Part 370’s exit moves the math to the FDIC, not the data work

The FDIC proposed on November 4 to stop making the largest banks compute deposit insurance in-house. The calculation moves to the FDIC. Clean ownership data, provisional holds and warnings to pass-through account holders stay with the banks.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

The FDIC proposed on November 4 to stop making the largest banks compute deposit insurance in-house. The calculation moves to the FDIC. Clean ownership data, provisional holds and warnings to pass-through account holders stay with the banks.

  1. The week On Wednesday, November 4, 2026, the Federal Deposit Insurance Corporation published a proposed rule in the Federal Register under the title “Large-Bank Deposit Insurance Determination Modernization” (91 FR 69642, RIN 3064-AG35).
  2. The calculation the FDIC would take back Federal law requires the FDIC to pay insured deposits “as soon as possible” after a bank fails.
  3. The data that stays with the bank Section 360.9 already requires a covered bank to produce depositor and customer data in a standard format after the close of any day’s business.
  4. The holds that come back Section 360.9’s second requirement is mechanical.
  5. The pass-through accounts the bank must warn The third obligation concerns accounts in which the bank’s records do not hold what the FDIC needs.
  6. Where the argument is weakest, and what to file by January 4 The strongest objection to our reading is that the FDIC is right on the numbers.
  7. With hindsight Little time has passed since the proposal appeared, and it is still only a proposal.

The week

On Wednesday, November 4, 2026, the Federal Deposit Insurance Corporation published a proposed rule in the Federal Register under the title “Large-Bank Deposit Insurance Determination Modernization” (91 FR 69642, RIN 3064-AG35). Its Board of Directors had approved the proposal on October 30. Comments must be received by January 4, 2027.

The proposal does three things. It would remove 12 CFR part 370, the FDIC’s rule titled “Recordkeeping for Timely Deposit Insurance Determination.” It would also rewrite 12 CFR 360.9, an older rule that already applies to a wider group of large banks, so that most of the banks leaving part 370 move under it. And it would index the dollar thresholds in section 360.9 to inflation.

The two rules have different histories. Section 360.9 dates from a final rule adopted in July 2008, after advance notices in 2005 and 2006 and a proposal in January 2008. It applies to an insured depository institution with at least $2 billion in deposits and either at least 250,000 deposit accounts or $20 billion in total assets. Such a bank must be able to give the FDIC depositor and customer data in a standard format after the close of any business day. It must also be able to place and release provisional holds on deposit accounts automatically.

Part 370 came later. The FDIC adopted it in 2016 for banks with two million or more deposit accounts, and revised it after an April 2019 proposal. It goes further than section 360.9. A covered bank must configure its own information technology systems to calculate the insured and uninsured amount in each deposit account, by ownership right and capacity. It must keep complete and accurate records for that calculation, produce output files in prescribed formats, certify compliance and accept periodic FDIC testing. A bank that certifies under part 370 is released from section 360.9, including its hold requirement.

The FDIC’s stated reason for the change is its own progress. The proposal says the agency has enhanced its systems for calculating deposit insurance since 2016, and plans further expansion. With those capabilities, it says, it no longer needs the largest banks to run the calculation in-house, provided they meet a strengthened section 360.9.

The scale is set out in the proposal’s economic analysis. As of March 31, 2026, the FDIC insured 4,287 institutions. It estimates that 148 would be affected: 31 currently under part 370 and 117 currently under section 360.9. Of the 31, the FDIC classes five as “Highest Complexity,” with more than 30 million accounts each and about $6.8 trillion in domestic deposits between them. It estimates net industry savings of about $19 million in the first year and about $32 million a year after that, and says these estimates probably understate the savings.

The proposal also announced a change that did not wait for comments. Effective immediately, the FDIC said, it will delete all depositor and customer data received in a section 360.9 or part 370 compliance review within 60 days of the review. It also said it will no longer require customers’ Social Security and tax identification numbers in section 360.9 compliance reviews and will assign synthetic identifiers instead.

This proposal is easy to file under deregulation, and the FDIC’s own summary invites that reading. Thirty-one banks would no longer have to run, test and certify a deposit insurance calculation engine. The agency’s estimate is tens of millions of dollars a year in savings, and it calls the estimate conservative.

Read by the people who run a large bank’s deposit platforms, the proposal says something narrower. It moves one function, the calculation, from the bank to the FDIC. It does not move the conditions that make the calculation possible. Those stay with the bank, and two of them, the standard data files and the provisional holds, come back in a form these banks have not been required to maintain since they certified under part 370.

Our argument is that chief information officers and chief data officers at affected banks should treat the change as a re-platforming of their failure-day obligations, not a retirement of them. The calculation engine can go. The ownership data, the hold mechanism and the depositor communications become more exposed, because the FDIC will now run its own engine on whatever the bank hands over.

The article takes the proposal in five parts. The first describes what the FDIC would take back and why it believes it can. The next three are the obligations that remain with the bank: the depositor data, the provisional holds and the pass-through accounts. They do not overlap, and together they cover what section 360.9 as proposed would ask of a larger bank. The fifth tests the argument against the strongest objections and sets out what a bank should put in front of the FDIC before January 4. Banks should confirm their reading of the proposal with counsel.

The calculation the FDIC would take back

What part 370 asked banks to build

Federal law requires the FDIC to pay insured deposits “as soon as possible” after a bank fails. The statute sets no deadline. The FDIC’s practice, the proposal says, is to give most depositors access to insured funds on the next business day.

For most failed banks that practice works because the FDIC can take the bank’s records and run its own determination. Part 370 grew out of a concern that this would not hold for the very largest banks. The proposal recalls the two problems the 2016 rule addressed: deposit records that might be inconsistent or incomplete at failure, and limits in the FDIC’s own systems when faced with millions of accounts.

The answer in 2016 was to make the banks do the work. A part 370 bank had to be able to compute, for each account, the insured and uninsured amounts by ownership right and capacity. Ownership right and capacity refer to the categories in the FDIC’s deposit insurance coverage rule, 12 CFR part 330. The account types the proposal mentions include single ownership, joint ownership, accounts of corporations, partnerships and unincorporated associations, and informal revocable trusts.

Computing that at scale is hard. The bank has to aggregate every account a depositor holds in the same category, apply the limit and assign the result back to accounts. We think the hard part is doing so across product systems that may not have been designed to share a customer key. The proposal describes the obligation as operating, maintaining, continuously enhancing and remediating IT systems able to calculate coverage within 24 hours of the FDIC’s appointment as receiver.

What the FDIC says has changed

The proposal’s core claim is that the FDIC’s own claims processing and determination capacity has grown since 2016. It says the agency “can now readily process many more insurance determinations than it could in 2016.” It also refers to planned expansion of those capabilities.

The proposal does not describe the FDIC’s systems in technical terms. A reader cannot tell from it how the agency’s engine scales, how it was tested against the largest banks’ files or what volume it was sized for. That gap matters, and the final section returns to it.

What the change saves, by the FDIC’s count

The economic analysis gives the most concrete picture of what part 370 cost. The FDIC sorts the 31 part 370 banks into three tiers by number of checking, savings and retirement accounts. Thirteen “Lowest Complexity” banks have fewer than six million accounts. Thirteen “Middle Complexity” banks have between six and 30 million. Five “Highest Complexity” banks have more than 30 million.

Annual savings from dropping part 370 are put at about $625,000 for a Lowest Complexity bank, $875,000 for a Middle Complexity bank and $2.125 million for a Highest Complexity bank. The FDIC also expects one new bank a year to cross the part 370 threshold under current rules, at an implementation cost it puts at about $3.5 million a year for each such bank. Removing part 370 would avoid that.

Against these savings, the 29 banks that would become “larger covered institutions” under section 360.9 would bear one-time transition costs. The FDIC estimates about $210,000 for each Lowest Complexity bank, $254,000 for each Middle Complexity bank and $1.7 million for each Highest Complexity bank, about $14 million across the industry. Ongoing section 360.9 costs are put at about $23,000 a year per bank, or 185 hours.

The net figures follow: about $19 million in the first year, about $32 million a year after that, and annualized savings of about $31 million over ten years at a 3.7 percent discount rate. The FDIC values labor at $125 an hour and says several times that its estimates likely understate both savings and costs.

Why those numbers mislead a technology budget

Those figures are labor-hour estimates, built from the FDIC’s current and historical burden estimates for the two rules. They are useful for comparing alternatives. They are a poor guide to what a bank’s technology organization will actually see.

We expect that much of what a part 370 bank built is not a separate system that can be switched off. The ownership-category logic, the customer matching and the account-to-customer relationships are likely to live inside the bank’s data warehouse, its customer master and its deposit systems. Those structures serve other purposes too: know-your-customer files, regulatory reports, marketing, fraud models. Retiring the calculation removes the testing and certification overhead. It does not unwind the data model, and a bank would be unwise to unwind it, for reasons the next section explains.

The transition estimates also use the FDIC’s historical hours for the Data Extraction and Reporting components of part 370. That is a reasonable proxy for producing section 360.9 files. It is a weak proxy for rebuilding provisional holds, which the FDIC gives 18 months and which this article treats separately.

The data that stays with the bank

The standard file, and the 26 fields that would become mandatory

Section 360.9 already requires a covered bank to produce depositor and customer data in a standard format after the close of any day’s business. The FDIC expects the files as soon as practicable and no later than the following calendar day. They must reflect end-of-day ledger balances as of the day requested.

For ordinary covered banks, the rule has always been a mapping exercise. The 2008 preamble, quoted in the proposal, said covered institutions are not required to collect or generate new information. They map existing data elements and internal codes into the standard formats. A field the bank does not hold can be left empty.

The proposal keeps that rule for ordinary covered banks. For larger covered institutions, those with at least $2 billion in deposits and two million or more accounts, it changes it. Twenty-six data fields would become mandatory. If the bank does not hold the information for a mandatory field, it must collect it and keep it in its deposit account records.

Twenty-one of the 26 fields are already required in part 370 output files. The other five exist in section 360.9 but are not explicitly required by part 370:

  • a deposit type indicator, domestic or foreign;
  • a customer ownership indicator at the account level;
  • the account title as it appears on the bank’s records;
  • a relationship code showing how each customer relates to the account;
  • a beneficiary code identifying agent, custodian or trustee relationships.

The FDIC expects the banks to hold this information already, because part 370 required complete and accurate records. That expectation is plausible for the data part 370 tested. Account title and relationship codes are a different matter. At some banks, titles may be free text typed at account opening, and relationship codes may have been defined for servicing, not for insurance. We expect mapping them to the FDIC’s values to be the kind of task that looks small in a burden estimate and can take months in a bank with several deposit platforms.

What the bank does not have to collect

The proposal carves out two groups of accounts from the duty to collect missing data. The first is accounts whose records disclose a relationship that might support additional coverage under 12 CFR 330.5 or 330.7, the provisions on recognizing deposit ownership and on agency or custodial accounts. The second is formal revocable trust accounts and irrevocable trust accounts insured under 12 CFR 330.10 and 330.12.

For these, the bank reports what it holds and need not gather more. The FDIC compares this to part 370’s alternative recordkeeping provision. The logic rests on part 330 itself: when a bank’s records disclose such a relationship, the details may be kept by the depositor or a third party instead of the bank.

A larger bank could also ask for a full or partial exemption from the mandatory-data requirement for a relatively small number of accounts, where meeting it would be impracticable or overly burdensome. While a request for exemption or extension is pending, the bank would not be treated as noncompliant on that point. The FDIC also says it intends to republish some part 370 frequently asked questions as section 360.9 guidance.

Two optional fields that only larger banks would see

The proposal adds two non-mandatory fields to Appendix C of part 360, for larger covered institutions only. One shows whether a customer owes a debt to the bank, which matters for the FDIC’s right of offset. The other shows, for government customers, whether the bank has pledged securities to cover any shortfall in deposit insurance.

Part 370 banks already keep this information, the FDIC says, so the fields should not be new to them. A bank that does not hold it need not collect it solely for section 360.9. Both fields may still involve joins that deposit teams do not own. Where loan systems hold the debt and treasury or collateral systems hold the pledges, a file that pulls from them needs an agreed customer key across three platforms, and a test that the key matches.

Why a calculation-free rule raises the stakes on data

Under part 370, a data error inside the bank had a chance to surface inside the bank. The bank’s own engine would aggregate, apply limits and produce results that the bank tested and the FDIC tested periodically. A wrong relationship code could show up as an odd insured amount on the bank’s own report.

Under the proposal, the bank sends raw fields and the FDIC’s engine does the arithmetic. The bank no longer sees the result of its data. An error in a customer identifier, an ownership indicator or a relationship code becomes visible only when the FDIC runs it, in a compliance review or on the day of failure.

That is, as we read it, the central reason not to dismantle the data controls built for part 370. A bank that keeps them gains an early warning the rule no longer provides. Practical steps, which are good practice rather than requirements in the proposal, include the following:

  • keep a shadow aggregation, even a simple one, that groups accounts by customer and ownership category and flags implausible results;
  • run data-quality rules on the 26 mandatory fields every month, as well as before an FDIC review;
  • make the customer master the owner of the unique customer identifier, and test that no individual holds two identifiers;
  • keep the lineage from each mandatory field back to its system of record, so an examiner’s question can be answered in hours.

The unique identifier deserves particular attention. The proposal says that in section 360.9 compliance reviews the FDIC will assign synthetic Social Security and tax identification numbers and then validate the bank’s process for giving each individual one unique customer number. We expect the FDIC’s engine to aggregate on that number. If one person carries two numbers, coverage may be overstated. If two people share one, it may be understated.

The foreign-deposit definitions

The proposal also rewrites the definitions of foreign deposits to match later changes in part 330. It adds the term “dually payable deposit”: a liability in a foreign office of a U.S. bank that would be a deposit if booked in the United States and is expressly payable here. Such deposits share deposit priority in resolution but are not insured. New terms for domestic deposit, domestic deposit account and foreign deposit account follow. For a bank with foreign branches, the deposit type indicator is no longer a label. It must reflect where and how each liability is payable.

The holds that come back

What a provisional hold does

Section 360.9’s second requirement is mechanical. A covered bank must be able to place provisional holds on deposit accounts automatically after failure, and release them as the FDIC completes its determination. The FDIC adopted the requirement, the proposal says, so it could give depositors liquidity soon after a failure while it finished the insurance determination.

The mechanism is simple to describe. Holds let the FDIC give depositors access to funds soon after a failure while some balances stay frozen. It then releases the holds as its determination concludes. Building the capability means the deposit systems must accept a hold instruction for millions of accounts in a batch, apply it correctly to each product and reverse it the same way.

Why some of these banks no longer have it

When the FDIC adopted part 370, it dropped the hold requirement for banks under that rule. The proposal gives two reasons: part 370 banks already faced heavy costs to build a calculation engine, and the FDIC expected holds to be less needed if determinations were faster. Under 12 CFR 370.8(d), a bank that certified part 370 compliance was released from section 360.9 for as long as it stayed under part 370.

The proposal now concludes that holds may still be relevant. Its reason is direct: under part 370, banks “still are not able to calculate deposit insurance for all accounts promptly upon failure.” Rescinding part 370 would put every larger covered institution back under the hold requirement.

Part 370 has its own requirement to restrict access to deposits until the FDIC makes its determination, which the proposal calls similar to, but not the same as, the section 360.9 hold. Some banks may have kept a capability close to the section 360.9 hold. Others may have let it lapse through core system migrations, product launches and staff turnover. The FDIC’s transition cost estimates do not separate holds from data work, which is a reason to build a bank’s own estimate.

The timetable, and what the 18 months buy

The proposal splits the implementation periods. A bank that is a larger covered institution on the final rule’s effective date would have six months to meet every section 360.9 requirement except holds. It would have 18 months to meet the hold requirement.

Other periods apply to other paths. A bank that newly becomes a larger covered institution after the effective date gets 18 months after the end of the second calendar quarter in which it meets the criteria. An existing covered institution that grows into the larger category gets 12 months after meeting the threshold for two consecutive quarters. The FDIC asks whether a merger that creates a larger covered institution should get 12 or 18 months.

We think 18 months is enough to rebuild a hold function if the work starts early. It is not generous for a bank with several deposit platforms, or with a core conversion already scheduled. Hold logic has to be built and tested on each system of record, and changes to core deposit systems can travel in a vendor’s or the bank’s own release cycle.

The exemption route

Section 360.9 already lets a covered bank ask to meet the hold requirement differently. It can propose memo holds or alternative account mechanisms, or seek an exemption for account systems where manual holds are feasible. The proposal leaves those provisions alone, and larger covered institutions could use them.

That route has a design consequence. A bank that wants to rely on manual holds for a small product system would, we expect, need to show they are feasible at failure, under time pressure, with reduced staff. We would expect an examiner to look for a written procedure, named roles, a tested run book and timing from a rehearsal.

Who owns the hold

At a large bank, the hold capability can fall between teams. Deposit operations may own the procedure, core platform teams the code, and a resolution-planning group the regulatory relationship. Part 370’s release from section 360.9 may have left the capability with no owner at all. The first practical step is to name one, with a test schedule and a budget line that survives the next reorganization.

As a matter of good practice, not a requirement in the proposal, the test itself should mirror failure-day conditions as closely as a bank can manage. That means a run on a full copy of production deposit data, not a sample. It means applying holds to every account in scope, timing the batch on each system and checking that available balances change as intended. It also means reversing the holds and confirming that no account is left frozen or wrongly released. Card authorizations, scheduled payments and sweeps should be included, because they interact with available balances in ways a simple test can miss. A bank that records these timings and exceptions will have the evidence an examiner asks for, and an honest view of whether 18 months is enough.

The pass-through accounts the bank must warn

Accounts with transactional features

The third obligation concerns accounts in which the bank’s records do not hold what the FDIC needs. The clearest case is a pass-through account. A deposit broker, a deposit placement service or another agent opens an account at the bank on behalf of many beneficial owners. Each owner may be insured as if the deposit were held directly, but only the agent knows who they are and how much each holds.

The proposal targets such accounts when they have “transactional features,” a term carried over from part 370. Summarized, an account has them if the holder or beneficial owner can move money to a third party by a method that may post on a later day. The posting day is later than the day the transfer began. Sweeps into such accounts at the same bank count too.

The FDIC’s concern is payments in flight. If insurance on these accounts cannot be calculated promptly, the proposal says, clearing and settlement of transactions already under way could be delayed or disrupted.

Contract or disclosure

Part 370, the proposal says, currently requires both of the actions below. The proposal asks for one of them. A larger covered institution would satisfy it in one of two ways.

  1. It could make a good-faith effort to sign contracts with the account holder. The contracts would oblige the holder to deliver beneficial-owner information to the FDIC within 24 hours after the FDIC’s appointment as receiver, in the format of a new Appendix I to part 360.
  2. It could give the account holder a written disclosure setting out Appendix I’s information and format. The disclosure would state that if the bank fails, the holder might not have access to the deposits until it delivers that information.

The bank would not have to validate the account holder’s ability to produce the file. Appendix I would be completed by the account holder, not the bank, and only at failure.

The exceptions

The proposal excludes several kinds of accounts with transactional features from this duty. They include mortgage servicers’ custodial accounts made up of borrowers’ payments and short-term real estate and title escrow accounts. Interest on Lawyers Trust Accounts and their equivalents are excluded, as are accounts held for employee benefit plans. So are accounts held for formal revocable or irrevocable trusts insured under the trust rules. The FDIC’s reasoning is that the agents behind these accounts already owe their principals good records and an understanding of coverage.

What this means for systems and contracts

The obligation reads like a legal task, but in our assessment most of its work is data and process. A bank first has to find the accounts in scope. That means flagging, in its deposit systems, which accounts are held in an agency capacity and which have transactional features under the rule’s definition. Neither flag may exist in the bank’s product catalog today.

Next comes the choice between contract and disclosure. Contracts give the stronger position at failure but require negotiation with counterparties who may hold accounts at many banks. Disclosures are easier to issue but put the burden of readiness on the holder. Either way, the bank needs a record of what was sent or signed, to whom and when, linked to the account. The proposal asks whether disclosure alone would be enough and what challenges banks expect.

Banking-as-a-service arrangements add a further complication. Where a bank holds accounts for a program manager that serves many end customers, the beneficial-owner ledger may live entirely outside the bank. A disclosure tells that partner what the FDIC will ask for. It does not tell the bank whether the partner could produce it.

Who bears the delay if the file never comes

The disclosure route is honest about one thing. If the account holder does not deliver its Appendix I file, its beneficial owners may wait for their money. The FDIC can pay them only once it knows who they are.

Those owners may be the bank’s customers in every sense except the legal one. They may see the bank’s name on statements or in an app. When access stops, we expect many would call the bank, or the failed bank’s successor, before they call the agent. A bank planning for this case should decide in advance what its own staff and channels would say, and should test whether its larger partners could produce the file at all. A short exercise with a sample of partners would show more than any disclosure log.

Where the argument is weakest, and what to file by January 4

The case that this really is a reduction

The strongest objection to our reading is that the FDIC is right on the numbers. The calculation engine was the expensive part of part 370, and it goes. Testing and certification go with it. The transition cost estimates, even if low, are small next to the annual savings for the five Highest Complexity banks.

That objection holds for the budget. It does not hold for the risk. Part 370’s heavy machinery gave banks a reason, and a regulator-tested method, to keep their ownership data in order. The proposal keeps the duty to hold complete data for 26 fields but removes the bank’s own view of what that data produces. Banks that cut the controls along with the engine will find out about errors later and in a worse setting.

The case that the FDIC is taking on concentration risk

A second objection runs the other way. If every large bank’s determination now depends on one FDIC system, that system becomes a single point of failure for the deposit insurance promise. The proposal offers no technical description of the system’s capacity or testing.

We think this concern is fair and belongs in comment letters. It is not a reason for a bank to oppose rescission, since the bank cannot run the FDIC’s engine for it. It is a reason to ask the FDIC to describe how it tests its engine against larger covered institutions’ files, how often, and what it would share with each bank about the results. A bank that knows how its data performed in the FDIC’s engine can fix problems before a failure, not after.

The case that ordinary covered banks are untouched

The proposal says it would neither add requirements nor add burden for the 117 banks already under section 360.9. Several changes still reach them. The rule would codify automatic release after three consecutive quarters below the threshold, which the FDIC already does by notice. It would index the thresholds to inflation. It would revise the foreign-deposit definitions. And the FDIC’s new data retention practice applies to their compliance reviews too.

Each of these is minor. The indexing change does mean a bank near the threshold should track the indexed values, not the round numbers it remembers, when it forecasts whether the rule will apply.

How the change differs by type of bank

The five Highest Complexity banks gain the most and face the largest transition, by the FDIC’s own figures. For them the savings are measured in millions a year. So is the transition cost, and the hold rebuild touches the most systems. In our view, their comment letters should carry weight because their data is the hardest test of the FDIC’s engine.

The Lowest Complexity banks face a different calculation. They have the smallest savings and the smallest transition costs. They may also have smaller resolution-planning teams. For them, the risk is that the change is treated as an accounting event and the data controls lapse with the engine.

Banks near the two-million-account line face a third problem: moving in and out of scope. The proposal notes that institutions have moved in and out of scope of one rule or the other over one or two quarters. It would release a larger covered institution from the extra requirements after three consecutive quarters below the line, and bring it back after two consecutive quarters above. A bank growing by acquisition should model its account count across the deal, because a merger can trigger the 12-month or 18-month clock.

Banks with foreign branches have a fourth concern, the deposit type indicator. Under the revised definitions, each foreign-office liability must be classed as dually payable or not. That classification depends on the account’s terms, which may sit in documents rather than in a system field.

The data retention change

The retention change deserves a separate note because it is already in effect. The FDIC used to keep section 360.9 review data until the next review, and part 370 test data for up to 120 days after its final management letter. It now says it will fully delete all depositor and customer data no later than 60 days after a section 360.9 or part 370 compliance review.

Dropping real Social Security and tax identification numbers from reviews reduces what a bank sends and what the FDIC holds. It also changes the test. The FDIC says it will assign synthetic identifiers and validate the bank’s process for assigning unique customer numbers. A bank should expect review questions on how it prevents and detects duplicate customer records, and should have the evidence ready.

What to put in a comment letter

The proposal lists numbered questions. For a larger covered institution, the most useful answers are concrete and drawn from its own systems.

  • On the 26 mandatory fields, say which are hard to populate and why, with counts of affected accounts if the bank can share them.
  • On holds, say whether 18 months is enough given the bank’s platforms and release calendar, and whether any core migration falls inside that window.
  • On pass-through accounts, say whether disclosure alone would reach the counterparties that matter, and how banking-as-a-service programs should be treated.
  • On FDIC capability, ask for a description of testing and for feedback to each bank on how its files performed.
  • On exemptions, describe the account types that would need one, so the final rule can name them.

Comments must reach the FDIC by January 4, 2027. A bank that intends to comment should give the drafting to the people who own the deposit data, with counsel reviewing, not the other way round.

With hindsight

Little time has passed since the proposal appeared, and it is still only a proposal. As of November 14, 2026, the comment period is open and runs to January 4, 2027. Part 370 remains in force, and so does the release from section 360.9 that it gives certified banks. Nothing in the proposal changes a bank’s obligations until a final rule takes effect.

One change was announced as effective immediately: deletion of depositor and customer data no later than 60 days after a section 360.9 or part 370 compliance review. The FDIC also said it will no longer require real Social Security and tax identification numbers in section 360.9 compliance reviews and will assign synthetic identifiers instead. Banks with reviews scheduled can expect that approach.

Several things are not yet known. The proposal does not say when the FDIC expects to finalize the rule. It does not list the part 370 frequently asked questions the FDIC plans to recast as section 360.9 guidance, and asks commenters which ones should be. It does not describe in technical terms the systems the FDIC says make the rescission possible. And the comments themselves are not yet in, so it is too early to say whether the threshold for larger covered institutions, the list of mandatory fields or the hold timetable will change.

The timetable in the proposal runs from the final rule’s effective date, which does not exist yet. If the rule is adopted as proposed, a bank that is a larger covered institution on that date would have six months for the data requirements and 18 months for the hold requirement. Banks should plan against those intervals without assuming a start date.

The proposal sits among other bank proposals with comment periods running into the winter. On November 13, the FDIC, the Federal Reserve Board, the National Credit Union Administration and the Office of the Comptroller of the Currency acted on another proposal. They extended the comment period on their proposed third-party risk management guidance to December 16. The original deadline had been November 16. For a bank whose deposit systems are run by a vendor, both proposals touch the same relationship: who builds the hold logic and the standard files, and on what schedule.

The comment file will be the next public evidence. The proposal says comments will be posted on the FDIC’s website, with personal information included as submitted. We will look in letters from the largest banks and their trade associations for how much of the 26-field requirement they regard as already met, and for how many banks let the hold capability lapse while under part 370. Neither number is public today.

What would show whether our reading holds is mostly in the future. If the final rule keeps the mandatory fields and the hold requirement, the banks leaving part 370 will be judged on data they send to an engine they cannot see. Those that kept their own aggregation and data-quality checks will learn about errors from themselves. Those that cut them will learn from the FDIC. We expect to return to the rule when the comment file is public and again when the FDIC acts on it.

Spotted an error? Tell us and we will check it and correct the article where needed.

  • FDIC
  • Part 370
  • deposit insurance
  • data quality
  • core banking