What are the opportunities for digitized regulation?

FinTech Global recently sat down for an interview with Clausematch and FINRA to find out the opportunities – and potential pitfalls – of digitized regulation. Watch the full interview here.

Explain what is the FINRA machine-readable rulebook initiative? How did it come to light and how does it work?

Haimera Workie, VP and Head of Innovation, FINRA

"FINRA's machine-readable rulebook initiative is really an effort that's designed to make the FINRA rulebook easier to navigate and understand, both with respect to individuals like you and me, but also with respect to computer systems. The initiative is composed of three parts.

The first involves the development of taxonomy. And by taxonomy, we mean a set of terms that can be legal terms, business terms, but key terms that help identify what each of the rules relates to. For example, one key term maybe disclosure, and so we would tag all the rules that contain a disclosure element with this tag. Another key term could be record keeping, and we'd likewise tag the rules with that tag, and so forth. Once you have the rules tagged with appropriate terms, then you can access the information in the same way that you would do with things like online shopping sites like Amazon. For instance, on Amazon, if you're interested in looking at certain type of chair, whether it be kind of a certain height, a certain brand or certain color, you could go to a pull down menu and kind of specify which types of criteria you're looking for that will pull down the set of chairs that meet your criteria. And the same way, once you have the rules that are tagged, then you can say, you know, you're a certain type of firm that cares about things like say, for example, if we're talking about record keeping and disclosure requirements, and the rules that contain those elements would show up and make it easier for you to kind of identify which rules that you may be concerned about.

 

Now, we went through the process of tagging 40 of the rules. And those 40 Rules represent about 50% of the page views that we get on FINRA. Now, in addition to this first element of kind of developing that taxonomy, and laying it over the rulebook, the other two elements really involved the delivery mechanisms for the taxonomy. So that can be easier to digest both from computer to computer as well as from a human to computer. The humans computer delivery mechanism is an enhanced search tool that we refer to as "first." This tool is available on vendors website and can be used to search rules based on the key terms. And as much similar ways I described with kind of the online shopping experience where you can tag certain types of items that you're concerned about. Whether they be you know, things like you know, the firm that you are, for example, query execution firm, you may care about debt securities, and so forth, you can tag those criteria and identify the rules that are there that apply to you.

 

The other delivery mechanism that we have is designed for a computer-computer interface, which is the API protocol that allows you to download the entire rulebook along with all the applicable tag terms that may be involved. What this allows you to do then is basically link up your own internal policies and procedures, potentially, with the tags that are contained within the rulebook. So for example, if there's a change in a requirement that relates to, you know, a record-keeping requirements, that tag would show up on the rulebook and a tag would potentially show up also on your policies and procedures. Therefore, you need to update those types of policies and procedures."

 

 

"We've received a lot of fairly positive feedback on the initiative, specifically around kind of areas, for example, for new firms that may not be familiar with the rulebook, so they can help identify what may apply in their specific situation, but also from kind of more sophisticated firms that allows them to help facilitate the automation of the compliance process."

 

"The impetus for this project really dates back to a special notice on innovation that we put out back in 2018. And there we asked a series of questions about the merits of developing a more digitized method for accessing rules. As part of that, there was quite a bit of support to explore this in more in depth kind of ways, which ultimately led to the development of the machine-readable rule book prototype, which we put out last year."

 

"As part of the special notice that we put out more recently last year, along with kind of the machine readable rulebook kind of different elements I spoke about, we had an economic analysis section, where we talked about kind of what the potential impact would be from a market standpoint, both in terms of the benefits that the industry would potentially be able to have kind of qualitatively as well as potentially quantitatively. And what we found there is that there's potentially a significant cost savings for firms with the use of these types of tools in terms of their compliance program and potentially an enhanced ability to have a more effective compliance program as well."

 

 

"2022 for Clausematch has been very interesting, because before 2022, we worked primarily with global financial institutions like global banks and Tier 1 financial institutions, banks and asset managers. But 2022 really was the year when we started getting a lot of interest from smaller financial institutions. And it just shows that compliance is becoming more and more important for even smaller financial institutions who might not have hundreds or thousands of compliance people and compliance departments, but actually, just a few people working on compliance, and they're implementing technology now to deal with issues.

 

On the other hand, we also have invested heavily into Research and Development (R&D) on the regulatory side. The combination of work was the release of the Knowledge Graph. It's the digital way to apply machine-learning to rulebooks of regulators. And that was the work we've done with one of the regulators, which was practically having subject-matter expertise from a regulator going through the rulebook, tagging the rulebook with the taxonomy terms, and various tags, and then applying machine-learning to that data set in order to tag the remaining of the regulatory data; this was done with very high precision. And actually, this is a one of a kind data set now released into the public in order for others to be able to take that data set, teach the machine, train the models, and then apply to the specific rulebooks that regulators would like to tag."

 

"A knowledge graph is a well-known term among data science people and developers. But a knowledge graph is a way to represent our graph database in a visual way. So in effect, what you're seeing when you're viewing the Knowledge Graph is how the data connects to each other across the dataset.

 

A simple example: take a term like loans or credit cards; it would be a node within the Knowledge Graph. And then you would see everything that is related to credit cards. For example, as a product, you will see the documents which are related to other taxonomy terms, and you can explore it in a visual way, how all of that data is related. Equally, for example, you can start from a credit card term, and see that there are five obligations in the specific regulatory document related to credit cards. And then if you go further, you can say that credit cards are also a product which is provided by the financial institution. And that financial institution has ten procedures and policies, which talks about how to issue cards, the impact of AML/ KYC on customers, and so on."

 

 

"We see huge opportunities for institutions to be able to consume regulation in a much friendlier way, and to understand what is applicable to them. I also think a bigger opportunity in the industry is that the knowledge graph allows you to start mapping internal processes, policies, procedures, controls, and even systems to the regulatory data set, which then allows you to understand the impact.

 

Imagine you've got a particular regulatory document changing. You can see the waterfall-effect on everything within the organization that is mapped to that specific regulatory change. That is what 99% of our customers are trying to achieve, but it's not easy. And it's not fully possible unless we digitize the rules."

 

 

"I think long-term, you've got one change in the regulation and you understand at the market-level the impact of the that new proposed change. That would be huge. That could drive regulatory policy in a much more uniformed way."

 

 

"From our perspective, to the extent that the industry wants to move towards digitized rule books or have a way of having greater regulatory intelligence associated with the rules, we want to be able to facilitate that in a way that's consistent with our broader mandate, which is investor protection and market integrity. We see this effort of development and a machine-readable rulebook as potentially doing that, because what it does is, it allows firms to potentially have a method for enhancing their compliance efforts in a way that's more effective, both from a cost perspective method, but also in a way that potentially allows them to devote resources to potentially higher-skilled levels of compliance.

 

As we've talked with firms about this, I think one of the things that they really pointed to is the desire to be able to make sure that they're not missing things. For example, if a new rule comes out, or a current rule gets changed, they want to be able to make sure that they're updating their policies and procedures in the appropriate manner. And this tool potentially allows them to do that in a more effective way."

 

"This is what the industry has been asking for a long time. To give you an example: a large asset manager operates in many countries and has compliance officers functioning in different regions. They spend hours looking at regulatory changes. That work is not necessary if the rulebooks and the data comes out of regulators in a digital form. 

 

The remaining process is understanding whether the regulation is applicable to you. Which departments, which systems, which policies does the regulation actually impact? Everything right now is manual. By simply changing a way to release the rules in a digital form, we are automating things in a huge way. In the current algorithms today, we can also see which parts of the new regulations may conflict with their current policies. This is all possible once we digitize regulation."

 

 

"I think that it's potentially a way forward. And for all the reasons that Jay and I both have spoken about, because it offers a number of benefits. But I think we would want to hear more from the industry regarding this as well, which is why we put out the request for comments last year as part of the rollout of this initiative. Specifically, in what ways would they actually use tools like this? How would would supplement or augment current efforts by the industry, whether it be vendors or in house efforts by firms? And what does the development of a tech assault taxonomy look like?

 

As you think about from one regulator to another regulator, and how those those interplay, we did try to think about some of these issues. And we actually worked with a vendor to look at how other rules may be tagged and make sure that we're trying to be consistent from a broader approach. But that's an area that I think we would appreciate comment on and some thoughts from the industry."

 

"I love that idea that validating rules from other regulators and other countries as well to harmonize them because that would make life easier for global institutions."

 

 

"I think there's potentially two ways: one is direct access. So if you are an investor, or you are a small company that's looking to become a broker dealer, having the machine- readable rulebook available to you allows you to potentially access the types of requirements that may apply in your situation. Obviously, ultimately, you may need to seek counsel in terms of making sure that you're in complete compliance, but at least this is a way to get started. It's conservative, it's very much a useful tool.

 

And for investors who may care about what are the appropriate things that apply, in this case, when I'm working with a broker or somebody else, it can be a tool that can be used to figure out what are the types of requirements that may apply in a more in-depth manner. These tools can also be used as a way to augment the ability of firms to  automate their compliance process. And as I was speaking about before, to the extent that they can save time, in terms of tasks that would have been done manually, they potentially can devote those resources into enhancing their overall compliance, which obviously has a direct benefit for investors."

 

"I think if we look at the big picture of why do we have regulations? We have regulations to make products and services safer for the public. When compliance programs don't follow regulations, then ultimately, that leads to misconduct and human error and unsafe products in the market. As a general view, by digitizing regulation and digitizing compliance, ultimately brings safer products to the public. That can be applied to any industry that has standards they need to follow. They are there to make sure we operate safely and I think there is a huge benefit to the public. 

 

I also think there is a huge potential for more innovation generally within the industry. Because if it is easier to understand the regulations as a new entrant, then it is c cheaper to start, it's cheaper to comply with regulations.  Then, you can direct the funds which you would have spent on compliance towards innovation into products and services."