The 2016 bank cyber ANPR floated a two-hour recovery clock; 2020 sound practices left tolerance to each board
On October 19, 2016, three banking agencies sought comment on cyber standards for large firms, including a tested two-hour recovery objective for sector-critical systems. The rulemaking was withdrawn in 2019, and 2020 sound practices had each board set its tolerance for disruption.
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
On October 19, 2016, three banking agencies sought comment on cyber standards for large firms, including a tested two-hour recovery objective for sector-critical systems. The rulemaking was withdrawn in 2019, and 2020 sound practices had each board set its tolerance for disruption.
- The week The release described the scope under consideration.
- The origin: a clearing and settlement clock The notice quoted the Sound Practices Paper’s definition.
- The event: corruption is not a power cut The notice’s sentence named three kinds of cyber event: disruptive, corruptive or destructive. [2] They are different problems.
- The reach: which systems and whose The higher tier was defined by systems, not firms. [2] A covered entity would have some systems under the sector-critical standards and the rest under the enhanced standards.
- The proof: tests and measures The notice said the two-hour objective would be validated by testing. [2] Testing programs would include severe but plausible scenarios. [2] For a corruptive event, a meaningful test would need to start from altered data and end with verified data.
- How the notice reached different firms For a holding company above the thresholds, the notice meant a board-approved, enterprise-wide cyber risk management strategy built into its overall business strategy. [2] It meant board approval of cyber risk appetite and tolerances. [2] It also meant the two-hour objective for any sector-critical systems it ran.
- Objections, in order The agencies did present it as a consideration and asked for comment. [2] But advance notices shape later proposals.
- What a bank’s technology and risk leaders could check today A firm could check whether its recovery objectives distinguish disruption, destruction and corruption.
- With hindsight The comment period was extended.
The week
On October 17, 2016, the Federal Register carried notice of an FDIC board meeting set for 10:00 a.m. on Wednesday, October 19, 2016. [3] Its discussion agenda had one item: an advance notice of proposed rulemaking on interagency enhanced cyber risk management standards. [3]
On October 19, 2016, the Federal Reserve Board, the OCC and the FDIC announced that they had approved that advance notice. [1] Their joint release said it invited comment on potential enhanced cybersecurity risk-management and resilience standards for large and interconnected entities under their supervision. [1] It said the standards would apply as well to services that third parties provide to those firms. [1]
The release described the scope under consideration. It named depository institutions and holding companies with total consolidated assets of $50 billion or more, and the U.S. operations of foreign banking organizations with total U.S. assets of $50 billion or more. [1] It also named financial market infrastructure companies and nonbank financial companies supervised by the Board. [1] It said the standards would not apply to community banks. [1]
The release said the standards would be tiered. A higher set would apply to systems that provide key functionality to the financial sector. [1] For those sector-critical systems, it said, the agencies were considering requiring firms to substantially mitigate the risk of disruption or failure due to a cyber event. [1]
The release said the agencies chose an advance notice so they could hear comments on all aspects before developing a more detailed proposal. [1] It said they also sought comment on methods to quantify cyber risk and compare it across firms. [1] Comments were due January 17, 2017. [1]
The notice itself set out five categories of standards. They were cyber risk governance, cyber risk management, internal dependency management, external dependency management, and incident response, cyber resilience and situational awareness. [2]
For sector-critical systems, the notice described two specific standards. The first was to minimize residual cyber risk by implementing the most effective, commercially available controls. [2] The second was a recovery time objective of two hours, validated by testing, to recover from a disruptive, corruptive or destructive cyber event. [2]
The notice explained where the two hours came from. It cited the Sound Practices Paper issued by the Board and the OCC with the SEC, which set a two-hour recovery objective for core clearing and settlement activities. [2] It said that paper’s scope was limited to resumption of clearance and settlement in wholesale markets. [2]
The notice proposed to borrow that paper’s thresholds to find sector-critical systems. It considered systems that support clearing or settlement of at least five percent of the value of transactions, on a consistent basis, in named critical markets. [2] Those markets were federal funds, foreign exchange, commercial paper, government and agency securities, and corporate debt and equity. [2]
It also asked whether the same five percent test should reach other markets, such as derivatives, and systems that maintain a significant share of U.S. deposits. [2] It named substitutability and interconnectedness as further factors. [2] The notice closed with 39 numbered questions, and its signature page records the Board’s order of October 19, 2016. [2]
The October 2016 advance notice on enhanced cyber risk management standards is a long document with five categories of standards. [2] One sentence in it carried a number that a firm could test against: a two-hour recovery time objective for sector-critical systems, after a disruptive, corruptive or destructive cyber event. [2]
Our argument is that the two-hour clock was the notice’s clearest idea and, in our reading, its weakest. It came from a post-September 11 paper about recovering clearing and settlement after a wide-scale disruption. [2] Applied to a corruptive attack, it measured speed. The notice’s own incident response section said recovery also had to reckon with corrupted data spreading through connected systems. [2] The OCC and the FDIC recorded the rulemaking as withdrawn in 2019. [5] [6] Their 2020 sound practices replaced a fixed clock with a tolerance each firm’s board sets. [8]
We take the clock in four parts:
- The origin: where the two hours came from and what they were built to measure.
- The event: what changes when the event is corruptive rather than physical.
- The reach: which systems and which third parties the clock would have covered.
- The proof: how a firm would have shown it met the clock, and how the agencies would have measured risk.
The parts do not overlap. The first is the clock’s history, the second the threat it was applied to, the third its scope and the fourth its evidence. After them we look at how the notice reached different firms, take the objections in order, and end with what a bank’s technology and risk leaders could check today.
This is a reading of public documents, not legal advice. Firms should confirm current supervisory expectations with counsel and their supervisors.
The origin: a clearing and settlement clock
What the Sound Practices Paper measured
The notice quoted the Sound Practices Paper’s definition. A recovery time objective is the amount of time in which a firm aims to recover clearing and settlement activities after a wide-scale disruption. [2] The goal was completing material pending transactions on the scheduled settlement date. [2]
That definition tells a reader what the clock was for. It was about settlement finishing on the day it was due. The threat in view was a wide-scale disruption, and the notice described the paper as addressing risks in the post-September 11 environment. [2]
What the notice proposed to change
The notice proposed to apply the clock more widely in two directions. It would reach other large, interconnected financial systems beyond core clearing and settlement organizations. [2] It would also apply to recovery from cyber events. [2]
The notice said the agencies were considering the two-hour objective with advances in technology and consistent with the two-hour objective in the Sound Practices Paper. [2] In our reading, the notice treated the number as portable from one threat to another.
Why the number was attractive
A number is easy to examine. A supervisor can ask whether a test met it. A board can ask whether the firm is on track. A firm can budget against it. The notice’s other standards were described in terms of frameworks, strategies and policies. [2] The two-hour objective was the exception.
The notice also tied the clock to testing. It said testing programs would include a range of scenarios, including severe but plausible ones. [2] Tests would challenge communications protocols, governance arrangements and resumption and recovery practices. [2]
A threshold measured in value
The notice’s test for sector-critical systems counted value. A system supporting clearing or settlement of at least five percent of the value of transactions in a named market, on a consistent basis, would qualify. [2] For deposits, it floated a significant share, such as five percent, of total U.S. deposits or balances due from other depository institutions. [2]
Value is a reasonable measure of what fails to settle if a system stops. It is a weaker measure of what goes wrong if a system keeps running on bad data. A system with modest value but many connections could spread corruption further than a high-value system with few. The notice recognized this by naming interconnectedness and substitutability as further factors and asking how to weigh them. [2]
What the number assumed
A recovery clock assumes the firm knows when the event began and what a recovered state looks like. After a site loss, both are clear. The building is gone, and recovery means the backup site is processing. The sector-critical passage did not say how those assumptions change for a cyber event. [2] The notice’s incident response category came closer, as the next part shows. [2]
What an advance notice is for
The agencies chose an advance notice so they could benefit from comments on all aspects before developing a more detailed proposal. [1] The notice ended with 39 questions, several on the clock itself. [2] An advance notice does not bind anyone. It shows what regulators are weighing and invites evidence.
That status shapes how a firm should have read the clock in 2016. It was a signal of what supervisors might later expect, and an invitation to explain what a clean recovery from corruption would take. A firm with test results could have answered the notice’s cost and feasibility questions with data. [2]
The event: corruption is not a power cut
Three words in one sentence
The notice’s sentence named three kinds of cyber event: disruptive, corruptive or destructive. [2] They are different problems. A disruptive event stops a system. A destructive event erases or damages it. A corruptive event changes data while the system keeps running.
For a disruptive or destructive event, a fast recovery from a clean copy is the right goal. For a corruptive event, speed can be dangerous. A firm that restores from a backup taken after corruption began restores the corruption.
Replication spreads the damage
The notice itself named the mechanism that makes corruption hard. In its incident response category, it said recovery and resilience strategies should address the potential for malware or corrupted data to replicate through connected systems or high availability solutions. [2]
That is a sharp point. Systems built for fast recovery, such as mirrored data centers and real-time replication, copy whatever is written. If what is written is corrupt, the copy is corrupt too. The tools that serve a time objective can defeat a clean recovery.
Contagion to others
The notice did not treat a firm’s recovery as its own business alone. It considered requiring covered entities to identify and mitigate the cyber risks they pose to sector partners and external stakeholders through interconnectedness, to prevent cyber contagion. [2]
It said resilience strategies and exercises should consider wide-scale recovery scenarios and minimize risks to or from interconnected parties. [2] For a corruptive event, that cuts against speed. A firm that resumes sending messages before it has confirmed its data is clean can pass the problem to counterparties.
Recovery point objectives
The notice paired time with data. For cyber attacks that may corrupt or destroy critical data, it said recovery strategies should achieve recovery point objectives based on the criticality of the data needed to keep the institution operational. [2] It cited the FFIEC IT Examination Handbook’s business continuity booklet, which it said called for plans addressing attacks that may disrupt access, corrupt data or destroy data or systems. [2]
A recovery point objective says how much recent data a firm can afford to lose. A clean recovery from corruption may require going back to a point before the corruption began, which may mean losing more data than planned. The notice set a two-hour time objective for sector-critical systems. [2] It did not set a matching recovery point objective for them, and it did not say what a firm should do when the two pulled apart. [2]
Knowing when the clock starts
After corruption, the first question is when it began. A firm may discover altered data well after the alteration. The two-hour objective, as written, was a time to recover from a cyber event. [2] The notice did not say whether the clock started at the event or at discovery. [2]
In our reading, that gap mattered more for corruption than for the other two kinds of event. A destroyed system announces itself. A corrupted one may not.
What recovered means for data
After corruption, recovered has to mean the data is right as well as available. That requires a known-good copy and a way to bring transactions since then back into line. The notice addressed part of this elsewhere. [2]
Among the enhanced standards, the notice said firms should establish protocols for secure, immutable, transferable storage of critical records. [2] That standard pointed at the right control for corruption: a copy an attacker cannot change. The two-hour objective, in our reading, did not sit easily beside it, because a clean restore and a fast restore can pull in different directions.
Records another institution could restore
The notice went further on critical records. It said preserving critical records in a large-scale cyber event is essential to confidence in the banking system and to resolution or recovery after a catastrophic event. [2]
It considered requiring covered entities to keep secure, immutable, off-line storage of critical records. [2] The records named included financial records, loan data, asset management account information and daily deposit account records with balances and ownership details. [2]
Those records would be formatted using defined data standards so another financial institution, a service provider or the FDIC could restore them in a resolution. [2] In our reading, this was the notice’s answer to corruption at its worst: if the firm cannot recover, someone else can rebuild its customers’ positions.
Failures across infrastructure
The notice also considered strategies for performing core business functions during multiple concurrent or widespread interruptions. [2] It named cyber attacks on several elements of interconnected critical infrastructure, such as energy and telecommunications. [2]
A two-hour objective that assumed power and networks were available would not hold in that scenario. The notice did not reconcile the two. [2] A firm planning against both would need to know which scenario the clock applied to.
Third parties and substitution
The notice asked about one route to faster recovery: replacing a failed provider. It asked whether firms could feasibly use defined data standards to increase the substitutability of third-party relationships and reduce recovery times. [2]
That question recognized that a firm’s recovery may depend on a provider’s recovery. If the provider was corrupted, the firm’s own two hours would depend on the provider’s clean copy, or on moving to another provider with compatible data.
What the notice asked commenters
The notice asked directly about the clock. It asked for comment on the appropriateness and feasibility of a two-hour objective for all sector-critical systems, and on the incremental costs. [2] It asked what impact the objective would have on firms’ use of third-party service providers. [2]
Those were the right questions. They also showed that the agencies did not yet know the answers. An advance notice is the stage for that uncertainty, and the notice said so. [1] [2]
The reach: which systems and whose
Covered entities
The notice considered applying the enhanced standards to firms with total consolidated assets of $50 billion or more on an enterprise-wide basis. [2] The release listed bank and holding company thresholds, foreign banks’ U.S. operations, financial market infrastructures and nonbank financial companies the Board supervises. [1]
Community banks were outside. [1] The notice said other financial entities, including community banks that were not covered, would continue under existing guidance, standards and examinations for third-party banking services. [2]
Sector-critical systems
The higher tier was defined by systems, not firms. [2] A covered entity would have some systems under the sector-critical standards and the rest under the enhanced standards. The five percent thresholds, borrowed from the Sound Practices Paper, would decide which. [2]
The notice admitted uncertainty here too. It asked whether firms had enough information to determine whether their systems were sector-critical. [2] It asked what factors should measure interconnectedness and how to weigh them. [2]
Smaller firms that provide critical services
The notice recognized that a smaller banking organization might provide sector-critical services, directly or through covered entities. [2] It asked what criteria should bring such an organization under the sector-critical standards, and how to weigh the costs against the benefits to the financial system. [2]
Third-party services
The release said the standards would apply as well to services third parties provide to covered firms. [1] The notice said services supporting a covered entity’s sector-critical systems would be subject to the same sector-critical standards. [2]
In our reading, this reach turned the two-hour clock into a contract term. A firm whose sector-critical system ran on a provider’s platform would need the provider to recover within the same window. The notice’s question about third-party impact pointed at that consequence. [2]
An inventory of every asset
The internal dependency standard would have required more than a list. The notice considered requiring an inventory of all business assets enterprise-wide, prioritized by criticality to the functions they support, the firm’s mission and the financial sector. [2]
The listing would include mappings to other assets and business functions, information flows and interconnections. [2] Firms would track connections among assets and cyber risk levels through the life cycle. [2] The notice named insider threats, data transmission errors and legacy systems acquired through mergers among the sources of risk. [2]
Monitoring outside parties in real time
The external dependency standard was more demanding still. The notice spoke of the ability to monitor in real time all external dependencies and trusted connections supporting the firm’s cyber risk management strategy. [2]
It considered requiring firms to keep a current, accurate and complete listing of all external dependencies and business functions, with mappings to supported assets. [2] Firms would prioritize monitoring, incident response and recovery of systems critical to the enterprise and the financial sector. [2]
Trusted connections
The notice included a category of risk that firms might overlook. It said external dependency management covers interconnection risks from non-critical external parties that maintain trusted connections to important systems. [2]
That reached beyond critical vendors. A small supplier with a network connection into a sector-critical environment would be in scope. For the two-hour clock, such a connection is a route by which corruption could enter a system the clock covered.
The life of a relationship
The notice listed the stages of a relationship its external dependency strategy would cover: due diligence, contracting and sub-contracting, onboarding, ongoing monitoring, change management and offboarding. [2] It said firms would need metrics to measure effectiveness in reducing external dependency risk. [2]
Recovery terms belong at the contracting stage. A provider that cannot commit to the firm’s recovery objective at contract signing is unlikely to meet it in an incident.
Internal and external dependencies
The notice defined the dependencies it wanted managed. Internal dependencies were the firm’s business assets, meaning workforce, data, technology and facilities, and the information flows among them. [2] External dependencies were relationships with vendors, suppliers, customers, utilities and other outside organizations. [2]
The definitions mattered to the clock. A recovery objective is met only if every dependency the system needs recovers inside it. A firm would need to know them all to know whether two hours was possible.
The proof: tests and measures
Testing as validation
The notice said the two-hour objective would be validated by testing. [2] Testing programs would include severe but plausible scenarios. [2] For a corruptive event, a meaningful test would need to start from altered data and end with verified data.
That is a harder test to design than a failover test. It needs a way to plant corruption in a test environment, detect it, find a clean point and reconcile. The notice described the scenarios generally and left the design to firms. [2]
Governance evidence
The proof would start with governance. The notice considered requiring a written, board-approved, enterprise-wide cyber risk management strategy incorporated into the firm’s overall business strategy and risk management. [2] It considered requiring the board to review and approve the firm’s enterprise-wide cyber risk appetite and tolerances. [2]
For a sector-critical system, the board would approve a strategy that included the two-hour objective. The board would then need evidence that tests met it. That chain, from board approval to test results, is the supervisory record the notice implied.
Independent functions and review
The notice considered integrating cyber risk management into at least three independent functions, with checks and balances. [2] It named the three lines of defense model as an example. [2]
Its incident response standard called for programs supported by policies, procedures, governance, staffing and independent review. [2] It said those programs would need escalation protocols linked to decision levels, procedures to contain cyber contagion, communication strategies and a way to feed lessons learned back. [2]
How examiners would have used it
The notice linked the standards to existing examination tools. It said the FFIEC’s Uniform Rating System for Information Technology, issued in 1978 and revised January 20, 1999, is used to assess IT risks at institutions, affiliates and service providers. [2] It said supervisory work on the external dependency standard could inform part of that rating. [2]
In our reading, a two-hour test result would have become an input to an examination rating. That raises the stakes of how the test is designed. A failover test that passes the clock would count; a corruption test that misses it would count against.
What a corruption test would record
A test that took the corruption case seriously would record more than elapsed time. It would record when the planted corruption was detected, which recovery point was chosen and why, and how many transactions had to be reprocessed. It would record whether replication carried the corruption anywhere else. [2]
Those records would show whether the firm met a recovery point objective as well as a time objective, which the notice’s incident response category asked firms to consider. [2] They would also show a board what a clean recovery costs in time. That is the number a board needs to set a realistic objective.
Measuring residual risk
The notice set out a second measure for the Board’s firms. The Board was considering requiring its covered entities, at the holding company level, to measure quantitatively their ability to reduce the aggregate residual cyber risk of sector-critical systems to a minimal level. [2]
The measurement would account for internal dependencies, external dependencies and trusted connections with access to sector-critical systems. [2]
Quantifying cyber risk across firms
The notice’s section on quantification was frank. It said the agencies were not aware of any consistent methodologies to measure cyber risk across the financial sector using specific risk management objectives. [2] It named two existing approaches it was familiar with and asked for comment on methods for inherent and residual risk. [2]
In our reading, that admission undercut the measurement standard. A firm cannot be held to reduce risk to a minimal level by a measure the regulators said did not yet exist in consistent form.
The form the standards would take
The notice described three approaches to implementation. [2] One combined a regulatory requirement to maintain a cyber risk management framework with a policy statement or guidance describing minimum expectations. [2] A second would impose specific cyber risk management standards by regulation, across the five categories. [2]
A third would set out, in more detail, the specific objectives and practices a firm would have to achieve in each area. [2] The agencies said they would weigh clarity, the effort needed, adaptability to the changing environment and cost. [2]
In our reading, the two-hour objective fit the third approach best. It was a specific objective. The notice did not say which approach it would carry. [2]
How the notice reached different firms
Bank holding companies above the thresholds
For a holding company above the thresholds, the notice meant a board-approved, enterprise-wide cyber risk management strategy built into its overall business strategy. [2] It meant board approval of cyber risk appetite and tolerances. [2] It also meant the two-hour objective for any sector-critical systems it ran.
The notice said cyber risk management would be integrated into at least three independent functions, such as the three lines of defense model, with checks and balances. [2]
National banks under heightened standards
For OCC-regulated covered entities, the notice said the cyber framework would be incorporated into the risk governance framework required by the OCC’s heightened standards at 12 CFR part 30, Appendix D. [2] In our reading, this would have placed cyber governance inside an existing set of OCC guidelines rather than in a new structure.
Financial market infrastructures
Payment, clearing and settlement firms were the original subject of the Sound Practices Paper’s clock. [2] For them, the notice extended a familiar objective to a new threat. The five percent tests were drawn from a paper written about their activities. [2]
Technology service providers
A core processor or payments provider serving covered firms would have faced the standards through its customers. [1] [2] The notice’s question about using defined data standards to increase substitutability would have reached these providers directly. [2]
Community banks
Community banks were outside the standards. [1] They still relied on the same providers. A provider that improved recovery for its large customers might improve it for small ones as well, or might offer different service levels. The notice did not address that effect. [2]
Asset managers and custodians within covered groups
The critical records the notice named included asset management account information. [2] An asset management or custody business inside a covered group would have needed immutable, off-line copies of those records in a format another institution could restore. [2] That is a data engineering task as much as a security one: agreeing formats, testing restores and keeping the copies current.
Supervisors and examiners
For supervisors, the notice would have meant new test evidence to review. It said supervisory work on the external dependency standard could inform part of the IT rating. [2] Examiners would have needed to judge whether a firm’s two-hour test was a fair test of a corruptive event, or only of a failover. That judgment turns on test design, which the notice left to firms. [2]
Nonbank financial companies
The release named nonbank financial companies supervised by the Board among the entities in scope. [1] Their systems might not run settlement in the named markets, so the five percent tests might not reach them. [2] The notice’s further factors of substitutability and interconnectedness were the route by which their systems could have been treated as sector-critical. [2]
Foreign banks’ U.S. operations
The release included the U.S. operations of foreign banking organizations with total U.S. assets of $50 billion or more. [1] Those operations would have needed recovery objectives for U.S. systems that might depend on group infrastructure outside the United States.
Firms between the thresholds
Firms just under the $50 billion threshold were outside the enhanced standards. [1] They could still provide services that larger firms depended on. The notice’s question about smaller organizations providing sector-critical services addressed that case. [2] A regional bank acting as a correspondent or processor for others could have been drawn in by function rather than size.
Objections, in order
1. The clock was only a starting point
The agencies did present it as a consideration and asked for comment. [2] But advance notices shape later proposals. A specific number in an advance notice frames the debate in a way that general principles do not. The clock deserved the scrutiny it invited.
2. Two hours had worked for clearing and settlement
It had a basis there, according to the notice. [2] The question was whether it fit a corruptive cyber event. The notice did not explain how a clock built for wide-scale disruption would apply when the data itself was in doubt. [2]
3. Fast recovery limits systemic harm
That is true for disruption. Settlement that fails to finish can spread to counterparties. But a fast restore of corrupted data can spread harm too, by settling the wrong amounts. In our view, a recovery objective for corruption needs a data-integrity condition before it needs a time.
4. The immutable storage standard answered the concern
It answered part of it. Immutable storage gives a firm a clean copy, and the notice also called for recovery point objectives for data that could be corrupted. [2] It did not say which governed when a two-hour time objective and a clean recovery point conflicted. [2] That ordering is the gap this article points to.
5. A board-set tolerance is too loose
There is a real case for a fixed number. Boards may set tolerances that suit their budgets. The 2020 paper tied a firm’s tolerance to existing regulations and guidance and to severe but plausible scenarios. [8] Whether that is enough is a supervisory question. It does avoid applying one number to different threats.
6. The withdrawal left a gap
The rulemaking did not move forward. [5] [6] But the agencies’ existing guidance continued, and the 2020 paper drew together existing regulations, guidance and statements on operational resilience. [7] The gap was in a binding number, not in supervisory attention.
7. The agencies could not have measured the outcome
The notice’s own section on quantification said the agencies were not aware of consistent methods to measure cyber risk across the sector. [2] That supports the objection for the residual risk measure. It does not apply to the clock, which a test could measure. The difficulty with the clock was what it measured, not whether it could be measured.
What a bank’s technology and risk leaders could check today
Recovery objectives by event type
A firm could check whether its recovery objectives distinguish disruption, destruction and corruption. A single objective for all three repeats the notice’s assumption. [2] An objective for corruption could include the steps to find a clean point and verify data before resuming.
Immutable copies of critical data
The 2020 paper says recovery from destructive malware, including ransomware, may include protocols for secure, immutable, off-line storage of critical data. [8] A firm could check which critical data has such a copy, how old it is and how it would be reconciled.
Tolerance for disruption, set and tested
The 2020 paper asks the board to articulate the firm’s tolerance for disruption, considering its risk profile and operational capabilities. [8] A firm could check that its tolerance is written, that it reflects severe but plausible scenarios, and that tests show whether the firm can stay within it. [8]
Provider recovery in the same terms
A firm could check whether its critical providers have recovery commitments that fit its own tolerance. The 2020 paper says information systems that support critical operations, including elements that depend on third parties, should have regularly tested recovery programs. [8]
When the clock starts
Finally, a firm could decide, in writing, when a recovery clock starts for a corruptive event: at the event, at detection or at declaration. The 2016 notice did not answer that question. [2] Each answer leads to different testing and reporting.
Where replication could carry corruption
A firm could map where real-time replication and high availability links would copy corrupted data from one system to the next. The 2016 notice named that risk directly. [2] The map shows where a clean copy must sit outside the replication path, and how a firm would cut a link during an incident.
Records in a restorable form
A firm could check whether its critical records could be restored by someone else. The 2016 notice considered requiring defined data standards so another institution, a service provider or the FDIC could restore records in resolution. [2] Even without a rule, a test restore by a second party shows whether the format works.
An inventory with mappings
A firm could compare its asset inventory with the notice’s description: all business assets, prioritized by criticality, with mappings to functions, information flows and interconnections. [2] The 2020 paper describes a firm identifying risk transmission channels by analyzing the interconnections and interdependencies of its critical operations and core business lines, considering third-party risks. [8] The gaps between the two descriptions and the firm’s actual inventory are where a recovery plan will fail.
With hindsight
The comment period was extended. On January 24, 2017, the agencies published a notice reopening comment until February 17, 2017. [4] They said the extension reflected the range and complexity of the issues and would give interested persons more time. [4] The notice recorded that the advance notice had been published in the Federal Register on October 26, 2016, at 81 FR 74315. [4]
No proposed rule appears in the OCC’s or the FDIC’s agenda timetables. The Spring 2019 Unified Agenda entry for the OCC’s RIN 1557-AE06 lists the rulemaking under completed actions. [5] Its timetable shows the advance notice, the comment period and its extension, and a final line reading Withdrawn 03/06/2019. [5] Its abstract says the rulemaking is not moving forward at this time. [5]
The FDIC’s entry for RIN 3064-AE45 in the same agenda also lists the rulemaking under completed actions. [6] Its timetable ends with Withdrawn 02/21/2019. [6] The FDIC entry’s timetable gives the extended comment period end as 02/24/2017, while its own abstract, the OCC entry and the January 2017 notice give February 17, 2017. [4] [5] [6]
On November 2, 2020, the Federal Reserve issued SR 20-24, distributing an interagency paper titled Sound Practices to Strengthen Operational Resilience. [7] The letter says the paper applies to firms with average total consolidated assets of $250 billion or more, or $100 billion or more with $75 billion in certain other measures. [7]
The letter says the sound practices are drawn from existing regulations, guidance and statements, and from common industry standards. [7] It says the guidance does not amend, expand or alter the agencies’ existing regulations or guidance. [7] The letter shows that it was revised on June 2, 2026, and that the attachment was revised to remove references to reputational risk. [7]
The paper does not set a two-hour objective. It asks a firm’s board, in setting risk appetite, to articulate the firm’s tolerance for disruption, considering its risk profile and the capabilities of its operational environment. [8] It says that tolerance is informed by existing regulations and guidance and by analysis of severe but plausible scenarios. [8]
The paper cites the 2003 interagency white paper among that existing guidance. It describes that paper as covering geographic diversity of data centers and recovery and resumption time objectives for firms in critical financial markets. [8] In our reading, the two-hour clearing and settlement objective stayed in its original place, while the wider cyber application proposed in 2016 did not go forward. [2] [8]
The paper addresses corruption more directly than the clock did. It says a firm should establish controls to safeguard the integrity and availability of critical data against destructive malware, including ransomware. [8] It says recovery may include protocols for secure, immutable, off-line storage of critical data. [8]
What held up from October 2016 was the immutable storage idea and the attention to third-party dependencies. [2] [8] What did not hold up was a single recovery clock for every kind of cyber event across sector-critical systems. [2] [5] [6] As checked on November 16, 2026, the Federal Reserve’s SR 20-24 page shows the revised paper as current. [7]
Spotted an error? Tell us and we will check it and correct the article where needed.