Extract data from FATCA and CRS filings

Banks and investment firms have to tell tax authorities about the accounts they hold for foreign residents, and the FATCA and CRS regimes are how they do it. FATCA is the US rule that identifies US persons holding accounts abroad; CRS is the OECD standard that more than 100 jurisdictions use to exchange the same kind of account information with each other. A compliance team at a financial institution produces a filing that pairs the reporting entity with a list of reportable accounts, each carrying the account holder's identity, tax residency, classification, and balance. What the filing turns on is identity and jurisdiction, not just balances. Each account holder is tied to one or more tax residencies by an ISO 3166 country code and a taxpayer identification number issued in that jurisdiction, and a mismatch or a missing TIN is exactly what a regulator looks for. The reporting institution carries a Global Intermediary Identification Number for FATCA, a BIC, and a home tax ID. An entity account adds controlling persons, each with their own residency and TIN, and the FATCA classification such as Passive NFE together with the CRS classification decide whether an account is reportable at all. High-value accounts trigger enhanced due diligence, and the account open and close dates set the reporting window. Talonic reads the FATCA/CRS filing and returns the reporting entity, the account holders, and the controlling persons with their residencies, identifiers, classifications, and account balances as structured fields. A filing from Meridian Bank AG, GIIN identified, for the period ending 2025-12-31, reports an account held by Lena Fischer, an Austrian tax resident with a German bank account, classified as a CRS reportable account with a balance of 1,284,500 EUR flagged high-value, and the record loads into a reporting system with the jurisdiction and classification fields intact, so a compliance analyst validates residency and reportability from data rather than rereading the form.

What gets extracted from FATCA and CRS filings

How extraction works for FATCA and CRS filings

FATCA and CRS filings are produced from core banking systems and reach a compliance team as PDFs, XML exports rendered to print, and scanned schedules, with no single global layout since the regimes leave format to each jurisdiction. Talonic classifies the filing and maps it to the regulatory-filing schema in the Field Registry, which separates the reporting entity from the account holders and the controlling persons and keeps each account as its own record. Tax residencies are normalized to ISO 3166 country codes, taxpayer identification numbers are held per person and per jurisdiction, and the FATCA and CRS classifications that decide reportability are captured as their own fields. Account balances are typed as numbers in their ISO 4217 currency, the high-value flag is read where stated, and the reporting period end and the account open and close dates are parsed to ISO 8601. Every field returns with a confidence score and a pixel-region pointer that conforms to DIN SPEC 91491, so a compliance analyst can verify a residency or a classification against the source filing before it is exchanged.

Sample extraction

A CRS filing schedule for a single reportable account

{
  "document_number": "CRS-2026-DE-004821",
  "document_date": "2026-05-28",
  "reporting_period_end": "2025-12-31",
  "reporting_entity": {
    "name": "Meridian Bank AG",
    "giin": "S1A2B3.99999.SL.276",
    "tax_id": "DE-811234567"
  },
  "account_number": "DE89370400440532013000",
  "iban": "DE89370400440532013000",
  "bic": "MERIDEFFXXX",
  "account_type": "depository",
  "account_status": "active",
  "account_opened_date": "2019-03-11",
  "account_holder": {
    "name": "Lena Fischer",
    "tax_residency": "AT",
    "tax_id": "AT-1234567890"
  },
  "fatca_classification": "Non-US Individual",
  "crs_classification": "Reportable Person",
  "reportable_account": true,
  "total_amount": 1284500,
  "currency": "EUR",
  "high_value_account_flag": true,
  "financial_institution_type": "bank",
  "kyc_completion_date": "2025-11-04",
  "next_review_date": "2028-11-04"
}

Frequently asked