Regulators acknowledge climate financial risks: Machine learning can help measure them
Scalable modeling approaches, like Terrafuse AI’s, are needed to measure the spectrum of climate risks for financial institutions
Climate change, if not addressed in an urgent manner, will pose a significant risk to society. While the impacts can be variable, under current policies the world is likely to warm up to 3C compared to pre-industrial times by the end of the century. For vulnerable countries, this may lead to a loss of Gross Domestic Product of up to 20% by 2050 and up to 64% by 2100.
That is, climate change will likely pose significant financial risks to firms and the economy. Climate-related financial risks refer to the set of potential risks that may result from climate change and that could potentially impact the safety and soundness of individual financial institutions and have broader financial stability implications for the banking system.
Climate change affects the financial system through two main channels. The first involves physical risks, arising from damage to property, infrastructure, and land. The second, transition risk, results from changes in climate policy, technology, and consumer and market sentiment during the adjustment to a lower-carbon economy.
To manage these risks, however, we need to first measure them. For example, a recent study estimates that real-estate portfolios could lose annual returns by as much as 40% by the end of the decade. To undertake climate risk analysis that can inform decision-making across the financial system, regulators and financial institutions need reliable, consistent, and comparable data and projections for climate risks, exposure, sensitivity, vulnerability, and adaptation and resilience.
Recognizing this need for transparency in climate risk disclosures, many protocols (e.g., TCFD and NGFS) have emerged over time. While ESG reporting has been around since 1997, via the establishment of the Global Reporting Initiative (GRI), the Taskforce for Climate Related Financial Disclosures (TCFD) was established in 2015 by the Financial Stability Board to incorporate climate risk in financial reporting. Next, the Network for Greening the Financial System (NGFS) was launched in 2017 by a group of central banks to share best practices in climate risk measurement and management. These are used to develop recommendations for more effective climate-related disclosures that promote more informed investment, credit, and insurance underwriting decisions and, in turn, enable stakeholders to better understand the financial system’s exposures to climate-related risks.
Furthermore, financial regulators are starting to recognize climate risks as systemic. In the United States, President Biden issued an Executive Order in 2021 to highlight an urgent need for assessing these risks, and to create a report on plans for measuring and managing them. The European Union, as an early mover, has already mandated in 2019 that the three financial regulators draw up a methodology to assess these risks.
As a result, these financial regulators are moving towards mandatory climate risk financial disclosures. The Securities and Exchange Commission (SEC) in the United States recently released climate risk disclosure rules, post a public review. In the UNFCCC COP in 2021, the UK announced mandatory climate risk disclosures starting April 2022. The European Union is also likely to announce these mandatory disclosures anytime now.
Given these upcoming changes, a key question is: What are the implications for businesses, including for financial institutions?
In this context, it is particularly relevant to focus on physical risk measurement and management, given lack of attention so far on acute risks that arise from increasing frequency and severity of extreme weather events, such as floods, fires, hurricanes, and droughts. In fact, a recent study estimates that the syndicated loan portfolios of the US banks face nearly half a trillion USD in losses due to these risks.
As the mandatory climate risk financial disclosure rule goes into effect, businesses and financial institutions will need to measure the following aspects of climate physical risks: hazards, exposures, vulnerabilities, and financial impacts. While the last two are in the purview of financial institutions, such as insurance companies, the first two create opportunities for technology startups. They require deep expertise in climate science and machine learning.
These startups, such as Terrafuse AI, Future Proof, and Climate AI use sophisticated artificial intelligence techniques. Terrafuse AI is unique because it combines the accuracy of physics-based models with the speed of artificial intelligence. These techniques allow for property-level resolution forecasting of climate and weather events, needed for accurate measurement of physical risks. For example, Terrafuse AI provides the likelihood of wildfire to burn any property in the coming year, which is key to assessing physical risks due to wildfire, and pricing of insurance products.
* * *
Gireesh Shrimali is a Research Scholar at Stanford University. Previously, he was the Director of Climate Policy Initiative’s India Program. He has taught at the Middlebury Institute of International Studies as well as the Indian School of Business. His research focus is on renewable energy finance and policy, the catalytic role of finance in getting to the 2C target, and provision of low-cost, long-term capital for clean energy transition. He holds a PhD from Stanford University, an MS from the University of Minnesota, and a BTech from the Indian Institute of Technology. He is an advisor to Terrafuse AI.