Read the video transcript for the MSc Global Finance webinar with Prof. Nick Motson, as he delivers a taster lecture exploring how enhancements made to the Environmental, Social and Governance (ESG) profile of a European fixed income portfolio impact its performance.
Watch the masterclass
- And our class today is some research that we've just very recently finished. Some of you may have attended a masterclass I did last year. And this one's along the same sort of lines. The one last year was about smart beta. And it was based on some research that was commissioned by Aon Hewitt, and subsequently with the Invesco.
And this piece of research was commissioned by Insight investments. And it's a very topical area. We're going to talk about ESG and fixed income. So kicking off, we've got to say what was our research question. And one of the key things or one of the reasons I stayed on at Bay's was that we've got-- I can keep this one foot in the real world and one foot in academia.
And we have people coming to us all the time asking us to do research for them. And most of the time we turn them down. And the reason we generally turn them down is they come to us with a question such as, can you prove that? And for us that's not really the way we want to do things. We're not for sale in terms of proving something that's going to be beneficial to somebody else.
But what we do really like is pieces of the research like this, and like the previous smart beta one. Which is, can you look at this question? And our question was, if you apply ESG screens or any ESG criteria to a fixed income portfolio, what effect does that have?
And very much our key thing was that we can apply the academic rigor to this. I work together with Andrew Clare and Anil Keswani, who both also teach on the global finance degree. Anil teaches asset management and Andrew teaches investment strategy. And we apply academic rigor, but we've all worked in the real world as well.
So we want to step outside the ivory tower and think about this, not just from the point of view of the academic approach. But we can use the tools that we've got and the skills that we've got to add some weight to any arguments that we put forward.
And for the outputs from this, we've got a working paper. We'll make sure that the slides are available, as a PDF afterwards. The link is there. And the working paper that we've put out a couple of weeks ago is there on SSRM for you to read if you're further interested in this topic.
So diving straight into what we did, and the first important thing is data. And those of you who've you've looked at academic research in the past, you'll realize that there is absolutely tons of academic research on equities. But there's very little academic research on much smaller proportion of academic research on fixed income.
And the reason for that is pretty straightforward. A, we don't have access to the data all of the time. And B, the data is far more complicated. And we were kindly provided with data by IHS Markit. And what Markit gave us was all of the component files, are the weights, the returns, and all the descriptive info for that iBoxx EUR corporate index going back to the early 2000s.
And what this gave us was about 7,000 bonds, and approximately 2000 different issuers. And the reason it's more complicated is if we do this on equities, then, for example, if we look at Volkswagen, there is one equity. But when I look in 2022, there are, I think, 48 different Volkswagen bonds within this index.
And if I was to pick BNP Paribas in our bank, there is one equity. But, again, there's 35 to 37 different bonds when you look over the last year. So it becomes a big messy animal dealing with this bond data. And we were very grateful to be provided with this in a reasonable format by market.
And then what we needed to do was combine this fixed income bond data with some ESG data. And we've got access to Refinitiv, so we've got a load of terminals here at the business school. And Refinitiv provide ESG scores for lots of companies.
And that ESG score, there is an overall composite score. I've given you the picture of how they break it down on the screen. But they give you an overall composite score, but they also give scores for ES and G individually, and also for controversies. I mentioned Volkswagen earlier on, Dieselgate was a controversy, so their ESG score would have gone down.
And we've got the three key pillars, which is ES and G. And within those pillars we can break that down even further. So, for example, if we're going to talk about G, which is the governance one, there is a score for community or human rights. And things like that.
So what we've done is we've analyzed all of the different issuers. We have to combine the ESG scores, which we get from Refinitiv, with the issuers from the bond data, and make sure that we can match all those up. And that was an ugly job for me to do with some Python using the Refinitiv API, and matching everything. But subsequently we end up having a picture of the scores of the issuers of these bonds.
Now, ESG is relatively-- I won't say it's completely new, but it's a relatively new phenomena in when we talk about the length of time that we've got data for financial markets for. And on this slide I can show you how many of the bonds in the index the market provided us with, how many of those have actually got ESG data. Because not all companies report it, or haven't reported it in the past. It's become much more common over time.
And what we decided we were relatively unscientific, when we look towards the end of 2021, we've got 80% plus of the bonds we've got data for. But we go back 10 years, and it's approximately 50%.
So we decided to only look at 10 years worth. If we go back to the financial crisis period 2007, 2008, you've got less than 50% of the bonds in the index have actually got any ESG data. So we can't measure anything with those.
So all of the results that I show you over the coming slides will be based off that 10 year sample period. Just to say what we then have to do is create a new benchmark. So we can't compare some filters that we're going to do with the market benchmark, the one provided by IHS Markit. We have to create our own benchmark, which is those bonds that have ESG data.
Because what we're going to do is we're going to exclude some bonds, we're going to only hold the highest rated ones, we are tilt towards the highest rated ones and away from the lowest rated ones, et cetera.
So we created a new benchmark. On this slide I'm just showing you that our new benchmark looks very similar. It's slightly different, it performs slightly better than the broad index. Maybe because those companies that don't give ESG data are the ones that have been weaker companies in the past. But everything will be compared. We're doing an apples to apples comparison here.
And if we quickly look at what the ESG scores have done over time, we can see there hasn't been a massive variation over the years. We can see that if anything, the governance and social scores have gone up a little bit, and the environmental score has gone down a little bit. But the ESG average for the index is not changed substantially over the 10 years that we're looking at.
Which is important because what we want to do is make sure that-- There are a couple of other people who've looked at this in the past, and we don't want to say-- If we're different from them, is it because our data is different or is it because we're looking at things in a slightly different way, which is what I'm going to show you in a minute. Because we are looking at things, maybe, slightly differently to how people have in the past.
So that's the data out of the way. What did we actually find? And how does ESG criteria affect your returns? If you're an investor. And we went into this with a very open mind. We kind of thought that a good result would be that tilting towards companies with higher ESG scores are excluding companies with low ESG scores, or are we doing things that we didn't like, or people maybe don't like.
If that doesn't affect your return, then that's good. You can do something that's socially positive without a cost. We weren't expecting to get higher returns, necessarily, we weren't actually hunting for a higher return. But we really wanted to assess, did it impact your returns? So the first thing we did was actually at the request of inside, because this is what many of their clients will do.
They'll come along and say, look, we want to exclude the bad stuff. We don't want to invest in tobacco companies, we don't want to invest in mining companies, we don't want to invest in defense companies, and we don't invest in oil and gas producers. Because all of those we consider them to be bad sectors, things that we'd rather not invest in.
And what we find is if you exclude those sectors, the effect is absolutely nothing, practically zero. And the return is identical, and the volatility is two basis points difference. And statistically we can't say that the two things are different at all. So holding everything or excluding these sectors really doesn't make any difference.
But the reason it doesn't make any difference is mainly because those sectors make up an extremely small part of the index. So our benchmark is over here on the right hand side, which is obviously 100% of the bonds.
But when we exclude those sectors, we're still holding nearly 94% of the bonds. So actually because those sectors are so small, then they don't really have that much of an impact if you exclude one.
So that was step one. Now if you decide that you don't want to hold tobacco and oil companies, it wouldn't have hurt you historically. We don't know whether it's going to hurt you in the future, but historically over this 10 year period it wouldn't have done.
So then we started to get a little bit more scientific and a little bit more academic about it, and pretty much any academic paper you read, where they're looking how does some factor affect the returns on some asset class. What they do is they split it into buckets, whether that be deciles, quartiles, quintiles, and look at the top one versus the bottom one. And so that's what we did.
And looking at the number of bonds that we had, we decided to split it into five groups. So we split it into quintiles. And we looked at the ESG combined score. So what's the total score across all of the factors for these companies, and what would it look like if we bought the quintile with the highest score, and sold the quintile with the lowest score. And we'd be rebalancing annually.
And we get a result which actually would have outperformed. You'd have outperformed to the tune of around 3% over the sample period, which is not insignificant. But one thing did surprise us, and that was that that outperformance really occurred before 2016.
If I had any prior ideas when we started this research, I was thinking, well, ESG has become a hot topic. Every time I open the FT, there's a story about ESG. That must mean that investors are diving into this, which means that probably there's more demand for bonds of companies that have got good ESG scores, and less demand for those they've got low ESG scores.
And those flows of people, buy one thing and not buying the other, should have driven returns higher over the period. But that's not what we see. It all happened five, six, seven, years ago was where the out-performance was. And over the last seven years or so there's really not been much of a difference between the two.
Now, if I was purely an academic, I could probably stop here. I've proven that high ESG scores outperform low ESG scores. But the devil is in the detail, and you need to get a little bit deeper into what's behind those results.
So on this next slide we did dig quite a lot deeper. And here we get a little bit of a surprise. And we can say that maybe the whole story isn't told by this academic way of top decile versus bottom decile. Because if you look across the table on the right hand side, two things become very clear.
The first one is that both the top decile and the bottom decile outperformed the benchmark. They're holding all of the bonds, holding the worst 20% with-- the 20% with the worst ESG scores. Their return was actually slightly higher than the benchmark. Admittedly, it's got slightly higher volatility, but, yeah, it did have slightly higher return.
The poor performing ones are kind of in the middle. And, in fact, if you look the second quintile, that quintile with the next 20% best scores, that did the worst. So just top versus bottom doesn't tell you anything. And, in fact, this is certainly not a linear relationship here.
But if we look where this out-performance comes from of that top decile, we can see that it is statistically significant. We've tested it, and we can reject the null hypothesis of no difference with 10% certainty, because it comes out as a 7% as a P value. And where the difference comes from is actually in the tail.
So we've got VaR and conditional VaR here. So what we're talking about is VaR is the left hand tail, it's the bottom 5% of returns, where is that cut off. And the C bar, the conditional bar is what's the average in that table.
And what we see is that the conditional bar, so those really the bad returns. That's what you've got less of, that's where the out-performance comes from. That tail is smaller for those companies with higher ESG scores. And that kind of makes sense, doesn't it? ESG is a risk.
The company gets caught doing something wrong, they're polluting or whatever else, they become more risky, and that's when their bonds do poorly. Or if they find out that they're manipulating their results of the emissions from their cars, for example, then that's a risk. And their ESG scores go down and their bonds perform poorly.
So that was the next piece-- I was saying that it's not linear. And you shouldn't just judge a book by its cover. The next thing we dug into is that there are going to be sector effects here, and it's impossible to avoid them.
And I'll maybe trying to explain that a little bit more. Because if we exclude bonds based on their ESG score, or we pick certain bonds based on their ESG score, the sector make up of that is going to be different from the sector makeup of the benchmark.
And that will affect the returns anyway, because some sectors do better than others at different points in time. And ESG scores vary within sectors and across sectors, as returns vary within sectors and across sectors.
And there's two ways to look at this. And this is where, again, I'm going to maybe disagree with some of the academic approach. Because many of the academic papers that we read when we were doing the desk research for this, what they would do is they'd say, look, we're going to normalize scores by sector. We're going to take the average score for a sector, and those that are above it get positive and those that are below there are going to get a negative.
And we're going to then filter on those ESG scores. The alternative way, and which is the way that we've done it, is say we're going to take ESG scores as given, we're not going to adjust them in any way. But we're going to look-- when we look at the returns afterwards, we're going to say how much of that return difference is driven by sector tilts, and how much of it is driven by ESG scores.
And we can calculate the sector tilts by just saying, we'll calculate sector indices, and we'll apply the weights that we've got using those sector indices and see how different that would be from the benchmark. So we went with option two, whereas much of the academic research goes with option one.
And why do we do that? Well, if you do it option one way, you could pick an arms manufacturer who has a higher ESG score than other arms manufacturers, rather than picking a wind farm that has a lower ESG score than other wind farms. And that's a big question.
Now, the question is why are you looking at this in the first place? You very much are looking at it because you want to do the right thing. And whereas if you're just measuring it within sectors, you may not be doing the right thing. You might pick the tobacco company that's got the highest ESG score, rather than picking the renewable energy company.
So we very much went with approach two. And how does that affect our results? Well, it actually strengthens them somewhat. So on this slide, you can see this is the top quintile with the top ESG scores, and this is the out-performance that we saw a couple of slides ago.
And this total out-performance, so the shaded area, the total out-performance, is driven partially by sector effects in blue. So the sector effect is positive. And part of it's driven by the ESG score effect, which is in green.
If you look out the right hand side, you'll see the sector effect is also positive for that bottom quintile. But the ESG effect is very negative. So as he strengthens our result, there is definitely people are being rewarded for being good in ESG terms, on the left hand side, and penalized for being poor, on the right hand side.
And the sector effect is just kind of bundled in there. But when you break out, you can just see what is the ESG effect. So overall this is good news. We're seeing companies who have better ESG credentials that have been rewarded, and those that haven't have been penalized. But thus far we've only looked at the overall ESG score.
And we also really haven't thought about what's an investor going to do about it, because investors aren't going to go along the top ones and short the bottom ones. Very few of them are. They're probably going to decide to just exclude some proportion.
And when we look at exclusions, the news doesn't get quite as good, because you really would have to exclude all but the top 20%. So you'd have to exclude 80% of the bonds in order to get a significant return effect from this. If you exclude 20%, 40%, or 60%, statistically there is no difference in return.
Now as I said at the beginning, that's not necessarily a bad thing. We're saying that it doesn't cost you, but you're certainly not seeing any statistically positive benefit from it either. And for a large investor, it's kind of difficult to just hold 20% of the benchmark. There are capacity issues with that, there's only 70 bonds out there, for example. So it does become a little bit more difficult.
So thus far we've looked at just the ESG combined score. I'm quickly going to blast through the E, the S and the G. And I'll spend a little bit less time because the pictures are the same. And as I said, you can have a copy of the presentation afterwards.
If we look at E, the environment, here the picture is very clear. The gray line is the benchmark, the green line is the top quintile, the red line is the bottom quintile. So for E, the effect is the strongest.
And maybe that's not a surprise, it's relatively easy to measure, and it's something that's a hot topic. I say hot topic, from London that was 40 degrees a couple of weeks ago. Everybody is focusing on environmental issues, and it's maybe not so much of a surprise that it's been priced.
And we can see that-- [COUGHING] excuse me, it's fairly linear, and it's definitely the top quartile outperforming the bottom quartile. And what we see is here we've got statistically-- the top quartile is statistically significantly different from the benchmark, significantly better. And the second, the bottom quintile is significantly worse.
So in terms of E, the results are fairly strong. Some of that effect is driven by sectors. Though it's not quite as strong in terms of a lot of the effect comes from the sector effect, as opposed to the ESG effect. And that weakens the result slightly. But certainly for E, it does look like it's priced. And mainly it was prior to 2018, which is what we saw for the whole ESG thing.
In terms of exclusions, once again, you have to exclude 80% before you actually make any significant difference. So you do have to hold a relatively small proportion of the bonds. So that's E. And effectively-- I see we are about 20 minutes in, and that kind of brings me to the end of the really good news. Because now when we start looking at S and G and some other stuff, the results certainly become less clear.
So turning to social and the S pillar. Here is less clear, both the top and the bottom quintiles underperform the benchmark, which is somewhat surprising. And, in fact, the second worst score, in terms of ESG, is the best performing. So the fourth quintile is the best performing quintile, which is kind of disappointing, really.
But both the top and the bottom underperform, but none of the differences really is statistically significant, apart from that second lowest. The rest we can't say there is a significant difference between the benchmark and the quintile.
When we break out the sector effects, though, we'll see that actually the sector effects here are hiding the ESG effect. Because the sector effect is very negative for the top quintile, which means the ESG effect is actually positive. And for the bottom quintile, it's practically zero, which means we've got a negative ESG effect.
So, maybe, you could say, if you were feeling generous, that there is an effect there. The last S-- and I'll come back to S at the end, because I do have some other things to say about it. When it comes to G, so governance, you would think that this one would be-- it's been around for a long time.
Now before we were even talking about ESG, people were talking about good corporate governance of companies. And you would expect this to show up relatively strongly, but, in fact, it doesn't.
Once again, both the top and the bottom quintiles underperformed the benchmark, which is somewhat of a surprise. And, in fact, what we see, the big surprise, is the second lowest is, once again, the best performer, and is statistically significantly different, is the only one that is.
It looks like G is not particularly well priced. And when we look at the sector effects, it doesn't really adjust our outcome either. It looks like both the top and the bottom quintile are poor. So, really, it looks like G is not rewarded or penalized good governance or poor governance, which is somewhat of a surprise to us.
I'm just looking, there's a question come up, and I will answer that question at the end. Now another question is, I said at the beginning, should investors-- can you really exclude 20%-- can you exclude 80% of the bonds in existence, and only hold 20% of them. It seems unlikely.
So we wanted to just test, and it was almost going back to our smart beta duet days. Could you tilt towards those with higher ESG scores and away from those with lower ESG scores? And to do that, we used a fairly simple methodology, which has been used in the equity world. And to some extent I think there are some fixed income indices that use this as well.
We standardize the scores by calculating a Z-score. We chop off the tails, and plus or minus three of those Z-scores. And then we map those into weights. And there are two ways of mapping into weights, MSCI use one way, and FTSE use a slightly different way. We tested both, the difference is minimal.
What do we find if you were to tilt or if you were to have tilted over the last 10 years was that maybe not a surprise. You were rewarded for the E piece. Though, the environmental score, you would have done well if you tilted towards E and away from poor scores.
And for overall, for the ESG combined score, not so much. You would have got a slightly higher return, but it's not significantly different. And not surprisingly, it doesn't work particularly well for S and G, considering our earlier results.
I'm just going to give you some pictures. You have only benefited from E or ESG, which is what I just showed you on the last page. Now, just to wrap up before I start taking the questions. I said that we like to step outside the ivory tower. To get an academic paper published what you need is a good result. And you can throw away all of the bad results.
There's been lots of criticism of financial research, for P hacking, and data mining, et cetera. And just so that we're not going to be accused of any of that, I want to show you some of the disappointing results.
And I'm going to pick out two of them. And I said I'd come back to the S, the social piece, which is kind of difficult to measure, to some extent. And I'm going to show you what, to me, was the most disappointing results of the whole thing. And that is, when you break down, I said that they were the three pillars, E, S, and G. And within them there are 10 different category query scores.
And within those category query scores, you can pick human rights, for example. Probably personally I think it's quite important, well, very important. Probably one of the most important things that companies should be doing.
And here we get the most disappointing result ever, which is that the companies with the highest human rights score, if you just hold the 20% of companies that have got the highest human rights score, then that would have cost you 50 basis points a year. So one of the biggest differences we get is statistically significant.
And the companies that do the best are those with a lower human rights scores. Now we should ask why that is, and it could be two possible reasons. One, it could be that it's not measured correctly. And I can see there's a question come up, and I'm going to try and answer that at the end.
It could be that we're not measuring this stuff correctly, or should I say Refinitiv are measuring correctly. Or it could be that the market is not pricing it. People are willing to invest in companies that have got poor human rights scores.
If that's the case, then I think that's pretty disappointing. And it's something that will probably change. And just because historically it hasn't made a difference, and in the future it may do. But in the same way as environmental didn't make a difference 20 years ago, and people now do put it into their criteria.
The human rights was a disappointing outcome. And simply community was a disappointing outcome. Because the community measures not just how nice they are to their local community, but how they are to the world as a whole as well. I think the definition is something about being a good citizen, and is measured across about 20 different criteria.
And here we find that the companies that are the worst, in terms of citizenship or community, actually perform the best. So that fifth quintile is by far the best performer, which, again, is somewhat disappointing. Whether it will be like that going forwards is something that people need to consider.
So just to run through our conclusions from the research, before I come to the questions that have come up on the way through. Our conclusions are, we're not wildly different from previous studies. I've given you a couple of examples of papers that have been published in the past. Now, generally, what studies have found is that higher ESG scores have historically given somewhat higher returns. Not massively, but slightly.
What we've added to it is quite a little bit more depth and a bit more practicality to it. Because if you just compare the top and the bottom quintiles, then you really don't get to see the whole picture. It could be that both the top and the bottom are actually worse, just happens that the top is better than the bottom.
I think it's really important that you disentangle the sector effects. And that comes to the question that's just come up on the chat. It's kind of very important that you think about it in the correct way, because you are doing this for a reason. And just because, as I say, a tobacco company that has a very high ESG score relative to other tobacco companies, doesn't necessarily mean that you want to invest in that company.
For an investor, so if you're a trustee of a pension fund and you're trying to invest ethically, the effect of excluding sectors, excluding companies with low scores, seems to be fairly marginal. To make a statistically significant difference you have to exclude up to 80%.
So it wouldn't have hurt you, it wouldn't have helped you either. But you can invest and sleep at night knowing you're doing the right thing. And finally, we looked at the tilts. And the tilts would have helped, especially the environmental. But, again, that was the only one that was statistically different.
View the Global Finance course page