Look, learn, predict

Financial services firms should ensure their approach to climate change risk assessment takes the complexity of those risks into account, says Neil Cantle
 

Climate change is now firmly on the risk agenda for financial services firms. This has, in part, been driven by the significant increase in regulatory focus and expectations in this area. Specifically, going forward, firms must now evaluate the anticipated financial impact of climate change, and are expected to include considerations of climate change within risk management frameworks.

A typical risk framework is anchored around the risk appetite associated with a firm’s key objectives. Having described the amount of risk that can be tolerated around these goals and the associated preferences for different risk types, the rest of the framework is designed to optimise the firm’s chances of delivering those goals. In order to integrate climate change into that framework, the first question is, ‘How might climate change impact my key goals?’

There is an immediate challenge here, in that the dominant narrative around encouraging firms to be better world citizens has so far centred on carbon emissions. At one level, this helps focus attention on one of the key drivers of climate change and has, at least, made the scale of the challenge more visible, giving investors a useful way to judge and then reward or punish firms for the extent to which they are ‘doing the right thing’. For many firms, however, reducing their carbon footprint will not fit easily with their existing strategic goals. There is a risk that trying to force it will turn meeting the goal into a pure compliance exercise. The risks that most financial services firms face from climate change are more complex than this metric alone.

Early efforts at assessment in the insurance world focused mainly on physical risk – the more tangible impacts to property, land, natural resources and so on. If your business involves insuring or financing assets that could be physically impacted, then you can hopefully make some sensible assessment of how their value and/or usability might be different in the future as a result of climate change. An often-discussed consequence of climate change is the impact of rising sea levels on coastal property, but extreme weather can also significantly impact inland properties through, for example, flooding – so the assessment is more involved than it might initially seem.

Complex connections

The Bank of England’s recent consultation in relation to the 2021 biennial exploratory scenario (find Milliman’s summary of the requirements at bit.ly/2QNdEYG) highlights the second order effects of this for banks. Homeowners who cannot secure insurance represent a higher risk for the banks that are providing them with mortgages. So, assessing the risk is made more difficult due to the highly complex way in which everything is interconnected, leading to second, third and higher order effects that can be challenging to identify and evaluate. An approach that recognises these interactions is required, so that important consequences are not missed.

As well as considering the direct effects of climate change itself, firms must take into account the many different influencing actions that could happen between now and the future date when the ultimate impact is felt. Which changes are already inevitable, and which might be averted if we act soon enough? There is sufficient international attention on climate change that ‘do nothing’ seems unlikely, but even if we do continue to take action, we don’t yet know precisely what world those actions will create, or what journey we will go through on the way to that end state. It currently seems plausible that there could be pain before gain, so your risk assessment during the coming years will not progress in a nice linear fashion.

Matters driven by climate change have also become somewhat conflated with social and governance matters. Well-intentioned activity under each of these banners does not necessarily lead to a desirable outcome for each of the others. In the context of a highly complex adaptive system such as the planet, the opportunity for unexpected and unintended consequences is very large. Given the need for firms to assess and disclose the financial impacts of climate change, they need to think carefully about where the effects of these other, interrelated factors should be accounted for.

A structured but agile approach

Physical risk assessment is therefore less easy than it might appear. But what about transition risk? This is considerably more difficult than physical risk – it is a period during which the perceived can be made real. Unlike physical risk, changes during transition tend to be driven more by how people feel about future physical risk rather than the impact of the physical risk itself. Asset valuations, for example, are made by judging the future demand for the products and services a company provides, and the ability of the firm to provide them in an economically viable manner. Those judgments can change at very short notice based on new information that markets had not previously priced in.

The main challenge with quantifying the financial impacts of climate change comes down to the fact that the number of possible outcomes is very large, and the ways in which those impacts might arise is varied and evolving over time. There is no nice historical dataset that can be leveraged to assess risks like this – their history is simply not relevant enough to the future. The go-to method is therefore scenario analysis, but this rather puts the onus back on the expert to come up with a situation that feels relevant and meaningful.

When confronted with something new and novel, such as climate change, it can be challenging to know whether the scenario you have come up with is at all representative of what could happen and to judge the extent to which it is a likely or unlikely situation. At the same time, the exercise is about more than just getting a number – the main benefit of scenario analysis is that you learn about how the risk could behave, how you might observe signals that it could be about to crystallise and what actions you might take pre or post-event to improve your outcome.

A good approach to tackling these kinds of complex and evolving risks is to adopt a structured but more agile approach – a fixed process that can be embedded and followed repeatedly, but which will generate new insights and new outputs as the risk evolves.

“By creating a rich picture of past and present dynamics you are forming a kind of ‘map’ that can be used to recognise emerging trends”

Building useful scenarios

The first stage requires the gathering of ‘what you know’ about the risk’s development so far, and any theories about how it might develop in the future. If you are assessing an asset in a particular sector, what economic data do you have? What information can you obtain about demand for the company’s products and services and how they might be impacted by climate change? How are their consumers’ attitudes changing as a result of climate change? What would it take for buying behaviours to change? Are there geographical or cultural differences that might be relevant to the perception of this company’s services?

This gives a more ‘timeless’ view of the mechanisms that have played out so far and the other dynamics that might be seen at future dates. By creating a rich picture of all these past and present dynamics you are forming a kind of ‘map’ that can be used to recognise emerging trends and their likely future pathways, helping you to create more relevant and meaningful scenarios. Importantly, by drawing a path from current dynamics to anticipated future ones, you are also able to bring future outcomes back to effects that can be seen ‘now’ and which drive immediate actions better than purely considering events that feel a very long way away.

The second stage of the process is to see if the data can help you notice things that you hadn’t spotted before in the evolution of the risk. In typical scenario analysis, once the ‘situation’ is described, the data part of the exercise is primarily about choosing the right parameters to evaluate the scenario. However, it is possible to collect and analyse data to challenge whether you have the right scenario in the first place. Look for non-linear, potentially time-lagged relationships and the degree to which the data is confirming that what you observe ‘makes sense’. 

This will enable you to judge the extent to which there is evidence to support the potential for a sudden and dramatic change – a tipping point. Without this step it is highly likely that you will underestimate the likelihood of an extreme situation occurring.

In the face of this evidence from observation, you can form a much better educated view about what might happen and ensure your scenario is realistic – both in terms of the elements it contains and the extremity of what happens next. Causal modelling techniques provide a structured opportunity for modellers to bring a scenario to life, and permit them to gain a deeper understanding of how the future might play out than they will get from the consideration of ‘a number’. Given that the future development of a risk such as the financial impact of climate change remains so fluid, it is important to be able to explore which factors would be consistent with particular outcomes – these then give you the insights needed to align your risk limit frameworks to be consistent with your risk appetite. As you learn more about the risk in the future, the indicators tracked can be adapted and evolved accordingly.

“When confronted with something new, it can be challenging to know whether the scenario you have come up with is at all representative”

Multiple interacting factors

Intimidating though it is, the modelling of a complex risk such as the financial impact of climate change is possible. The approach, however, is different to the types of long-range scenario analysis favoured by strategy consultants of the past. In situations with high complexity and adaptation, the scenario approach needs to involve multiple interacting factors that are allowed to do just that – interact. The outcomes ‘emerge’ from the interaction and cannot be specified neatly upfront. The creation of a meaningful scenario and a set of assumptions for detailed modelling requires firms to first understand and describe what could emerge from the forces anticipated to be at play. Laying out a ‘look, learn, predict’ process enables scenarios to be created and analysed in a structured, repeatable way, but without oversimplifying what is being modelled. Ironically, using an approach rooted in complexity creates results that make more sense!
 

Neil Cantle is principal at Milliman

Image credit | Shutterstock