
+            "non_qualified_dividend_income",
             "capital_gains",
             "filing_status"
           ]
@@ -53,21 +61,65 @@
         }
       ],
       "outputs": [
-        "salt_deduction",
-        "medical_expense_deduction",
-        "charitable_deduction",
-        "interest_deduction",
-        "qualified_business_income",
+        "employment_income_before_lsr",
+        "self_employment_income_before_lsr",
+        "taxable_interest_income",
+        "qualified_dividend_income",
+        "non_qualified_dividend_income",
+        "tax_exempt_interest_income",
+        "short_term_capital_gains",
+        "long_term_capital_gains_before_response",
+        "non_sch_d_capital_gains",
+        "taxable_private_pension_income",
+        "taxable_ira_distributions",
+        "social_security_retirement",
+        "social_security_disability",
+        "social_security_dependents",
+        "social_security_survivors",
+        "charitable_cash_donations",
+        "charitable_non_cash_donations",
+        "real_estate_taxes",
+        "home_mortgage_interest",
+        "student_loan_interest",
+        "rental_income",
+        "estate_income",
+        "farm_income",
+        "miscellaneous_income",
         "partnership_income",
         "s_corp_income",
         "partnership_self_employment_net_earnings",
-        "partnership_s_corp_loss"
+        "first_home_mortgage_balance",
+        "second_home_mortgage_balance",
+        "first_home_mortgage_interest",
+        "second_home_mortgage_interest",
+        "first_home_mortgage_origination_year",
+        "second_home_mortgage_origination_year"
       ],
       "nonnegative_outputs": [
-        "salt_deduction",
-        "medical_expense_deduction",
-        "charitable_deduction",
-        "interest_deduction"
+        "employment_income_before_lsr",
+        "self_employment_income_before_lsr",
+        "taxable_interest_income",
+        "qualified_dividend_income",
+        "non_qualified_dividend_income",
+        "tax_exempt_interest_income",
+        "long_term_capital_gains_before_response",
+        "taxable_private_pension_income",
+        "taxable_ira_distributions",
+        "social_security_retirement",
+        "social_security_disability",
+        "social_security_dependents",
+        "social_security_survivors",
+        "charitable_cash_donations",
+        "charitable_non_cash_donations",
+        "real_estate_taxes",
+        "home_mortgage_interest",
+        "student_loan_interest",
+        "first_home_mortgage_balance",
+        "second_home_mortgage_balance",
+        "first_home_mortgage_interest",
+        "second_home_mortgage_interest",
+        "first_home_mortgage_origination_year",
+        "second_home_mortgage_origination_year"
       ],
       "notes": "PUF detail is a primary-source donor and must stay separate from benchmark comparisons. IRS disclosure aggregate rows are replaced from raw PUF aggregate totals before uprating; Forbes top-tail synthesis is not enabled."
     },
@@ -533,6 +585,22 @@
           "domain": "is_aca_ptc_eligible",
           "weight": "tax_unit_weight"
         },
+        {
+          "kind": "calibrate_binary_assignment_joint_targets",
+          "variable": "takes_up_aca_if_eligible",
+          "count_targets": [
+            "irs_soi_premium_tax_credit_returns_by_state"
+          ],
+          "amount_targets": [
+            "irs_soi_premium_tax_credit_amount_by_state"
+          ],
+          "optional": true,
+          "preserve_true_anchors": true,
+          "preserve_true_anchor": "reported_has_subsidized_marketplace_health_coverage_at_interview",
+          "domain": "is_aca_ptc_eligible",
+          "count_weight": "household_weight",
+          "amount_weight": "weighted_assigned_aca_ptc"
+        },
         {
           "kind": "compute_ratio",
           "output": "selected_marketplace_plan_benchmark_ratio",

packages/populace-build/src/populace/build/us_runtime/__init__.py

@@ -38,6 +38,11 @@
     load_asec_h5_tables,
     pool_asec_sources,
 )
+from populace.build.us_runtime.cps_carried import (
+    CPS_CARRIED_FORMULA_OWNED_COLUMNS,
+    CPS_CARRIED_PERSON_INPUTS,
+    derive_us_cps_carried_inputs,
+)
 from populace.build.us_runtime.demographics import (
     AGE_BANDS,
     DEMOGRAPHICS_SCHEMA_VERSION,
@@ -69,6 +74,8 @@
 )
 from populace.build.us_runtime.puf_support import (
     BASE_ASEC_SUPPORT_CHANNEL,
+    PUF_TAX_DETAIL_DEFAULT_PERSON_OUTPUTS,
+    PUF_TAX_DETAIL_DEFAULT_TAX_UNIT_OUTPUTS,
     PUF_TAX_DETAIL_SUPPORT_CHANNEL,
     US_PUF_SUPPORT_STAGE_NAME,
     clone_us_frame_for_puf_support,
@@ -107,6 +114,8 @@
     "BuildConfig",
     "AsecSource",
     "BASE_ASEC_SUPPORT_CHANNEL",
+    "CPS_CARRIED_FORMULA_OWNED_COLUMNS",
+    "CPS_CARRIED_PERSON_INPUTS",
     "SimpleTaxExpenditureReform",
     "ReformValidationSpec",
     "REFORM_VALIDATION_SCHEMA_VERSION",
@@ -138,13 +147,16 @@
     "US_SOURCE_MANIFEST",
     "US_SOURCE_STAGE_SPECS",
     "US_STAGE_NAMES",
+    "PUF_TAX_DETAIL_DEFAULT_PERSON_OUTPUTS",
+    "PUF_TAX_DETAIL_DEFAULT_TAX_UNIT_OUTPUTS",
     "PUF_TAX_DETAIL_SUPPORT_CHANNEL",
     "US_PUF_SUPPORT_STAGE_NAME",
     "US_STATE_INCOME_TAX_TARGET_SPECS",
     "US_STATE_INCOME_TAX_TARGET_REFERENCES",
     "compile_us_fiscal_target_registry",
     "build_pooled_asec_unit_frame",
     "clone_us_frame_for_puf_support",
+    "derive_us_cps_carried_inputs",
     "disaggregate_us_puf_aggregate_records_from_manifest",
     "hard_target_package_aliases",
     "impute_us_puf_tax_detail_support",

packages/populace-build/src/populace/build/us_runtime/asec_pool.py

@@ -177,6 +177,7 @@ def build_pooled_asec_unit_frame(
         tax_unit_mode=tax_unit_mode,
         strata=strata,
     )
+    frame = _attach_household_attributes(frame, ("state_fips",))
     return frame, pooled.metadata
 
 
@@ -279,6 +280,18 @@ def _remap_source_year(
     result["source_household_id"] = result["PH_SEQ"].to_numpy()
     result["source_person_id"] = _source_person_id(result)
     result["source_row_id"] = np.arange(len(result), dtype=np.int64)
+    if "GESTFIPS" in household:
+        state_by_household = (
+            household[["H_SEQ", "GESTFIPS"]]
+            .drop_duplicates("H_SEQ")
+            .set_index("H_SEQ")["GESTFIPS"]
+        )
+        result["state_fips"] = (
+            result["source_household_id"]
+            .map(state_by_household)
+            .astype("int64")
+            .to_numpy()
+        )
     result["PH_SEQ"] = result["PH_SEQ"].map(household_map).astype("int64")
     result["household_id"] = result["PH_SEQ"]
     result[ASEC_PERSON_WEIGHT_COLUMN] = (
@@ -287,6 +300,42 @@ def _remap_source_year(
     return result, next_offset
 
 
+def _attach_household_attributes(frame: Frame, columns: tuple[str, ...]) -> Frame:
+    person = frame.table("person")
+    available = [column for column in columns if column in person]
+    if not available:
+        return frame
+
+    household = frame.table("household").copy()
+    tables = {entity: frame.table(entity).copy() for entity in frame.entities}
+    for column in available:
+        values = person.groupby("person_household_id", sort=True)[column]
+        low = values.min()
+        high = values.max()
+        unequal = low.to_numpy() != high.to_numpy()
+        if unequal.any():
+            bad = low.index.to_numpy()[unequal][:5].tolist()
+            raise ValueError(
+                f"Household attribute {column!r} must be constant within "
+                f"household; household id(s) {bad} carry unequal values."
+            )
+        household[column] = (
+            pd.Series(household["household_id"])
+            .map(low)
+            .astype(person[column].dtype)
+            .to_numpy()
+        )
+        tables["person"] = tables["person"].drop(columns=[column])
+    tables["household"] = household
+    return Frame(
+        tables,
+        frame.schema,
+        {entity: frame.weights_for(entity) for entity in frame.weighted_entities},
+        frame.strata,
+        mass_log=frame.mass_log,
+    )
+
+
 def _globalize_source_unit_column(person: pd.DataFrame, column: str) -> pd.DataFrame:
     if column not in person.columns:
         return person