To monitor, or not to monitor. This has long been the question for U.S. financial services with regard to voice surveillance. While in the U.K. fines for voice surveillance failures have made headlines, U.S. obligations appear to be less clear, and enforcements a little more nuanced, resulting in voice surveillance often considered a ‘nice to have’ rather than a compliance necessity.
For U.S. swap dealers, however, recently regulatory movement suggests that these requirements may now be more clear-cut.
The story of voice surveillance so far
Historically, U.S. regulation has focused chiefly on the supervision of written communication or eComms. The implementation of the Dodd Frank Act in 2010 saw a gentle turn in the tide, initially empowering the Commodity Futures Trading Commission (CFTC) to require that swap dealers preserve voice calls. Alongside this, generic swap dealer supervision rules were established. It was only a matter of time before both recordkeeping and supervision requirements were melded together.
Regulatory rules, messaging, and enforcement action over the past five years shows that voice supervision requirements now stand firmly alongside recordkeeping requirements for swap dealers and securities-based swap dealers.
Regulators including the CFTC and National Futures Association (NFA) have issued a series of fines to swap dealers who failed to supervise voice communication. While these fines may not have caught media attention in the same way that other ‘off-channel comms’ enforcement has, the implications are still severe.
Regulatory whispers: enforcement action for voice surveillance failures
2019 – CFTC issues $13m fine to TP ICAP for failure to supervise brokers’ conduct on voice calls
On September 13, 2019, the CFTC issued Tullett Prebon Americas Corp (TP ICAP) with a $13 million fine for “failing to supervise employees and making false or misleading statements to CFTC staff”. In two orders, issued jointly, TP ICAP was charged $11 million for violating CFTC regulations and for its “failure to supervise brokers’ conduct on its U.S. Dollar Medium Term Interest Rate Swaps Desk”, as well as $2 million for making false or misleading statements.
In particular, the CFTC found that TP ICAP:
“failed to implement procedures to monitor for and prevent the types of false or misleading statements that are the subject of this Order” […] and therefore failed to adequately supervise activities relating to its business as a Commission registrant in violation of Regulation 166.3, 17 C.F.R. § 166.3 (2018)”.
By means of remediation, the CFTC ordered TP ICAP to enhance its “review of voice-brokering by such means as increased random sampling”.
2021 – NFA issues $150,000 fine to Tullett Prebon Financial Services for failing to review oral communications
On June 1, 2021, the NFA issued TP ICAP with a $150,000 fine for its failure to keep full, complete, and systematic records of transactions related to its business dealings, and for its failure to supervise employees. The NFA noted that TP ICAP had a “deficient program to oversee employees’ recordkeeping activities” and had “inadequate procedures to review and supervise” the communications of its employees.
The NFA had found that TP ICAP was only reviewing oral communications where an issue had been flagged in written communications. Otherwise, oral communications were not subject to periodic reviews. Moreover, the firm’s policy had failed to consider the above order from the CFTC to increase random reviews of voice brokering activities.
2024 – CFTC issues BNY Mellon $5m fine for failure to implement policies and procedures to monitor voice communications
In this instance, the CFTC found that over a five-year period, BNY Mellon had failed to correctly report millions of swap transactions as required by CFTC regulations. As well as this, it found that
“BNYM also failed to properly supervise its business as a Swap Dealer in that BNYM had no written policies or procedures to: (i) monitor the voice communications of its Associated Persons (“APs”); and (ii) monitor the e-communications of its APs that communicated in languages other than English, in order to ensure its APs were complying with the Act and Regulations.”
Of particular interest here, is the CFTC’s focus on non-English translation, adding:
“Most if not all of BNYM’s APs utilized telephonic communications for the same purpose. During the Relevant Period, BNYM had no written policies and procedures in place, including use of a foreign lexicon, for monitoring the e-communications of its non-English speaking APs for compliance purposes. In addition, BNYM had no written policies and procedures in place to monitor the voice communications of its APs for compliance purposes.”
Notably, while the earlier CFTC and NFA fines explicitly mentioned increased random sampling as a method to improve supervision of voice calls, the CFTC’s 2024 order is silent on the matter. It is likely that this is a recognition that available technological solutions have advanced significantly since those earlier fines, and the regulator is aware that it is now possible to perform effective automated voice surveillance without the need for random review.
What about the SEC?
While enforcement action has so far been limited to the CFTC and NFA, the Securities and Exchange Commission (SEC) has not been entirely voiceless on the matter. In fact, in a Risk Alert issued in January 2024, the SEC noted that a reasonably designed supervisory system for security-based swap dealers should include the review of oral communications where they have been recorded.
“Security-based swap dealers have an overarching obligation in Rule 15Fh-3(h)(1) to establish and maintain a supervisory system that is reasonably designed to prevent violations of applicable federal securities laws and the rules and regulations thereunder relating to the security-based swap dealer’s business as a security-based swap dealer. Therefore, if a security-based swap dealer records oral communications with counterparties or potential counterparties, the security-based swap dealer generally should consider providing for the supervisory review of such communications. Similarly, if a security-based swap dealer chooses to provide certain disclosures required by Rule 15Fh-3(b) orally, the security-based swap dealer should consider how it will supervise these oral communications.”
As such, if the firm records voice calls, then they should have some supervision over those and if it doesn’t record voice calls then it must demonstrate how it is able to supervise activities that occur over the phone.
What does this mean for swap dealers and voice surveillance?
Regulatory action serves to show that the CFTC and NFA and even the SEC to an extent have an expectation that registered swap dealers should be capturing and surveilling voice communication alongside other eCommunication data. As well as this, it looks as though the CFTC expectation of surveillance tools to surveil in multiple languages, perhaps extending to voice too. Such firms will need to implement voice surveillance solutions to meet regulatory demand.
For securities-based swap dealers who may previously have believed voice surveillance to be out of scope for them, given there is no explicit regulatory requirement to record voice calls – it may be time to think again. The SEC’s Risk Alert also raises questions as to whether the SEC may apply the expectation (if you record then we expect you to surveil) to broker dealers, who have similar supervision requirements, but also do not have a requirement to record voice calls.
Historically the CFTC has permitted the use of random sampling in voice calls as a stop-gap, however, firms will need to show that any solution that they implement is ensuring that they are compliant with relevant regulations. Given the pace of technological development, organizations that opt for random sampling will likely find themselves falling foul of regulatory expectation sooner, rather than later.
Does random sampling leave too much to chance?
Random sampling or ‘dip sampling’ is the process of monitoring voice communication at random. This process is incredibly manual, requiring that individuals spend long periods of time listening to calls to establish whether or not misconduct or risk exists. It is unlikely this sampled approach would be considered sufficient if deployed on written communications. Downfalls of random sampling include:
- Inefficient – the time of compliance professionals would be better spent on reviewing true risk or misconduct.
For example, the statistical rule of three says that if you assume 1 in every 1000 calls contains something of relevance to Compliance (e.g. Complaint, misconduct, etc.) in order to achieve 95% confidence that you will find at least one issue you would need to review 3000 calls. Assuming the average call length is 2 minutes, you would need to listen to 100 hours of calls to find a single issue. This equates to someone listening to calls for 6 hours every day for 16 days, with no certainty of finding true positives.
- Prone to human error – individuals have limited attention spans and can be distracted, leading to potential risks being missed.
- Incomplete reviews – by its very nature, random sampling only examines a small percentage of voice communications, leaving the remainder completely untouched. With AI-enabled surveillance products however, all voice communications are monitored.
- Unable to understand non-English languages – oftentimes firms will need to hire translators to listen to non-native voice calls, meaning the cost of monitoring is increased, or that non-English calls are ignored despite containing potential risk
- Outdated – new, AI-enabled technology can automatically transcribe and translate voice calls and flag risks or risk areas, so reviewers need only analyze parts of conversations, rather than lengthy dialogue in its entirety
Regulatory focus on voice surveillance to meet global requirements, from MAR, MiFID II, and SEC Rules, is increasing. Audio transcription is critical to robust voice surveillance to meet these regulatory obligations and ensure watertight, compliant communication. Identify misconduct, reduce false positives, and mitigate the risk of gaps in your communication records with Global Relay’s AI-enabled voice-to-text transcription solution.
