Climate change scenarios used by investors in their financial projections “were not created with investors in mind”, and projections based upon them “could lull people into a false sense of security”, according to Professor Riccardo Rebonato, scientific director at the EDHEC-Risk Climate Impact Institute.
The Institute’s research looking at Dynamic Integrated Climate-Economy (DICE) models concluded they have been inappropriately used for scenario analysis. As a result Professor Rebonato and his team are developing a solution that attaches approximate probabilities to climate scenarios in an effort to reflect the uncertainty of making investment decisions based on climate change.
ESG Clarity caught up with Professor Rebonato to find out more.
Could you introduce me to the research you’ve conducted and the work that the EDHEC-Risk Climate Impact Institute are undertaking?
The Institute’s goal is twofold. One is to look at the expected impact of climate change on asset prices, and the other is to look at how finance in general can help support the green transition. In my role as scientific director, I take an active role in following the first one of those two goals.
One of the things we have observed is that the scenarios used by investors, although of very high quality, are not created with investors in mind, but with policymakers in mind. In particular, they completely avoid any probabilistic assessment for different scenarios, which makes it very difficult for investors to use them because how are they meant to assess whether any given scenario is something worth taking seriously? So, creating scenarios with probabilities attached is our big research project.
When people hear ‘scenario analysis’ and ‘stress testing’, many people tell me that they’ve been doing these types of things for decades without probabilities attached. That’s true, and their expert knowledge is fine when it comes to market scenarios and predicting market events such as interest rate rises. But, when it comes to climate change, we’re in completely uncharted territory. We don’t have this intuitive understanding of what we should worry about or not, and that is why attaching a rough probability becomes so important.
Can you explain some of the drawbacks of current scenario analysis methods?
First and foremost, as I said, is the absence of probability. There are the Shared Socioeconomic Pathways (SSP) narratives and Representative Concentration Pathways (RCP), with the present framework having you join them together to work out emissions and concentration scenarios to form climate policies. This is quite opaque, and the users of these scenarios don’t always realise what is implied by some of these combinations, and sometimes the cost of those policies can be much higher than what is currently spent on education, defence or perhaps even healthcare.
Given the current fiscal and debt situation of most countries, it is very unlikely that this combination of scenarios will come true, and therefore, the fact that they do not have a probability attached to them is very misleading, because investors are left with the five SSP narratives and the six RCPs for a combined total of 30 possibilities, and it’s just not realistic to assume that they are all equally as likely.
Secondly, the narratives are very detailed. For example, one of the narratives is specifically on resurgent nationalism, and there are others on social, economical and political resilience. But there are only five of them, and so cannot possibly cover all eventualities. So, what we’re trying to do is to make less detailed, specific narratives more focused on economic variables like GDP per person, energy intensity, emissions intensity, etc. that are consistent with what we are observing in the real world. That’s not a criticism of what has been done with the SSP/RCP system, it was a breakthrough at the time. But every breakthrough can be improved upon.
I imagine some asset managers might feel uncomfortable making decisions based on variable data, is that a potential issue?
The data variance is definitely not an issue, it’s an integral part of the information we want to convey because it gives us an idea about the uncertainty that we have.
Let me give you a concrete example of how knowing uncertainty can change your decision. If you’re faced with investment A and investment B, with respective returns of 3.5% and 4%, it’s a no-brainer, you’d choose 4%. If, however, investment B was centred around 4%, but it could return between 1% and 7%, some people would decide not to take the risk and go for the guaranteed 3.5% from investment A instead.
Understanding that uncertainty is a crucial part of the decision-making process. After all, it’s not for nothing that modern finance is known as an exercise in mean variance.
Could you tell me more about the modularity of the solution you’re working on, and whether it would be freely available and easy to replicate for other analytics firms?
We’re an academic institution, so yes, our goal is to make everything open source and accessible.
With all these types of models, what we’re actually looking at is a series of modules which can be slotted in and out. For example, you could have one that provides economic output, one that links economic output to emissions, etc. Each one of those modules can be taken out and substituted. I think of it as a rotation, or an evolution, more than a wholesale change, because you can substitute aspects of the model. It’s not a matter of a revolution, just a matter of making things more useful to investors, with our ultimate clients being among strategic investors, sovereign wealth funds, pension funds and others.
What impact are you hoping this could have on risk modelling, company valuations, etc.?
To arrive all the way down to the company level is a long path. The first port of call is asking what impact this will have on the economy as a whole. If you think about it, share price is the discounted sum of all the dividends, with ‘dividends’ being what the economy has produced, divided up between the providers of capital. So, the next step is to see at the aggregate level, if I were to hold the market portfolio – let’s say MSCI World – how is this valuation impacted by climate change? Going down to this sectoral level is the second step.
But, for me, I’m a top-down kind of guy, and it’s that first step – asking about the impact on economic output – that is the most important.
