--- title: "Why FFIEC CDR Bulk Data Falls Short for Bank Peer Analysis" description: "FFIEC CDR bulk data is the right source for bank regulatory analysis, but it has real gaps that distort peer benchmarks and ALCO metrics. Here is what to know." canonical_url: https://www.bankregreports.com/blog/ffiec-cdr-bulk-data-problems/ source: BankRegReports published: "2026-09-29" updated: "2026-10-16" --- A community bank CFO pulling together an ALCO package has a reasonable expectation that publicly reported Call Report data should produce consistent peer benchmarks. It does not always. The same NIM figure can look materially different depending on how the underlying FFIEC data was processed: whether [units](/blog/ffiec-data-unit-inconsistencies/) were normalized correctly, whether prior-period amendments were captured, and whether the [peer group](/blog/bank-peer-group-analysis/) was filtered to institutions on the same form variant. These are not exotic edge cases. They are routine data problems that produce benchmark drift in the 10–40 basis point range, which is enough to change a board conversation about margin performance. This post explains where those problems come from and what they mean for bankers who rely on peer comparison data for ALCO, credit committee, and regulatory exam preparation. ## What Does the FFIEC Central Data Repository Actually Contain? The FFIEC CDR is the public repository for bank Call Report submissions: FFIEC 031, 041, and 051, along with the [UBPR](/blog/how-to-read-ubpr-report/) built from them. [FR Y-9C](/blog/what-is-the-fr-y-9c/) holding company data is published separately by the Federal Reserve through the National Information Center (NIC). It covers thousands of FDIC-insured institutions going back decades, and it is the authoritative source for the numbers that show up in examiner workpapers, UBPR peer group reports, and every serious bank benchmarking product on the market. The official source is the [FFIEC](https://www.ffiec.gov). For community banks (the $300M to $10B asset class that makes up a little over half of FDIC-insured charters), the CDR contains the raw material for every metric your ALCO committee tracks: net interest margin, efficiency ratio, CRE concentration, nonperforming loan ratios, capital adequacy, and deposit costs. The question is not whether the data is there. The question is whether it arrives in a form that actually produces correct peer comparisons without significant cleanup work. The answer, in most cases, is no, and the reasons matter to bankers, not just to the analysts building the systems. ![Median NIM by Asset Tier - Community Banks, Q2 2026](figures/ffiec-cdr-bulk-data-problems-1.svg)
Histogram of net interest margin across every FDIC-insured U.S. bank
The FFIEC CDR units trap turns a 3.5% NIM into 350% or 0.035%. Computed correctly, net interest margin across every bank sits in a narrow band, proof the normalization matters. BankRegReports analysis.
## The Units Problem: Why Your NIM Might Be 100x Off FFIEC [MDRM](/blog/mdrm-codes-developer-guide/) codes (the item identifiers that label every field in a Call Report) carry different unit conventions depending on the schedule, the form variant, and in some cases the vintage of data. UBPR ratios arrive as percent values (a net interest margin of 3.21% is stored as 3.21), while UBPR dollar items, like the CDR Call Report bulk files, are reported in thousands of dollars. Mix either with a whole-dollar source such as the NCUA 5300 without checking and a balance can be off by a factor of 1,000. The true integer-scaled-by-100 case is next door: NCUA 5300 rate fields store 5.66% as 566. There is no single reference table that resolves this for every code in every quarter. The Federal Reserve's item type metadata is the closest approximation, but it has gaps, particularly for data predating the XBRL reporting transition, where backfilled historical data sometimes carries the scaling conventions of the era it was originally reported rather than current conventions. For a banker reviewing an ALCO package, this surfaces as a peer benchmark that looks wrong but is hard to challenge without knowing why. A NIM peer median of 332% instead of 3.32% is obvious. A peer median that is off by 10–15 basis points because a subset of peer institutions had their ratios scaled incorrectly before averaging is not obvious. It looks like your bank is performing slightly better or worse than the peer group when the divergence is actually a data artifact. ## Amendment Handling and the Silent Restatement Problem Banks amend Call Report filings. The CDR public data reflects the most recent submission for each institution and quarter, which means a filed amendment replaces the original data without any visible indication that a restatement occurred. For most routine amendments, this is not a problem. For amendments that correct a capital ratio, restate loan loss reserves, or revise nonperforming asset balances, it creates silent discontinuities in trend analysis. Consider what this means for a net charge-off trend. If a bank restated Q2 2023 NCOs upward as part of a CECL methodology correction, and you are pulling quarterly trend data for that institution's peer group without tracking amendment history, your Q2–Q3 2023 trend line contains a jump that represents a restatement, not a change in credit performance. Peer group averages that include restated data mixed with point-in-time data from institutions that have not yet restated will produce benchmark comparisons that drift in unpredictable ways. The UBPR, which is the examiner's primary peer comparison tool, addresses this by publishing quarterly snapshots that are anchored to specific reporting periods. For ALCO teams building internal peer comparisons directly from CDR data, amendment handling is an explicit design requirement that the raw CDR bulk download does not satisfy. ## Form Variant Routing: The 031 vs. 051 Distinction Call Reports come in three form variants. FFIEC 031 covers banks with foreign offices. FFIEC 041 covers domestic banks above a certain asset threshold. FFIEC 051 is the streamlined form for smaller institutions. The schedules differ in material ways: some line items appear on 031 but not 051, some are reported at different granularities, and some concentration metrics are computed differently across variants. A peer group comparison that pools institutions across all three form variants without accounting for schedule differences will produce averages where some fields are populated for some institutions and structurally absent for others. This is not a data quality failure in the traditional sense; the data is correct. But averaging across a field that exists for 60% of your peer group and is absent for 40% does not produce a valid peer median. It produces a number that looks like a peer median but systematically excludes smaller institutions from the denominator for certain metrics. For community bank peer comparisons, particularly benchmarks that span the $300M to $10B range, form variant routing is a threshold issue. The $300M–$1B segment skews heavily toward FFIEC 051 filers. The $1B–$10B segment is more mixed. Any platform that does not explicitly route by form variant before computing peer statistics will produce comparisons that are biased toward the form structure of larger institutions in the peer group. ![CRE Concentration by Asset Tier - U.S. Banks, Q2 2026](figures/ffiec-cdr-bulk-data-problems-2.svg) ## What This Means for Peer Comparison Accuracy The practical consequence of these data problems is that peer benchmarks built directly from CDR bulk downloads require significant validation before they are presentation-ready. For the metrics that drive examiner conversations, such as NIM, CRE concentration relative to the 300% of risk-based capital screening level in the interagency guidance, nonperforming loan ratios, and capital adequacy against the well-capitalized minimums, the margin for error is narrow. Examiners arrive with UBPR data already prepared. The UBPR applies unit normalization, form variant routing, and peer group construction using standardized methodology. If your internally prepared peer benchmarks diverge from UBPR figures, the examiner's first question is about why, and "we built it from CDR bulk data" is not a satisfying answer if you cannot trace the divergence. The current benchmarks for the $1B–$10B community bank peer group give some context for where the thresholds matter most. ROA sits at roughly 1.2–1.3% for typical performers (median 1.29% at June 30, 2026); below 0.80% generates board questions and below 0.60% generates examiner interest. CET1 capital ratios have a median around 13% for community banks, well above the 6.5% well-capitalized minimum, but banks that have compressed capital through aggressive buybacks or rapid loan growth and sit below 10% will draw examiner attention regardless of whether they technically meet the regulatory minimum. CRE concentration above 300% of risk-based capital triggers the interagency guidance on commercial real estate concentration risk, not a hard prohibition, but an expectation of enhanced stress testing and board-level oversight that examiners will test directly. For all of these metrics, a peer comparison that is off by 30–40 basis points due to unit normalization errors or form variant pooling is not a minor rounding issue. It changes the position of your institution relative to peer medians in ways that affect how you present performance to the board and how you prepare for examiners. ## How BankRegReports Addresses the Problem The [BankRegReports platform](https://www.bankregreports.com) is built specifically to deliver normalized, peer-comparable bank data to bankers who need it in business terms, not to analysts who want to build data infrastructure. The platform ingests FFIEC Call Reports, UBPR ratios, Federal Reserve NIC and Y-9C filings, FDIC institution data, and NCUA 5300 credit union data, applies unit normalization and form variant routing across all of them, and handles amendment tracking so trend data reflects consistent reporting periods. The output is accessible through a web interface at [bankregreports.com](https://www.bankregreports.com): search any FDIC-insured institution, pull peer group comparisons by asset tier or custom peer selection, and export to Excel. For ALCO teams, that means peer benchmarks that are already normalized and can drop directly into existing templates. For credit committees, it means CRE concentration data that is computed on the same basis as the examiner's workpaper. For CFOs preparing board packages, it means efficiency ratio and ROA comparisons that use consistent peer definitions quarter over quarter. ## Deposit Beta and the Rate Cycle Context One area where data quality problems compound quickly is deposit beta analysis, which has been the central ALCO question for most community banks since the Fed began hiking in early 2022. Community banks in the $1B–$10B range ran cumulative deposit betas of 35–55% through the 2022–2023 hiking cycle, with significant variation by funding mix and competitive market. Banks with higher concentrations of commercial DDA lagged their retail-heavy peers by 10–15 percentage points on deposit cost increases, a real performance difference that showed up in NIM outperformance through mid-2023. Measuring deposit beta requires clean, amendment-aware quarterly cost-of-funds data across a multi-year window. It also requires the ability to compare your trajectory against peers who are genuinely comparable by funding structure, not just by asset size. A $2B bank funded 60% by retail CDs competes in a different repricing environment than a $2B bank funded 40% by commercial operating accounts, even though they land in the same asset tier. Peer comparisons that do not account for this produce beta benchmarks that are directionally correct but analytically thin for ALCO purposes. The next phase of the rate cycle, as the Fed holds or cuts and banks manage for duration risk on the asset side, will put the same analytical pressure on asset repricing timelines. Community banks that built out deposit beta tracking during the hike cycle are better positioned to model asset beta and NIM recovery scenarios for the next 6–8 quarters. Those that are still working from spreadsheets with manually downloaded CDR data will be slower to produce the scenario analysis their boards are going to ask for. --- The benchmarks and peer data referenced in this post are available through the [BankRegReports platform](https://www.bankregreports.com), no data engineering required. Search any FDIC-insured bank, pull peer group comparisons, and, on the Pro plan, export to Excel for your ALCO package or board presentation. ## About this document This is the markdown twin of the page named in `canonical_url` above. Source filings are public record; the cleaning, normalization, cross-source linkage, and derived metrics are BankRegReports' own. Attribute with a link to the canonical URL. Markdown twins are published for editorial content only. Institution data pages do not have them. For data at scale use the API, which is documented at https://www.bankregreports.com/api/ and specified at https://www.bankregreports.com/openapi.json. It returns typed JSON with explicit units and periods, and one call replaces hundreds of page fetches. ## Sitemap - [Glossary](https://www.bankregreports.com/glossary/): every metric, with formula and thresholds - [Blog](https://www.bankregreports.com/blog/): analysis and explainers - [API documentation](https://www.bankregreports.com/api/): the programmatic way to get this data - [XML sitemap](https://www.bankregreports.com/sitemap.xml): the complete crawlable index