SOI Table 1.4 income-source + Table 2.1 mortgage-total measures (populace#299 Build H)

packages/irs_soi/table_1_4/source_package.yaml

@@ -611,3 +611,173 @@ record_sets:
           national calibration source for self_employed_pension_contribution_ald.
         legal_vintage: tax_year_{year}
         value_scale: 1000
+      - measure_id: capital_gain_distributions_returns
+        label: Returns with capital gain distributions reported on Form 1040
+        ordinal: 17
+        column: AJ
+        column_by_year:
+          2021: X
+          2022: AJ
+          2023: AJ
+        concept: irs_soi.returns_with_capital_gain_distributions_1040
+        unit: count
+        aggregation: sum
+      - measure_id: capital_gain_distributions_amount
+        label: Capital gain distributions reported on Form 1040
+        ordinal: 18
+        column: AK
+        column_by_year:
+          2021: Y
+          2022: AK
+          2023: AK
+        concept: irs_soi.capital_gain_distributions_1040
+        unit: usd
+        aggregation: sum
+        source_concept: irs_soi.capital_gain_distributions_1040
+        concept_relation: broad_match
+        concept_authority: ledger-us
+        concept_evidence_url: https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-complete-report-publication-1304-basic-tables-part-1
+        concept_evidence_notes: >
+          IRS SOI Table 1.4 "Capital gain distributions reported on Form 1040"
+          is the capital-gain component filers report directly on Form 1040
+          without a Schedule D. PolicyEngine-US carries this as the
+          non_sch_d_capital_gains input (the non-Schedule-D leg of the
+          net-capital-gains split), so Ledger treats the SOI line as a broad
+          match to that component.
+        legal_vintage: tax_year_{year}
+        value_scale: 1000
+      - measure_id: estate_trust_net_income_returns
+        label: Returns with estate and trust net income
+        ordinal: 19
+        column: BX
+        column_by_year:
+          2021: BL
+          2022: BX
+          2023: BX
+        concept: irs_soi.returns_with_estate_and_trust_net_income
+        unit: count
+        aggregation: sum
+      - measure_id: estate_trust_net_income_amount
+        label: Estate and trust net income
+        ordinal: 20
+        column: BY
+        column_by_year:
+          2021: BM
+          2022: BY
+          2023: BY
+        concept: irs_soi.estate_and_trust_net_income
+        unit: usd
+        aggregation: sum
+        source_concept: irs_soi.estate_and_trust_net_income
+        concept_relation: broad_match
+        concept_authority: ledger-us
+        concept_evidence_url: https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-complete-report-publication-1304-basic-tables-part-1
+        concept_evidence_notes: >
+          IRS SOI Table 1.4 "Estate and trust — Net income" is the positive leg
+          of Schedule E Part III estate and trust income. PolicyEngine-US carries
+          the signed estate_income input; Ledger treats this SOI line as a broad
+          match to the positive component of that input.
+        legal_vintage: tax_year_{year}
+        value_scale: 1000
+      - measure_id: estate_trust_net_loss_returns
+        label: Returns with estate and trust net loss
+        ordinal: 21
+        column: BZ
+        column_by_year:
+          2021: BN
+          2022: BZ
+          2023: BZ
+        concept: irs_soi.returns_with_estate_and_trust_net_loss
+        unit: count
+        aggregation: sum
+      - measure_id: estate_trust_net_loss_amount
+        label: Estate and trust net loss
+        ordinal: 22
+        column: CA
+        column_by_year:
+          2021: BO
+          2022: CA
+          2023: CA
+        concept: irs_soi.estate_and_trust_net_loss
+        unit: usd
+        aggregation: sum
+        source_concept: irs_soi.estate_and_trust_net_loss
+        concept_relation: broad_match
+        concept_authority: ledger-us
+        concept_evidence_url: https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-complete-report-publication-1304-basic-tables-part-1
+        concept_evidence_notes: >
+          IRS SOI Table 1.4 "Estate and trust — Net loss" is the negative leg of
+          Schedule E Part III estate and trust income (reported as a positive
+          magnitude). PolicyEngine-US carries the signed estate_income input;
+          Ledger treats this SOI line as a broad match to the magnitude of the
+          negative component of that input.
+        legal_vintage: tax_year_{year}
+        value_scale: 1000
+      - measure_id: other_income_net_income_returns
+        label: Returns with other income (net income)
+        ordinal: 23
+        column: CN
+        column_by_year:
+          2021: CB
+          2022: CN
+          2023: CN
+        concept: irs_soi.returns_with_other_income_net_income
+        unit: count
+        aggregation: sum
+      - measure_id: other_income_net_income_amount
+        label: Other income (net income)
+        ordinal: 24
+        column: CO
+        column_by_year:
+          2021: CC
+          2022: CO
+          2023: CO
+        concept: irs_soi.other_income_net_income
+        unit: usd
+        aggregation: sum
+        source_concept: irs_soi.other_income_net_income
+        concept_relation: broad_match
+        concept_authority: ledger-us
+        concept_evidence_url: https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-complete-report-publication-1304-basic-tables-part-1
+        concept_evidence_notes: >
+          IRS SOI Table 1.4 "Other income — Net income" is the positive leg of
+          the Form 1040 Schedule 1 other-income line (miscellaneous income not
+          reported elsewhere). PolicyEngine-US carries the signed
+          miscellaneous_income input; Ledger treats this SOI line as a broad
+          match to the positive component of that input.
+        legal_vintage: tax_year_{year}
+        value_scale: 1000
+      - measure_id: other_income_net_loss_returns
+        label: Returns with other income (net loss)
+        ordinal: 25
+        column: CP
+        column_by_year:
+          2021: CD
+          2022: CP
+          2023: CP
+        concept: irs_soi.returns_with_other_income_net_loss
+        unit: count
+        aggregation: sum
+      - measure_id: other_income_net_loss_amount
+        label: Other income (net loss)
+        ordinal: 26
+        column: CQ
+        column_by_year:
+          2021: CE
+          2022: CQ
+          2023: CQ
+        concept: irs_soi.other_income_net_loss
+        unit: usd
+        aggregation: sum
+        source_concept: irs_soi.other_income_net_loss
+        concept_relation: broad_match
+        concept_authority: ledger-us
+        concept_evidence_url: https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns-complete-report-publication-1304-basic-tables-part-1
+        concept_evidence_notes: >
+          IRS SOI Table 1.4 "Other income — Net loss" is the negative leg of the
+          Form 1040 Schedule 1 other-income line (reported as a positive
+          magnitude). PolicyEngine-US carries the signed miscellaneous_income
+          input; Ledger treats this SOI line as a broad match to the magnitude of
+          the negative component of that input.
+        legal_vintage: tax_year_{year}
+        value_scale: 1000

packages/irs_soi/table_2_1/source_package.yaml

@@ -566,3 +566,34 @@ record_sets:
         aggregation: sum
         value_scale: 1000
         expected_cell_type: number
+      - measure_id: home_mortgage_interest_returns
+        label: Returns with total home mortgage interest deduction
+        ordinal: 37
+        column: CW
+        column_by_year:
+          2021: CK
+          2022: CW
+          2023: CW
+        source_column_id: home_mortgage_interest_returns
+        expected_column_header_row: 8
+        expected_column_header: Number of returns
+        concept: irs_soi.returns_with_total_home_mortgage_interest_deduction
+        unit: count
+        aggregation: sum
+        expected_cell_type: number
+      - measure_id: home_mortgage_interest_amount
+        label: Total home mortgage interest deduction amount
+        ordinal: 38
+        column: CX
+        column_by_year:
+          2021: CL
+          2022: CX
+          2023: CX
+        source_column_id: home_mortgage_interest_amount
+        expected_column_header_row: 8
+        expected_column_header: Amount
+        concept: irs_soi.total_home_mortgage_interest_deduction
+        unit: usd
+        aggregation: sum
+        value_scale: 1000
+        expected_cell_type: number

tests/test_ledger_bundle.py

@@ -27,7 +27,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "aggregate_duplicate_key_count": 0,
         "entity_count": 8,
         "error_count": 0,
-        "fact_count": 38971,
+        "fact_count": 39173,
         "geography_count": 1053,
         "period_count": 16,
         "semantic_duplicate_key_count": 12,
@@ -36,7 +36,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "source_package_count": 58,
         "warning_count": 1,
     }
-    assert len(rows) == 38971
+    assert len(rows) == 39173
     assert rows[0]["aggregate_fact_key"].startswith("ledger.aggregate_fact.v2:")
     assert rows[0]["semantic_fact_key"].startswith("ledger.semantic_fact.v2:")
     assert source_packages["source_package_count"] == 58
@@ -52,7 +52,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "cms-aca-oep-state-level-2025",
         "jct-tax-expenditures-2024",
     ]
-    assert coverage["fact_count"] == 38971
+    assert coverage["fact_count"] == 39173
     assert coverage["counts"]["by_source"] == {
         "bea": 445,
         "bfp_economic_outlook": 5,
@@ -69,7 +69,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "hhs_acf_tanf": 110,
         "ici": 12,
         "hmrc": 193,
-        "irs_soi": 33535,
+        "irs_soi": 33737,
         "jrc_euromod_be": 18,
         "kff": 52,
         "nbb_national_accounts": 1,
@@ -211,15 +211,15 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "month:2024-12": 260,
         "month:2025-12": 255,
         "tax_year:2022": 5886,
-        "tax_year:2023": 28187,
+        "tax_year:2023": 28389,
         "tax_year:2024": 34,
     }
     assert coverage["counts"]["by_geography"]["country:BE"] == 31
     assert coverage["counts"]["by_geography"]["nuts1:BE1"] == 6
     assert coverage["counts"]["by_geography"]["nuts1:BE2"] == 17
     assert coverage["counts"]["by_geography"]["nuts1:BE3"] == 6
     assert coverage["counts"]["by_geography"]["commune:11002"] == 1
-    assert coverage["counts"]["by_geography"]["country:0100000US"] == 1861
+    assert coverage["counts"]["by_geography"]["country:0100000US"] == 2063
     assert coverage["counts"]["by_geography"]["state:0400000US06"] == 217
     assert (
         coverage["counts"]["by_geography"]["congressional_district:5001700US0601"] == 56
@@ -234,7 +234,7 @@ def test_build_bundle_writes_merged_consumer_contract(tmp_path):
         "institutional_sector": 13,
         "pension_plan": 2,
         "person": 3682,
-        "tax_unit": 33541,
+        "tax_unit": 33743,
     }
     assert not coverage["duplicates"]["aggregate_fact_keys"]
     assert len(coverage["duplicates"]["semantic_fact_keys"]) == 12

tests/test_ledger_soi_targets.py

@@ -127,7 +127,7 @@ def test_build_soi_table_1_4_wage_facts_from_packaged_source():
     }
 
     assert validate_facts(facts).valid
-    assert len(facts) == 340
+    assert len(facts) == 540
     assert (
         facts_by_concept_and_range[("irs_soi.returns_with_total_wages", "all")].value
         == 128_591_050

tests/test_ledger_source_package.py

@@ -267,7 +267,7 @@ def test_source_package_path_builds_valid_soi_table_1_4_facts():
 
     assert package.package_id == "soi-table-1-4"
     assert len(cells) == 8109
-    assert len(facts) == 340
+    assert len(facts) == 540
     assert validate_facts(facts).valid
     assert facts[0].source.source_table == "Publication 1304 Table 1.4"
     assert (
@@ -580,8 +580,8 @@ def test_national_soi_source_package_aliases_validate_fixture_counts():
         "soi-table-2-1": {
             "record_set_count": 1,
             "row_count": 1,
-            "measure_count": 37,
-            "source_record_count": 37,
+            "measure_count": 39,
+            "source_record_count": 39,
             "source_region_count": 1,
         },
         "soi-table-2-5": {
@@ -681,7 +681,7 @@ def test_soi_table_2_1_package_builds_itemized_deduction_details():
     assert validate_source_rows(rows).valid
     assert validate_source_cells(cells).valid
     assert validate_facts(facts).valid
-    assert len(facts) == 37
+    assert len(facts) == 39
 
     charitable = values_by_record[
         "irs_soi.ty2023.table_2_1.itemized_all_returns.all.charitable_amount"

tests/test_ledger_suite.py

@@ -137,13 +137,13 @@ def test_build_source_suite_supports_soi_table_1_4(tmp_path):
     assert summary["counts"] == {
         "artifact_count": 1,
         "agent_acceptance_error_count": 0,
-        "concept_alignment_count": 2,
-        "constraint_count": 612,
-        "consumer_fact_count": 340,
-        "fact_count": 340,
+        "concept_alignment_count": 7,
+        "constraint_count": 972,
+        "consumer_fact_count": 540,
+        "fact_count": 540,
         "lineage_coverage": 1.0,
         "source_cell_count": 8109,
-        "source_record_count": 340,
+        "source_record_count": 540,
         "source_region_count": 1,
         "source_row_count": 0,
     }