FDIC Quarterly Banking Profile archive: four decades of industry aggregates, delivered as one panel

The phrase sounds like a filing cabinet; behind it sits the regulator's complete quarterly verdict on American banking, unbroken since the quarter ended March 31, 1986. Each quarter, aggregate income statements, balance sheets and asset-quality measures are compiled across every FDIC-insured institution - roughly 4,700 active charters - then re-cut by asset size group and by community versus non-community status. Nothing gets dropped along the way: more than 160 quarters remain in the record, through the latest verified release covering Q1 2026.

Every numerator and denominator behind the "the banking sector is thriving / in trouble" take you have ever read lives in this one run. In Datadory's catalog it is the FDIC Quarterly Banking Profile, scoring 9 out of 10 on the catalog rubric and filed under diversified banks; the glossary definition settles terminology before anyone starts arguing about it.

The archive framing is the commercially useful half. Trend work needs depth, not just the newest quarter, so Datadory treats the full run as one asset: every quarter stacked onto the last, definitions held constant, delivered daily, weekly, or hourly - your call. A forty-year panel arrives assembled rather than stitched.

Two views travel together in every quarter. The first is the scorecard the press quotes. The second is the structural map of the archive's analytical core - seven themed grids of aggregate line items, every row carrying a period and an asset-size breakout:

ARCHIVE CORE - seven themed grids, one shape end to end 01 Ratios by asset size groups 02 Ratios by community vs non-community banks 03 Balance sheet 04 Loan performance 05 Quarterly income 06 Annual income 07 Small business and farm loans

BREAKOUT DIMENSION - carried on every aggregate row asset_size_group: All Insured Institutions | Assets < $100 Million | Assets $100 Million - $1 Billion | larger bands ```

Five themes organize everything: earnings, asset quality, net interest margin, capital and problem banks. Ten core fields define the aggregate grid - five income-statement aggregates (total interest income, net interest income, provision for loan and lease losses, total noninterest income, total noninterest expense), the bottom-line net income figure, two asset-quality counters (net charge-offs and the count of unprofitable institutions), the number of institutions reporting, and the asset_size_group enum that turns one flat industry number into a curve across size bands. Monetary amounts arrive in USD millions, typed, with definitions attached.

Everything else a quarter carries - full balance-sheet detail, loan-performance splits, small business and farm lending lines, community-bank ratio sets, deposit-insurance-fund trend series - maps into exactly the same shape as additional fields on request. A worked version of the dictionary sits on the product page and travels beside every delivery.

• Temporal: quarterly observations from Q1 1986 through the current quarter - more than 160 quarters, one of the longest continuous runs in US banking statistics - plus annual rollups for long-cycle work.
• Weight: the whole machine-readable run stays analytically light, small enough to hold in memory, so a backtest runs against the full history rather than a sample of it.
• Shape: constant. One value per line item per period per breakout, unchanged across vintages, so a margin study built on 1990s quarters uses the same table as a current-quarter refresh. No vintage archaeology, no reconciliation project.
• Cycles: the run spans the savings-and-loan aftermath, the dot-com break, the global financial crisis and the 2023 regional episode - enough repetition to calibrate a model instead of narrating a memory.

Longer horizons sit nearby rather than inside the archive. Through the Federal Reserve Economic Data Portal, Z.1 flow-of-funds series reach back to 1945 and H.15 rates to 1959; cross-border positions open at 1977-Q4 in BIS Locational Banking Statistics, where we put the two banking aggregates head to head. The profile sits squarely among multi-decade series rather than trailing them - and delivered together, the long context and the quarterly detail share one join key: the calendar quarter.

Cohort structure is the archive's distinguishing feature. Every ratio line is reported for All Insured Institutions and then again inside asset size bands - Assets < $100 Million, Assets $100 Million - $1 Billion, and larger bands above them - while a dedicated Ratios by CB vs. NCB grid separates community banks from non-community banks. A standalone community-bank performance cut repeats the split every quarter. Who counts as a community bank in the first place is settled by the community bank classification framework the series is built on.

• Is the sector getting better or worse? Net interest margin, provisioning, charge-offs and bottom-line profit as one national series since 1986 - direction established before anyone argues about a single bank. The vocabulary is pinned in the net interest margin entry.
• How does one bank's quarter compare with the tide? Community and non-community subtotals give the baseline a management team's claims get measured against.
• Where does credit stress sit by size? Problem-bank counts and loan-performance aggregates by asset-size band show whether stress is a tail phenomenon or a broad wave - and the bank failure history dataset picks up the story when stress becomes resolution.
• What does a full cycle look like? Four decades of earnings, margin and problem-bank data, one consistent shape, ready for model calibration.

What aggregates cannot answer is equally sharp: anything about one named institution. There the drill-down takes over - the FDIC BankFind Suite API carries certificates, locations and institution-quarter financials reaching toward the 1930s. The two layers reconcile cleanly because both descend from the same supervisory universe.

The archive answers "what happened to the industry?"; the rest of the stack answers who moved, loan by loan, and why. Six records cover most pairings:

A practical stack falls straight out of those roles: profile grids for the industry line, HMDA rows for the origination volume behind it, H.15 for the rate path, then FFIEC cuts to name the banks that moved the aggregate. All of it arrives on one delivery contract rather than six.

1. Investors and quant researchers tracking margin, provisioning and profitability cycles across all insured institutions since Q1 1986 - the base-rate series any single bank's trajectory gets judged against (investors and quants use cases).
2. Market researchers and consultants anchoring reports in official aggregates cut by institution size rather than vendor estimates (market researchers).
3. Competitive intelligence teams testing a rival's "we beat the industry" line against community and non-community subtotals before crediting management (competitive intel workflows).
4. Data scientists turning each quarter into tidy features keyed on period and asset-size band, ready to merge onto bank-level panels (data scientists).
5. Journalists, academics and students quoting the regulator's own quarterly scorecard with definitions behind every field.
6. Sales and growth teams walking into bank pitches already fluent in earnings season - so their buyers do not have to be.

Files, feeds, or straight into your warehouse. Daily, weekly, or hourly - your call.

Every delivery carries the complete field dictionary, the sample rows and the coverage statement - mapped and typed before it reaches you, identical columns and join keys whichever channel you take.

Snapshots pin by release, so a report re-run next month reproduces against the same vintage instead of drifting with aggregate revisions; newer quarters slot in as their own vintages, never as edits underneath you. The upstream wrapper a statistical release ships in stops being your problem the moment the rows are typed.

The rhythm is worth naming. The archive itself adds one observation set per calendar quarter, but the panels around it move faster - institutional filings weekly, rates daily - and aligning all of them on one delivery contract is the point of taking the stack from a single vendor. Changing frequency later is configuration, not migration.

Name the periods, size bands and line items when you request a sample - the last eight quarters of margin and provision aggregates cut to the bands your model prices, or the full problem-bank series if stress mapping is the job. Real rows come back shaped to that specification, field dictionary and coverage statement attached.

You keep the sample, the dictionary and the coverage statement regardless of what happens next. Production follows the exact shape you evaluated - the thing you tested is the thing that ships.