Diversified Financial Services · Federal Reserve
Federal Reserve MDRM Data Dictionary
Datadory delivers diversified financial services data covering the Federal Reserve MDRM Data Dictionary: the Board's Micro Data Reference Manual, which resolves every eight-character item code behind FFIEC Call Reports, FR Y-9C holding company filings and hundreds of sibling regulatory forms into plain-language definitions, item types and validity windows - roughly 1.33 million dictionary rows spanning items collected since 1956 - delivered as typed rows shaped to your schema.
API, files, or your warehouse. Daily, weekly, or hourly.
- geo
- United States - institutions filing Federal Reserve and FFIEC regulatory reports
- How far back
- Reporting status filters reach back to 1956; individual items carry Start Date and End Date validity windows
- How fine
- One row per data item per reporting form per validity period - roughly 1.33 million rows across hundreds of forms
What is the Federal Reserve MDRM Data Dictionary?
The decoder ring the whole regulatory-reporting stack depends on. The Federal Reserve MDRM Data Dictionary is the Board of Governors' Micro Data Reference Manual: the authoritative lookup that turns each eight-character MDRM Identifier into a definition, a validity window, a confidentiality flag and an item class. Four letters name the collection - RCON and RCFD for Call Reports, BHCK, BHDM and BHCA for FR Y-9C schedules - and four digits name the item, 2170 being the canonical Total Assets.
Coverage runs wide as well as deep: FFIEC 031 and 041 Call Reports, the FR Y-9C holding company report that captures diversified financial holding companies, FFIEC 002-019, FFIEC 101/102, FR 2644, the TIC series and NCUA 5300 among hundreds of forms, together yielding roughly 1,332,936 dictionary rows. Entries are searchable by item number, item name, series or form name, and filterable by item type, confidentiality and reporting status reaching back to 1956.
Datadory resolves those codes ahead of time, delivering typed rows that map filing columns to meanings without a manual lookup step.
What do the sample rows show?
The anatomy of an entry, twice over. Both rows carry item code 2170 - Total Assets, class F for financial/reported, no confidentiality restriction - but they differ exactly where the dictionary insists they must: the mnemonic and the form. One pairing sits on FFIEC 101 with a window opening September 30, 2016 and running open-ended; the other belongs to the Thrift Call Report.
Read the pair as a warning worth paying for: keys in this world are composite, and validity windows are part of the data, not an afterthought. Any pipeline that treats 2170 as globally unique, or ignores the start and end dates when joining historical filings, will mislabel columns silently - which is precisely the failure mode Datadory resolves upstream of your models.
Which fields does the dictionary define?
Nine columns carry most of the payload, verified during the August 2026 cataloging pass. Two deserve special care in any pipeline:
- ItemType - the letter deciding what a value is: J projected, D derived, F financial/reported, R rate stored as a decimal, S structure, E examination/supervision, P percentage stored as a percent value. Confusing R with P corrupts every ratio you compute.
- Start Date and End Date - the item's validity window, with open-ended items running to 12/31/9999. This pair decides whether an item even existed on your filing date.
Beyond the nine below, a mnemonic-level SeriesGlossary column and the remaining README-documented metadata elements ride along on request - tell us at sample time and they ship with the extract.
Where does coverage run, and at what grain?
Geography - United States institutions filing Federal Reserve and FFIEC regulatory reports: commercial banks on the Call Reports, holding companies on the FR Y-9C, credit unions through NCUA 5300, and the international exposures captured by the TIC series.
Temporal - reporting-status filters reach back to 1956, and every item carries its own start and end dates, so the dictionary describes not just what is collected today but the full lineage of what has ever been collected, including retired codes.
Granularity - one dictionary row per data item per reporting form per validity period. Multiply hundreds of forms by their item inventories and you arrive at roughly 1.33 million rows: the complete vocabulary of American bank regulatory reporting.
How is the data delivered?
API, files, or your warehouse. Daily, weekly, or hourly.
You pick the channel and the cadence; the field definitions above travel unchanged through all three. What Datadory adds is the unglamorous work between lookup and load: resolving mnemonics to their forms, honoring validity windows against your filing dates, casting R-type decimals and P-type percentages correctly, and carrying the confidentiality flag so downstream publishing never trips over an item that was never yours to show.
Who uses this data, and for what?
A dictionary earns its keep wherever someone meets a filing feed cold:
- Bank equity and credit research - analysts decoding FR Y-9C and Call Report feeds map every BHCK and RCON column to its meaning before trusting a single trend line, and the dictionary is the map.
- Regulatory capital and stress-testing teams - capital ratios, risk-weighted assets and schedule RC items all resolve here, with item classes telling you which values are dollar figures, rates or percentages.
- RegTech and data-vendor schema mapping - building a normalized bank-fundamentals product means maintaining a crosswalk of thousands of item codes; this dictionary is the crosswalk's source of truth, windows included.
- Academic banking research - panel studies built on decades of Call Reports depend on knowing which items existed in which periods, which is exactly what the validity windows encode.
- Due diligence and counterparty review - when a filing must be interpreted quickly and correctly, the difference between a derived item and a reported one matters.
Which personas get the most value?
Ranked by how directly one resolved row settles the day job:
- Developers & Data-Product Builders - schema mapping is the product when your pipeline ingests regulatory filings; a complete, typed code dictionary removes the maintenance burden that usually comes with it.
- Investors & Quants - bank holding company screens live and die on correctly decoded FR Y-9C items; the ItemType column alone prevents the decimal-versus-percent errors that quietly poison ratio screens.
- Data Scientists & ML Engineers - feature engineering on bank panels needs stable definitions across decades; validity windows tell you where a feature's support begins and ends.
- Journalists & Academics - citing a figure accurately means knowing what the filing actually asked for, in the form's own words.
What should I know before requesting a sample?
Three things, stated up front. First, scope your forms: the dictionary spans hundreds of collections, so tell us whether you are decoding Call Reports, FR Y-9C schedules, thrift filings or the TIC series, and the sample arrives pre-filtered to those mnemonics.
Second, choose your window policy: current-items-only keeps the vocabulary small, while full historical windows preserve retired codes for backfilled panels - both are supported, but they are different deliverables.
Third, decide how confidentiality flags should surface in your extract: passed through as a column, used to exclude items, or both. Say the word and the sample reflects it.
Which datasets pair with this one?
Three complements cover what a dictionary alone cannot:
- Federal Reserve Economic Data Portal - Board Releases - the Board's published releases alongside the vocabulary that defines their underlying items.
- FDIC Bank Data & Statistics - institution-level banking statistics that slot neatly against decoded Call Report items.
- Federal Reserve Yield Curve (Feds Daily) - a rates overlay for interest-income analysis built on decoded balance-sheet items.
For the contrast case, see MDRM Data Dictionary vs Treasury Resource Center Data Center: one decodes regulatory filings, the other catalogs fiscal publications.
Field dictionary
Every field below is documented against real records. The full dictionary ships with the sample.
| field | type | definition | example |
|---|---|---|---|
Mnemonic | string | First four characters of the eight-character MDRM Identifier, naming the reporting form or data collection - RCON and RCFD for Call Reports, BHCK/BHDM/BHCA for FR Y-9C schedules. | BHCK |
Item Code | string | Final four digits identifying the specific data item; codes repeat across forms whenever the reported concept matches. | 2170 |
Start Date | date | First calendar date on which collection of the item begins - the near edge of the item's validity window. | 9/30/2016 |
End Date | date | Last calendar date on which the item is collected; open-ended items run to 12/31/9999. | 12/31/9999 |
Item Name | string | Human-readable label assigned to the item number or code. | TOTAL ASSETS |
Confidentiality | enum | Binary Y/N flag for whether the item may be shared publicly; definitions publish either way, so confidential items remain visible in the vocabulary. | N |
ItemType | enum | Letter classifying the item: J projected, D derived, F financial/reported, R rate stored as a decimal, S structure, E examination/supervision, P percentage stored as a percent value. | F |
Reporting Form | string | The presentation of a specific data collection for regulatory reporting - the form on which this item appears. | FFIEC 101 |
Description | text | Explanation of what should be reported for the item, generally taken from the reporting form's instructions. | "The sum of all asset items. This item must equal 'Total Liabilities; Limited-Life Preferred Stock and Equity Capital (3300)'." |
Coverage at a glance
| dimension | value |
|---|---|
| Geography | United States - institutions filing Federal Reserve and FFIEC regulatory reports (banks, holding companies, credit unions, cross-border TIC reporters) |
| Temporal | Reporting status reaches back to 1956; every item carries its own Start Date and End Date validity window |
| Granularity | One dictionary row per data item per reporting form per validity period - roughly 1.33 million rows across hundreds of forms |
Questions buyers ask
What exactly is an MDRM identifier?
Eight characters doing one job. The first four letters name the collection - RCON and RCFD for Call Reports, BHCK, BHDM and BHCA for FR Y-9C schedules - and the final four digits name the item, 2170 being Total Assets. Concatenate them and you hold the key that every downstream filing feed, crosswalk and vendor schema ultimately references.
How much ground does the dictionary cover?
Roughly 1,332,936 dictionary rows across hundreds of reporting forms: FFIEC 031 and 041 Call Reports, the FR Y-9C holding company report, FFIEC 002-019, FFIEC 101/102, FR 2644, the TIC series and NCUA 5300 among them. Because rows are keyed per item per form per validity period, the count reflects lineage, not duplication.
Why do the ItemType letters matter so much?
They decide what a value is before you touch it: J is projected, D derived, F financial/reported, R a rate stored as a decimal, S structural, E examination or supervisory, P a percentage stored as a percent value. Treat an R as though it were a P - or the reverse - and every ratio built on it shifts by two orders of magnitude.
How far back does item-level coverage go?
Reporting-status filtering reaches back to 1956, and each item carries its own start and end dates inside that span. Practically, that means the dictionary documents not just today's vocabulary but retired codes too - which is what makes decade-spanning panel studies possible without silent column drift.
What role does the confidentiality flag play?
Each item carries a Y/N marker for public shareability. Definitions publish regardless, so the vocabulary stays complete: you can see that a confidential item exists, where it sits on its form, and exclude it deliberately downstream rather than discovering it by accident.
Does the dictionary replace the reporting forms themselves?
No, and it is not meant to. The dictionary defines the items; the forms arrange them into schedules with their own instructions and edits. Used together they keep schema mappings honest - the dictionary tells you what BHCK2170 means, the form tells you where it sits and what must reconcile against it.
Notes on this record
- Four letters, four digits The entire key structure of American bank regulatory reporting fits in eight characters: a mnemonic naming the collection, an item code naming the datum. Learn the pattern once and every form reads faster.
- Same code, different uniforms Item 2170 is Total Assets on the Thrift Call Report, on FFIEC 101 and elsewhere - but only the mnemonic-plus-code pair identifies anything. Composite keys are the rule here, not an optimization.
- Validity windows are the trap Start Date and End Date ride on every row, open items running to 12/31/9999. Join against historical filings without honoring them and columns get mislabeled silently - the worst kind of bug.
- R versus P Rates are stored as decimals, percentages as percent values, and the ItemType letter is the only thing standing between you and a hundredfold error. Casting rules belong upstream of models, which is where we put them.
- Verified, not assumed Definitions were checked against the manual in the August 2026 pass - the same standard met by 1,495 of the 1,744 datasets (85.7%) in the catalog.
See the rows before you pay anything.
Name this dataset and we send real records from it — scoped to the fields you asked for.