FFIEC CDR Bulk Data Download

It is the population edition of U.S. bank regulation's most-filed document. The FFIEC Central Data Repository - the Federal Financial Institutions Examination Council's publishing arm - collects quarterly Call Reports from every commercial bank in the country and computes the Uniform Bank Performance Report from them, and the bulk products serve those filing populations whole: not one bank at a time, every bank at once.

The product shelf runs in two families. Call Report extracts - single period, balance sheet, income statement, and past due across four periods - carry what banks reported, on forms FFIEC 031 (banks with domestic and foreign offices), 041 and 051 (domestic-only banks, split by asset size). UBPR cuts - ratio single-period, ratio, rank and stats four-period - carry what the examiners computed from those reports: hundreds of ratios with peer-group ranks and percentiles beside them. Single-period products take one reporting-period end date; four-period products take a calendar year and return all four quarters.

For custody-bank analysis the headline item is Schedule RC-T on fiduciary and related services, the schedule holding the trust-asset and custody line items that separate true custody banks from the rest of the population - and it rides inside the same files as everything else. Files arrive tab-delimited (Excel compatible) or XBRL, zipped, with the caveat that the XBRL option has received only limited instance-validation testing. This page sits on Datadory's application software shelf because that is where analysts hunting structured external feeds actually look; the FFIEC source profile holds the institutional story.

Two spines, one key. The Call Report row carries a reported figure; the UBPR row carries a computed one; both hang off the same certificate number and quarter-end:

# Call Report extract - one institution, one schedule item, one quarter-end
Financial Institution Name : State Street Bank and Trust Company
FDIC Certificate Number    : 12345
Reporting Period End Date  : 2026-06-30
MDRM / Schedule-Item code  : RCT   (Schedule RC-T - fiduciary and related services)
Value                      : 41000000   (thousands of dollars)

# UBPR cut - same institution, calculated ratio in place of a reported amount
Financial Institution Name : State Street Bank and Trust Company
FDIC Certificate Number    : 12345
Reporting Period End Date  : 2026-06-30
UBPR Ratio / Rank          : 1.24

Read aloud: a bank files a number for the fiduciary-services schedule as of June 30, 2026, and the value lands in thousands of dollars with the schedule and line encoded in the RCT code. The UBPR side takes the same identity pair and attaches ratios, ranks and percentiles computed against the filer's peer group. Because both sides share the key, a custody-market map can hold reported trust assets and peer-relative profitability in one table without a reconciliation layer. These rows illustrate the delivered shape from the dictionary above; your sample arrives in exactly this schema, cut to the banks and quarters you name.

Six fields anchor the core, and each earns its place. Identity comes as a pair - certificate number for joining, legal name for reading. Time arrives as a quarter-end date. The MDRM / schedule-item code is the semantic payload of the Call Report side: one string encoding which schedule and which line within it, which is how a filing that runs hundreds of line items collapses into a long-format table. Value carries the figure itself in thousands of dollars for monetary items. The UBPR columns round out the dictionary with calculated ratios plus the rank and percentile columns that make peer comparisons one sort operation deep.

One honesty note: the published bulk files do not enumerate their own column headers, so these definitions are Datadory's reading of the structure - exact header names get confirmed against real files when your sample is prepared. Everything beyond the core six - full schedule axes, four-period UBPR panels, peer-group attributes - is listed as additional fields on request beneath the table.

Geography: the whole U.S. commercial banking system - national banks, state member banks, insured nonmember banks - including foreign-office filers on form FFIEC 031. There is no sampling layer between you and the population; every filer in the selected period ships.

Temporal: quarterly, anchored to the last calendar day of each quarter, which is when Call Reports are filed. Selection runs either way you want it: one reporting-period end date for the single-period products, or a calendar year for the four-period products, which return all four quarters in one artifact.

Granularity: one row per institution per schedule item on the Call Report side, one row per institution per ratio on the UBPR side - long format, thousands of filers deep, hundreds of millions of dollars scaled to thousands. Scale reference: a single-period population spans hundreds of megabytes compressed, which is why samples are cut to named banks, schedules and quarters rather than shipped wholesale.

API, files, or your warehouse. Daily, weekly, or hourly.

Raw bulk artifacts reward nobody: hundreds of megabytes of tab-delimited or XBRL per selection, headers to decode, four-period products to stitch. Datadory delivers the same populations as tables keyed on certificate number and quarter-end - through an API for on-demand pulls, files for batch work, or loaded straight into your warehouse for teams joining bank figures against their own positions. Cadence is your call: quarterly rhythm suits most bank analysis, weekly or hourly matters when a monitoring screen needs the newest available population the moment it exists. Start with a sample scoped to your banks, schedules and date windows, then scale to recurring delivery once the joins prove out.

• Custody and fiduciary market mapping - Schedule RC-T splits fiduciary and related services and the trust-asset line items that define the custody tier, so the whole fiduciary business of every chartered bank can be sized from one population file.
• Peer benchmarking with UBPR ranks - Ratio, rank and percentile cuts place any filer inside its UBPR peer group on hundreds of ratios with examiner-computed numbers.
• Credit-cycle and delinquency tracking - The past-due product returns four periods per request, making delinquency drift observable quarter over quarter across the entire population.
• Panel construction for models and backtests - Rows key on certificate number and quarter-end, so multi-year panels concatenate instead of requiring reconciliation.
• Market-sizing for vendors targeting banks - A whole-population file doubles as an addressable-market census, segmentable by size, balance-sheet shape and fiduciary footprint.
• Regulator-grade citation - Every figure traces to a filing the bank itself submitted, which is why researchers and journalists prefer it over aggregator copies.

Investors & quant researchers lead the fit: reported statements and peer-relative ratios, population-complete, with identifiers clean enough to join against holdings or market data (investors view). Market researchers & consultants get trust-and-custody market maps and bank-tier sizings straight from Schedule RC-T (market researchers view). Risk & treasury analysts get past-due trajectories and UBPR percentiles for counterparty and correspondent watchlists. Data scientists & ML engineers get a stable identifier-plus-quarter-end key that makes feature engineering a groupby (data scientists view). Competitive intel & product teams segment the population to size institutional customer bases (competitive intel view), and journalists, academics & students get figures with citation-grade provenance (journalists view).

Within the catalog, the nearest sibling is FFIEC Central Data Repository - Public Data Distribution, the portal view of the same repository; this record is the population edition, and the scored head-to-head against an unrelated shelf-mate lives at CDR Public Data Distribution vs F-Droid. FDIC BankFind Suite API answers institution-attribute questions - locations, charters, identities - that this dataset assumes rather than provides, and FDIC Quarterly Banking Profile gives industry aggregates where this gives bank-level rows. On the mortgage side, FFIEC HMDA data browser and dataset downloads covers loan-level lending disclosures from the same council. If the question is 'every bank, every quarter, reported and peer-ranked', this is the record.

Three things, stated plainly. First, scale: a single-period population spans hundreds of megabytes compressed, so samples are cut to the banks, schedules and quarters you name - which is also the faster route to an answer than receiving a file the size of a small operating system. Second, header confirmation: the bulk files do not self-document their column names, so the exact headers ship alongside your sample, resolved against real files. Third, two families, one key: decide early whether your question wants reported Call Report figures, computed UBPR ratios, or both joined on certificate number and quarter-end - the sample can carry either or both. Name the banks, schedules and windows, and real rows return in exactly the schema shown above.