Assessing Financial Risks from Physical Climate Shocks : A Framework for Scenario Generation

Scenarios developed to assess the financial risks from physical climatic events do not fully capture such shocks, finds this World Bank paper from three distinguished sustainable finance academics. The study identifies five areas in which material contribution to physical climate risks to the financial sector are not consistently included in current scenarios, a deficiency that could lead financial institutions to underestimate the potential scale of climate risks.

Climate-related scenarios are used in stress testing and other risk management exercises by both financial institutions and regulators.

Sudden and severe physical climate shocks are most likely to generate material shocks to the financial sector in the near term, the authors suggest, yet are not explicitly considered within the core scenarios developed by the Network for Greening the Financial System.

The paper examines approaches to assessing these financial risks by central banks and supervisors, reviewing current empirical evidence on the economic and financial impacts of physical climate shocks on the banking sector. Five specific gaps or “risk drivers” are identified: extreme weather events; uncertainties in climate models; compound scenarios; indirect economic impacts of shocks; and feedback between the real economy and the financial sector.

Turning to how these gaps can be better addressed within scenarios for physical climate-related financial risk assessment, the authors draw from disaster risk finance analytics and catastrophe risk models to propose a “realistic disaster” approach to developing scenarios. The framework combines model-based projections, expert judgment and the best available science to identify and prioritise key country-specific risk drivers and transmission channels.

The approach allows the definition of a set of “informative, relevant yet pragmatic climate scenarios that span the space of plausible future outcomes,” the authors say, helping overcome challenges in data scarcity and constraints on the availability of models.