BLOG
Professor Jean-Francois Mercure, Director of Exeter Climate Policy
Dr. Jesse Abrams, Senior Impact Research Fellow
Recent events and experience made us start to think quite critically about the ability of for-profit organisations to stay on course with their mission when working on climate related risk issues in financial services. In this post we want to explore issues arising with the governance of climate-related risk and opportunity assessments: who generates evidence for whom under which mandate, and with which (conflicts of) interests? As a spoiler, we consider it increasingly unlikely that for-profit data providers can resist the pull from providing tools for critical appraisal towards selling risk-washing as a service. Not because they’re dishonest, but because their business model depends on it as long as they operate as privately owned, licensed black boxes that resist external scrutiny.
When a data provider provides for profit data to asset managers, especially when the purpose is regulatory compliance, we see a clash between two opposing mandates. On the one hand, the asset manager has the mandate, by fiduciary duty, to minimise risks for a target level of returns. The opposing mandate is that the asset manager is required by regulation to disclose self-assessed climate-related and other risks publicly. Against this backdrop, a for-profit climate data provider has the objective to sell data to maximise profit. But the asset manager has a choice over which data provider to use. The obvious incentive that results is that the data provided magically ‘contains’ the level of declared climate-related risk for the asset manager such that climate-related disclosures do not raise any red flags, but count as credible. This is possible because the estimation of climate-related risks, and the creation of climate scenarios, has, to date, no accepted methodology, and therefore is full of subjectivity. And of course, for-profit data providers do not usually release their methodologies with sufficient detail for external scrutiny, as they provide this under expensive licenses. Asset managers are generally more interested in the outcomes than in the methodology. The methodology would of course be of more interest to the interested public and the financial regulator than to asset managers themselves.
Disinformation in climate policy, politics and economics is not new. This has been studied by B. Franta (Oxford) in what regards the role of economists to limit perceptions of risks and inflate perceptions of costs of low-carbon transitions. Greenwashing has also led to a new type of litigation, especially in the US.
As we know from Taleb in The Black Swan, risk can be vastly underestimated by fund managers, and by extension, by rating agencies. Asset managers are under severe pressure to take risks and hope for the best. This became obvious of course in 2008 with sub-prime mortgage-derived financial assets, which were rated low risk and selling exponentially up until the moment when suddenly they were not anymore, and the entire market crashed. This is particularly salient in the case of climate-related risks, an area of analysis that is but nascent. Climate change is perceived as something that happens decades from now. Partly, this is a side-effect of Mark Carney’s famous Tragedy of the Horizon speech, where the emphasis was on the idea that climate impacts can only be expected to occur beyond the typical 3-5-year horizon of asset managers.
This is of course not true. Climate change impacts are felt now. They will become worse later, but they are sufficiently devastating now to be material for finance. Everyone in Europe has become, if they were not already, acutely aware of this during the recent heat wave in June. World Weather Attribution found the event would have been ‘virtually impossible’ fifty years ago, with a comparable heatwave in June 1976 running 3.5°C cooler. France recorded around 1,000 additional deaths and Spain over 1,000 excess deaths in its second-hottest June on record. Climate change is happening now and it is killing people now.
The other myth that resulted from Carney’s speech is that the magnitude of climate impacts will not be high or destructive within the financial time horizon. The 2022 floods in Pakistan tell a different story, where 30% of the country went underwater and 30% of public finance was absorbed by the response, requiring emergency loans later on. This could occur anywhere in the world. It is really just a matter of luck: the statistics are heavily fat-tailed. Every year, there is a not so small and increasing chance in most countries that such a high magnitude disaster occurs. The risks estimated by data providers will vary vastly depending on what’s considered in scope by them or by the asset manager commissioning them. This isn’t speculation. A peer-reviewed study of transition risk metrics across nine major providers, covering 1,565 firms in the MSCI World Index, found strong divergence in the transition risk assigned to the same company depending on which provider’s methodology was used. Put a single firm through nine different black boxes and you get nine different answers about how exposed it is. That is not what a mature risk discipline looks like.
Transition risk estimations possess perhaps even more subjectivity, and also have their choice of scope. Were we to have a sudden devaluation of oil-related assets, due to a sudden realisation in financial markets that the transport sector has acquired too many electric vehicles, how big should we expect the correction to be? Corrections like this work like earthquakes, we simply cannot predict the magnitude of the next quake. But we can calibrate the frequency-magnitude distribution. There is really no particular reason to expect a climate risk quake to have the same magnitude as the 2008 crisis: it could easily be much larger relative to the global economy. Otherwise, this would be like saying that earthquakes in countries in active regions cannot be larger than the largest quake they’ve seen in the historical record. Nobody would find this convincing.
In some ways the Strait of Hormuz crisis was predictable, but the magnitude of the oil crisis and its duration could not have been foreseen with any kind of accuracy. Risk assessment is not a prediction game, it’s a stress-testing game. But should the Hormuz situation be considered a relevant transition risk for climate risk reporting, or should it be seen as out of scope? We think it should, but that’s subjective. Oil companies make windfall profits out of such oil crises, while heavy energy users and sovereigns risk insolvency. What’s in scope risks depending on who you are and what you are happy disclosing.
The regulatory gap is visible. TCFD disclosure was voluntary and principles-based; its successor, ISSB’s IFRS S2, has made climate disclosure mandatory in a growing number of jurisdictions. But neither regime requires the methodology behind a provider’s numbers to be disclosed alongside the numbers themselves. Regulators mandated the output. They didn’t mandate seeing how the sausage is made.
To be fair to the industry: competition among for-profit providers has produced faster iteration, better user interfaces, and more granular, higher-resolution datasets than most academic groups could realistically maintain. That’s a genuine public good. But innovation in presentation is not the same as innovation in rigour, and a competitive market does nothing to solve the underlying incentive problem. If anything, competition for client business intensifies the pressure to deliver numbers the client wants to see, rather than numbers that are robust.
When assessing risk, the objective is to stress-test a system to find vulnerabilities. The objective is not, a priori, to conclude that the system is safe. Hence stress-testing requires some honesty and transparency. But for-profit organisations operate under business secrecy. So the game that financial regulators have set for banking institutions and corporations when requiring self assessment and disclosure of climate-related risks remains incomplete for as long as it does not include a strict requirement for transparency to allow public scrutiny.
This is in most ways similar to fire safety. Everyone sees the obvious conflict of interest that arises if manufacturers of construction materials collude with for-profit organisations mandated to test those materials for fire safety. Allowing for-profit companies to outsource to other for-profit ‘friendly partners’ to mark their homework as a mechanism to regulate fire safety is certainly looking for trouble. As we know, this can very much lead to housing towers burning down and lots of casualties. It would obviously be safer if the regulator wasn’t driven by a profit motive. The manufacturer outsourced the risk washing to a colluding for-profit partner.
For-profit organisations are, after all, not guardians of the public, as they will say so themselves. For investment funds, or public liability companies, this is enshrined in the law by fiduciary duty. As far as our experience goes in the private sector, the profit motive generally trumps any other mission. Effectively, it is quite hard to defend any other mission, as the bottom line will be ‘we need to sell product if we are to survive’, to which nobody can argue.
Our key point is this. In the financial sector, if fund managers buy climate-related risk assessments or scenario datasets from for-profit private providers, and if those assessments are not open for public scrutiny, then the above conflict of interest arises, with the incentive for risk washing. These data providers are under pressure to provide assessments that contain the perceived climate risk borne by those funds commissioning them, because the risk is that the asset managers don’t buy the data if they don’t agree with what’s in the tin. As we have by now seen too many times, we hear “I don’t care what’s in those scenarios, as long as someone will pay for them”, or “Key for me is that – to meet the market – we have to prioritise outcome intuitiveness over modelling accuracy. There is no value in being technically correct if the clients simply don’t buy it.”, or, “we’re only a small company and we’re struggling to survive, it’s a tough market out there and we have to offer a unique product”. This is all, in a way, legitimate for a for-profit company to say.
Designing scenarios for climate financial risk assessment on the basis of client requirements, rather than based on science, equates to a legitimation of climate-related risk-taking. And this legitimisation process becomes so much worse and dangerous if it is done under an apparent academic collaboration, and universities have to be wary of this. A for-profit data provider may not be able to resist, the pressure to make money is too high, especially if they have raised capital.
Due to climate-related risk disclosure requirements, what we may be seeing is the emergence of a profitable market for risk washing. Some of it may be full of good intentions, but it makes little difference for as long as this is done without allowing detailed public scrutiny. It appears to be in too few organisations’ interest to disclose real climate-related risks. Meanwhile, a market for making up data is highly profitable. It is also in the interest of all those involved to believe that climate risks are small relative to other risks, and to search for the dataset that supports this. Many of my colleagues across a range of non-profit and public institutions tell me the same, we are not the only ones.
It is hard to conclude anything other than that data providers for climate-related risk assessment should not have for-profit status, and should instead be driven by other values. The challenge is, central banks and financial conduct authorities around the world will not have the capacity to monitor this. The Network for Green the Financial System, the network of central banks focused on climate change, has gone a long way developing scenarios to test corporate and systemic financial resilience. The scenarios were created by academic modellers that had no financial interests, on a very meagre budget if any at all. That’s largely why central banks trust them.
These scenarios have in turn become criticised quite widely in corporate circles, almost out of habit. It is hard to resist the suspicion that this criticism largely originates from the fact that financial institutions have no control over the content of those scenarios, because they are narratives created without a profit motive. They cannot be adjusted to fit the needs of their users. Their subjectivity does not belong to corporates, but to the public sector.
What would good practice actually require? At minimum: methodology published in full, not licensed piecemeal. Assumptions and scenario inputs that a third party can independently reproduce. No commercial license gating the scrutiny of a number used for public disclosure. And audit by a body with no stake in the outcome. This is what we already expect of financial audit more generally. This isn’t a hypothetical ask. Publicly-funded, non-profit modelling groups already do this well in adjacent fields: the climate science community itself operates almost entirely this way, with CMIP model output, code, and assumptions published for anyone to scrutinise and challenge. There’s no reason climate financial risk assessment should be held to a lower standard of openness than the climate science it’s built on. It doesn’t have the marketing budget of a commercial data provider, and it won’t always give a client the answer they were hoping for. That’s rather the point.
Our recommendation is this. We urge central banks to issue guidance for developing scenarios for climate-related risk assessment along transparency principles, allowing disclosures to be audited and scrutinised by the greater interested public. We recommend that some standardisation of the procedure is developed in order to minimise the scope for picking methodologies with respect to desired outcomes. We also recommend that methods be developed through collaboration between actors from public institutions, the education and non-profit sectors. That seems to us to offer a way for climate-related risk assessment to remain reliable and relevant. Otherwise, the moral hazard created by for-profit data providers will have no historical precedent, as it facilitates increasing climate risk-taking by asset managers, rather than effective risk management.
