How Call Report Forms Change, and How to Track It

If you maintain a call report time series long enough, something breaks. A column that was populated for forty quarters goes null. A new MDRM code appears with no history behind it. A ratio shifts at a quarter boundary by more than any bank’s fundamentals moved.

None of that is a bug in the data. It is the form changing, and the change was announced in public, months in advance, in documents almost nobody reads.

Why the Form Changes at All

The call report is a federal information collection, so any change to it runs through the Paperwork Reduction Act. That statute requires agencies to justify what they collect, to solicit public comment before changing it, and to obtain approval from the Office of Management and Budget. Every added item, deleted item, and revised definition passes through that process.

The changes themselves originate from a few places: a new accounting standard that alters what a line means, a new capital or liquidity rule that needs data to supervise, a burden-reduction initiative that removes items from small filers, or a supervisory concern that emerged and now needs measurement.

The Notice Sequence

The process is predictable enough to plan against.

A 60-day Federal Register notice proposes the change and opens the comment period. This is the earliest formal signal, and it typically arrives six months to a year or more before the change takes effect. The notice describes the proposed items in detail and is usually accompanied by redlined draft forms and instructions.

A 30-day Federal Register notice follows, summarizing the comments received, stating what the agencies decided, and submitting the collection to OMB for approval.

OMB approves the revised collection. The FFIEC then publishes final forms and instructions for the effective reporting period.

The change takes effect as of a specific report date, and banks file the revised form for that quarter.

The lag between first notice and effective date is the useful part. If you follow the 60-day notices, you know roughly a year ahead which items are moving.

The Documents That Tell You What Moved

Four documents carry the actual change, and they are easy to find once you know they exist. (Source: FFIEC Reporting Forms.)

The instructions update. Each cycle’s instruction book is accompanied by a short update document listing the instruction pages that were revised for that report date. It is a one-page read and the fastest way to see which schedules moved without diffing a 35,000-line book. Note what it does and does not do: it names the pages that changed, not what the change says. For that, go to the revised pages themselves or to the Federal Register notice.

Supplemental instructions. A short quarterly memo, shared across the FFIEC 031, 041, and 051, covering topics that need attention this cycle. It carries transition guidance, reminders on newly effective accounting standards, and answers to the questions the agencies are fielding most often.

Federal Register notices. The proposal and final notices themselves, which give the reasoning. When you need to know why an item changed rather than just that it did, this is the source.

The transmittal from your primary federal regulator. The FDIC transmits call report revisions to its supervised institutions by Financial Institution Letter, the OCC by bulletin, and the Federal Reserve through its own channel, following the same fragmented structure described in the guide to supervisory guidance channels.

One caution on the redlined drafts. A large share of what the FFIEC publishes alongside a form is proposal-cycle material: draft forms and draft instructions attached to a proposal that may be modified or dropped before it takes effect. A redline is not evidence that a change happened. Check whether the proposal reached a final notice and an effective date before treating it as real.

What Actually Breaks in the Data

Four failure modes, in rough order of how often they cause trouble.

An item is discontinued. The MDRM code simply stops appearing after a given quarter. Nothing in the data marks the discontinuation; the series ends. Code that assumes the column exists throws, and code that fills forward silently produces a flat line that looks like a real observation. This is the one that does damage quietly.

An item is added. A new code appears with no history. Any calculation using it produces nulls for every prior period, and a peer comparison spanning the boundary compares banks that reported it against banks that could not have.

An item is redefined. The code stays the same, the values stay populated, and the meaning changes. This is the worst case because nothing at all is visibly different. The evidence is in the revised instruction pages for that cycle and in the Federal Register notice that adopted the change. CECL adoption is the well-known example on the allowance line; there are many smaller ones.

A threshold moves. Several schedules are completed only by banks above an asset or activity threshold. When the threshold changes, a cohort of banks starts or stops reporting an item, and the population of the series changes without any individual bank changing behavior. Cross-sectional statistics computed across that boundary are not comparable.

To all four, add the amendment problem: banks amend prior-period call reports after filing, so a value pulled today for a quarter three years ago may differ from what you pulled last year for the same quarter. Some of the extraction consequences are covered in the guide to FFIEC bulk data problems.

Practical Habits for a Long Series

A few things make this manageable.

Read the instructions update every quarter. It is a one-page document naming the revised pages. Fifteen minutes a quarter, following it through to the pages it names, avoids most of the surprises.

Store what you pulled, with the pull date. If you overwrite history each refresh, you lose the ability to explain why a number moved. Keeping the vintage of your extract turns an unexplainable discrepancy into a two-minute check.

Never fill forward across a discontinuation. A null after a discontinued item is the honest answer. A carried-forward last value is a fabricated observation, and it will be reported as a fact by whatever sits downstream.

Mark definitional breaks in the series itself. If you know the allowance definition changed at a specific quarter, that boundary belongs in your metadata, not in the analyst’s memory. Anyone reading a twenty-year chart deserves to see where the definition moved.

Pin the instruction vintage to the report date. The instruction book that governs a 2011 filing is a different document from the one governing a 2026 filing, and the FFIEC posts historic reporting forms for every quarter back to the form’s introduction, plus the instruction book for each cycle in which it was reissued. When you need to know what a 2011 item meant, read the 2011 instructions, not the current ones. The general structure of those books is covered in the guide to call report instructions and the glossary.

That last point generalizes. The MDRM code is stable across vintages. The meaning behind it is not. Treating a code as a permanent definition is the assumption that produces the largest and quietest errors in bank time-series work.

---

Frequently Asked Questions

How often does the call report change?

The FFIEC revises call report forms and instructions on a regular cycle, with changes typically taking effect at a quarter boundary. Substantive revisions go through the Paperwork Reduction Act notice process, which usually runs six months to a year or more from the first Federal Register proposal to the effective reporting date, so most changes are announced well in advance.

Where can I see what changed in the call report this quarter?

The FFIEC publishes an instructions update document each cycle that redlines the changes from the prior version, along with a supplemental instructions memo shared across the FFIEC 031, 041, and 051. The Federal Register notices for the revision explain the reasoning, and each agency transmits the revisions to its own supervised institutions by letter or bulletin.

Why did an MDRM code stop appearing in my data?

Most likely the item was discontinued in a form revision. When an item is removed, its code simply stops appearing after the last reporting period it applied to, and nothing in the data marks the end of the series. The instructions update for the relevant cycle identifies the instruction pages that were revised.

Can a call report item change meaning without changing its code?

Yes, and this is the hardest case to detect. Items are redefined while keeping the same MDRM code, so the column stays populated and nothing looks different. The evidence is in the revised instruction pages for the cycle in which the change took effect, and in the Federal Register notice that adopted it. A time series crossing such a change is not comparing like with like.

Are redlined draft call report forms the same as final changes?

No. A large share of published redlines are proposal-cycle drafts attached to a Federal Register proposal that may be modified or dropped before adoption. A redline indicates a proposed change, not an effective one. Confirm that the proposal reached a final notice and has an effective reporting date before treating it as a real change.

---

BankRegReports publishes call report, UBPR, FDIC, Federal Reserve, NCUA, and SEC EDGAR data for every U.S. bank and credit union, with peer benchmarks and quarterly history. Look up a bank