CLS Blue Sky Blog

What’s in Your Climate Score?

An increasing number of consultancies, financial technology firms, data providers, and investment advisory groups offer information about localized physical climate risks like floods, hurricanes, and wildfires. In recent years, mainstream financial services companies have acquired many of the early climate-analytics startups and weather-modeling firms. The list includes: The Climate Service, bought by S&P; Four Twenty Seven and RMS, both bought by Moody’s; Acclimatise, bought by Wills Towers Watson; South Pole, bought by proxy adviser ISS; Planetrics, acquired by McKinsey; Jupiter Intelligence, partnered with Boston Consulting Group and the U.S. government; Rhodium, which licenses its physical risk model to asset manager BlackRock; and Carbon Delta, whose “Climate VaR” model was bought by MSCI.

In the accelerating “climate intelligence arms race,” companies increasingly claim to use better science, better methods and models, and more comprehensive data than their competitors.[1] But the proprietary nature of these products make them hard to compare, let alone oversee. And the value chain of climate services extends from government institutions collecting weather data and running supercomputers down to consultancies that assess and communicate localized risk. So, it can be difficult for an end-user to determine the source of data  and the value a given provider added. While there is enormous demand from the private sector for useful risk assessment, there is little in-house capacity to evaluate products. Academics have flagged that private consultancies can over-claim what science is able to predict at small geographic scales or over particular time periods.[2] Worryingly, the physical-risk scores produced by leading ESG firms have been found to have little correlation with one another.[3]

The outcomes of these risk assessments have significant consequences in the economy. They are used to allocate equity capital through active and passive investments. They influence insurance premiums, which affect housing prices and cause demographic shifts. They are considered in municipal bond ratings, thereby serving as a mechanism for prioritizing city and county infrastructure projects that protect certain neighborhoods over others.[4]

In my forthcoming article, I call for greater oversight of these risk scores and providers. I also call for a substantial increase in investment in the government’s own resources for assessing and communicating near-term climate risk.

Despite the federal government’s linchpin role in developing resource-intensive global climate models, it has arguably lagged the private sector in the production and broad dissemination of “usable” climate science aimed at helping individuals and entities decide how to adapt to climate change.[5] I argue that this lag has significant consequences for regulators’ ability to monitor and police these providers of information about financial risk. Outside of the United States, scientists with expertise in climate extremes warn that financial regulators are making some of the same mistakes as private climate-services providers, including designing risk-disclosure rules that require more precision than climate models can deliver.[6] Indeed, this winter the Federal Reserve rolled out a pilot climate-scenario analysis examining hurricane risk on the east coast that asked banks an impossible science question.[7]

As financial regulators increase their ability to monitor climate risks, other groups within the federal government work to provide better public information on climate risk. I argue that these two efforts, one focused on financial risk and the other on resilience and adaption in the U.S. economy, must be better integrated at both the policy and technical levels. The National Science Foundation and the National Oceanographic and Aeronautics Administration agree that more research funding is needed to assist the insurance industry in pricing near-term climate risk. Nevertheless, there is little oversight over how entities like Fannie Mae and the Federal Housing and Finance Administration license the climate risk data underlying their own longer-term risk assessments from non-governmental third parties.

As I explain in detail in my article, projecting long-term climate risks – like those affecting a 30-year mortgage – requires a combination of data outputs and complex methods. These assessments take significant training and modeling resources to produce, and as they are projections far in the future, the only way to validate them is to explore their methods and data inputs. In academia, peer review and data-transparency requirements serve this role of checking the climate forecasts we rely upon. But as these questions of future risk leave academic journals and enter underwriting policies, the private sector would like to keep their methods and data proprietary and hidden. The insurance industry is vigorously protesting the Federal Insurance Office’s recent effort to collect data on which U.S. regions may be exposed to physical hazards or lack of available insurance.

But the severity and systemic nature of climate risk means that these insurance models potentially have macroeconomic significance; in our interconnected world, we increasingly need to know one another’s exposure. Whether a luxury condo is resilient to hurricane risk depends, in part, on its location and projected climate conditions, but it also depends on the resilience of the surrounding municipality. A building that has zero flood risk is not very useful if all the roads leading to the building are washed out or if the city in which the condo sits lacks the capital to adapt.

Advocates and science experts alike have long called for a “National Climate Service” (NCS), arguing that location-specific climate risk information is a public good, akin to the National Weather Service’s free forecasting and provision of data. I agree with NCS advocates that the private sector cannot be relied upon to provide climate services equitably or reliably, and that the industry requires far more oversight. Importantly, all private climate services rely on upstream climate data and models that were collected and produced by an enormous network of public institutions. We should not hesitate to require pseudo-regulatory third parties like ratings agencies and insurers to tell us how they are applying this public resource, our collective knowledge about the future.[8]


[1] Jesse M. Keenan, A Climate Intelligence Arms Race in Financial Markets, 365 Science 1240, 1240 (2019).

[2] Tanya Fiedler et al., Business Risk and the Emergence of Climate Analytics, 11 Nat. Climate Change 87, 91 (2021);

[3] Linda I. Hain et al., Let’s Get Physical: Comparing Metrics of Physical Climate Risk, 46 Fin. Rsch. Letters 1, 3 (2022).

[4] Savannah Cox, Inscriptions of Resilience: Bond Ratings and the Government of Climate Risk in Greater Miami, Florida, 54 Env’t. & Plan. A: Econ. & Space 295, 295 (2021).

[5] Adam H. Sobel, Usable Climate Science Is Adaptation Science, 166 Climatic Change 1 (2021).

[6] A.J. Pitman et al., Acute Climate Risks in the Financial System: Examining the Utility of Climate Model Projections, 1 Env’t. Rsch.: Climate 1, 4 (2022); Irene Monasterolo et al., The Good, the Bad and the Hot House World: Conceptual Underpinnings of the NGFS Scenarios and Suggestions for Improvement, No. 2302, 2302 (Banco de España Feb. 2023).

[7] R. Saravanan, Hurricane Fed: The New Climate Stress Test for Banks, MetaModel Blog (Mar. 1, 2023),

[8] Cf. Mariana Mazzucato, The Entrepreneurial State: Debunking Public Vs. Private Sector Myths 179-194 (2013) (arguing that the public should question whether we get enough in return when we allow the private sector to capitalize on and corner research undertaken by public institutions).

This post comes to us from Madison Condon, associate professor at Boston University School of Law. It is based on her forthcoming article, “Climate Services: The Business of Physical Risk,” available here.

Exit mobile version