Is Decarbonization Gaining Momentum? Exploring ICE Climate Data Insights - MATLAB
Video length is 28:42

Is Decarbonization Gaining Momentum? Exploring ICE Climate Data Insights

Monika Sabolova, ICE
Girish Narula, ICE
Marshall Alphonso, MathWorks

This session examines whether global decarbonization efforts are progressing by analyzing the latest trends using ICE's climate transition finance data. Monika Sabolova and Girish Narula from ICE, together with Marshall Alphonso from MathWorks, will present key findings on carbon emissions, intensities, and net-zero alignment across corporate portfolios. Learn how this data is effectively processed and visualized using MATLAB®, providing actionable insights for professionals focused on advancing sustainability initiatives.

Published: 22 Oct 2024

All right. Thanks, everyone, for joining us. We're going to be talking about, "Is Decarbonization Gaining Momentum?" It's all about data and modeling, and so I'm pretty excited to join Monika on this presentation here.

So we're going to kick things off here with, quickly, I thought this might be an interesting way to frame things. I'm going to start with the view of the MathWorks. MathWorks has been doing a lot of really interesting work in terms of disclosures and actions.

And one of the things that we've learned since 2009, if you see here, some of the actions we took were we set up a lot of solar panels to power a lot of the infrastructure that's driving a lot of our cloud development and AI work. And so what we learned is what we need is very high-quality data about emissions and targets, as well as a flexible sort of modeling platform that gives us the ability to quickly adapt to the changing nature of climate change. We're very excited to join Monika and the ICE team today, because we're here to talk about the data piece and how that's coupled together with the modeling piece.

So we're excited to share with you that now, available in R2024a, we actually have a connector to the ICE data that I was able to leverage, and I'm going to be showing you a demo a little bit later on that uses the data to tilt portfolios. So with that said, I'm going to hand it off to Monika to start with the introductions around the data piece and go from there.

Fantastic. Thank you so much, Marshall. And I hope everyone can hear me well, as well?

Yes, I can hear you, Monika.

Yes, we can.

Thank you so much. Thanks, Marshall, for the introduction. And really, it's a pleasure to be part of this webinar today and talk to you about some of the key features when it comes to ICE climate transition finance data, which is now integrated in the MathWorks system. So as part of the ICE global financial data offering, we also provide extensive climate data. And I thought it may be useful to really provide a brief overview of some of those data and metrics that clients can use for a variety of different use cases, such as climate risk analysis or scenario analysis.

So the key pillars of the climate transition finance data offering is our emissions and targets data. In addition, MathWorks can actually utilize also forward-looking metrics and data, such as temperature scores or forward-looking emissions projections as well. So in terms of the emissions data, we offer GHG protocol-defined emissions data that are either reported or inferred.

We provide emissions data for all of the scopes. So scope 1, scope 2, as well as all the 15 categories of scope 3 emissions. We collect all of the data manually, and we actually analyze the data globally for over 9,000 companies. And once the data is analyzed and really verified to ensure that it's of highest quality, we then use that data set to model data to 30,000 companies globally.

And so that basically represents our standard data set. So this data is then mapped to over 1.4 million securities. And really, this robustness in terms of the mapping is one of our strengths. So we have the capability to actually also work with clients when it comes to increasing the coverage. So it doesn't really just stop here. It's not just off the shelf. We often work very closely with our clients, whether that's some bespoke project or simply increasing the emissions inference.

So when it comes to the decarbonization targets piece-- sorry, Marshall. If you can go back-- just to finish on that part of the decarbonization targets data, we provide a comprehensive information about companies targets, along with information such as whether the target has been approved by SBTI or not. So really, the data is robust. Again, we cover over 30,000 targets that span across 5,500 companies. And very briefly, in terms of the use cases-- so again, the use cases are very, very different, but mostly, clients are using the data for transition risk analysis, net zero analysis, climate stress testing, scenario analysis, among others.

So if we can move to the next slide. Thank you. So just to elaborate on that coverage point, I thought it would be useful to provide a further breakdown when it comes to the coverage, in terms of looking at different emission scopes, perhaps regions, and also data quality. So as you can see on the slide, our coverage is really global. So we aim to cover the major investable indices. So we can see that in Europe, US, and Asia-Pacific, we broadly have a similar coverage. So those regions, they kind of fall into a similar bucket.

We can also see that currently, the coverage in Africa, Latin America, and Italy, the Middle East, is the lowest. So that comes to the fact that the coverage is lower in terms of companies are not reporting as much or simply are not part of the major indices. But also, what's encouraging is that most of the companies-- I would say more than half of the companies-- are reporting very good quality data, as you can see on the chart. What we define as good-quality data is really the GHG protocol definition. So if a company reports emissions data and the emissions cover at least 95% of global company operations, that classifies as complete, good-quality data. But what's also interesting is that 26% of the highest quality data, as you can see on the slide, is also third-party assured.

So to continue on that encouragement point, what's also interesting is that there's an improving trend in terms of emissions reporting when comparing the current data set to the previous year. So this is, again, driven by, first of all, continuing to increase the coverage of the data. But also, this is due to the fact that companies are simply reporting more in terms of their emissions data, and also targets data. In particular, we can see that there is an increase in the complete, high-quality, reported scope 1 and 2 data, as you can see on the slide.

So to move on to the next point and talking about the data quality point. So there are a number of data features that are really important to our clients, and it's been already alluded by Marshall. So clients are really interested in the coverage, of course, and the quality of the data, but also transparency of the data. And we really excel at all of those areas. So we offer global cross-asset class coverage as mentioned. We can also expand the coverage using our inference model.

And I think it's maybe worth using in a particular example here, just to bring that to life. So for example, we are very proud that we collaborated and worked very closely with European Central Bank on their economy-wide climate stress testing. And we actually provided ECB with emissions data for over 4 million companies, that the data has been used in the published report that is available online. And again, we're very proud to be part of that project. So this is just an example, in terms of filling the gaps and inferring emissions data for companies that don't report, for example.

When it comes to the quality of the data-- so again, I mentioned it very briefly, but our analysts are actually trained to manually collect the data. The data goes through very robust validation process. Again, the quality assurance process is very detailed. And we basically want to ensure that the data is of highest quality. So in some cases, we even engage with companies directly-- For example, if we feel that they reported perhaps an extra zero emissions figure, or perhaps if the data has been challenged.

And now, on the transparency point-- this is very important to us, and to our clients-- to understand whether the data has been inferred or reported, the completeness of the disclosed emissions data, but also information, such as, were the emissions data approved by third parties? So this is all very relevant. And as briefly mentioned, we can also provide forward-looking emissions projections data and temperature scores data. So on the point of forward-looking emissions data, those basically take into consideration current emissions, but also historical emissions and the targets of companies, and then we take into consideration sectoral trends and regional trends of the selected scenario. We currently support NGFS, IPCC and IEA scenarios.

So the emissions performance of the portfolio of a company can be then compared against a climate scenario, which can be then translated also into a single metric, which some of you may be familiar-- sorry, Marshall. Yeah. Thank you. Some of you may be familiar with the implied temperature rise metric. So just to briefly mention, there is a usefulness for this metric, in case clients are interested, to really simply compare a company to another company or a portfolio to another portfolio using a single metric, such as implied temperature rise. Thanks, Marshall. If we can move on to the use cases slide. Thank you.

So again, I know that Marshall will spend much more time talking about the actual application of the data in the system, but I do want to spend a little bit more time talking about the use cases. Again, I work very closely with our clients. I'm the product engagement manager. That means that I basically engage in terms of how the clients are using the data and also our platform to understand how I can support them collecting any feedback and things like that. And so the primary use cases really span across different types, and the kind of application of the data is kind of represented through those use cases.

So as I already mentioned, the ECB example-- we work with a number of other central banks, but also commercial banks. So for example, commercial banks, they may have loan books for which they want to really have the highest quality and fully reported data, but they also want to fill in gaps when it comes to companies, such as private companies or small and medium enterprises, that simply don't report emissions. And so here is where ICE steps in to support those type of use cases.

In addition, we also work, of course, with asset managers and asset owners, different types of investors. They, again, use our data for, often, climate risk analysis, et cetera. And then in regards to the pension funds-- so again, we work with a lot of pension funds. And in particular, we work with them on supporting them with their TCFD regulatory reporting. So again, this is another angle of a use case, I would say.

And last but not least, we don't only engage with financial institutions. We also work with universities or international financial organizations or policy organizations, and they often use the data for research purposes. So with this, I hope I provided a good overview of the data. I think it's really important to really highlight the coverage, of course, the transparency, the quality piece of that.

And now I'll pass it to Marshall to let him to demonstrate, really, how the data has been integrated in the MathWorks system, and again, how it can be used by the clients. So thanks, Marshall.

Thank you. Thank you so much, Monika. Just to highlight Monika's points, there's two big components to it-- there's the data quality, transparency coverage, and we'll use that as part of the demo. But as we start thinking about future steps, there's a lot of really interesting things to consider when thinking about the multi-period problem, thinking about forward-looking projections, and how we think about utilizing this data in practice. So I can share some of those insights at the end. All right.

So we're going to start with a quick example here. So this has been shaped over many years of experience, working in the asset management space. There are many clients out there that are now starting to say, from a capital markets perspective, how do we build towards momentum? Like I mentioned in the beginning, MathWorks is starting to do a lot more corporate disclosures around climate change, and then we're starting to leverage a lot of the data to try and figure out where the momentums are, and where we should start enhancing the energy supply to the MathWorks.

In addition to that, when we start thinking about our portfolios-- like client portfolios at financial institutions-- there's a big question that's coming up with retail with serving these high net worth individuals, our retail clients. How do we tilt a portfolio towards, maybe, if the portfolio thematically wants more green energy in their portfolio, how do you tilt towards that?

And what does it really mean to be green? Now, thanks to the data from ICE, I was able to add a lot more color to the question of, what does it mean to talk about emissions, what it means to talk about targets. And with the frameworks from SBTI and others, we're able to really think about the modeling in a lot more detail.

There have been a number of challenges as we start thinking about this thing. I'll just break it up into three buckets here. When we deal with data, it's going to be a very critical component, as a feeder, in here. But I'm going to jump right into the modeling piece. And the last piece I'll just quickly touch on is, how do you present this and allow for that interactive discussion with clients? And that's a very critical component for a lot of our clients that we work with, and I'll show you an example of that.

So before I jump into the actual demo itself, I'll just frame it a little bit. When we start thinking about, simplistically, where do we put our money, a client might be asking that question. They have to consider tax benefits. They have to think about a variety of constraints around the problem. They have to think about, which asset class I'm using. But these days, one of the most important components, especially in this environment in which we're dealing with climate change, is, how do we precision-tune the portfolio for that specific client?

So clients out in the Middle East, they have certain kinds of expectations around what they think about climate change. When we start thinking about Europe, they have certain kinds of regulatory constraints around what it means to look at precision tuning. Here in the United States, in North America, where we're dealing with asset managers that are basically thinking about a variety of constraints. And as you think about the constraint space, to really tilt the portfolio, it starts looking more and more mathematically complicated to deal with. And so this example is about portfolio optimization, leveraging the data that ICE has provided us, and then hopefully tilting towards the portfolio benchmark that we're targeting.

Now, I wanted to capture some of the data that I was able to leverage. And also, on the left-hand side, you can see a little bit of the formulation here. This is not the only formulation. Like I mentioned, the modeling piece is very adaptive, depending on the client, depending on the institution, depending on the sector you're operating in, depending on the country you're operating in as well. And so as we start thinking about climate constraints, I'm being a little bit loose on the generic concept of what it means to talk about a climate score.

But what I've done for this example here is I've actually been able to take in the ICE data and come up with a way to aggregate all of this into a climate score that we can tilt towards or away from. Now, this is not recommending a certain way of aggregation. I'll talk more about SBTI. I'll at least mention some of the SBTI frameworks of how to merge corporate emissions data with some of the slides, but what I'm going to be focusing on here is more around the modeling piece.

All right. So let's dive into the application itself. And the first thing you'll notice is the modeling platform looks, maybe, different than maybe what you're used to. You'll notice that everything is a web page. The modeling platform I have here, I'm actually doing a lot of my development online, which is actually kind of cool.

I started my development on the desktop, and this application that you're going to see here, I built there, and then I now am completely working in the cloud. And the nice thing about that is that gives me the ability to quickly run high-performance compute as a first step. So I've broken out the steps that we're going to go through in this modeling exercise to mirror what traditional asset managers go through, the first step being, all right, let's pick an investable universe.

An investable universe takes a lot of thinking. There is the constraints around what my emission scores are. What's the Scope 3 impact? What's the reported emission score? What's the PCAS score on the data for the corporates? As well as traditional more financial metrics, such as what are my expected returns, what's my expected volatilities, as well as the covariates and how everything sort works together to be able to even take the next step on setting up the benchmark portfolio.

But before we get into building the temperature-optimized portfolio, a very important next step once you've defined your investable universe-- and you notice, every time I click it, it's actually choosing, which are highlighted in blue, a random set of securities, because I'm not going through the exercise of thinking through which ones I should pick for the actual universe itself-- but I just want to show you that as we're sort of stress-testing this thing, I'm allowing to pick a random set from the securities below. And I think there's around 600 and something securities of high-quality data that I was able to source from ICE, with the ICEans.

The next step here is the second big step. Great. We've chosen an investable universe. It looks like I'm picking this one, this one, this one, this one, and it keeps going down the list right here. The next big step is saying, what is my benchmark portfolio and what does that look like? So I'm going to click Set Up my Benchmark Portfolio. And for now, I'm just generating an efficient frontier.

There is some really nice analytical infrastructure out of the box with the portfolio object that basically allows me to generate a frontier. And I'm just going to choose a Sharpe ratio portfolio, as sort of my benchmark portfolio, which you can see here, which is the weights SR. And that becomes our benchmark and our target portfolio. In the ideal case, you'd want to choose securities that may be a little bit more representative for the client that you're targeting. But for now I'm just going to use a benchmark portfolio.

Now, mathematically, on the top left, just to formulate the problem, I am doing a tracking error optimization, for those who want to get into the weeds of it, along with the constraint around saying my ESG of that optimal portfolio needs to exceed some sort of-- sorry, I apologize. Not ESG, but in my case, since we're not dealing with the S and the G, we're dealing only with the environmental piece, we're saying that the climate score, however you define and couple that together, exceeds some sort of climate threshold. So that's my basic constraint, along with the idea that we're a long-only portfolio and the sum has to be equal to 1. So no leverage in this portfolio.

So if I click Optimize, now it's actually running live. And we're not just running it for one portfolio. We're running it for a variety of potential thresholds on the climate score. So using the ICE data, I was able to create these emission scores, impact scores-- normalized, in this case-- but on the right here, you can kind get a sense of the raw data that's been aggregated to form my total temperature score.

And let's now take a look at, for this benchmark portfolio, what does the performance look like? And the first thing I see here is, as I decide to take on more ESG climate scoring-- and keep in mind, this is just a random portfolio. This is not meant to be a communication around the direction of the industry. But what I can see here is my expected return is sort of flat, and then after I start taking on a certain excess amount of climate score-- which, remember, is it's a pretty sophisticated aggregation of a lot of information, like I showed you on the slide-- you can see a drop in our performance of this optimized portfolio.

Typically, what we found-- obviously, these are just a few data points right here-- what we find is sometimes, this drop in performance is due to maybe some of the overconcentration, because you're forcing higher climate scores. You're basically saying, look, I the only available things in my investable universe is forcing me into certain concentrations, which is basically causing my performance to drop, because the user's emphasizing that I want more climate score.

And the last thing I want to point out right here is the tracking error and the deviation from the portfolio itself is quite significant, allowing for, basically, an evaluation similar to what I do with the performance, but then saying, look, I'm taking on some extra ESG score, but then it allows myself to think about the deviations from that. Now, when it comes to concentration, you can see here the concentrations when we think about temperature.

Now, given the time constraints, I want to be sensitive to that. So I will switch back over and highlight a few points right here. When we start thinking about the climate score trade-offs, this is just one way of framing the problem, in the sense that we're talking about a tracking error type optimization. But our modeling frameworks are quite flexible and allowing us to model a variety of other kinds of constraints that are not captured in that simple formulation.

And a lot of our clients are coming to us with more and more sophisticated-- there's examples in which we're talking about hundreds of lines of constraints in production that had a time frame of convergence under five seconds for convergence. So there's a lot of really powerful optimization capabilities. But part of the battle is getting the right high-quality data, as well as thinking hard about how you formulate the problem. That's where we can really help, where we have some very, very strong mathematicians, numerical mathematicians, on our team, and we can really help with some of the modeling piece.

There's a piece that I want to allude to. We've implemented-- and I'm happy to point you to a webinar that goes into a lot more detail-- implemented some of the traditional scoring methods on the temperature scores around the portfolio itself-- WATS, as an example, is the SBTI 1-- as well as thinking hard about how do we-- this is not from SBTI-- how do we think about merging the target data with corporate GHG emissions data? And when we get such high-quality coverage that ICE is providing us in terms of for all the ICEans that we're able to provide, we were able to then leverage that in terms of creating a larger picture of not just a single temperature score, but also get an analysis that looks like what you're seeing below-- disaggregating it by country, by region, disaggregating it by sector.

So the modeling platform that we have here gave me the flexibility to quickly adapt. And I want to mention one thing right here-- I was able to take this modeling and very, very easily deploy this. Oops. I can do that here. MS. Oops.

And so what I did was I basically took this, I ran this little application, this little wrapper I wrote for myself, and it's taking the application that I built, and it's now deploying it live onto the web. It's a fairly quick deployment process.

But once you have that deployed-- I'll give it a second here. There you go. So we've just deployed it. And here's the application itself.

So now this gives me a live application that I've just updated, based on that, with the same exact capabilities as I saw before. So this is now becomes this interactive dashboard that allows me to interact with my clients fairly quickly. So I get to focus on the modeling piece, which I love to do, and then share my results.

I'm going to hand it off to Monika, but I wanted to-- yeah, I'll hand it off to Monika here, for a second. But I'm going to pull up a slide, Monika, that might be useful, and I'm happy to talk about it to help us transition to any of the questions that might come up. I took one of the blogs that was put out by ICE, and I loved some of the key insights that Ian had shared around mirroring some of the ideas.

There's a good trend going on. There is an increase in terms of reporting. At least in the North American companies that I deal with, we're starting to see more and more-- at least I can speak for The MathWorks, where we're spending a lot of time thinking about how we calculate Scope 1, Scope 2, as well as global Scope 3-type emission communications, as well as we're seeing a nice trend right here with the declining absolute emissions, and then for the reporting companies as well. So it was a good positive trend for implied temperature rise falling.

So with that, I'll hand it off to Monika for maybe if you want to have any other insights you want to add. Otherwise, we can transition into questions.

Sure. Thank you so much, Marshall. This is very insightful. I appreciate we have maybe a minute and a half left for our time. But I'm glad that you also pointed out today this fantastic thought leadership, which was produced by my colleague, and indeed is exactly what you mentioned. So we've done an analysis.

And it is interesting and encouraging to see that not only we are seeing increase in terms of reporting-- so more companies are reporting their emissions data and their targets data-- but yes, when we conducted an analysis on putting together a portfolio, a diversified portfolio, where companies reported complete scope 1 and scope 2 data, at least one category of the 15 categories of scope 3, what we observed is that, really, there is a trend of reduction in emissions. But yeah, the message is very clear. We see increase in reporting and we see some trends in terms of also that reduction in emissions. So yeah. I appreciate we have maybe one minute left, so I'll pass this now to Akshay.