What’s powering the latest innovations in thematics? S&P DJI’s Head of Thematic Indices, Ari Rajendra, and Invesco’s Head of EMEA ETF Equity Product Management, Chris Mellor, discuss the rise of thematic investing and how AI and NLP technologies are sharpening the tools tracking transformative trends.
[TRANSCRIPT]
Jenny Ellice:
Thematic strategies continue to gain traction, with recent analysis reporting that global assets and thematic funds have doubled to 269 billion U.S. dollars in the five years to the end of June 2024. But what's driving this growth and how are deeper data sets and technological advancements helping facilitate the next generation of thematic indices? I'm Jenny Ellice, and I'm joined today by Ari Rajendra of S&P Dow Jones Indices and Chris Mellor of Invesco to discuss how new AI capabilities, including S&P DJI's Kensho natural language processing technology, are sharpening the thematic toolkit to help investors seeking to access long-term trends systematically. Well gentlemen, it's great to have you with us today.
Chris Mellor:
Lovely to be here.
Jenny Ellice:
Ari, I want to start with you, and why are thematics so interesting right now and how do they relate to indices and their development?
Ari Rajendra:
Sure, Jenny. Thematic indices are at an exciting juncture. We've seen a surge in ETFs, and part of that is fueled by the advancement in technology and data, which has made the creation of thematic indices more transparent and efficient. One of the most interesting developments is the use of AI tools, particularly in natural language processing, or NLP, to analyze unstructured data, to identify companies associated with a specific theme. Now this capability also has the advantage of identifying companies that may not yet be generating revenues but pivotal to emerging themes. It is also invaluable to identifying themes that are new and emerging, that may not otherwise be identifiable by conventional taxonomies. At S&P Dow Jones Indices, our global Kensho indexing platform exemplifies this innovation. It leverages in-house data and enhanced NLP models to parse through global filings to identify companies that are aligned to its compelling themes of tomorrow. Now the combination of data, technology and expertise has the potential to unlock opportunities in terms of allowing investors to access new themes that are shaping tomorrow.
Jenny Ellice:
And Chris, following on from what Ari just said, what are you seeing from market participants in thematic investing?
Chris Mellor:
Yes certainly thematic ETFs is one of the fastest-growing areas of the ETF market in general, as you indicated before, globally. It's also true in Europe. ETF assets in thematic ETFs currently stand at around about USD 60 billion. That's grown fifteen-fold over the last decade, three times faster than the pretty heady growth in the European equity ETF space as well. If you look at what a thematic is doing and why investors are interested in them, ultimately they're looking for longer-term growth potential that comes from structural themes. That may be coming from demographics, it may be geopolitical. It may indeed and very often is coming from technological development. And the obvious example today is the impact of AI and its potential to transform the world around us. It's worth remembering that while we're looking for long-term structural growth potential, there will be cyclicality in both performance and demand from thematic products. The most obvious example is if we think back to 2020 and the COVID crisis, the impact that that had on slowing growth expectations, and the rapid response from central banks cutting interest rates, resulted in significant inflows into the thematic growth space. Around about 20% of all inflows into equity ETFs in Europe went into thematics during that 2020 period. Obviously we then had a reversal as growth started to recover, interest rates started to go back up, and 2022 and 2023, we saw much slower demand for thematics. Where we are today, I guess we're rolling over on the cycle again, we're back into the world of cutting interest rates and perhaps worries about longer-term, or shorter-term growth, and therefore, the long-term growth potential of thematics is making them more interesting once again.
Jenny Ellice:
Well, Ari mentioned Kensho, so what stood out to you about the Kensho model and data set?
Chris Mellor:
So Invesco is one of the world’s biggest ETF asset managers, and we obviously have the option to work with, and indeed do work with, many index providers across all asset classes and across a wide array of strategies. The reason we chose to work with S&P DJI and Kensho when we were looking to launch three new thematic products, focusing on very current themes, so AI, cybersecurity and defense innovation, was the advantages that Ari has talked about, from both the deep data sets that they have, as well as the leading NLP processing approach that helps to capture those opportunities in thematic areas. If you think about how an ETF works in capturing exposure to a theme, the key question we're asking is how best to identify companies that have meaningful exposure to that theme. And in real terms, there's only three ways you can go about that, broadly speaking. The first is good old-fashioned earnings and revenue, sort of similar to a sector-type index. That tends to be more useful for a more well-developed theme. The second way is to work with industry experts in a particular area of the market, which is excellent for sort of particular niche themes. But the other, third way is obviously NLP and using harnessing technology to help identify exposures. And in particular, the NLP approach here is capturing exposure across a wider universe in an efficient way, but also allowing exposure to perhaps those sort of areas of the market a little earlier than the previous two examples I gave.
Jenny Ellice:
So finally, Ari, let’s delve a little bit deeper into NLP. How does it work exactly, and how does it relate to your indices?
Ari Rajendra:
Certainly. So, as I mentioned earlier, NLP is a form of AI that allows us to go through unstructured texts. There's essentially a vast amount of information that is not necessarily neatly categorized. And what the global Kensho platform does, it essentially goes through, it focuses on company filings and other official documents to gain deep insights on the products and services that the company is involved in. Now this used to be a very manual process prior to the advent of NLP. Now this is automated, and it enhances the efficiency associated with this process. And now how does it work, in terms of how does it work, the NLP models are trained using industry models, which are essentially structured libraries of terms and phrases that help define a theme. The real advantage is the technology enables us to identify themes and companies contributing to them, and this unlocks potential that traditional methods may overlook.
Jenny Ellice:
Well, it's certainly interesting to see these exciting developments come to life. Ari, Chris, really great to have you in the studio today.
Ari Rajendra:
Thank you, Jenny.
Jenny Ellice:
Well to learn more about S&P DJI's indices and the topics discussed today, visit us at the link below.
spglobal.com/spdji/thematics
