A 2008 rule let big banks leave data fields blank; the 2016 Part 370 proposal made each gap need FDIC approval

The February 26, 2016 proposal said large banks reporting under a 2008 rule could leave fields null, and their files were frequently incomplete. For banks with two million or more accounts it demanded complete ownership data, a calculation run on it, approved gaps and certification.

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 February 26, 2016 proposal said large banks reporting under a 2008 rule could leave fields null, and their files were frequently incomplete. For banks with two million or more accounts it demanded complete ownership data, a calculation run on it, approved gaps and certification.

  1. The week The proposal targeted banks with two million or more accounts.
  2. The data: knowing every depositor’s right and capacity Deposit insurance is calculated per depositor, per ownership right and capacity, not per account. [1] The FDIC must aggregate all deposits a depositor holds in the same right and capacity and then apply the $250,000 limit. [1] That aggregation requires knowing that two accounts belong to the same person and that they fall in the same category.
  3. The calculation: what the bank’s systems had to do The proposal set a performance standard.
  4. The exceptions: where the bank could not know enough Some deposits are insured on a pass-through basis.
  5. The proof: certification and testing A certification is a statement by the bank that its systems and data meet the rule.
  6. How the proposal reached different institutions The proposal applied to FDIC-insured institutions with two million or more deposit accounts. [1] Others met it through their relationships with those banks.
  7. Objections, in order The final rule estimated the cost to covered institutions at $386 million and called it small relative to their revenues and expenses. [2] The 2026 proposal estimated that rescinding Part 370 would save the industry about $33 million a year. [4] Both figures are the FDIC’s own estimates and depend on its assumptions.
  8. Checks a bank data or technology leader could run Measure the share of deposit accounts linked to a single, consistent identifier for each owner, across all deposit systems.
  9. With hindsight The comment period ran longer than first set.

The week

On Friday, February 26, 2016, the Federal Deposit Insurance Corporation published a proposed rule titled Recordkeeping for Timely Deposit Insurance Determination. [1] The notice was dated February 17, 2016, by order of the agency’s Board of Directors, and the notice set a comment deadline of May 26, 2016. [1]

The proposal targeted banks with two million or more accounts. They would have to keep complete and accurate data on each customer’s ownership interest by right and capacity, for every account. [1] They would also have to be able to compute insured and uninsured amounts for each owner, for the agency to use if the bank failed. [1]

The core of the notice was a confession about an older rule. In July 2008 the agency had adopted section 360.9, which asked large banks to supply standard data files and to be able to place holds on accounts. [1] Those banks were allowed to populate the fields using only preexisting data elements. [1] If a bank did not keep the information for a field, it could enter a null value. [1]

The notice said what came of that. As a result of this discretionary approach, it said, the standard data files were frequently incomplete. [1] It said years of work with the banks covered by the 2008 rule had led the agency to conclude that its data sets and other requirements were not sufficient for the largest failures. [1] It said the 2008 rule had not been as effective as had been hoped. [1]

The notice described the kinds of gaps it found. It named multiple accounts under one name at different addresses, and different names at one address. [1] It named accounts opened through a broker or custodian where only the agent appears, and trust beneficiaries not recorded in a structured way. [1] It said trust records were often on paper or scanned images that required manual review. [1]

It blamed consolidation for part of the problem. From 2004 to 2014, it said, the largest number of accounts held at a single institution rose 119 percent. [1] Banks that grew by acquisition had inherited legacy systems with missing or inaccurate records, sometimes incompatible with the buyer’s platforms. [1] The notice said those banks could still run daily operations without cleaning up the inherited files. [1]

It also recalled the speed of the last crisis. It said that since 2009 the agency had been called on to resolve 47 institutions within 30 days from the start of the resolution process to closing. [1] It noted that the condition of Washington Mutual and Wachovia had deteriorated very quickly, and cited their final Call Reports showing 42 million and 29 million accounts. [1]

The notice followed an advance notice published on April 28, 2015, whose comment period ended on July 27, 2015. [1] The agency said it received 10 comment letters and held six meetings or calls with banks, trade groups and software providers. [1] It projected that 36 institutions would be covered, each holding between 2 million and 85 million accounts. [1]

The most revealing sentence in the February 2016 proposal is about the rule it was replacing. The FDIC’s 2008 rule let large banks fill standard data fields using only preexisting data, and use a null value when they lacked the information. [1] The FDIC said that, as a result of this discretionary approach, the files were frequently incomplete. [1]

Our argument is that the 2016 proposal was, at bottom, a lesson in data standards: a field that may be left blank will be left blank. The proposal closed that door in four ways. [1] It required complete and accurate data, made the bank prove the data worked by computing on it, allowed gaps only by FDIC approval with notice to depositors, and backed it all with certification and testing. [1] For bank data leaders, that design applies well beyond deposit insurance.

We take the rule in four parts:

  1. The data: what a covered bank had to know about each depositor.
  2. The calculation: what the bank’s systems had to do on the night of failure.
  3. The exceptions: where the bank could not know enough, and what it had to do instead.
  4. The proof: how the FDIC would confirm a bank was ready.

The parts do not overlap. The first is records, the second processing, the third the gaps and the fourth assurance. After them we look at how the proposal reached different kinds of institution, take the objections in order and end with checks a bank’s data or technology leader could run.

The data: knowing every depositor’s right and capacity

Why coverage depends on data the bank may not hold

Deposit insurance is calculated per depositor, per ownership right and capacity, not per account. [1] The FDIC must aggregate all deposits a depositor holds in the same right and capacity and then apply the $250,000 limit. [1] That aggregation requires knowing that two accounts belong to the same person and that they fall in the same category.

The proposal gave examples of what goes wrong. The FDIC might find multiple accounts under one name at different addresses, or different names at the same address. [1] Accounts opened through a broker or custodian might name only the agent. [1] Trust beneficiaries might not be recorded in a structured way. [1] Each gap meant manual work after failure, and possibly delay.

Legacy systems after mergers

The proposal tied the data problem to how the largest banks grew. Banks that grew through acquisition inherited the legacy deposit systems of the banks they bought. [1] Those systems might have missing and inaccurate account data, or might not be compatible with the acquirer’s systems, leaving multiple deposit platforms. [1]

The FDIC observed that the largest institutions could still run their banking operations without spending to integrate those systems or update the acquired files. [1] In our reading, that observation explains why the rule was needed. A data problem that does not hurt daily operations has no internal sponsor. A failure would expose it all at once, on the worst possible night.

The unique identifier

The proposal’s answer started with identity. It used the term unique identifier for a number associated with an individual or entity that a bank can use to monitor its deposit relationship with only that individual or entity. [1] It said a Social Security number, taxpayer identification number or other government-issued number, such as a passport or visa number, could be used if the bank used it consistently and only for that person. [1]

Each depositor and each party in a deposit relationship would need a unique identifier in the bank’s system, “within two years after the effective date of a final rule.” [1] In our reading, this was the hardest data task in the proposal for an institution built by acquisition. The same customer may appear in several legacy systems under different keys.

The ownership code

The second element was a code. The proposal would require a covered bank to assign an account ownership right and capacity code to each deposit account within two years. [1] The code tells the calculation which coverage category applies: single, joint, business, trust and so on.

Where a bank lacked the information to assign the code or identify the owners, the proposal said it would need to obtain it. [1] If it could not, it could apply to the FDIC for an extension or exception where permitted. [1]

The categories behind the code

The FDIC’s 2026 proposal gives a sense of which categories are routine. It said a large bank should, in the normal course of business, already have the information for several kinds of account. [4] It listed single and joint ownership accounts, accounts owned by a corporation, partnership or unincorporated association, and informal revocable trust accounts such as payable-on-death accounts. [4]

The harder categories were the ones the 2016 proposal named: pass-through accounts and formal trusts. [1] For those, the information may sit with an agent or a trustee rather than the bank. The code is easy to assign when the facts are in the bank’s records and hard when they are not.

Complete and accurate

The key phrase was complete and accurate. The 2008 rule let banks use only preexisting data and leave fields null. [1] The proposal would require covered banks to maintain complete and accurate depositor information for all deposit accounts. [1]

That moved the standard from reporting what the bank happens to have to maintaining what the FDIC needs. For a data team, it is the difference between an extract and a data quality program. The extract can be built in months. The program has to change account opening, maintenance and system migration so that the data stays complete.

The presumption in the records

The proposal restated a principle of the coverage rules. The FDIC must presume that deposits are owned as the institution’s deposit account records show. [1] If the records disclose a relationship that provides a basis for additional insurance, its details must be ascertainable from the institution’s records or from records kept by the depositor or a designated party. [1]

The proposal also said deposit insurance would not be withheld just because fiduciary details were missing from a covered bank’s records. [1] The consequence of missing data was delay, not loss of coverage. That distinction runs through the whole rule: Part 370 was about speed, and the bank’s data decided how fast the FDIC could move.

Data quality as a side benefit

The FDIC noted that commenters on its 2015 advance notice saw benefits beyond resolution. [1] The benefits commenters identified included better compliance with anti-money laundering and Bank Secrecy Act requirements, because banks would need to identify the beneficial owners of various types of accounts. [1]

In our reading, that point shows why a completeness duty pays twice. A bank that knows exactly who owns each deposit, under one identifier, can use that knowledge for customer due diligence, sanctions screening, privacy requests and marketing consent. The same data that serves the FDIC on the night of failure serves the bank every other day.

The calculation: what the bank’s systems had to do

Within 24 hours

The proposal set a performance standard. A covered bank’s IT system would need to calculate deposit insurance coverage for each deposit account and adjust account balances within 24 hours after the FDIC’s appointment as receiver. [1] Commenters had predicted that the calculation would take at least 24 hours even if limited to single, joint and business accounts. [1]

The FDIC had considered asking only for closing night deposits. Its 2015 advance notice had asked what accounts should count. [1] The proposal recorded commenters’ suggested definitions, one of which would have covered about 70 percent of a commenter’s accounts and 55 percent of balances from its core systems. [1] It chose to cover all accounts instead. [1]

Why speed mattered

The proposal gave five reasons for prompt payment. It said prompt payment maintains public confidence, lets depositors meet their financial obligations and avoids possible systemic consequences when one of the largest institutions fails. [1] It added that delay could reduce the failed bank’s franchise value, raising the cost to the Deposit Insurance Fund. [1] Prompt payment would also reduce the chance of disruptions to check clearing and direct debits during resolution. [1]

The FDIC also said it strives to make most insured deposits available by the next business day after a bank fails, although the statute sets no specific time period. [1] In our reading, the 24-hour standard for covered banks was built to meet that practice at the largest scale.

The bridge bank option

The proposal linked speed to the FDIC’s resolution choices. To establish a bridge depository institution, which it called a likely strategy for large complex institutions, the FDIC must generally be able to determine insured and uninsured deposits rapidly. [1] It said having that option lets it transfer parts of a failed bank to smaller institutions rather than sell the whole bank to another large bank. [1]

In our reading, that is the policy reason behind the data requirement. Without fast, accurate insurance data, the FDIC’s options at a large failure narrow. The proposal said confidence that any size bank can fail without systemic disruption would reduce the moral hazard that might lead the largest banks to take excessive risks. [1]

Why the bank, not the FDIC

The proposal was candid about the reason. Using the FDIC’s IT system at a bank with millions of accounts would require transmitting massive amounts of deposit data. [1] The time needed to transmit and process it would be a significant impediment to a timely determination. [1] So the proposal asked banks to upgrade their systems so that the FDIC could perform the determination on all accounts without a data transfer. [1]

Commenters had objected that this transferred the FDIC’s statutory responsibility to the banks. [1] The FDIC disagreed. It said it would use the bank’s capabilities only after failure, so it would not be delegating its responsibility. [1] Record keeping, it said, is the covered institutions’ responsibility. [1]

In our reading, that exchange shows the design choice exactly. The determination stayed the FDIC’s. The machine that would make it moved into the bank. That placement was a judgment about where the computing could be done fastest in 2016, not a principle of deposit insurance.

What the system must do

The final rule made the functions explicit. As the current text reads, a covered institution’s system must be able to calculate coverage for each account under the FDIC’s coverage rules and generate output records in a prescribed format. [5] It must also be able to restrict access to deposits until the FDIC decides, and debit the uninsured amount. [5]

Each function is a change to core banking processes. Restricting access touches payments, cards and online banking. Debiting uninsured amounts touches the general ledger and customer statements. These are not functions a bank uses on any normal day, which makes them hard to keep working.

Holds, restrictions and debits

The 2008 rule had required covered institutions to be able to place and release holds automatically if the FDIC could not make a determination by the next business day. [1] Part 370 went further: the bank’s system would calculate coverage, restrict access and debit the uninsured amount from each account. [1] [5]

The difference is between pausing and settling. A hold freezes a balance while someone works out the answer. A debit applies the answer. A system that can debit the uninsured portion of millions of accounts in one run must be right, because the customer sees the result as a change to their balance.

Keeping it working

A capability used only on the night of failure must still survive every system change. The proposal said more frequent FDIC testing might be needed for banks that make major acquisitions or undertake major IT system conversions. [1] That is a recognition that the calculator could break whenever the bank changed anything around it.

The 2026 proposal named this cost directly. It said removing Part 370 would relieve covered banks of operating, maintaining, continuously enhancing and remediating deficiencies in systems capable of calculating coverage within 24 hours. [4]

Batch windows and timing

Commenters raised operational timing. The proposal recorded a comment that a bank deposit servicer would normally require 24 hours’ notice to run batch processing. [1] A failure announced at the end of a business day leaves little room for that.

For a technology leader, the 24-hour standard is a capacity plan as much as a data plan. The calculation has to run alongside, or instead of, normal end-of-day processing, on systems sized for it, with staff available to run it.

The exceptions: where the bank could not know enough

Pass-through and trust accounts

Some deposits are insured on a pass-through basis. When an agent or custodian holds deposits for others, coverage passes through to each actual owner. [1] In those cases, the proposal noted, the only party named in the bank’s records may be the agent or custodian. [1]

The FDIC considered two options for such accounts in its advance notice. [1] One was to place temporary holds on the whole balance at failure and have the agents submit owner information later, a process the proposal called labor-intensive and generally requiring that depositors’ access be restricted. [1]

The exception process

The proposal gave covered banks four kinds of relief to request. [1] They were an extension of deadlines, an exception from information requirements for certain accounts, an exemption if all its deposits were fully insured, and release if its account count fell. [1]

Exceptions had conditions. A bank could apply if it lacked the information for certain accounts, had asked account holders for it and received no response. [1] It could also apply if it had a reasoned legal opinion that the information was protected from disclosure by law. [1]

Information that changes too fast

A third ground for exception addressed churn. A bank could apply if the information needed to calculate coverage changed so frequently that continual updating would be neither cost effective nor technologically practicable. [1] The FDIC said it anticipated that the rate of change would need to be daily or near daily. [1]

An application for an exception would need to describe the affected accounts, the number of accounts and the total dollars on deposit in them. [1] For a data team, that means the exception itself requires good data. A bank cannot ask to be excused from knowing its depositors without knowing how many accounts it is talking about.

The pending file

The final rule settled how unresolved accounts would be handled. For accounts kept under the alternative recordkeeping requirements, a bank had to keep enough information to populate a pending reason field in a pending file produced by its system. [2] Where the beneficial owner of an instrument could not be identified, the bank would keep only the pending reason code, to show more information was needed. [2]

In our reading, the pending file is the honest part of the design. It lists what the bank does not know, account by account, and why. A bank that can produce an accurate pending file knows the edges of its own data, which is a core aim of any data governance program.

Telling depositors about delay

An exception came with a duty to the customer. The proposal would require a covered bank, as a condition of an exception, to notify affected account holders that payment of deposit insurance may be delayed until the needed information is provided to the FDIC. [1]

The proposal also made clear that an exception did not reduce coverage. It said the rule would not change the coverage standards in part 330, and that a bank’s inability to obtain the information, or an approved exception, would not reduce pass-through coverage. [1] The exception affected timing, not the amount. [1]

Prepaid cards

Prepaid cards were a named problem. The proposal discussed closed-loop and open-loop cards and payroll cards associated with an employee’s name, address and a unique identifier. [1] It noted the identifier might be a passport number for a foreign national. [1] It said banks could request an extension or exception for accounts, including various prepaid cards, where depositor information would truly be unavailable or infeasible to collect. [1]

Thresholds and movement in and out

Coverage turned on account counts reported in the Call Report. A bank would be covered if it had two million or more deposit accounts for the two consecutive quarters before the effective date, or later for any two consecutive quarters. [1] It could apply for release after three consecutive quarters below two million. [1]

The FDIC said the threshold followed from the greater challenge the largest institutions pose and their lower implementation cost per account. [1] It said it did not contemplate imposing these requirements on community banks. [1] It also said it was improving its own systems for banks below the threshold. [1]

The proof: certification and testing

Annual certification

The proposal would require covered institutions to certify compliance annually. [1] It said a failure to meet the requirements would result in enforcement action under section 8 of the Federal Deposit Insurance Act. [1]

A certification is a statement by the bank that its systems and data meet the rule. It needs evidence behind it: test results, data quality measures and an account of exceptions. The officer who signs it relies on the teams who produce that evidence.

FDIC testing

The FDIC described its testing plan. It expected one initial on-site testing visit, then visits no more often than annually once the initial test succeeded. [1] It proposed no testing during the two-year implementation period. [1] Banks would run their own in-house tests annually and give the FDIC verification, a summary of results and a certification that the functionality could be implemented. [1]

Commenters asked for clear testing standards and preferred off-site testing with attestation. [1] They said even basic testing would take a minimum of 12 hours and many staff. [1] The FDIC acknowledged that testing would add demands on staff and systems and said it had tried to minimize the burden. [1]

Enforcement behind the certificate

The proposal tied the certificate to the FDIC’s general enforcement power. It said failure to meet the requirements would result in enforcement action under section 8 of the Federal Deposit Insurance Act. [1] It also said a bank that did not obtain the information or an exception could face sanctions under that section. [1]

That made the certification a supervisory matter. A bank that certified on thin evidence risked more than a failed test. Boards and audit committees had reason to ask what stood behind the signature.

What a test proves

A test proves that the calculation ran on the day of the test, on the data as it stood. It does not prove the data was right. A complete test program pairs the run with data quality checks on a sample of accounts against source documents.

That pairing is the bridge between the two halves of the rule. The calculator can only be as good as the data. A bank that tests the calculator and not the data has tested the easier half.

Testing after change

The proposal said more frequent testing might be needed for banks that make major acquisitions, experience financial distress or undertake major IT system conversions. [1] Each of those events is a moment when the data or the calculator could break.

For a bank, that list doubles as a trigger list for its own testing. A conversion plan, an acquisition integration plan and a recovery plan for financial stress should each include a Part 370 test before the change is declared complete.

Implementation time

The proposal set its main deadlines “within two years after the effective date of a final rule.” [1] Commenters had said banks would need at least four years with potential extensions. [1] The FDIC kept two years and pointed to the extension and exception process for banks that needed more time. [1]

How the proposal reached different institutions

The proposal applied to FDIC-insured institutions with two million or more deposit accounts. [1] Others met it through their relationships with those banks.

A large bank built by acquisition

This was the case the FDIC described: multiple deposit systems inherited through acquisitions, with missing or inaccurate data. [1] The unique identifier requirement forced a decision about which system is the record for each customer. [1] The data program was the main cost, and the FDIC’s final rule later estimated almost half of the rule’s total cost would go to legacy data clean-up. [2]

A regional bank approaching two million accounts

A growing bank could cross the threshold through growth or a merger. [1] Its planning question was when, because the clock started when it became covered. [1] A merger plan that would cross the line needed the Part 370 build in its integration budget.

A bank with deposit-broker or fintech partners

A bank taking deposits through agents or partners held pass-through accounts whose owners might appear only in the partner’s records. [1] It needed either the owner data on its own books or the alternative arrangements the rule allowed, plus notices to account holders where an exception applied. [1]

A broker-dealer with a bank sweep program

A broker-dealer that sweeps customer cash into bank deposits is an agent in the FDIC’s terms, and its customers may be insured on a pass-through basis. [1] The receiving bank’s Part 370 obligations reached into the broker’s records. The broker became the source of the data its bank partners had to certify they could obtain.

A trust department or trust company

Trust accounts were a named difficulty: the proposal said beneficiary data might be missing and trust records were often on paper. [1] A bank trust department became a data supplier to its own institution’s deposit systems.

A prepaid card program manager

Program managers that hold cardholder data outside the bank were in the same position as deposit brokers. [1] The bank’s ability to identify cardholders depended on the manager’s records and on the contract between them.

A bank with many small accounts

The threshold counted accounts, not assets. [1] A bank with a large number of small accounts could be covered even if its balance sheet was modest. The proposal allowed a covered institution to seek an exemption if all the deposits it takes are fully insured, and said a relatively small covered institution that met that condition could apply. [1]

A core processing vendor

The FDIC met with software providers during the advance notice stage. [1] Banks that run deposits on a vendor’s core platform depend on that vendor to deliver the calculation, the hold and debit functions and the output formats. A bank’s certification therefore rests in part on its vendor’s roadmap and testing.

A bank’s privacy and security teams

The proposal said a Social Security number, taxpayer identification number or other government-issued number could serve as the unique identifier. [1] Using such numbers as a join key across every deposit system spreads a sensitive field into more places. The privacy and security teams had a stake in how the identifier was stored, masked and accessed in the systems built for the rule.

A community bank

The FDIC said it did not contemplate imposing the requirements on community banks. [1] A community bank still met the underlying problem through section 360.9, if it was covered by that rule’s thresholds, and through the FDIC’s own determination process. [1]

Objections, in order

1. The FDIC was outsourcing its job

Commenters made this argument. [1] The FDIC answered that the banks would supply data and capability, and the FDIC would still make the determination. [1] Its 2026 proposal would have the FDIC calculate insurance using its own internal IT systems, which in our reading shows the placement was a choice about capacity rather than about responsibility. [4]

2. Two years was not enough

Commenters asked for at least four years. [1] The final rule gave three. [2] In 2019 the FDIC added an optional one-year extension for the original covered banks, saying many would otherwise need to expend significant resources on an expedited basis. [3]

3. The cost was too high for a once-in-a-generation event

The final rule estimated the cost to covered institutions at $386 million and called it small relative to their revenues and expenses. [2] The 2026 proposal estimated that rescinding Part 370 would save the industry about $33 million a year. [4] Both figures are the FDIC’s own estimates and depend on its assumptions.

4. Data quality was the bank’s business anyway

In part it was: commenters saw anti-money laundering benefits. [1] But the 2008 rule had let banks report null fields, and the FDIC said the resulting files were frequently incomplete. [1] The proposal turned a business preference into a regulatory duty with a certification behind it. [1]

5. The rule would never be used

A rule built for a rare event is judged by readiness, not use. The FDIC cited the speed of the Washington Mutual and Wachovia deteriorations, and said that since 2009 it had resolved 47 institutions within 30 days from the start of resolution to closing. [1] The readiness case rested on how little warning there might be.

6. Account count was the wrong test

The FDIC conceded that the number of accounts is only one dimension of the problem. [1] It said it is the most readily measured one, and that it is highly correlated with complex account relationships and multiple deposit systems. [1] It said the threshold captured one half of one percent of insured institutions but those where prompt determination was hardest. [1] Commenters had proposed other lines, including $10 billion in assets and 100,000 accounts. [1]

7. Customers will not supply the information

The proposal expected that some would not. It allowed an exception where a bank had asked an account holder for the information and the account holder had not responded. [1] The bank then had to tell those account holders that payment of their insurance could be delayed. [1] The burden of the gap fell on the customer’s timing, with notice, rather than on the bank’s compliance.

Checks a bank data or technology leader could run

Identifier coverage

Measure the share of deposit accounts linked to a single, consistent identifier for each owner, across all deposit systems. The 2016 proposal’s unique identifier is the reference point. [1]

Ownership code completeness

Measure how many accounts carry a valid ownership right and capacity code, and how many are unknown or defaulted. [1]

Pass-through inventory

List every program where deposits are held through an agent, broker, fintech or custodian, and where the owner data is kept. Confirm the contract requires the partner to deliver it on demand. [1]

Migration controls

Add deposit insurance data fields to the acceptance tests for every core system conversion and acquisition integration, so data quality does not drop with each change.

Batch capacity

Confirm that the systems that would run a failure-night calculation, holds and debits have the capacity to do so alongside normal processing, and that the staff who would run them are named.

Exception register

Keep a register of account types for which the bank relies on account holders or partners for owner data, with account counts and balances, since any exception request must state them. [1]

Identifier protection

Where a government-issued number serves as the owner identifier, confirm it is protected in every system that uses it for matching, with access limited to the processes that need it. [1]

Board reporting

Report deposit insurance data quality to the board or its risk committee on a schedule, with the identifier and ownership code measures above and the count of accounts in the pending file. A measure the board sees is a measure someone owns.

Partner contracts

For each deposit partner, confirm the contract states the format and deadline for delivering owner data after a failure. The 2016 proposal and the 2026 proposal both rely on account holders and agents delivering information quickly. [1] [4]

Regulatory watch

Track the November 2026 proposal. If it is finalized, plan the transition to section 360.9’s formats and provisional holds, and keep the depositor data the revised rule would still require. [4]

Evidence for certification

Assemble the evidence behind each annual certification in one file: test results, data quality measures and the exceptions in force. The signing officer should be able to see it.

With hindsight

The comment period ran longer than first set. In a notice published on May 26, 2016, the FDIC extended it to June 27, 2016, saying it wanted to give the public more time, particularly on the estimated cost of compliance. [6]

The FDIC’s Board dated the final rule November 15, 2016, and it was published on December 5, 2016, effective April 1, 2017. [2] It required covered institutions to configure their IT systems to calculate insured and uninsured amounts by ownership right and capacity, and to maintain complete and accurate information for each deposit account, except as otherwise provided. [2] The final rule estimated it would apply to 38 institutions, which together held more than $10 trillion in assets and over 400 million deposit accounts. [2]

The final rule changed the timeline and some mechanics. It said a covered institution “has three years after the effective date for implementation.” [2] It allowed alternative recordkeeping for pass-through and certain trust accounts, with extra provisions for accounts with transactional features. [2] Where the proposal had planned FDIC visits no more often than annually, the final rule said testing would occur no more frequently than on a three-year cycle, absent material changes. [1] [2] It estimated the cost to covered institutions at $386 million, an average of 93 cents per deposit account. [2]

The compliance date proved hard to meet. The FDIC’s July 30, 2019 amendments said the original covered institutions faced an April 1, 2020 compliance date. [3] The FDIC said many of them would be unable to meet it without expending significant resources on an expedited basis. [3] It added an optional one-year extension, to no later than April 1, 2021, upon notification. [3]

As checked on November 16, 2026, 12 CFR part 370 remains in the Code of Federal Regulations. [5] Its current text still requires a covered institution’s system to calculate coverage, generate output files, restrict access and debit uninsured amounts within 24 hours after the FDIC’s appointment as receiver. [5]

In November 2026 the FDIC published a proposal, dated by its Board on October 30, 2026, that would remove part 370 and fold the banks it covers into a revised section 360.9. [4] As checked on November 16, 2026, it is a proposal with comments due by January 4, 2027, and part 370 remains in force. [4] [5] The FDIC’s analysis, based on Call Report data as of March 31, 2026, counted 31 institutions currently subject to part 370. [4]

In our reading, the 2026 text is a test of the blank-field lesson, and the lesson held. For banks with at least $2 billion in deposits and two million or more accounts, the proposal would require complete and accurate data for 26 mandatory data fields. [4] If such a bank lacked the data for those fields, it would have to collect and maintain it, except for certain types of accounts. [4]

For other fields, the 2026 text returns to the 2008 approach. It says a larger bank would provide non-mandatory fields in the standard format only if it already has the information, and need not collect anything it does not already have to populate them. [4] In our reading, the FDIC drew the line exactly where the 2016 proposal located the failure: fields it must have are mandatory, and fields it can live without may stay blank.

What held up is the 2016 diagnosis that optional data standards produce incomplete data. [1] The 2019 extension and the final rule’s own cost estimate show how much legacy clean-up that diagnosis implied. [2] [3] In our view, a bank data leader can take the design straight into other programs: name the fields that must be complete, measure them, approve every gap and certify the result.

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

  • FDIC
  • Part 370
  • deposit insurance
  • depositor data
  • pass-through accounts
  • bank resolution