Design choices for a Paris-aligned benchmark

Paul: [00:00:01] Welcome to Index Ideas from FTSE Russell. I'm Paul Amery, your podcast host. In this podcast we explore index ideas that can help you address real world investment themes and challenges. We look into how indices are built and why. As a reminder, you can't invest in an index, and the concepts that we discuss in this podcast are not investment advice. Any reference to potential investment strategies is therefore intended for informational and educational purposes only. In this episode of Index Ideas, I'm joined by Andreas Schroeder, who is Head of Index Design EMEA at FTSE Russell. Andreas, welcome to Index Ideas.

Andreas: [00:00:38] Hi Paul, thank you very much for having me here.

Paul: [00:00:40] So Andreas, you recently co-authored a paper at FTSE Russell on building an index to meet the requirements of an EU Paris-aligned benchmark. That paper is available on the FTSE Russell website and you can find it in the show notes. So how do EU climate benchmarks work?

Andreas: [00:00:57] Yeah. So to be labelled an EU, or to be allowed to be labelled as an EU Paris-aligned benchmark, the regulator has set out a very specific set of requirements. Namely, we need to achieve a reduction with respect to emissions compared to the underlying universe, namely of 50%. So we need to halve the emissions that the underlying universe is producing. We have got also a requirement that the index itself has to decarbonise 7% year-on-year from that initial level of 50%. So the worse of the two, the more severe of the two—decarbonisation—needs to be achieved. And in addition, the regulator has set out the requirement to have some way of forward lookingness in those metrics. This is very vague, right. And so we needed to translate that into a more tangible set of metrics. So we used something [from our partnership] with the Grantham Institute, namely the Transition Pathway Initiative, which provides two data sets. One is the management quality, which reflects how company disclosures are being done: are they sufficient on emissions or how the company management is managing the transition? That's set number one. Set number two is the Transition Pathway Initiative carbon performance score. This is how the company is aligned towards the 2050 net zero objective. Meaning do they already decarbonise sufficiently or if they are deferring the burden of decarbonisation further and further, the closer that we come to 2050? And the third signal that we've got is the green revenues, a proprietary data set that FTSE Russell has put together, which reflects what percentage of the company's revenue comes from green sources.

Paul: [00:02:54] Right. So you've got quite a complex set of requirements. You've got some regulatory goals. You know, when you start with a blank sheet of paper and design an index that combines all these requirements, what are the possible approaches that you can take?

Andreas: [00:03:07] Yeah. Before we go into that, it's very important to point out that when we solve for those requirements, right, you will have a very big-time exposure to industries, countries, and you will also hit the capacity limits. So in order for us to solve for that in a sensible way, we do need to impose some constraints. So you have got a very strong set of objectives, but also a strong set of constraints. And those two bits together, they basically dictate which methodologies you can use. And you're completely right. So in your question asking which methodologies we can use, generally we’ve got three types of methodologies. One is filtering. That's very simple, very transparent, where you go and you exclude the worst stocks based on whatever characteristic, or you leave only the best stocks in. So that's number one. Then there is tilting. That's the one which FTSE Russell historically prefers. This is where you are not as exclusive and inclusive, but you gradually change the weights of individual stocks to preference the favourable stocks and downweight the unfavourable ones. And then the optimisation is the third one, which is very interesting in this case, that you basically throw the objectives and the constraints together and rely on an optimiser to do all the hard lifting.

Paul: [00:04:32] So what are the pros and cons of these approaches when building the sustainable index? How would you summarise the pros and cons of those approaches?

Andreas: [00:04:42] Great question. So when we're looking at the three methodologies, they by themselves are very clearly distinct, going from very explainable and less flexible—that's the filtering—to very complex and opaque, but very flexible—that's the optimisation—and tilting is somewhere in between. And when we basically move to the approaches specifically that we can choose for solving for the Paris-aligned benchmarks, we cannot use filtering, basically because it is not flexible enough. We need to rule it out. But what we can do is we can go and use target exposure, which is a form of tilting. And that is what I'm talking [about] in the paper, one of the approaches. The other two are optimisation based: one is tracking error optimisation and the [other is] turnover optimisation.

Paul: [00:05:33] Right. So just so I understand this, you're saying that the simplest approach, the filtering approach doesn't work for this type of index. So you are left with either tilting or optimisation.

Andreas: [00:05:43] That is correct.

Paul: [00:05:44] And when you're optimising you can either optimise to minimise the tracking error or you can optimise to minimise the turnover.

Andreas: [00:05:52] Correct. And there are other flavours as well, but those are the ones which we are looking at in the paper.

Paul: [00:05:57] Obviously people can go and read the paper, but what did you eventually decide upon and why?

Andreas: [00:06:03] Yeah. So we decided to go with the target exposure methodology for a variety of reasons. The clients that we've worked with or we consulted before constructing the index, they preferred a couple of things that need optimisation, neither of the two optimisation techniques we're providing. So when we looked at how you can differentiate the outcomes, so basically the indices, usually you go with something that traditionally is easy to measure: which is tracking error, turnover, concentration and capacity. So tracking error and concentration clearly is the preferred metric by the optimisation. Turnover is the best for the turnover optimisation. But capacity is the best for target exposure. Now if you leave it at this, you can argue one way or another which way to go. For our clients, robustness and transparency were the key metrics, so we wanted to select a methodology that would highlight the robustness and the transparency.

Paul: [00:07:12] So Andreas, you mentioned robustness and explainability. What do you mean by those two terms and how do you measure them?

Andreas: [00:07:19] Yeah. So robustness deals with the fact that if you give the index a little bit different set of data, because all data comes with a little bit of noise, and particularly, sustainability data is not crisply managed. How the index methodology translates this noise in the inputs into the index itself, into the weights. So a methodology which is amplifying the noise is less robust. And the methodology which is not amplifying the noise or reducing the noise is more robust. So we can show that target exposure in this case is much less sensitive or more robust compared to optimisation. And similarly on explainability, what we've published recently—and again, there is a podcast and a paper available for both of those—so, for the explainability, we published something what we call weight explainer, which allows for the target exposure methodology very clearly to identify how the index weight was produced, starting with the benchmark and doing the individual tilts to individual signals. So this is definitely something you can watch out the space in the publication, because it is a great way to visualise the impact of signals on the final index weight of the stock.

Paul: [00:08:35] I guess ideally an index should be very robust and very explainable, but there must be some trade-off then between those two?

Andreas: [00:08:42] Not between those two, but so the trade-off happens between tracking error and robustness and transparency, or complexity and robustness and transparency. As we talked about before, optimisation is more flexible. It can solve much better for tracking error. But it will be too complex to be robust and explainable. Right. So this is the trade-off.

Paul: [00:09:06] In general is it true to say that there are more choices to make when building this type of index—when it's a sustainable investment index—or are these decisions true for any kind of index, whether it's a smart beta equity or other types of non-standard index?

Andreas: [00:09:25] Absolutely correct. So it is pretty much the same conclusions across the board. The problem that we are facing with the Paris-aligned benchmarks is that we are packing so many different requirements into the index that the differences become very clear. So the more objectives and the more constraints the index methodology needs to solve for, the bigger the advantage of the tracking error is for the optimised tracking error solution. The lower the turnover will be for the optimised turnover solution and the better the capacity and robustness and transparency will be for the target exposure solution.

Paul: [00:10:07] Right. Well thank you. That's given us a very helpful overview of how you go about building this type of index. I know you work with a number of colleagues in index research and design. Can you give us a feel for what other research areas you and they are currently looking at and building indices based on?

Andreas: [00:10:25] We research also different allocation methodologies. So be it filtering or target exposure or optimisation. We recently have created our own risk model, which is state of the art, which is at the heart of optimisation. So for tracking error optimisation, we use our risk models. We also work a lot on innovative ways of reporting, and so maybe we can look into the weight explainer that we mentioned earlier.

Paul: [00:10:55] Yes, I know your colleague Hannah Layman is coming on a podcast in the New Year to talk about the weight explainer. So that's an interesting area to go into a bit more detail on.

Paul: [00:11:05] And finally, Andreas, where can listeners to the podcast go to get more information about your team's work?

Andreas: [00:11:12] Yes. So we periodically publish blogs and papers and podcasts like this on the website. I hope we can blend in the link in this podcast. Or you can find it maybe in the messages.

Paul: [00:11:25] Andreas, thank you very much for joining me. And that's it for this episode. If you've enjoyed listening to Index Ideas, then please do follow us and give us a rating or review on your podcast app of choice. And if you'd like to get in touch with the show, you can do that via the email fmt@lseg.com. But for now, from me, Paul Amery, goodbye.