Skip to content

FFIEC NIC / NPW Institutional-Structure Data Guide

Overview

The National Information Center (NIC) is the Federal Reserve's central repository of structure and identifier data for every banking organization the Federal Reserve and the other federal banking agencies supervise. Where the FFIEC financial collections (Call Reports, FR Y-9C, FR Y-14, etc. — see COLLECTIONS_CATALOG.md) tell you what a bank reported, the NIC data tells you what the entity is, where it sits in a holding-company structure, and how it was born, merged, or failed.

The Board of Governors publishes a non-confidential extract of the NIC database as the NPW (National Information Center — "NPW" / institution-search) Bulk Data Download at:

The data is officially non-confidential public data. The first public iteration covers the Attributes, Relationships, and Transformations tables documented here; further detail is available via FOIA.

Authoritative field source: Board of Governors of the Federal Reserve System, Bulk Data Download — Data Dictionary and Reference Guide, Document Version 2.0, Rev. 4/25, prepared by the National Information Center. The machine-readable schemas and code lists in this repository (csv/NIC_*.csv) are transcribed verbatim from that PDF.

The Five Bulk Files

The download is delivered as five pipe/CSV files. Record counts below are from the FFIEC release and confirmed by on-disk row counts (wc -l minus header):

File Records Description
CSV_ATTRIBUTES_ACTIVE.CSV 61,080 Currently open & active institutions (current characteristics)
CSV_ATTRIBUTES_CLOSED.CSV 160,494 Last instance of closed / failed institutions (historical)
CSV_ATTRIBUTES_BRANCHES.CSV 172,819 Last instance of branches whose head office is Active or Closed
CSV_RELATIONSHIPS.CSV 284,934 Ownership-relationship history between entities
CSV_TRANSFORMATIONS.CSV 58,749 Mergers, splits, sales, failures
TOTAL 738,076 —

Table-family roles:

  • Attributes (3 files) — every entity has exactly one Attributes row carrying its legal/common descriptors, classification, regulatory status, geography, identifiers, and lifecycle dates. The three files split by status: Active (open & active), Closed (last instance of a closed or failed institution), and Branches (last instance of branches under an Active or Closed head office). Full schema: csv/NIC_ATTRIBUTES_SCHEMA.csv.
  • Relationships — the time-series history of ownership/control between two entities (parent → offspring). Full schema: csv/NIC_RELATIONSHIPS_SCHEMA.csv.
  • Transformations — the event log of mergers, purchases & assumptions, splits, sales, and failures. Full schema: csv/NIC_TRANSFORMATIONS_SCHEMA.csv.

All code lists for every coded field across all three tables are in csv/NIC_CODE_LISTS.csv (40 code lists, 274 values).

Update cadence: The Federal Reserve refreshes the public NPW bulk extract periodically (the file set is regenerated from the live NIC database; many derived fields such as PRIM_FED_REG are recomputed nightly inside NIC). Treat each download as a point-in-time snapshot and re-pull for the latest structure; the dictionary version (here v2.0, Rev. 4/25) governs the schema.

The Entity Model

ID_RSSD — the master key

Every entity in NIC has exactly one ID_RSSD ("RSSD ID"): a unique one-up number with a check digit, assigned by the RSSD software when the entity is first entered.

  • It carries no inherent attribute information.
  • It never changes and is never reused.
  • It belongs to one entity for that entity's entire life cycle and persists after liquidation or closure.

This makes ID_RSSD the universal join key across all NIC tables and the bridge to the FFIEC financial collections (the Call Report / FR Y-9 / FR Y-14 panels all key on RSSD). See IDENTIFIERS.md for how RSSD relates to FDIC Cert, OCC charter, NCUA, LEI, CUSIP, Tax ID/EIN, and ABA routing numbers (and which of those are reusable — RSSD is not; ABA routing numbers are reusable across non-concurrent entities).

Primary keys per table

Table Primary key
Attributes ID_RSSD + D_DT_START
Relationships ID_RSSD_PARENT + ID_RSSD_OFFSPRING + D_DT_START + RELN_LVL
Transformations ID_RSSD_PRECESSOR + ID_RSSD_SUCCESSOR + D_DT_TRANS

Spelling caution: the dictionary prose says "predecessor," but the actual Transformations column name is ID_RSSD_PRECESSOR (the Fed's own typo). Use the misspelled form in queries.

Date-field convention

Every integer date field DT_* (format YYYYMMDD) has a paired DB2-DATETIME field D_DT_* (format 'YYYY-MM-DD HH:MM:SS', must be quoted in SQL). The sentinel 99991231 means a non-terminated record/entity (i.e., still current). The Attributes and Relationships rows are valid over a [DT_START, DT_END] window; filter DT_END = 99991231 to get the current state.

The parent → offspring relationship model

Relationships rows connect a Parent (ID_RSSD_PARENT, the controlling entity) to an Offspring (ID_RSSD_OFFSPRING, the owned/controlled entity). The key qualifiers:

  • CTRL_IND — whether the parent controls the offspring (1 = controlled, 2 = non-controlled). For regulated relationships (REG_IND = 1) control means legal authority: ≥25% of voting securities, control over a majority of directors, or a controlling influence per Reg Y §225.2(e) / 12 CFR 574.
  • RELN_LVL — relationship level: 1 = Direct (parent is the immediate holder), 2 = Indirect (an intervening company ties the offspring to the top-tier U.S. BHC or FBO), 3 = 2G3 (repealed 1996, no longer reportable), 4 = Debt Previously Contracted.
  • PCT_EQUITY / PCT_EQUITY_BRACKET — the parent's percent of equity/voting control. Exact for BHC/bank/FBO direct holders; reported as a bracket (100*, 80 - <100, >50 - <80, 25 - 50, < 25, 0) for non-banking offspring.
  • EQUITY_IND / REG_IND / FC_IND — basis of control, whether the relationship is regulated, and whether the offspring is financially consolidated in the parent.

Because both direct and indirect rows are stored, you can build a complete holding-company tree: indirect rows tie every controlled entity to the top-tier parent, while direct rows give you the immediate ownership edges.

The transformation / merger-event model

Transformations is an event log. Each row links a Predecessor (ID_RSSD_PRECESSOR — in a merger, the non-survivor) to a Successor (ID_RSSD_SUCCESSOR — the survivor) on an effective date DT_TRANS, classified by TRNSFM_CD:

TRNSFM_CD Event
1 Charter Discontinued (Merger or Purchase & Assumption); one charter ends, no failure
5 Split (predecessor transfers 40–94% of assets to newly-formed successor(s); both continue)
7 Sale of Assets (40–94% to an existing successor; both continue)
9 Charter Retained (≥95% of assets transferred; charter continues under a new ID_RSSD)
50 Failure, Government Assistance Provided (predecessor fails and ceases; disposition by FDIC/RTC/NCUA/other)

For non-failure mergers where the survivor is a commercial bank, non-deposit trust company, or industrial bank (CHTR_TYPE_CD 200/250/340), ACCT_METHOD records pooling-of-interests (1) vs purchase/acquisition (2).

Worked Examples

A. Trace a BHC's subsidiaries

  1. Find the holding company's ID_RSSD in Attributes (e.g. by NM_LGL), confirming BHC_IND = 1 (or FHC_IND = 1 for a financial holding company).
  2. In Relationships, select rows where ID_RSSD_PARENT = <the BHC's RSSD> and DT_END = 99991231 (current) and CTRL_IND = 1 (controlled).
  3. Use RELN_LVL = 1 for the immediate (direct) subsidiaries.
  4. Use RELN_LVL = 2 to pull every entity indirectly controlled anywhere in the structure tree (because indirect rows tie each entity to the top-tier parent).
  5. Join each ID_RSSD_OFFSPRING back to Attributes on ID_RSSD to get the subsidiary's name (NM_LGL), charter type (CHTR_TYPE_CD), entity type (ENTITY_TYPE), and regulator (PRIM_FED_REG).
  6. To count just the U.S. banking subsidiaries, restrict offspring to BROAD_REG_CD = 1 — this is exactly how NIC derives the Attributes field BANK_CNT.
-- Current banking subsidiaries of a given BHC
SELECT a.ID_RSSD, a.NM_LGL, a.CHTR_TYPE_CD, a.PRIM_FED_REG
FROM relationships r
JOIN attributes a ON a.ID_RSSD = r.ID_RSSD_OFFSPRING
WHERE r.ID_RSSD_PARENT = :bhc_rssd
  AND r.DT_END = 99991231
  AND r.CTRL_IND = 1
  AND a.BROAD_REG_CD = 1;

B. Identify a bank failure

There are two complementary signals; use both for completeness:

  • Attributes (REASON_TERM_CD) — on the head-office row in CSV_ATTRIBUTES_CLOSED.CSV: 4 = failure, entity continues to exist (resolved by FDIC/RTC/NCUA/State/other, no Transformation row) and 5 = failure, entity ceases to exist (a Transformation row may or may not be present). So REASON_TERM_CD IN (4, 5) flags failed institutions. (Contrast 1 = voluntary liquidation and 2 = closure, which are not failures.)
  • Transformations (TRNSFM_CD = 50) — Failure, Government Assistance Provided: the predecessor failed and ceased, with disposition arranged by a resolution authority.
-- Failed institutions
SELECT ID_RSSD, NM_LGL, DT_EXIST_TERM, REASON_TERM_CD
FROM attributes_closed
WHERE REASON_TERM_CD IN (4, 5);

-- Government-assisted failures (event view)
SELECT ID_RSSD_PRECESSOR, ID_RSSD_SUCCESSOR, DT_TRANS
FROM transformations
WHERE TRNSFM_CD = 50;

Link the resulting RSSDs to FDIC failure records via ID_FDIC_CERT, and to historical bank panels (e.g. pre-1990 series) via an RSSD crosswalk. Note that the Fed's RSSD system generally does not cover institutions closed before ~1960, so very old failures will not have an ID_RSSD.

Field & Code Reference

The complete, transcribed-verbatim reference lives in the machine-readable CSVs:

File Contents
csv/NIC_ATTRIBUTES_SCHEMA.csv All 74 Attributes columns (identifiers, classification, regulatory status, geography, names, dates, termination)
csv/NIC_RELATIONSHIPS_SCHEMA.csv All 22 Relationships columns
csv/NIC_TRANSFORMATIONS_SCHEMA.csv All 6 Transformations columns
csv/NIC_CODE_LISTS.csv 40 code lists, 274 values (ENTITY_TYPE, CHTR_TYPE_CD, BNK_TYPE_ANALYS_CD, BROAD_REG_CD, ORG_TYPE_CD, EST_TYPE_CD, INSUR_PRI_CD, PRIM_FED_REG, REASON_TERM_CD, TRNSFM_CD, RELN_LVL, etc.)

A few high-traffic code lists are summarized below; the CSV is authoritative.

CHTR_TYPE_CD (Charter Type): 200 Commercial Bank · 250 Non-deposit Trust Co · 300 Savings Bank · 310 Savings & Loan Assoc · 320 Cooperative Bank · 330 Credit Union · 340 Industrial Bank · 400 Edge/Agreement Corp · 500 Holding Company only · 700 Securities Broker/Dealer · 720 Other Non-Depository · 110 Government Agency/GSE · 550 Insurance · 610 ESOP/ESOT · 710 Utility/Co-generator · 0 N/A (branches).

PRIM_FED_REG (Primary Federal Regulator): FRS · OCC · FDIC · NCUA · FCA · FHFA · OTS (valid only through 2011-07-21). Can change over an entity's history; null for branches/non-bank subs; derived nightly.

ENTITY_TYPE: 44 derived codes (NAT, SMB, NMB, FSB, SSB, SAL, FCU, SCU, BHC, FHD, FHF, SLHC, IHC, FBO, FBH, FBK, DBR, IBR, EDB, EDI, etc.) — see CSV.

Verification note: the source-facts digest used during this rebuild labeled ENTITY_TYPE as "49 codes," but the NPW PDF (and the digest's own enumeration) list exactly 44 codes (AGB … USB). This guide and the CSV use the PDF-verified 44.

Relationship to Other Data

  • FFIEC financial collections — join on ID_RSSD to attach Call Report / FR Y-9C / FR Y-14 financials to NIC structure. See COLLECTIONS_CATALOG.md.
  • Identifier crosswalks — ID_FDIC_CERT → FDIC SDI/failure data; ID_OCC → OCC; ID_NCUA → credit unions; ID_LEI → GLEIF. See IDENTIFIERS.md.
  • Historical bank panels — pre-FFIEC institutions (closed before ~1960) have no RSSD and require name/state fuzzy crosswalking.

Data Sources

  • Bulk data & institution search: https://www.ffiec.gov/NPW
  • Authoritative dictionary: Board of Governors / NIC, Bulk Data Download — Data Dictionary and Reference Guide, Document Version 2.0, Rev. 4/25 (the source of every field and code in csv/NIC_*.csv).

Notes

  • ID_RSSD is the single most important field in the entire dataset: stable, unique, never reused — the spine for linking structure to financials and to external identifiers.
  • Use DT_END = 99991231 to filter for current records in Attributes and Relationships.
  • Branches (EST_TYPE_CD > 1) point to their head office via ID_RSSD_HD_OFF and carry 0/N/A in many entity-level fields (CHTR_TYPE_CD, ORG_TYPE_CD, BHC_IND).
  • Failure analysis should combine Attributes REASON_TERM_CD IN (4,5) with Transformations TRNSFM_CD = 50 — neither alone is complete.