FHLBank Membership Roster & Delta Monitor

FHLBank System membership roster and quarter-over-quarter delta tracker, sourced directly from FHFA's own published member file (fhlb_members_q<N><YYYY>.xlsx). Covers all 11 FHLBank districts (Atlanta, Boston, Chicago, Cincinnati, Dallas, Des Moines, Indianapolis, New York, Pittsburgh, San Francisco, Topeka) and every member type FHFA tracks: commercial banks, savings institutions, credit unions, insurance companies and CDFIs.

Every row carries the standard cross-reference join keys published on the source file -- cert (FDIC certificate number), fed_id (Federal Reserve ID), ncua_id (NCUA charter number) and naic_id (insurance identifier) -- so this is a clean join target for fdic-ncua-health-rollup (bank/CU financial health) and any other CERT- or NCUA-keyed data source.

Who this is for

• Correspondent banking / capital markets desks tracking which institutions carry FHLBank advance access, or watching for new entrants and exits in a district.
• KYB / vendor-onboarding teams who need an authoritative FHFA-published cross-walk from an institution name to its FDIC CERT / Federal Reserve ID / NCUA charter number.
• Fintech and correspondent-lending BD teams building a target list of FHLBank members by state, charter type or district.
• Bank/credit-union competitive intelligence teams tracking membership changes -- mergers, charter conversions, new entrants -- as a leading signal.

Two modes

Roster mode (default)

Returns a filtered listing of current FHLBank members from the most recently posted quarterly file (FHFA hosts several recent quarters simultaneously; this actor automatically walks back up to 4 quarters if the very latest one is not yet posted).

{ "mode": "roster", "states": ["CO"], "maxResults": 500 }

Every roster row also carries is_new_since_last_run -- whether this member was absent from a cross-run baseline stored in your account's key-value store. On your very first run this is null for every row (there is nothing to compare against yet), never true.

Delta mode

Resolves the prior quarter's file too (a second, real FHFA file -- not a guess) and returns only the join/exit/attribute_change EVENTS between the two:

{ "mode": "delta", "eventTypes": ["join", "exit"], "states": ["TX"] }

• join -- a new FHFA_ID present in the current quarter, absent from the prior one.
• exit -- an FHFA_ID present in the prior quarter, absent from the current one (the member left the open-membership roster; FHFA's file does not say why).
• attribute_change -- present in both quarters, but one or more of member_name, city, state, zip, district, mem_type, char_type, cert, fed_id, ncua_id or naic_id differs. changed_fields names which; previous_values holds the old value for each.

Because FHFA hosts multiple quarters' files at once, delta mode computes a real two-file comparison on your very first run -- it does not need to wait for a second scheduled run the way a single-file source would.

Example output row (delta mode, a real "exit" event)

{
  "mode": "delta",
  "fhfa_id": "54201",
  "member_name": "Comerica Bank",
  "city": "Dallas",
  "state": "TX",
  "district": "Dallas",
  "mem_type": "Commercial Bank",
  "char_type": "State",
  "cert": "00983",
  "fed_id": "0060143",
  "mem_date": "2008-02-14",
  "member_tenure_days": 6754,
  "delta_event_type": "exit",
  "delta_current_quarter": "q22026",
  "delta_prior_quarter": "q12026",
  "file_quarter": "q12026"
}

Input fields

Field	Type	Description
mode	string	"roster" (default) or "delta".
states	array	US state/territory filter -- 2-letter code or full name, e.g. "CO" or "Colorado".
charterTypes	array	Filter by Federal, National, State, or None.
districts	array	Filter by FHLBank district name.
memberSince	string	YYYY-MM-DD only -- see "Date format" below. Anything else FAILS the run.
quarter	string	Pin an exact quarter, e.g. "q22026". Leave blank for the latest posted.
eventTypes	array	Delta mode only -- subset of join, exit, attribute_change.
maxResults	integer	Row/event cap for this run (default 1000).

Date format -- memberSince accepts YYYY-MM-DD and nothing else

Accepted: 2020-01-15, and the unpadded 2020-1-1 (padded for you). Blank or omitted means no cutoff at all, which is not an error. Every other form fails the run before any file is downloaded, and bills nothing.

memberSince is string-compared against membership dates already normalised to ISO, so a wrong format does not error -- it silently returns the wrong row set and labels it a successful filtered answer. Measured against a 3-row fixture (membership dates 2018-06-15, 2020-03-01, 2022-02-10) where the true answer for a 2020-01-01 cutoff is 2 of 3:

You pass	Returned before this fix	Why
01/15/2020 (US)	3 of 3 -- the whole set, as if no cutoff were applied	"01/..." sorts below every "20xx-.." date, so the comparison is true for every row
Jan 2020	0 of 3 -- a confident "no members joined since Jan 2020"	"Jan..." sorts above every "2..." date, so the comparison is false for every row

Two more worth knowing, both now rejected: 2020-1-1 returned 1 of 3 (unpadded, so it sorts below every zero-padded month and silently drops January through September), and 2020-13-45 returned 1 of 3 -- a plausible-looking subset from an impossible date. Slash forms are refused on purpose rather than guessed: 05/06/2025 is genuinely ambiguous between US (May 6) and EU (June 5) order, and guessing trades a loud, fixable error for a silent one-month window shift.

Output fields

Field	Description
mode	roster or delta.
fhfa_id	FHFA member ID, zero-padded to 5 digits. The delta join key.
district, member_name, city, state, zip	As published by FHFA.
mem_type, char_type	Closed-vocabulary institution/charter type.
cert	FDIC certificate number, when FDIC-insured. Join key to FDIC-sourced actors.
fed_id	Federal Reserve ID, zero-padded to 7 digits.
ncua_id	NCUA charter number. See ncua_id_looks_placeholder.
ncua_id_looks_placeholder	True for the rare all-zero "00000" value (very likely a not-yet-assigned placeholder, never guessed silently).
naic_id	Insurance-company identifier.
appr_date, appr_date_basis	Approval date + how it was derived (parsed/blank/unparseable/absent).
mem_date, mem_date_basis	Membership date (century-pivoted from FHFA's 2-digit-year format) + basis.
member_tenure_days	Days since mem_date.
file_quarter	Which quarter's file this row's member data is from.
is_new_since_last_run, delta_baseline_status	Roster mode: cross-run "new" flag + how it was determined.
delta_event_type, delta_current_quarter, delta_prior_quarter	Delta mode: which kind of event, and the two quarters compared.
changed_fields, previous_values	Delta mode, attribute_change rows only.
delta_transition_already_reported	Delta mode: whether this exact quarter pair was already reported by an earlier run (informational; the delta is always recomputed fresh).
source_url, source_resolution_method, source_file_last_modified, source_row_count	Provenance: exactly which file, how it was found, and its vintage.
matched_rows_total, results_truncated	So a capped answer is never mistaken for a complete one.
monitored_at	UTC timestamp of this run.

Use as an MCP tool

This actor is callable by AI agents (Claude, Cursor, etc.) via mcp.apify.com -- the input/output schemas above are written to be self-describing to an agent choosing a tool. A typical agent prompt: "Which FHLBank Dallas members joined or left between Q1 and Q2 2026?" maps directly to {"mode":"delta","districts":["Dallas"],"eventTypes":["join","exit"]}.

FAQ

Does this cover every FHLBank district? Yes, all 11.

What does "exit" mean -- did the institution fail? No. FHFA's open-membership file does not record a reason for a member's departure (merger, acquisition, voluntary withdrawal, or termination are all possible). exit means the FHFA_ID is no longer in the open-membership roster as of the current quarter; it is a fact, not an inference.

How current is the data? FHFA posts a new quarterly file on a multi-week lag after each quarter ends (verified: the April-June 2026 file posted in mid-August 2026). Every row carries source_file_last_modified and file_quarter so you always know the vintage.

Can I get the same member cross-referenced against FDIC/NCUA financial data? Yes -- join on cert (banks/thrifts) or ncua_id (credit unions) to fdic-ncua-health-rollup.

Pricing

$12 per 1,000 result rows (Pay-Per-Event, Result event only), with the standard graduated volume discount schedule.