Add UK firm calibration targets

arch/core.py

@@ -43,6 +43,7 @@
     "government",
     "dwelling",
     "institutional_sector",
+    "firm",
 }
 ALLOWED_AGGREGATIONS = {
     "sum",

arch/source_package.py

@@ -157,9 +157,13 @@
     "ssa-ssi-table-7b1-2024": Path("ssa/ssi_table_7b1_2024"),
     "hhs-acf-tanf-caseload-2024": Path("hhs_acf/tanf_caseload_2024"),
     "hhs-acf-tanf-financial-2024": Path("hhs_acf/tanf_financial_2024"),
+    "hmrc-vat-firm-targets-2024-25": Path("hmrc/vat_firm_targets_2024_25"),
     "kff-marketplace-effectuated-enrollment": Path(
         "kff/marketplace_effectuated_enrollment"
     ),
+    "ons-uk-business-firm-targets-2025": Path(
+        "ons/uk_business_firm_targets_2025"
+    ),
     "usda-snap-fy69-to-current": Path("usda_snap/fy69_to_current"),
 }
 SOURCE_ARTIFACT_CACHE_ENV = "ARCH_SOURCE_ARTIFACT_CACHE_DIR"

arch/sources/cells.py

@@ -935,6 +935,8 @@ def _delimited_scalar(value: str) -> Scalar:
     stripped = value.strip()
     if not stripped:
         return None
+    if "_" in stripped:
+        return stripped
     numeric = stripped.replace("$", "").replace(",", "")
     if numeric.lstrip("-").isdigit():
         return int(numeric)

arch/sources/rows.py

@@ -906,6 +906,8 @@ def _delimited_scalar(value: str) -> Scalar:
     stripped = value.strip()
     if not stripped:
         return None
+    if "_" in stripped:
+        return stripped
     numeric = stripped.replace("$", "").replace(",", "")
     if numeric.lstrip("-").isdigit():
         return int(numeric)

db/data/hmrc/vat_firm_targets_2024_25/hmrc_vat_firm_targets_2024_25.csv

@@ -0,0 +1,18 @@
+target_group,band_id,band_label,value
+vat_registered_count_by_turnover,negative_or_zero,"Negative or zero annual turnover",211400
+vat_registered_count_by_turnover,1_to_threshold,"GBP 1 to VAT registration threshold annual turnover",683700
+vat_registered_count_by_turnover,threshold_to_150k,"VAT registration threshold to GBP 150,000 annual turnover",280400
+vat_registered_count_by_turnover,150k_to_300k,"GBP 150,000 to GBP 300,000 annual turnover",338600
+vat_registered_count_by_turnover,300k_to_500k,"GBP 300,000 to GBP 500,000 annual turnover",186700
+vat_registered_count_by_turnover,500k_to_1m,"GBP 500,000 to GBP 1,000,000 annual turnover",184100
+vat_registered_count_by_turnover,1m_to_10m,"GBP 1,000,000 to GBP 10,000,000 annual turnover",240500
+vat_registered_count_by_turnover,greater_than_10m,"Greater than GBP 10,000,000 annual turnover",45800
+vat_registered_count_by_turnover,unknown,"Unknown annual turnover",159300
+vat_liability_by_turnover,negative_or_zero,"Negative or zero annual turnover",-2910
+vat_liability_by_turnover,1_to_threshold,"GBP 1 to VAT registration threshold annual turnover",-300
+vat_liability_by_turnover,threshold_to_150k,"VAT registration threshold to GBP 150,000 annual turnover",2420
+vat_liability_by_turnover,150k_to_300k,"GBP 150,000 to GBP 300,000 annual turnover",4830
+vat_liability_by_turnover,300k_to_500k,"GBP 300,000 to GBP 500,000 annual turnover",3970
+vat_liability_by_turnover,500k_to_1m,"GBP 500,000 to GBP 1,000,000 annual turnover",6840
+vat_liability_by_turnover,1m_to_10m,"GBP 1,000,000 to GBP 10,000,000 annual turnover",29520
+vat_liability_by_turnover,greater_than_10m,"Greater than GBP 10,000,000 annual turnover",132800

db/data/hmrc/vat_firm_targets_2024_25/manifest.yaml

@@ -0,0 +1,23 @@
+source_id: hmrc-vat-firm-targets-2024-25
+source_name: HMRC Annual UK VAT Statistics 2024 to 2025
+publisher: HM Revenue and Customs
+source_page: https://www.gov.uk/government/statistics/value-added-tax-vat-annual-statistics
+files:
+  2024:
+    filename: hmrc_vat_firm_targets_2024_25.csv
+    source_url: https://www.gov.uk/government/statistics/value-added-tax-vat-annual-statistics
+    source_table: Annual UK VAT Statistics 2024 to 2025 VAT trader population and net VAT liability by turnover band
+    sha256: 8bea6fdc9a5e397950fbc5bc58b6f26586e02ff76622b21b949592520a7d6d79
+    size_bytes: 1675
+    storage:
+      r2:
+        provider: r2
+        bucket: arch-raw
+        key: raw/hmrc/hmrc-vat-firm-targets-2024-25/2024/8bea6fdc9a5e397950fbc5bc58b6f26586e02ff76622b21b949592520a7d6d79/hmrc_vat_firm_targets_2024_25.csv
+        uri: r2://arch-raw/raw/hmrc/hmrc-vat-firm-targets-2024-25/2024/8bea6fdc9a5e397950fbc5bc58b6f26586e02ff76622b21b949592520a7d6d79/hmrc_vat_firm_targets_2024_25.csv
+    source_urls:
+      - https://www.gov.uk/government/statistics/value-added-tax-vat-annual-statistics
+    notes: >-
+      Curated national extract from HMRC Annual UK VAT Statistics 2024 to 2025.
+      Net VAT liability rows retain source values in GBP millions and are scaled
+      to GBP by the Ledger source package.

db/data/ons/uk_business_firm_targets_2025/manifest.yaml

@@ -0,0 +1,22 @@
+source_id: ons-uk-business-firm-targets-2025
+source_name: ONS UK Business, Activity, Size and Location 2025
+publisher: Office for National Statistics
+source_page: https://www.ons.gov.uk/businessindustryandtrade/business/activitysizeandlocation/datasets/ukbusinessactivitysizeandlocation
+files:
+  2025:
+    filename: ons_uk_business_firm_targets_2025.csv
+    source_url: https://www.ons.gov.uk/businessindustryandtrade/business/activitysizeandlocation/datasets/ukbusinessactivitysizeandlocation
+    source_table: UK Business, Activity, Size and Location 2025 enterprise turnover and employment size bands
+    sha256: 929717d7cd58dbd615b8383462c3c9da961663f8c50a16a666c58a2d12df0cf3
+    size_bytes: 900
+    storage:
+      r2:
+        provider: r2
+        bucket: arch-raw
+        key: raw/ons/ons-uk-business-firm-targets-2025/2025/929717d7cd58dbd615b8383462c3c9da961663f8c50a16a666c58a2d12df0cf3/ons_uk_business_firm_targets_2025.csv
+        uri: r2://arch-raw/raw/ons/ons-uk-business-firm-targets-2025/2025/929717d7cd58dbd615b8383462c3c9da961663f8c50a16a666c58a2d12df0cf3/ons_uk_business_firm_targets_2025.csv
+    source_urls:
+      - https://www.ons.gov.uk/businessindustryandtrade/business/activitysizeandlocation/datasets/ukbusinessactivitysizeandlocation
+    notes: >-
+      Curated national extract from the ONS UK Business 2025 workbook. Rows
+      retain published enterprise counts by turnover and employment size band.

db/data/ons/uk_business_firm_targets_2025/ons_uk_business_firm_targets_2025.csv

@@ -0,0 +1,15 @@
+target_group,band_id,band_label,value
+turnover_band,0_49k,"GBP 0 to GBP 49,999 annual turnover",387285
+turnover_band,50_99k,"GBP 50,000 to GBP 99,999 annual turnover",525225
+turnover_band,100_249k,"GBP 100,000 to GBP 249,999 annual turnover",873705
+turnover_band,250_499k,"GBP 250,000 to GBP 499,999 annual turnover",390055
+turnover_band,500_999k,"GBP 500,000 to GBP 999,999 annual turnover",244225
+turnover_band,1000_4999k,"GBP 1,000,000 to GBP 4,999,999 annual turnover",232795
+turnover_band,5000k_plus,"GBP 5,000,000 or more annual turnover",81325
+employment_band,0_4,"0 to 4 employees",2137200
+employment_band,5_9,"5 to 9 employees",300645
+employment_band,10_19,"10 to 19 employees",156590
+employment_band,20_49,"20 to 49 employees",84595
+employment_band,50_99,"50 to 99 employees",29335
+employment_band,100_249,"100 to 249 employees",14835
+employment_band,250_plus,"250 or more employees",11415