Abstract
Investors and regulators require reliable estimates of physical climate risks for decision-making. While assessing these risks is challenging, several commercial data providers and academics have started to develop physical risk scores at the firm level. This paper compares six physical risk scores, which rely on either model-based or language-based methodologies. We find a substantial divergence between these scores, also among those based on similar methodologies. Risk scores also lead to different rankings within and across sectors. We show how this divergence causes problems when testing whether financial markets are pricing in physical risks. Our results imply that financial markets may currently not be able to adequately account for the physical risk exposure of corporations using available risk scores. We identify six primary sources of uncertainty that drive the divergence of scores and offer recommendations for improving firm-level physical risk scores.
Highlights
The physical impacts of climate change pose major economic risks
To illustrate the potential consequences of the significant disagreement in risk assessments, we examine the robustness of the sorting mechanism to the use of different scoring models
Our results show that currently available metrics of firm-level physical climate risk diverge substantially
Summary
The physical impacts of climate change pose major economic risks. According to Swiss Re Institute (2020), weather-related events such as storms, floods, droughts, and wildfires caused global annual economic losses of USD 212bn on average over the last decade, and climate change is expected to lead to more frequent and more severe extreme weather events in the future. financial institutions aim to integrate these additional risks in financial decision-making. Fiedler et al (2021) argue that the scientific base for climate risk scores is still developing, raising the question of whether the use of these scores sufficiently improves financial decision-making. To explore this question, we compare six scores. Language-based approaches include a commercial score from Truvalue Labs (L1), and two academic scores, Firm-level Climate Change Exposure (L2) (Sautner et al, 2020a) and BERT-based climate risk measure (L3) (Kölbel et al, 2020). For model-based scores, we compare scores that aggregate different climate hazards and assess physical risk under an RCP8.5 scenario for a time horizon until 2050.4 Language-based scores were averaged over time to facilitate comparison (Table A.1)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
