The Road Ahead: Municipal Data & Technology Developments

From AI to machine learning to electronic trading platforms, the municipal market has been integrating these technologies to improve workflows and expand transparency. The municipal market is ripe for growth in these areas and our panelists will discuss how participants can integrate technology to help create a more efficient marketplace.

Transcription:

Tim Stevens
All right. Well, good afternoon, everyone. My name is Tim Stevens. I'm the president and co-founder of Lumesis, a financial technology company. And on my panel today, I've got Unmesh Bhide, chief product officer for Pricing Direct. We've got Abhishek Lodha with us, who is vice president of product development at Assured Guaranty. And I've got John Murphy, known in the market as Murph affectionately, who's a director at PFM for Investor Relations at PFM/Munite.

00:00:43:05 - 00:01:09:19
Tim Stevens

So I look forward to our conversation this afternoon. We're talking about technology and  several different focal points. I'd like to thank all my panelists today for joining me. I do believe, honestly, that we're at a really interesting time in our market for the application of technology. Both the adoption of technology and the creation of new technologies across our industry.

00:01:10:11 - 00:01:39:13
Tim Stevens

If you think about it, banks, institutional investors, municipal advisors,  issuers themselves, all the different participants in our market, at no time in history need more than to bring greater efficiencies to their workflows, improve their decision making, enhance their ability to compete with their market peers. At the same time, in our market ecosystem, there's never been greater availability of data.

00:01:39:14 - 00:02:05:06
Tim Stevens

Yes, there are still some issues with data locked up and unstructured, but we've never had more access to data than we do today. And then just basic technology computing power, cloud computing, the increased availability of artificial intelligence and blockchain technologies, a lot of really, really helpful tools to help us further our endeavors in this market. Most market participants as a technology vendor.

00:02:05:14 - 00:02:37:10
Tim Stevens

Yes, some are afraid of technology and worry about what it's going to do to their positions and their workflows. But most, most do embrace it as a way to improve the market that we all care so much about. Some of us in this room, I'm sure, remember bearer bonds and remember paper tickets that you used to punch with a clock on the trading desk and and so many others like the blue list to find secondary bonds.

00:02:38:17 - 00:02:59:21
Tim Stevens

And while some of those might be nostalgic to think about, I think all of us would agree that the market's better off today with book entry securities and with the ATS’ and electronic ticketing systems and all those other things that have made things just much more efficient. So today we're going to talk about several different focal points.

00:03:00:12 - 00:03:20:11
Tim Stevens

The first one we're actually going to start with as it relates to technology and data is on the pricing side. So I'm going to start with Unmesh for my first question here. And so the first question really is how is Pricing Direct itself, really helping firms create efficiencies in their workflows for pricing?

00:03:21:03 - 00:03:52:19
Unmesh Bhide

So, I mean, if you look at it in terms of valuation perspective, there are million plus bonds in muni world. And if you think about Vector only, you talked about AI as a kind of driver. I mean I mean, AI has been around since the nineties when I was still starting. And I think the real important part is in the last I would say five to seven years, is how the hardware and the software has helped AI to move forward and how it is actually embraced in the financial industry more and more.

00:03:52:28 - 00:04:13:26
Unmesh Bhide

And I think that's something the first important part of that. Now we price close to 2.8 million securities. But I think the thing that we had to think about it, it's big data in MuniWorld we have 60,000 plus issuers. So in that scenario, you had to start thinking about how are you going to cohort that number of securities.

00:04:14:05 - 00:04:58:14
Unmesh Bhide

And I think in that sense, it is very interesting that the way machine learning specifically kind of comes in kind of grouping of the security is very important. I give you a very good example. If you go back to March 2020, when the markets were difficult, if you want to call that way here the correlations between different sectors of the muni market like highlighted what it says education higher able to and say airport bonds but you never think about they have a similarity but it will show you there is a similarity because of COVID 19

00:04:58:14 - 00:05:30:04
Unmesh Bhide

So then the correlations that you see and that you observe you think as a pure analyst thinking about it is always a difficult aspect and the way we think about it in Pricing Direct is really pure human or pure machine is not as good as human plus machine. And I think if you go back to you in general technology-wise like we always talked about 80 20 rule for the ages, the way we look at now with email is we are talking like 95 viral and the extreme of how you are content that five was.

00:05:31:03 - 00:05:44:07
Tim Stevens

Interesting. So traditionally the market has really priced on a yield to worst basis. Is that what you're doing at Pricing Director you're considering option adjusted spread analysis. What are you doing there?

00:05:44:10 - 00:06:09:06
Unmesh Bhide

Yeah. So we are looking at OAS as a primary option adjusted basis. I mean in the muni market year to us has been the standard and still coded that way. But I think from an analytic perspective, it makes more sense. And I think the question really comes into play is if you go back to again March 2020,  because that's where you are always trying to back test how your valuation works.

00:06:09:18 - 00:06:35:28
Unmesh Bhide

And I think in that case, the lawyers kind of make a very big difference. The other part, I would say in the muni market is very I would say from  is a very fragmented in a sense, for example, the bond coupon say 4% and 5% versus let's call it 2% or 3%. The reason being a discount bond was a premium bond, though, there is a big differential off West.

00:06:36:12 - 00:06:49:25
Unmesh Bhide

So that is something that you can capture through the OAS model and you know, like obviously there are reasons for the tax and all those things. But at the same time, from your values and perspective, OAS helps in that there.

00:06:50:02 - 00:06:56:22
Tim Stevens

And any other interesting nuances perhaps that you've seen in the market like between taxable and tax exempts.

00:06:56:24 - 00:07:17:09
Unmesh Bhide

So I think the other interesting part, I think going into June of 2023 is going to be also migration of the LIBOR to SOFR. So I think from a political perspective right I think you can look at a you know like for the longest of the time we always talk about percent of treasuries as one example.

00:07:17:15 - 00:07:48:14
Unmesh Bhide

You know you also look at from me to the swapper or you can now look at the swapper. So in a way, as we move forward, that is the most important part. Now if you're trying to do a cross sector or relative value analysis say against the corporates versus taxable munis, you need to then think about it, how you're running your model in terms of volatility, perspective, because, you know, like the volatility that you might observe in the corporate world, implied volatility might be different than your implied volatility in muni world.

00:07:49:01 - 00:07:56:10
Unmesh Bhide

So the under their approach to look at it is how do you do your calibration of wall in your way. Smart is also.

00:07:56:12 - 00:08:28:29
Tim Stevens

Interesting. Thank you. Thank you. And I'll I'll add a little bit to that from my perspective at Lumesis. We've been really focused on on pricing recently in the past few years and the products that we've brought out to the market and really our focus has been in two different parts of the market, primary market pricing, where we have essentially a scale writer that builds a custom baseline scale for an individual deal that's coming to market based on its structural credit characteristics.

00:08:29:11 - 00:08:57:24
Tim Stevens

And where efficiencies are found by our users really on the desk, it's helping them address all the scale requests that they get right. Countless scale requests from bankers, oftentimes indicative and many different permutations. Right. It's never just one structure, it's a different coupon stack that sits with and without call. It's taxable tax exempt. And then the deal doesn't happen.

00:08:57:24 - 00:09:22:16
Tim Stevens

And four weeks later, they're back with the same credit, looking to generate scales again and see how the market has moved. So the ability to generate scales quickly and accurately in a nice, systematic way is how we've been helping there. The other way is really to help underwriters ensure that they're seeing all the comps in the market that are relevant right when you're building out of scale.

00:09:22:23 - 00:09:46:01
Tim Stevens

And that's particularly helpful when you're looking at a deal that's perhaps in a part of the market, a market segment that you're not particularly active in. And also being able to utilize secondary market data. So that's not that's been hard, I think, for a lot of firms to crack. Right. Just tremendous amount of secondary market transactional data.

00:09:46:15 - 00:10:22:21
Tim Stevens

But being able to bring all of that in spread IT database and pull the relevant trades to a particular pricing and use those in a systematic, consistent way is what we're helping with there. And then building in time saving efficiency features like maturity structure builders, we've integrated I propose public calendar so that our competitive underwriter, for example, can simply click a button and literally click a button and we draw a baseline dial specific custom yield curve based on transactional data that's that's readily observable in the market.

00:10:23:18 - 00:11:03:21
Tim Stevens

So that's, that's how we're helping on the primary market side and bringing technology to pricing there. On the secondary market side, we're really focused on a couple of different things. And this is primarily if you think of sales desk or traders, whether it's traders on the buy side or the sell side of the focus, there is a secondary market pricing analysis on an individual bond so a similar type of concept where you're using the structural and credit features of particular bond to source comps and to get an understanding of where the market is for that bond, whether or not that bond has said trades, obviously looking at that, but then looking at relevant comps based on

00:11:03:21 - 00:11:29:26
Tim Stevens

its features and then a trade ticker ing system which can be highly customized based on the market segment that you're looking at. So for example, if I wanted to look at single-A rated private higher ed in the Midwest with maturities 10 to 12 years from now and a 3% coupon, you can build, you know, a very specific ticker to look at live trade activity or historical.

00:11:30:05 - 00:11:40:17
Tim Stevens

So different ways that we're helping at Lumesis on the pricing side. Any other points anybody wants to make on the pricing side?

00:11:40:29 - 00:11:57:14
John M. Murphy

Can I just ask Unmesh one question. When when you're using your OAS pricing in the model does that ever conflict with just, say, mutual funds that are using their own OAS model in their pricing and how they in their valuation mean.

00:11:57:29 - 00:12:23:11
Unmesh Bhide

You know, it's lingo in the industry. There are so many ways models and like how you implement your that go your interest rate generation bats and everything so what we do is basically use a price and that's what I was talking about implied volatility to kind of look at it or you can look at from the implied spread based on a fixed wall if you're looking the other way around.

00:12:23:11 - 00:12:44:21
Unmesh Bhide

So at the end of the day it clears up on a price basis realistically early and that's where, you know, when we get the questions from the clients in terms of, you know, different models are predicting the richness of cheapness will be different. But we try to kind of have a dialog mode on the price last year to US business.

00:12:45:13 - 00:12:49:27
John M. Murphy

I apologize. I probably should have yesterday that my own edification.

00:12:49:27 - 00:13:12:21
Tim Stevens

No problem. So I know everybody that will be shocked to hear about this, but our next focal point is, yes, change in technology. It's been on every single panel, of course. So we wanted to talk about a couple of different aspects of ESG. Obviously, from a data perspective, start with Abhishek and get his thoughts. We obviously have heard a lot today about how there's fragmentation.

00:13:12:21 - 00:13:21:26
Tim Stevens

There isn't a lot of standardization. So I just wanted to get a sense from you, Abhishek, where are you seeing ESG data coming from? What are people using in the.

00:13:21:26 - 00:13:43:26
Abhishek Lodha

Market for sure. Firstly, I want to say that I like fragmentation. I think it leads to a lot more opportunities at market, especially because it's it's again, you know, someone said in the previous panel that there's a lot to unpack when you talk about ESG it's a small term, but it's a big spectrum, right? And provide having more providers.

00:13:43:26 - 00:14:00:17
Abhishek Lodha

And I think, you know, my last comment was around 20 or so different data providers, including rating agencies. So we're seeing a lot of movement happening. The way we think about it is, you know, very simplistically as two spectrums on the ESG side as well. You've got the risk management side, but you have a lot of data providers.

00:14:00:17 - 00:14:29:00
Abhishek Lodha

So the most tangible example is environmental and climate risk. So you see vendors like risk. I assess we're providing a lot of information on that space, but I think there are a lot more opportunities when it comes to understanding what that means for credit migration right at the end of the day. And that's what we're trying to get from from an ESG or environmental aspect A lot of people are trying to sort trying to crack that.

00:14:29:00 - 00:14:50:17
Abhishek Lodha

My previous job before and we were trying to do that from a data science, we use a lot of machine learning back then to figure out what environmental risk means. But again, it's, you know, you can't apply that to the muni market or 50,000 issuers in a blanket with the other spectrum outside of risk management where I see a lot more tailwind is values investing.

00:14:50:27 - 00:15:15:12
Abhishek Lodha

Right and that's where the market is is quite right because you see a lot of pressure coming from top down for institutions. So you see, you know, corporate social responsibility is for large institutions. They're trying to figure out how they can incorporate more of those ESG mandates flowing from Europe of course in the US public finance market. And the other is really the small market and impact investing, right?

00:15:15:12 - 00:15:39:06
Abhishek Lodha

You're seeing smaller denomination of accounts. Each one wants some sort of customization and flexibility and ESG tells a beautiful story for a lot of these people. So I think there's a lot more opportunity for data vendors in that space where you can start providing better stories and information around impact. And it could be a either use of proceeds, which is again, a big topic.

00:15:39:06 - 00:16:00:29
Abhishek Lodha

We've we've all looked at reference data from all the large vendors out there and one of the challenges that it kind of boxes is issuances into one single category. But there are tons of them, right? I mean, education, right? When a school bond is issued it could be used for building, you know, providing a lunch program. It could be used for building computer labs.

00:16:00:29 - 00:16:21:12
Abhishek Lodha

Right. All of these are different metadata. When you think of it from an investor standpoint. So unpacking that, creating that data is a huge opportunity. And and on the values investing. The other is which again, I've seen some providers do is identifying good players in the market. Right. So if you think of issuers as enterprises, how are they behaving?

00:16:21:12 - 00:16:30:03
Abhishek Lodha

Are they utilizing their resources in the right way based on how underserved their communities are? I think there's a lot of data that's now being generated out there as well.

00:16:30:28 - 00:16:44:27
Tim Stevens

Excellent. And if you were to think about you know, the application of technology in the ESG space, how how do you see technology? You know, how can it be used better to better segment the market itself from using that lens?

00:16:44:27 - 00:17:09:22
Abhishek Lodha

I think, you know, a lot of this ESG analysis, fortunately or unfortunately, is in the labs of credit analysts. Right. They're always they have to justify either to the investors or to the portfolio managers why this is a good investment through the lens of credit plus ESG. So the way we think about it is, you know, at the end of the day, credit and ESG needs to come together and the way technology can help.

00:17:09:22 - 00:17:32:19
Abhishek Lodha

And this is something that we've done exercises in the past as well is rank stacking issuers, rank stocking issuers based on ESG metrics and then identifying on a relative basis, maybe on pricing or on credit rating which ones fall within the clients portfolios. This is from an array perspective at a should we use it more to understand our portfolio risks?

00:17:33:16 - 00:17:53:17
Abhishek Lodha

Because environmental risk is a real fundamental risk for us because we're taking the deal for the life of it. So how do we take environmental risk and identify where we are clustering our portfolio and what makes sense, you know, going forward? So that's how technology, big data helps us really rank, stack and you know, almost like build a custom scoring model.

00:17:53:17 - 00:18:16:02
Tim Stevens

Sure makes sense. Thank you. Next thing I thought we talked about this morning, there were, you know, again, on every panel, there was discussion about ESG and at different points in time, there's been discussions on pricing related to ESG. Right. And if you're seeing any price differential this morning, heard a lot of no's last panel. Right. There was some talk about the Green Hill.

00:18:16:08 - 00:18:24:11
Tim Stevens

So having good measure, great opportunity. What are you seeing in the data in terms of whether or not there's a premium and where that might exist in the market?

00:18:24:12 - 00:18:56:29
Unmesh Bhide

So I think in terms of the employees, I think if you look at in the tax exempt market, right now, there is like, you know, if you look at some of the recent trade data with MTA bonds, I know in the last panel they talked about MTA being one of the big issues in that if you look at one of the recent trades where they were taxable as well as I mean green versus non green bonds with a similar maturity, you know, there is not much spread there I mean, if you look at, you know, in a, you know, taxable market also is not there.

00:18:56:29 - 00:19:20:24
Unmesh Bhide

I mean, that's kind of the very interesting aspect because if you compare that to the corporate market where you see a significant premium and if you go into a European sovereign issuers or European issuers you will see that much more in terms of the bias from Europe perspective, you know, they are looking for a green bonds perspective. So that's I think the one aspect.

00:19:20:24 - 00:19:40:16
Unmesh Bhide

And I think the other interesting aspect in terms of premium buys also will be which we we you know, we are looking at the non-US market, but at the same time, immune is too is really if you look at just a lot of the green bonds, but then you are looking at a social bonds and also sustainable bonds.

00:19:40:24 - 00:20:07:13
Unmesh Bhide

So I think how does that comes into play? I think next year or two will be interesting because as the issuance comes into play from various different means parties and whatnot, I think that's going to drive. And I think at the same time, the bigger state issuers would actually try that different because you need a actual, you know, amount outstanding to really drive that market forward.

00:20:07:16 - 00:20:17:19
Tim Stevens

Is it a challenge given the lack of standardization is it a challenge to kind of label those bonds? Right. How do you determine how you bucket these when you're doing that price analysis?

00:20:17:20 - 00:20:34:26
Unmesh Bhide

Yeah, I think the I mean, I think it goes back to into that. I mean, I think if you look at you, there's a lot of talk about greenwashing, too. So in the V, first you need to figure it out is really it's a greenwashing audit, a real green bond. I mean, there was in the previous panel talking about the labeling.

00:20:35:05 - 00:21:10:25
Unmesh Bhide

So I think that's the first aspect of it. And I think the more challenge I want to see is like on a sustainable model because in that case, you are actually trying to have a goals meeting next five years or ten years. So the tracking of those goals is also going to be and I think if you look at it from a big data perspective, I mean, the disclosures, if there are new disclosures are in standard format as a mail person, I would say that will help because I think if you are looking at a million bonds, you cannot look at that disclosure systematically.

00:21:10:25 - 00:21:13:25
Unmesh Bhide

And that's what I would go back to. The technology will help.

00:21:13:27 - 00:21:35:17
Abhishek Lodha

Yeah. And just to add on to the labeling red, like we've seen Green Zone labeling is still a very small sliver of the primary market. Right. It's extremely small. So how do you generate enough information to back, distort and look at the other and then the 999,000 bonds that are outstanding. Right. And be able to label them, which is where again, technology is going to play a big role.

00:21:35:18 - 00:22:02:24
Abhishek Lodha

I know we're going to get on the topics of LP and stuff that have been discussed in the past and and maybe we're going to discuss it today. But there's a huge opportunity for players outside of just looking like, you know, outside of just third party verifiers or verifiers themselves to look at the secondary market because there are so many funds out there that like funds have increased exponentially, which are ESG focused values focused, impact focused, but they're not getting the data that they need.

00:22:02:24 - 00:22:21:15
Abhishek Lodha

Eventually, a lot of them end up looking at loan classifications, higher education. That's it. But if you're issuing I don't know, for, you know, building a stadium, is that actually good use of money or the fact that and this is a stupid example I keep giving, but you know, a tobacco bond that funds a cancer research center, is that good or not?

00:22:22:04 - 00:22:31:08
Abhishek Lodha

Right. I mean, it's it's it's all it gets subjective, but there needs to be enough data to be generated for investors to then take that and apply their own thesis on top of it.

00:22:31:16 - 00:22:50:28
Tim Stevens

Excellent. Right. Thank you. So we're going to move away from ESG now. Switch the focus over to investor relations. So I'll I'll address this to Murph. Murph. Can you give us a sense of what's what's the real thrust of of the mini platform and how are you bringing efficiencies specifically to municipal issuers in that space?

00:22:51:20 - 00:23:10:08
John M. Murphy

First, I want to go back to ESG I have my own opinions and a fair amount of them voiced on this. But just the one thing on ESG I will say is there are note like investors, you have SMEs that are different. You have insurance companies have a different point. Mutual funds have a different view. Some are top down, some are bottom up.

00:23:10:08 - 00:23:26:21
John M. Murphy

Some demand, you know, impact numbers and scores. And others just want to know that you have a rating or, you know, you thought about it. So, you know, the standardization is key, I think, from a stance of terminology. Yeah. But I think this might come out in the MST or B requests for information that there that is what I think.

00:23:26:21 - 00:23:54:17
John M. Murphy

You know, we don't want regulation. We want, you know, some standardization so people can compare apples to apples all the time. So that's it for my ESG rant. But, you know, from from a standpoint of technology, I work at DFM, PFM developed a platform, an investor relations platform for and and it's the issuers outreach platform to investors. So it's a one stop shop of information, all pertinent information from that issuers put on a platform that investors can view.

00:23:54:17 - 00:24:15:20
John M. Murphy

And that includes everything. That's what I call the three, the three, the three legs of the stool. You have manual disclosure, which is, you know, 15 to 12 stuff, and then you have the voluntary disclosure and then also then you have the outreach. So those are the three things investor relations that I think need to be addressed. And then the, the outreach portion is muni well issuers pay for it.

00:24:16:04 - 00:24:43:23
John M. Murphy

They put up their information so they have better relationships with their, their buyers and their investors. Of who they may be. And they, they get a fair amount of information of, you know, tracking and where things and who's visiting their sites and what's what's going on. But the information they provide investors I think is really key because if you can find everything you need to make an investment decision on one page, it's there and we cover the whole 50,000, you know, plus muni issuers.

00:24:44:17 - 00:25:16:04
John M. Murphy

And we have multiple clients that are providing full information on the platform. But the information's there and you can set up portfolios. It really is a unique concept. It's a platform, it's a portfolio, it works in a lot of different ways. But for issuers it's their, their way of reaching the market and having it passively on someone's desk and in their portfolios and how they view it every, every day with, you know, like I said, I was one of the when I was at Fidelity, I kind of had answers to questions in some surveys for the folks at DFM on the public power side about what I would want if I were, you know, if if

00:25:16:04 - 00:25:34:20
John M. Murphy

an investor, you know, there was an investor relations site for some of the larger utilities out there. And I gave them my input and a fair amount of that is in there. So other investors had input into the site as well. So if you don't know what it is, come out and visit us outside. But we have a booth out there, if not just, you know, give me a call.

00:25:34:20 - 00:25:52:02
John M. Murphy

But it, it is the one spot. It's like this this market really needs. And in a lot of different ways, you talk about the ecosystem we have these nice little providers of great services all around, all around the business, whether it's for an FAA or whether for investors, whether it's for issues, whether it's for underwriters. There are so many great tools out there.

00:25:52:02 - 00:26:00:17
John M. Murphy

It's harnessing them all and putting it together to make it easier and more efficient for people on that. On the Muni platform, I think that's what we've done for investors and for issuers.

00:26:02:02 - 00:26:11:05
Tim Stevens

How about, John, are you seeing are you seeing access of the system by retail investors? I would imagine a big focus is institutional. Are you seeing leasing yeah.

00:26:11:11 - 00:26:32:00
John M. Murphy

One of the unique things is that if you sign up, if you're an investment professional, you sign up, you get credentials and you can build your own portfolios and that gives you access to your dashboards. But you have also access to individual, individual names. But we also, for our clients, when they're bringing new issues and on their investor relations sites, we have a retail link so anybody can click on it and go right to their page.

00:26:32:00 - 00:26:42:25
John M. Murphy

And that's been a huge source of a lot of our hits for for our clients. So retail is accessing it while also the institutional side, so both sides can can see it.

00:26:43:03 - 00:27:12:01
Tim Stevens

Interesting. Thank you. Any any other thoughts on investor relations? We'll move on ESG. ESG. So now we're just going to go to the broad subject of data and data aggregation, which you can spend probably two sessions just talking about that in our market. We know there's a tremendous amount of data in the market. Right. Again, never, never been more frankly, but still a fair amount that's unstructured, locked up in documents.

00:27:12:28 - 00:27:22:07
Tim Stevens

So, Abhishek, could you get your thoughts on, you know, what is the market do and what are you seeing happening out there? In terms of aggregating that data point?

00:27:22:10 - 00:27:48:06
Abhishek Lodha

Sure. I'd start by there's there's obviously a lot of unstructured data that still needs to be unpacked and unlocked right so we've all looked at, you know, it was I just give a funny and I started my muni career seven years ago, so not that long. And I was I was sitting in the audience and someone in the panel said that look at the technological transformation right in 2014 is like we can now control it into PDFs, right?

00:27:48:08 - 00:28:11:20
Abhishek Lodha

That's the point I'm trying to make is that we've come a long way since then and a lot data has been generated from unstructured data. Financial data is now available from four or five different sources. A lot of institutions are now looking into it, which is an intelligent document processing so that they can generate their own tabular data A lot has been done, but again, there's a, there's a long way to go.

00:28:12:14 - 00:28:36:13
Abhishek Lodha

So data is now available in a lot of ways, right? You know, the governments and federal governments have done a phenomenal job in terms of availability of data. The problem arises in accessibility. And the way I differentiate is it's available on websites. It's available in documents. But it's not centralized or stitched together. So I see a lot of firms actually now starting to focus out there because there's been a democratization of data tools.

00:28:36:13 - 00:29:00:05
Abhishek Lodha

Right. So a lot of you must have heard of, you know, tools like Tableau or Power BI. Right. A lot of larger institutions that are onboarding that. And it's almost like I call it by to build a kind of a situation where they're buying these tools that are available with which they can now unpack a lot of the data that they are either sitting on or it's available and stitched together so that they can evaluate their credit better.

00:29:00:05 - 00:29:26:24
Abhishek Lodha

So a lot of movement is happening around out there, but I think there's still a long way to go. Right? Larger institutions can afford it. I think medium and smaller, it's coming to that in terms of being able to access that data. And one, you know, every analyst I come across at least, you know who's younger than me, I you know, I tell them, hey, you know, stop learning, Excel, learn Python, because that's going to be the next tier of credit analysis for them because it's very usable.

00:29:26:24 - 00:29:30:09
Abhishek Lodha

So big data is now a lot more accessible than what it was.

00:29:30:09 - 00:29:59:12
Tim Stevens

Yeah, yeah. Yeah, I agree. And that's actually a good segway into into artificial intelligence. Because that's really what we're talking about here. And just speaking from our own experience that we miss this, we really started experimenting with artificial intelligence, probably about eight years ago, and it was really focused on this problem of data extraction. And what we've seen is just really a fundamental huge shift in the marketplace in terms of the availability of tools and services.

00:30:00:19 - 00:30:29:22
Tim Stevens

I was going to use that term. You stole it. Democratization. It's true. It's absolutely true. You see big firms like Amazon Web Services coming out with tools and services to really support not not Python developers. And not data scientists, but actually business analysts be able to utilize tools and build artificial intelligence models to answer questions in datasets that they have.

00:30:30:14 - 00:31:09:08
Tim Stevens

So, for example, Amazon has a product called Sage Maker One of the things when you're working on a AI, you're trying to determine several different things. Obviously, there's the whole data component, getting the data available and cleansed and ready to go and then there's a selecting the right model for the problem that you've got. And Amazon now has tools that help a business analyst determine what's the optimal model for the problem that you're looking at, the data center that you're looking at, there's the things called hyper tuning of the parameters in your models, and they've got processes used to help find the optimal set of of a few parameters.

00:31:09:08 - 00:31:38:08
Tim Stevens

So a lot of incredible changes that that arc is going to make this technology more available to a broader audience at our shop. Of course, we're trying to build or we're building artificial intelligence tools that have to scale right now, have to be industrial strength and used by our clients in our platforms right but there's an awful lot I think you're going to be able to do at your desktop.

00:31:38:20 - 00:32:08:20
Tim Stevens

All right. And using the computing power of the cloud really in accessing and accessing all of that for for our experience, what we've done in terms of how we've used it, it's really in a couple of different ways. Fundamentally, one is actually making sure that our own proprietary data is clean. So we use it essentially as a quality assurance tool to look for perhaps anomalies in the data and to improve data quality.

00:32:09:02 - 00:32:31:16
Tim Stevens

And the second is in our scale writing platform that I talked about, I mentioned those scales are built off of transactional data. But what do you do if there are points in your curve where there simply aren't market transactions? Other either primary or secondary. And so what we do is we fill in those holes in the yield curve by replacing what we're doing.

00:32:31:16 - 00:33:01:21
Tim Stevens

We're essentially replacing traditional interpolation methodologies like natural cubic spline with artificial intelligence, where we're able to based on that, again, structural characteristics of the transaction that you're looking at, have an understanding or prediction of what the curve shape should look like, not necessarily the nominal level, but the shape and so you can then splice that shape into the rest of your curve that is built by market transaction activity.

00:33:01:21 - 00:33:26:19
Tim Stevens

So those are just a couple of different ways that we're using. And at one, assess and we're frankly getting ready to embark on more hiring additional people really to push into that space. So earlier there was a panel about A.I. and and technology, and they think it's going to have a significant impact on our market. 90% of everyone said, yeah, it's going to have an impact.

00:33:26:19 - 00:33:55:00
Tim Stevens

And it is it's certainly here to stay. I think you're going to see more and more of it. So. So it wouldn't mesh for pricing direct. I know that you use A.I. extensively to to build your your pricing models in periods of volatility like that. We're in rates backing up. How do you control for that 100? How do your and models and machine learning models control for that?

00:33:55:06 - 00:34:17:01
Unmesh Bhide

So I think the important aspect, right, I think when you're building the models is really you know, I think maybe I'll let me step back for a minute in a sense, like when you're having a team of people that you are doing, it's a really a combination of what we call evaluators plus data scientists as a team together.

00:34:17:01 - 00:34:49:02
Unmesh Bhide

Because I think the the first problem that normally you see is a data scientist, you know, like to work in a vacuum versus a business working in their side. I think the first thing that you need to do is a kind of a good combination because that's going to drive your success of the project. That's number one. Number two is really the data scientist should know for the may not be a subject matter expert, but they should have a basic understanding of the markets.

00:34:49:17 - 00:35:12:05
Unmesh Bhide

And the same way the evaluators are need to know the basics of the email or machine learning specifically. And I mean, we don't expect them to build the models, but at least having a good intuition about it. Right. So that's number one. The other part that I want to see going into the next step is when you're building the models right?

00:35:12:05 - 00:35:34:25
Unmesh Bhide

At the same time, you do need to look at what are the features of the models which are going to work and like going back to the back. Testing is very important. So for example, you know, we do this back testing on a daily basis. So I think that is a good feedback mechanism in terms of how our prices are doing compared to the traded prices.

00:35:35:09 - 00:35:56:01
Unmesh Bhide

That's one aspect. The other aspect is when you're building the features, right? I think one of the things that is more useful, right? I mean, yes, the rates that are rising now, I mean, you need to start to look back, you know, like the way we talk about it, like, you know, people talk about like building in the March meeting in the Fed will likely to be a 50 basis point.

00:35:56:01 - 00:36:18:11
Unmesh Bhide

And people are not discussing this for a long time. But I think if you go back to 1994, I mean that was like we had a fairly high of 50 basis point in sequence and you know, that kind of created a first mortgage crisis so if you think about that we, you know having a historical look back to build the models number one, the other aspect is the liquidity kind of thing.

00:36:18:11 - 00:36:39:16
Unmesh Bhide

So you have to start looking at a liquidity, how you capture liquidity because that was another factor of March three 2019 themselves liquidation of the portfolio So that's where you look at that aspect of it. And I think the last part I would say, you know when you're developing the models, I think you have to think about it.

00:36:39:27 - 00:36:56:28
Unmesh Bhide

You know, like we are using say a like artificial neural network with like is that the base model? You need to kind of like look at different models like you know, I think, you know, the other part in just E.i. since we're talking about like I think we are blind, you know, we are using that in our liquidity scores.

00:36:56:28 - 00:37:12:10
Unmesh Bhide

We are looking at in terms of corporate, right? I mean, in the V, when you're looking at E.i., you're going to like look at more of the two part problem. One is the classification problem. And the second one is your regulation problem and then you can go from there.

00:37:12:14 - 00:37:36:16
Tim Stevens

Yeah, I completely agree that the whole the whole discussion from when mesh in terms of how you go about doing this completely agree. And that's our experience as well. You need that data scientist who really understands the appropriate models to be used, how to tune those parameters and subject matter. Experts write that to help in defining the problem and helping to come up with the solution.

00:37:37:09 - 00:37:48:01
Tim Stevens

And then you need a heck of a lot of data analysts as well just to again prepare data and also look at results, validation backed test and pressure test all that.

00:37:48:17 - 00:37:50:13
John M. Murphy

So it sounds like job security for me.

00:37:51:18 - 00:38:09:07
Tim Stevens

There you go. Yeah. So maybe Abhishek, maybe turning to staying with in the topic of artificial intelligence, but thinking about it in terms of of credit analysis, right. Where are you seeing air use and how do you think it can help in the in the world of credit?

00:38:09:15 - 00:38:36:03
Abhishek Lodha

I think I think it's it's a wonderful use case, right? I mean, we have muni as ten times more credits than the corporate market. Right. And our analysts on average are following anywhere between three 50 to a thousand credits per analyst and that's based on the markets already as we've done. I want to clarify first do we and or rather reiterate on the way demand automation spoke about AI and machine learning credits.

00:38:36:08 - 00:39:01:25
Abhishek Lodha

It's not a replacer it's an enhancer to our analysts. Right. You know we we are our philosophy is, you know man machine symbiosis. Right because of the fragmented nature of the market, the analyst is always in the driver's seat. Right. So our use of technology, maybe data feeds RTI is, you know, I mean, I guess my my ultimate objective is to build unmanned suits for every credit on the street where they are still in the driver's seat.

00:39:02:07 - 00:39:23:14
Abhishek Lodha

But, you know, they have the power of technology and data to support their decision making as fast as possible and as accurate as possible. So on the credit analysis side, there are two big uses of and a lot of the big uses of A.I. that we're seeing. The more tangible one is on the content generation side. So we spoke about unstructured data.

00:39:23:24 - 00:39:43:18
Abhishek Lodha

It lies in PDFs and lies in the form of sentences. It lies in the form of large blobs of text or tables. Within PDFs, you have ADP, you have an LP which is being used to extract this information, and the feedback loop is again based on the analyst. So at assured, we've been doing muni insurance for 30 years now.

00:39:43:29 - 00:40:03:07
Abhishek Lodha

We have probably around 40 analysts, you know, tens of thousands of credits that we look at every year. We've got a lot of this data sitting which is used in terms of training to identify those key pieces of information. The other aspect that you see is sentiment analysis, right? So you read the DNA and this is something that's happening on the corporate side a lot.

00:40:03:07 - 00:40:24:02
Abhishek Lodha

You have applications but more obviously is popular among the muni world. They're doing a lot of work in the corporate world. You have alpha sense and to you they're all looking through documents to identify key piece of information to drive, you know, your credit opinion. So content generation is big. The second phase of that application would be analytics, right?

00:40:24:02 - 00:40:44:01
Abhishek Lodha

To be actually able to apply or machine learning to predict credit migration. Right. Default risks are obviously low, but credit migration is a real risk for a lot of us. On the asset management side, I think predicting that would be the next frontier. But I think first we need clean data for that. So that's where it needs to be used first.

00:40:44:17 - 00:41:06:02
Tim Stevens

Yeah, not great comment and I'll echo that again. Like in terms of air being used as an assistance to the human, I mean, it's just it's huge. That's how we, we drive our data quality, right? There's so there because of the thousands and thousands of issuers and million bonds, you've got to drive the people to where the problems are.

00:41:06:10 - 00:41:24:14
John M. Murphy

So Tim, let me ask you one question. How do you get this air technology into the hands of multiple users? It's going to have to build it, have a team of 40 data scientists and an interface, you know, business model that interfaces together a valid model that I can actually use and operate. Or is it a you know, is there something is it all bespoke?

00:41:24:14 - 00:41:26:29
John M. Murphy

Is it all off the shelf? And I can just plug and play.

00:41:27:12 - 00:41:48:14
Abhishek Lodha

I think I mean, I'll speak from my experience, but, you know, I've built data science groups where you know, data science was inherent as part of the team. So we had our credit team was basically and a lot of firms have heard recently larger phones, I'll call it invest in sciences, right? Where your your credit research team combines data analysts, data scientists and credit analysts.

00:41:48:14 - 00:42:07:17
Abhishek Lodha

They work together. A large institutions can afford that. Obviously, because they have data science claims which are shared across multiple different asset classes for smaller and medium ones. What we're seeing is AI as a service available now, just like how you have software as a service. Right. There are vendors out there which will, you know, almost like plug and play.

00:42:07:17 - 00:42:23:14
Abhishek Lodha

You have a database of information. They come in, they plug their models, they will be consultants for the first three to six months and help you almost like build your models and, you know, build your investment thesis, validate that. And over time, it becomes just a plug and play instrument.

00:42:23:15 - 00:42:24:13
Unmesh Bhide

Yeah, I mean, that's right.

00:42:24:13 - 00:42:24:17
Tim Stevens

Yeah.

00:42:24:23 - 00:42:58:03
Unmesh Bhide

Yeah. I mean, obviously at JP Morgan, we can have a lot of dedicated resources. But I think if you're you're thinking the a small piece of why it's I think, you know, you can always use a W. S kind of scenario where as you need more compute power to kind of do a backtesting or build your model Australian thing, you can use that approach also where as long as I mean the other thing I think, you know, I would go back to years also like a five ton is a non I would do not that technical like there is not syntax and all those things you don't need to worry about that aspect.

00:42:58:09 - 00:43:12:01
Unmesh Bhide

So that's kind of help. I mean what I really was mentioning mean we are hiring, you know new analysts are the thing is, you know, do you know Python is one of the basic I mean we do a Python test for that matter.

00:43:13:03 - 00:43:39:10
Tim Stevens

Yeah. And just just further that I mean our role as an institution is to find and solve problems in the market. Right? We have clients that range from very small municipal advisory shops, issuers to the largest banks bracket firms on the street. So our ability to solve common problems with this type of technology is part of the way that you bring up the answers to the market through.

00:43:39:11 - 00:43:53:09
Tim Stevens

I watch in the time I'll move on to blockchain, spend a little bit of time there. Some are just curious as to what you're seeing in the market in terms of interesting applications of blockchain, if you've seen any.

00:43:53:18 - 00:44:15:08
John M. Murphy

Well, well, when I'm having advertising for other people, there are some there are some folks on the direct lending side and placement side that are trying to, you know, I wouldn't say infiltrate, but, you know, get a foothold in the market and, you know, kind of avoid, you know, the simple contract between an issuer and an investor and all the things that are tied into the blockchain that make that so much more efficient to use.

00:44:15:08 - 00:44:30:14
John M. Murphy

But then with that, everybody else in the room is out of a job. Right? So it's a simple contract, all the legal terms of with it. And there it is. It's very difficult to actually get that technology out to the by side because if you look at every mutual fund, you have to have a custodian. They have to hold bonds somewhere.

00:44:30:14 - 00:44:55:22
John M. Murphy

They are just going to be holding it with, you know, X, Y and Z and not in a custodial relationship and how they figure out all those details of how it works. Down the chain. I'm sure we get there because I think the technology is fantastic, but it just using the technology and working it through the legacy reporting systems and things that are in place with insurance companies and mutual funds, maybe not so much with hedge funds or what have you, but or sneeze, but it's there.

00:44:56:03 - 00:45:20:09
John M. Murphy

And, you know, you wouldn't believe until something goes wrong in a if you do a pricing of a bond and if something goes wrong, ten, ten, ten levels down, it might be affect some reporting somewhere and if you think about that and how it integrates and how that's supposed to work, you know, someone's out of a job for that on the mutual fund side because you price the fund, the SEC is all over you and then you just you know, you're not in that space yet for this to be clean and to be implemented through that whole stack.

00:45:21:09 - 00:45:25:16
John M. Murphy

So yeah, I think it has tremendous, you know, a tremendous future.

00:45:25:22 - 00:45:25:27
Tim Stevens

Yeah.

00:45:25:28 - 00:45:45:14
John M. Murphy

It's just how do we integrate it into the into the marketplace in a in a non disruptive way? I mean, it's supposed to be disruptive technology and all that stuff. But yeah, tell it, you know, tell somebody is not getting paid or they're not getting paid because of something happened and went wrong and because, you know, the process wasn't right or converted correctly.

00:45:45:21 - 00:45:46:22
John M. Murphy

I think that's the challenge.

00:45:48:00 - 00:45:50:01
Abhishek Lodha

And I'm sorry.

00:45:50:01 - 00:45:51:01
Tim Stevens

You know, please, please.

00:45:52:00 - 00:46:10:15
Abhishek Lodha

I think, you know, with blockchain, obviously, you know, again, just like ESG it is a very hot topic in the market and there's a lot of pilot work happening in other asset classes. So let's let's kind of take a step back on what the advantages of blockchain are. We. You know, I've lost a lot of money in cryptocurrency, so I won't go on that.

00:46:10:15 - 00:46:36:11
Abhishek Lodha

But on the blockchain side, you know, there have been a lot of studies in Europe which basically say on the primary mission site, it can lead to cost savings or execution cost savings of 35 to 65%. That's huge. Right. Obviously the challenge you have so you know, settlements become faster. You can reduce the denomination of bonds. You know, I, you know, like to confess, I work for neighborly.

00:46:36:11 - 00:46:57:17
Abhishek Lodha

At one point a lot of people know about the mini bond thing that was there, that there's huge applications. But the challenges are fold is right. Number one is regulatory challenge rate. So the pilots that have been done in Europe have been focused around central banks doing this. Right. So you're almost like at the hub of the network, right?

00:46:57:17 - 00:47:25:10
Abhishek Lodha

So that's number one. We don't have that yet. Second is a technological challenge building blockchain networks is intensive. It takes a lot of processing power. I mean, we talk about big Bitcoin blockchains. Bitcoin blockchains can maybe execute three to five transactions a second rate. Those all blockchain networks are still at this point, right? You've got newer networks like Solano and Cardona, Cardona, I think, which our executing thousands of networks.

00:47:25:10 - 00:47:46:20
Abhishek Lodha

So that technological challenge needs to be overcome. The third is transition cost rate. How do you get large institutions, as you said, to move towards new technology? There's a cost to that. There's a cost in general, right? Any technology or data, people talk about data being cheap and that's fair, but there's ancillary cost around it as well. So is that artificial intelligence or so is the blockchain?

00:47:47:03 - 00:48:04:12
Abhishek Lodha

And the fourth is network effects, right? If there aren't enough people on the network and the value of that network or blockchain is close to zero. So there are a lot of challenges that are there with blockchain, but eventually it's definitely going to get there because the cost savings, both on the investors and the issuer side are going to be significant.

00:48:04:12 - 00:48:23:11
Abhishek Lodha

And at least my personal thesis is there may not be one universal public blockchain network, but they're going to be private blockchain implementations across the board. Right. So DDC, right, executes, you know, and goes through all the transactions, right? I mean, they're they're the central hub out there. They have something called, I think, digital securities management system, right.

00:48:23:11 - 00:48:42:25
Abhishek Lodha

Where they're piloting how to transact using their own blockchain networks. And if they can bring their transaction costs down or settlement down from two to three days to one day, that's huge. And that gets trickled in to investors and issuers. So we'll see a lot of this private blockchain network, which will help, you know, larger players bring down costs for all the other players in the market.

00:48:43:07 - 00:48:56:20
Tim Stevens

Excellent. We are way out of time. We're we're out of time along on long on topics. Any questions from anyone before we. Sure you get a microphone over here, please? I've got a couple.

00:48:59:28 - 00:49:35:11
John M. Murphy

Well, thank you. It's a very interesting set of discussions. I'm glad you mentioned A.I., because that's where I have my question of I'm in the business of manually extracting data from processes. That's I do the old fashioned way. And I'm curious, when you're using an application to do just what I'm doing, what happens when you have something which the model is completely different from most other sets of data and also how do you deal with really error that which of course never happens to anybody else in this room.

00:49:35:11 - 00:49:43:00
John M. Murphy

But does they find that the numbers of the bottom of the page are the wrong numbers? How does that get with an attack going to be handled?

00:49:44:07 - 00:49:44:26
Tim Stevens

Anybody want to.

00:49:44:27 - 00:50:11:00
Abhishek Lodha

Yeah, I can I can I can start. So we actually did an exercise on that as well. So implementation of the AI is, is a workflow problem. So, you know, it's not just a model. You know, you put it in you push the data, you know, you're inputting documents from one site and extract data from the other. You want to have a workflow of a feedback to operate in the way we do it as we like the first thing we do as we segment the market, we're not going to treat all issues the same.

00:50:11:00 - 00:50:32:19
Abhishek Lodha

So a classic example is yeah, and would be great for issuers that are following us. Yeah. Because now you've created a standard input, right? There's a clean you know, it's almost like garbage. You don't want to be in a situation where it's garbage in garbage out rates. So you want to start with the clean you know, on fragmented parts of the market and start building your model first and then kind of expand.

00:50:32:19 - 00:50:54:10
Abhishek Lodha

So you start out there, but as soon as the data is generated so as using so it's it's the extraction is not just biology, it's what we call an ADP, right? So it's intelligent document processing. It takes the document, it breaks it into digits. Right? So you take each and every word or no, break it into a data point or metadata, and then you start extracting that out.

00:50:54:17 - 00:51:20:24
Abhishek Lodha

And then the EAI, make sure it classifies it correctly, whether, you know, if it's going to cash, cash equivalents or cash and investments, that all goes into one category of cash, right. So there's a template on the other side where it categorizes and pulls the data, at least from a credit standpoint. So if something does not fall within those categories, there's a feedback you where it gets thrown out into a set of credits that are numbers that the analyst needs to look at.

00:51:21:28 - 00:51:38:24
Abhishek Lodha

A lot of there are a lot of data companies do the same thing and then once they do that, once you go in and look at these errors and fix them, now the software knows what are the errors that have been fixed, which I need to be aware of the next time I do this whole process. So that's essentially how you implement any sort of EIA instead of just doing a blanket.

00:51:38:27 - 00:51:39:10
Abhishek Lodha

I'm sorry.

00:51:39:10 - 00:52:03:26
Unmesh Bhide

Yeah. I mean, like I have to like the basic that this is like a natural language processing LP technique you want to look at specifically right where you have a temp list basically. So just those templates, you're basically training your machine to learn those templates I need that those you're if you're outside those templates, that becomes a new template like depending on how different it is and that process remain iterative.

00:52:03:26 - 00:52:27:05
Unmesh Bhide

I mean, like some other times, I think if you look at like a Tom Cruise boat as a good example, like and like you're just reading those terms through an LP and reading. I mean that I think the one important aspect is from a disclosure perspective as well as from a user perspective. If we have a standardized data, that's what I think the biggest jump will be using technology.

00:52:27:05 - 00:52:54:20
Tim Stevens

Yeah. And the only addition, I completely agree with what's being said. Also, it's the human validation element right there. You have to have human analysts that are validating some of that output, right from the models and understanding statistically if the model continues to improve and it's you're not you're never done training your models, you're constantly retraining your models and taking some piece of it and validate it with people.

00:52:54:24 - 00:53:13:28
Abhishek Lodha

And the ultimate objective, I think, is not to get to 100% of document gathering, but rather have the analyst focus on that. So high yielding stuff, right? Like, I mean, we had a short you know, we keep joking around, but like isn't it better for our analyst to focus on all the high yield names instead of spending their time on the plain vanilla standard stuff?

00:53:13:28 - 00:53:36:13
Abhishek Lodha

Sometimes, obviously, we do our due diligence across the board and very, very rigorous, but how can we use technology so that we can shorten the time it takes for them? To do the same quality of analysis on the plain vanilla stuff so that now it's resource utilization rate, we're all to do more with less. That's it, right? How do we focus our analysts towards the right credit at the right time is really what the objective should be.

00:53:36:14 - 00:53:39:15
Tim Stevens

That's it. Another question. Yeah.

00:53:42:09 - 00:54:06:00
Abhishek Lodha

I'm Jane. I'm from Carrboro and Abhishek, actually. Really good to see you. You were actually on a panel on a similar topic not too long ago. So my question is about taking a step back and thinking about the data being even produced or tracked or maintaining a first place. And I have no doubt that larger to medium size issuers will over time be better and better at this.

00:54:06:00 - 00:54:35:05
Abhishek Lodha

And they can work quite well with the kind of analytics and technology that all four of you have worked on. But does that mean that the smaller issues, which relates to Abhishek, you mentioned there like there's a magnitude more issuers in the muni market and say the corporate market, are they going to be left behind? Like what, if anything, can analytics and technology do to make sure the entire muni market has data that then can be analyzed?

00:54:35:07 - 00:54:59:10
Abhishek Lodha

Yeah, it's it's a complex question. I'd start with I'm going to segway a little bit towards Govtech as a market. Govtech is a $400 billion market at this point, right? There are a lot of players out there. Like we like a lot of times UPS. And again, my personal prejudice may be, but like a lot of times when I hear conversations in, in the muni market, it's all focused towards issuers should do this issuers should do that.

00:54:59:21 - 00:55:17:10
Abhishek Lodha

But there are not a lot of applications being built where issuers can do that at a fraction of a cost rate. Like no one wants to be bad at disclosure, right? Or it's just that they don't have the right tools to do it in a lot of cases. So I think there's a huge opportunity, again, in the Govtech space where tools can be provided.

00:55:17:10 - 00:55:41:15
Abhishek Lodha

I'm sure you guys have a lot of like tools like where Kiva are open. Gov. Gov. Government, they are trying to build better financial management tools for governments and one of the things I keep telling a lot of these guys when I meet is that needs to trickle as a data feed to investors, right? Eventually, because the more efficient you are as an issuer the better it is for you and the better it is for the investor because they get clarity on the data.

00:55:41:15 - 00:56:03:22
Abhishek Lodha

So I think there's I would pivot towards a bigger opportunity out there where, you know, Govtech can solve a lot of these problems may be financial management, debt management. You know, Muni is trying to do that same, right? Like they're trying to create a universal platform so that the swiveling from an issuer standpoint is is gone. Similar tools can be built for other things as well.

00:56:03:22 - 00:56:19:25
Abhishek Lodha

ESG is another great example, right? We we talk about disclosures from issuers, but most issuers, I can guarantee you don't know what's happening at a grassroots level within their own communities. So if they start getting better tools to do that, I'm sure disclosures eventually will improve and our data costs will improve.

00:56:21:27 - 00:56:29:18
Tim Stevens

Great. Well, thank you. Thank you to my panelists. So appreciate your participation. And thank you all for your kind attention. Thank you. Thank you.