A New Tool for Compliance: FINRA’s Machine-Readable Rulebook Initiative
FINRA's new Machine-Readable Rulebook is designed to enhance firms' compliance efforts, reduce costs and aid in risk management, with a lot to gain for firms of all sizes and various business models.
On this episode, Afshin Atabaki, Associate General Counsel with FINRA's Office of General Counsel, and Haime Workie, Vice President of FINRA's Office of Financial Innovation, join us to delve into what exactly the Machine-Readable Rulebook is, how it works, and how you can start taking advantage of all it has to offer.
Resources mentioned in this episode:
Listen and subscribe to our podcast on Apple Podcasts, Google Podcasts, Spotify or wherever you listen to your podcasts. Below is a transcript of the episode. Transcripts are generated using a combination of speech recognition software and human editors and may contain errors. Please check the corresponding audio before quoting in print.
00:01 - 00:20
Kaitlyn Kiernan: FINRA's new Machine-Readable Rulebook is designed to enhance firms' compliance efforts, reduce costs and aid in risk management, with a lot to gain for firms of all sizes and various business models. On this episode, we hear what exactly the Machine-Readable Rulebook is, how it works, and how you can start taking advantage of all it has to offer today.
00:20 – 00:30
00:30 - 00:54
Kaitlyn Kiernan: Welcome to FINRA Unscripted. I'm your host, Kaitlyn Kiernan. I'm pleased to welcome two guests to the show today to discuss FINRA's Machine-Readable Rulebook Initiative. Joining us for the first time is Afshin Atabaki, Associate General Counsel with FINRA's Office of General Counsel. And joining us once again is FINRA Unscripted pro Haime Workie, Vice President of FINRA's Office of Financial Innovation. Afshin and Haime, welcome to the show.
00:54 - 00:55
Afshin Atabaki: Thank you, Kaitlyn.
00:55 - 00:56
Haime Workie: Thanks for having us.
00:56 - 01:04
Kaitlyn Kiernan: So, Afshin, as the newcomer on the show, can you kick us off by introducing yourself and telling us a little bit more about what you do at FINRA?
01:04 - 01:30
Afshin Atabaki: Sure. Thank you, Kaitlyn. I'm Afshin Atabaki, a Special Advisor and Associate General Counsel in FINRA's Office of General Counsel. Our group is primarily responsible for regulatory policy at FINRA. That includes rulemaking and guidance. I've been at FINRA for about 22 years. I'm responsible for a wide range of subject matters. They include books and records, registration and qualification requirements, and others. And I'm also co-lead on this project along with Haime.
01:31 - 01:34
Kaitlyn Kiernan: Great. Thanks, Afshin. And Haime, how about you?
01:34 - 02:09
Haime Workie: So, I'm Haime Workie, I head up FINRA's Office of Financial Innovation. The office was formed about four years ago, and it's designed to facilitate innovation in a manner that strengthens FINRA's underlying mandate, which is investor protection and market integrity. And this support for innovation takes many different forms, including engaging with securities industry participants on fintech-related topics such as artificial intelligence, cloud computing and blockchain technology. One recent example is, of course, the Machine-Readable Rulebook which we undertook in partnership with Afshin and his colleagues at the Office of the Chief Legal Officer here at FINRA.
02:11 - 02:21
Kaitlyn Kiernan: Yes, the Machine-Readable Rulebook definitely seems innovative. So, at a high level, can you kick us off by telling us what exactly the Machine-Readable Rulebook Initiative is?
02:21 - 03:20
Haime Workie: Yeah, it's a lot of words to try to digest and to figure out what it means. At its basic principle, it's an effort designed to make the FINRA Rulebook easier to navigate and understand, both for individuals as well as computer systems. And this is all done with the idea that we want to be able to facilitate compliance and make it more efficient within the securities industry. The initiative really has three separate parts. The first part is the development of a taxonomy. And that's really a set of terms, whether they be business, legal or otherwise. They're used to identify topics that a rule relates to, for example, a rule may relate to a disclosure item, although it may not say disclosure, and will be tagged as such. And then the other two parts are really delivery mechanisms to make this kind of rulebook that's been tagged with these key terms easier to digest both for humans and for computers. So, for computer-to-computer interaction, we have our API protocol and for human to computer interaction we have an enhanced search tool which we've called FIRST.
03:21 - 03:24
Kaitlyn Kiernan: Thanks, Haime. And what was the impetus for this project?
03:25 - 03:51
Haime Workie: The impetus for this project dates back to a Special Notice on innovation that we put out back in 2018, where we asked the merits of developing a machine-readable rulebook and talked about what a taxonomy was and what it potentially allowed people to do. Based on this Notice that was put out, there was a lot of support for FINRA to explore this area and this ultimately led to development of the Machine-Readable Rulebook prototype that was put out last year.
03:51 - 04:01
Kaitlyn Kiernan: And anyone who's looked at FINRA's Rulebook knows it's quite large and extensive. So, how did you pick the 40 rules that you started with for this initiative?
04:02 - 04:46
Haime Workie: It's important to note when we tagged the Rulebook, we didn't tag the entire Rulebook when we started. We started off with 40 rules and these rules were selected because we wanted to have an iterative process whereby we solicit feedback as we look to go to tag potentially the entire Rulebook. Now, the 40 rules represents about 50% of the page views that users or others who go on to FINRA's website and view the FINRA Rules. And this metric was developed using Google Analytics. But in addition to those hard metrics that we determined, we also convened an expert group of compliance folks that focus on the securities industry to get input and solicit their feedback on what rules we select as part of the initial 40 rules.
04:47 - 04:56
Kaitlyn Kiernan: Afshin, Haime mentioned things like taxonomy and tagging. Can you explain how the taxonomy works and what it means exactly to tag a rule?
04:57 - 05:50
Afshin Atabaki: So, basically the taxonomy is a method of classification or categorization. Yes, we have a lot of rules. Some of them pre-date my 22 years at FINRA, so you have different folks writing in different styles. And within these rules you have all these various requirements. And what we were looking at is, is there an effective way to call or pull from these rules terms, concepts—which we refer to as tags—and deconstruct the rules so that it makes it easier for users to be able to find what they're looking for. So, we mentioned this is a prototype, and as part of the prototype, we went to these 40 rules, each one of them, we identified and listed the various terms and concepts that appear in them, terms of art, regulatory terms, concepts, requirements. And then we categorized them and that's the taxonomy part. We did this to make it more effective in terms of the searchability.
05:51 - 05:56
Kaitlyn Kiernan: And how does the taxonomy allow for enhanced search?
05:57 - 08:01
Afshin Atabaki: The idea was now you have all these terms and concepts that we've pulled from these rules, how do you organize them in a manner that's usable, browsable? To do so, we created two main topics for the tags: summary topics and detailed topics. The summary topics include eight broad categories. They have general applicability. We picked Firm Type, as an example. That's based on Member Supervision's categorization of firm types. So, we use an existing classification method and applied it here. Security Type—it’s binary, at this level, we pick equity and debt, but a fundamental level. Another example of one of these eight categories is Obligations and Duties. The requirement to maintain books and records, that's an obligation so, that falls under that category. And then these categories also include various subcategories, which also cover these high-level, fundamentally core concepts, subject matters, requirements. They're all embedded there. Haime mentioned disclosures. That's one of them. We created a subcategory for disclosures. Does this rule have a disclosure or disclosure-like requirement? Know Your Customer is another one. Example KYC, Anti-Money Laundering, AML. Another example, private securities transactions. These are all concepts and terms that we use. They're embedded within the rule. We just pulled them out and categorized them to make it easier to find. The other area I mentioned, Detailed Topics. Those are divided into three main categories of classification: Defined Terms, Related Laws, Rules and Regulations and Terminology, these terms of art that we use and other terms. It's a pretty broad category that one, terminology. But Detailed Topics, basically, are these very granular topics and terms. And they're going to provide additional depth to the search function. At its core, the development of the taxonomy allows you to navigate through this ruleset—again, as you mentioned, it's a vast, conceptually large ruleset—in a more intuitive way. We believe it will in fact make it easier to find those rules that are relevant to what you're looking for when you search our Rulebook. And in that sense, it's an enhancement to that current search functionality.
08:02 - 08:21
Kaitlyn Kiernan: The taxonomy sounds foundational to this project, but the initiative also included the development, as Haime mentioned, of two delivery methods for the Machine-Readable Rulebook. There's the FINRA Rulebook Search Tool, or FIRST. That's one of the methods. Maybe we can start there. How does that work?
08:22 - 10:35
Afshin Atabaki: I mentioned the Summary Topics and Detailed Topics reside somewhere—on our website. And the search tool is the primary method by which you navigate through the tags, and it sits on the same web page as the FINRA Rules, it's under the FINRA Manual. You refer to it as the FINRA Rulebook Search Tool, FIRST. The Summary Topics and Detailed Topics I just mentioned, those are browsing functions in the search tool. You can browse by either of those main categories. So, you go there to the search tool, you see a tag that's relevant to your search under these main topics, you select it and the rules that relate to that tag will appear in the results on the side. You can then click on any of those rules that appear in the results, it will take you to a page that includes the rule text, and then you'll see the tags on that page as well, on the rule text page, and those are the tags that we've applied to each of these rules. It's rule-specific, but that's how you get to them. You can also do a reverse search, by the way. If instead of landing on the search tool page, you land on a page that has the rule text, that includes the tags that we've applied, again, this is limited to 40 rules, so, be mindful of that, but you can land on that page. Let's say Rule 1220 is one of the rules that we've tagged, it's a registration rule. And you see the tags, you can click on any of those tags that appear there because you're curious about which other rules have that same tag associated with them. You click on that and then it'll immediately take you back to the search tool page and then show you all the rules that have that same exact tag, so it's a reverse lookup. The other thing you can do is you can do a free text search in our existing search box, which has always been on our website. And if you look for any of the tag terms, you're trying to find something, but you're not sure exactly what the exact phrase is, but you get it close enough and what it will do is pick up the tagged rules that have that tag in them. I do want to note here that one of the important things about the search tool, and I think Haime mentioned it as well, is that it's informational. We have a disclaimer on the search tool site that it is for informational purposes. You should not and cannot rely on it for regulatory compliance advice. Ultimately, you still need to understand the rules. You have to read them, determine the relevance of the rule to your business and practices. However, this is a big however, we believe this tool could significantly help your search efforts. It's the first step in your search, and hence it's called FIRST.
10:35 - 10:47
Kaitlyn Kiernan: It sounds like this is not just helping people find rules, but also helping them get a better understanding of how they relate to each other, how different obligations tie together with these tags and themes.
10:47 - 11:01
Afshin Atabaki: Yeah, it shows you the relationship between things, sort of the purpose of the web, because if you read something out of context, you may not fully capture its meaning. But when you see it with all the other associated rules, you're going to get a better picture.
11:02 - 11:08
Kaitlyn Kiernan: And can you walk us through an example of how a user might start with a broad topic and end up narrowing it down?
011:09 - 13:29
Afshin Atabaki: Yeah, take the Summary Topics as an example. The search tool allows you to select any of those tags under the Summary Topics, and then right there you're automatically, by selecting it, you're going to narrow the set of rules associated with that tag. And it also allows you to select multiple of these tags within the Summary Topics to further narrow the results. Just a quick example. You're looking for rules that apply to retail firms. Once you do that, you're going to get a narrow set of rules that are associated with that specific tag. It's among the 40 rules that we've tagged. Again, it's not the entire Rulebook, but it's a prototype. This is something that firms, I think, have been asking for a long time. How do we know which rules apply to our business type? Well, here's a step towards that. If you want to further narrow the results, let's say to include account opening, servicing or termination, this is a concept which we've come up with—I say we've come up with, I mean the industry understand it's a concept, but we've come up with the term and applied it to the rules to make it easier to find those rules that have an account opening, servicing or termination component. We've included this under a tag that we labeled as Account Information and Management. It sits under the Summary Topics, again it's a fundamental concept. So, let's say you're looking for those types of terms. You're also interested in KYC, Know Your Customer, and AML requirements. Those have an account opening component and you select these. And by the way, KYC and AML sit under Obligations and Duties, but you select these terms altogether and it does a multi search for you. And now you have a narrower set of rules that have these three classifications or three objects: Retail Firms, account opening services and termination, KYC or AML requirements. A traditional search function would not be able to do that because a rule might not expressly have the term "retail" in it, though that applies to retail firms. The same is true again for account opening, servicing or termination, it's a term that we've come up with, it's not in a rule, but right there this will help your understanding of the rule, are other rules that have these requirements and how they all come together. You can also do the same for Detailed Topics. Detailed Topics are more granular. As an example, you're searching for rules that specifically relate to investment company securities. It's a very narrow term. You can browse, you select that tag from a list of Investment Products, itt's under terminology. And right there, you'll get those results. Here are all the rules out of the 40 that relate to investment company securities.
13:30 - 14:10
Haime Workie: I like to think of this as almost bringing the principles of online shopping to the FINRA Rulebook. So, you have the online search tools that are out there, for example, for shopping sites such as Amazon. And it's the new year so people are probably thinking about exercising. So, let's say you want to get a treadmill. You can pick out the brand of the treadmill. You can pick out the speeds you want it to go to. You can pick out the colors and then you can get a set of treadmills that are specific to what you're looking for. In a similar way, what we're trying to do, and as Afshin was describing, is allow you to have the FINRA Rulebook, but then figure out what it is that you actually want based on the terms that you're concerned about and figure out which portion of the Rulebook actually may apply in your circumstance.
14:11 - 14:16
Kaitlyn Kiernan: But the benefit is then you won't have the FINRA Rules following you around the Internet until you buy that treadmill.
14:17 - 14:18
Afshin Atabaki: Yeah, right.
14:19 - 14:21
Haime Workie: We try to avoid that aspect of online shopping.
14:22 - 14:29
Kaitlyn Kiernan: So, Afshin, you mentioned this is a prototype. What makes it a prototype? Do you plan to make additional developments to the tool?
14:29 - 15:25
Afshin Atabaki: We've applied this to 40 rules, but those 40 rules account for a significant portion of the rules that users typically view. So, though a prototype, and limited in number, these particular tags that we've selected for these 40 rules will give us a broad insight at the way users are going to navigate our Rulebook. And that's the experience we're looking for. We want folks to use this to test whether the tool has practical, real-world value beyond the concept. Once we evaluate all of the feedback that we have received, then we're going to have to determine whether to apply the tags to the remaining FINRA Rules and how to do that. That's part of this as well. It's pretty resource intensive. I use a line from Jurassic Park. We figured out, could we and now the question is, should we go further? And that's the purpose of the prototype. The question is, does the industry find this of value? And I think we've done some of that preliminary assessment. We need to understand broader with practical use, whether the industry finds this useful.
15:26 - 15:38
Kaitlyn Kiernan: The FINRA Rules will find a way, to borrow from Jurassic Park. So, there's also an application program interface or API. How does that part work?
15:38 - 16:24
Afshin Atabaki: This is beyond my level of expertise, but I'll try here. So, basically the API, the application programming interface, is what makes the rules machine-readable. We developed the API to allow anyone, including member firms, to ingest the tags applied to each of these 40 rules in an aggregate manner through a, and again, this is techspeak, JSON file, which then makes it machine-readable on the recipient side. The API is available through the FINRA API Developer Center. It's part of the same package as the FINRA Rulebook API, so, firms may already have that package. There is a fee associated with the API and you can get more information about it on our API Development Center. But it's really the machine-readable aspect of this initiative.
16:24 - 16:40
Haime Workie: The API process has a fee attached to it, but there's no additional fee associated with the Machine-Readable Rulebook. The terms are tacked on, so anybody who subscribes to the general API protocol would be able to receive the tag terms that are embedded within the Rulebook.
16:41 - 16:44
Kaitlyn Kiernan: And why develop the two different delivery methods?
16:45 - 17:10
Afshin Atabaki: They serve different purposes. The search tool is the public facing side of this, it allows any user, for free, to browse and select the various tags applied to each of the rules. The API allows you to transform the tags into a machine-readable format. But a firm might decide, look, we're going to use the API to be able to map the tags to our internal processes. So, that's the intrinsic value of the API versus a search tool.
17:12 - 17:22
Kaitlyn Kiernan: Haime, as part of the development of this initiative, FINRA conducted an economic impact analysis, which happens with a lot of new rulemaking. Can you share what we found from that?
17:22 - 18:14
Haime Workie: As Afshin mentioned, this is not an area of my expertise, but this was done by the staff from FINRA's Office of the Chief Economist who really took the lead and did a good job of trying to figure out what the economic impact would be in terms of the development of the Machine-Readable Rulebook. And this analysis is discussed in the 2022 Special Notice that came out in conjunction with the tools, the FIRST tool, as well as the API that got rolled out in October. The analysis revealed that FINRA member firms of various sizes, business models, stand to gain from FINRA's Machine-Readable Rulebook Initiative. By one estimate using a survey that was put out, there's indications that there may be savings to the industry as much as 8% of the FINRA-related compliance costs. But just as importantly as the cost savings, there's also the potential for a more efficient and effective compliance programs for firms as well.
18:15 - 18:23
Kaitlyn Kiernan: 8% certainly seems substantial. And are there any other anticipated benefits to firms or their vendors worth highlighting?
18:23 - 19:03
Haime Workie: One of the things it's important to think about as we're rolling this out: This is one of the potential tools within the larger regtech ecosystem. Regtech stands for regulatory technology. It’s really designed to make the compliance functions potentially more automated and more streamlined. But this is important not just in the sense of streamlining the compliance functions but freeing up resources for compliance personnel to engage in higher level compliance-related issues that potentially has greater added value that would allow us to further both our mission and the firm's goal, which is protecting investors and making sure the integrity of the market is such that the compliance policies and procedures the firms have support that integrity.
19:04 - 20:17
Afshin Atabaki: For one, it could significantly reduce this concept of the time to compliance lag. We typically spend a lot of time trying to find the applicable rule requirement. And we also spent a lot of time once we find it to determine whether a change or an existing rule impacts us. If you can find the relevant rule requirement and find it more quickly, you could enhance compliance efficiency. It can also free up your compliance resources, allow you to apply those saved resources to other efforts. And then if you subscribe to the API, you could also see whether a particular tag has been added or is changed as a result of a rule change. And you can then more quickly also identify whether that rule change impacts you. Just a quick example. As a regulatory attorney, you often get the assignment, 'hey go find every rule that applies to this concept.' And I'm going to pick disclosure, that could take weeks, even months, to identify each of those rules. I've done those exercises. If the term Disclosures was tagged like we have here, to all the applicable rules, again, we're focused on 40 here as part of this initiative, but if it's done and all you need to do is select that tag and with one click, you get the relevant rules. That's something that at least in my career, I've never come across. That to me is a significant benefit.
20:18 - 20:21
Kaitlyn Kiernan: Thanks for that example. That does sound like a big benefit.
20:22 - 20:24
Afshin Atabaki: It's a game changer in terms of searching and search functionality.
20:25 - 20:27
Haime Workie: So, I'm sensing some battle scars for Afshin.
20:28 - 20:33
Afshin Atabaki: Yeah, no disclosure assignments. No more.
20:33 - 21:36
Haime Workie: It's really hard to predict the future, Kaitlyn. But I think one thing is that market participants are likely to build on whatever we create, and it's likely to have it expand the impact. So, in terms of maybe where it's applied, for example, potentially beyond the Rulebook, other types of areas, but also in terms of what the use cases they have for it may be. As we talked about the API protocol, the idea behind API protocols—you have machine to machine communication, so that automates the compliance function—so, you can have the ability to have change management. For example, let's say there's a rule that changes that gets tagged with Record Keeping. That tag would get overlaid potentially with the policies and procedures that firms have that are also tagged Record Keeping, so, they would know they need to change some recordkeeping aspect as a result of that rule change. And facilitating those types of interactions is something that I think where you could have a lot more automated and streamlined approach to compliance that obviously would still require some level of human interaction to make sure everything's working appropriately but can free up resources.
21:37 - 21:39
Afshin Atabaki: Yeah, there's still going to be job security for lawyers.
21:41 - 21:49
Kaitlyn Kiernan: So, beyond the economic impact analysis, have you heard any anecdotal information from firms who have been using the new tools so far?
21:50 - 23:17
Haime Workie: As we're going through this process, I think it's important to note that we spoke to a number of different industry participants, both in terms of an expert group that we stood up to help consult with us as we were developing this tool, but also to potential users. And one of the areas that we heard that may be helpful is, for example, for startups. Let's say before even a firm decides to become a FINRA member, they may engage in certain types of startup activity where they want to know what potential rules would apply. Well, using the type of FIRST Search Tool that Afshin described, they could potentially figure out, let's say, ‘I want to be a trading and execution firm concerned about debt securities, and I want to find out what are the disclosure requirements that may apply in this context.’ So, they can find those rules and they can build their own process around that before they would eventually have to go to lawyers and so forth to help them stand up the business, but at least it gives them an area where they can tinker and experiment and figure out what rules would apply in that context. So, we've heard firms that are thinking about potentially becoming broker-dealers in the future say that this would be a really helpful tool for them to have. The other is, say a large firm that's out there that has a compliance side and a business side, and if the people on the business side were able for themselves figure out what are the set of rules that may apply in their context, that may help narrow the set of questions they need to ask the compliance folks in terms of what they need to do. And again, this can help free up compliance resources and make it easier to get out what the business desired goal is.
23:18 - 23:26
Kaitlyn Kiernan: To start wrapping up, there is currently an active Request for Comment about this initiative. What are you hoping to learn?
23:27 - 24:25
Afshin Atabaki: So, the comment period was extended to February 21st, 2023. We did that because it was around the holidays when the original comment period expired. The Special Notice requests comment on a variety of topics, including, if we do pursue this for the remaining rules, what's the best method? Do we partner with someone, including open-source folks? Or is there a specialized vendor out there that can assist with their natural language software? What tools can we use? And also, for me, the main takeaway from the comments hopefully will be, look, this has practical, real-world impact and we should pursue it. That's what I'm looking for. We did explore this topic with industry groups, with our committees, with many, many folks. But we've never had the full public view of the tool until we published a Special Notice as well as launching the tool itself. And we're really looking for that kind of feedback.
24:25 - 24:26
Kaitlyn Kiernan: Haime, anything you'd want to add?
24:27 - 25:07
Haime Workie: Yeah. One of the things that we tried to do as we were developing this is we worked with an outside vendor to make sure the rules and the tags that we were using were compatible with systems that may already exist out there that vendors are using or firms themselves have built. I would be really curious to hear from people about how they view the taxonomy that we have built, how usable it is for those who may already have taxonomies themselves, would help set up an industry standard or otherwise, are there other processes we should use in terms of developing a taxonomy? And then, as Afshin mentioned, thinking creatively about ways that we can carry this forward and working with the industry as we carry it forward. So, thoughts around that would be great as well.
25:07 - 25:15
Afshin Atabaki: Another reminder, comment period expires February 21st, so there is enough time to submit your comments and we will look for those.
25:16 - 26:02
Kaitlyn Kiernan: Yes. And we'll link to the Special Notice in the show notes so you can go check that out and check out all the questions that the team would love to hear your thoughts on. But that's it for today's episode of FINRA Unscripted. Haime and Afshin, thank you so much for joining me to tell me more about the FINRA Machine-Readable Rulebook Initiative. Again, listeners, you can go to our show notes for more information. And if you don't already, be sure to subscribe to FINRA Unscripted wherever you listen to podcasts. If you have any questions about today's episode or ideas for upcoming episodes, you can email us at [email protected].
Today's podcast was produced by me, Kaitlyn Kiernan, coordinated by Hannah Krobock and engineered by John Williams. A special thanks to Alex Khachaturian. Until next time.
26:02 – 26:07
26:07 - 26:35
Disclaimer: Please note FINRA podcasts are the sole property of FINRA and the information provided is for informational and educational purposes only. The content of the podcast does not constitute any FINRA rule or amendment or interpretation of such rules. Compliance with any recommended conduct presented does not mean that a firm or person has complied with the full extent of their obligations under FINRA rules, the rules of any other SRO or securities laws. This podcast is provided as is. FINRA and its affiliates are not responsible for any human or mechanical errors or omissions. Parties may not reproduce these podcasts in any form without the express written consent of FINRA.