Lessons learned in climate scenario analysis by the Federal Reserve: a pathway to robust modelling of climate risk in credit portfolios

• The Federal Reserve Board’s (FRB’s) pilot climate scenario analysis exercise showed a large range of credit impacts depending on the climate pathway, event and portfolio analysed.
• There was substantial variation both in the share of impacted loans and in the Probability of Default (PD) impacts, including variation arising from participating banks’ methodological choices.
• The most severe average credit impacts in the analysis were found in the CRE portfolio for physical risk.
• Given the variation in modelling approaches, banks can use the analysis results to help identify how to move to best practice.
   

The FRB’s climate scenario analysis

In order to understand and enhance climate risk management practices in banks, the FRB arranged an exploratory climate scenario analysis (CSA) exercise for the loan portfolios of six of the largest US banks: Bank of America Corporation, Citigroup Inc., The Goldman Sachs Group, Inc., JPMorgan Chase & Co., Morgan Stanley and Wells Fargo & Company. Earlier this month, the Fed published a selection of aggregated scenario analysis results, along with anonymized commentary on the various risk assessment approaches and methodologies applied by the institutions. The exercise was exploratory in nature, and does not have consequences for bank capital or supervision.

The exercise mandated assessment of climate-related risks in corporate (C&I), commercial real estate (CRE) and residential real estate (RRE) credit portfolios across two separate modules covering transition and physical risk. Participating banks reported climate-adjusted credit risk metrics, as well as qualitative details around their risk assessment methodology, governance, and risk management procedures.

The scenario analysis showed a large range of impacts depending on the climate pathway and event. The most severe average credit impacts in the analysis were found in the CRE portfolio for physical risk.

Across physical and transition risk scenario analyses, a key dynamic revealed was the significant variation in the share of impacted loans and the degree of variation in financial impact metrics such as PD. This variation depended on multiple factors including the specific climate scenario, how the bank represented the scenario or event (physical risk), and the granularity of the analysis.

First, let’s have a look at the range of results.

The physical risk analysis involved representing two severe climate scenarios. The first, selected by the regulator, was a Northeast hurricane. The second, called the idiosyncratic shock, was chosen by the bank to represent an extreme hurricane, wildfire, or extensive flooding scenario. Both scenarios were applied to CRE and RRE portfolios.

Results for the most severe Northeast hurricane scenario showed that around 20% of CRE loans and 50% of RRE loans were impacted on average. Average PDs increased by about 40 basis points (bps) for CRE and 10bps for RRE. For the idiosyncratic shock, which varied among participants based on their exposures and choices, average PDs increased by around 260bps for CRE and 110bps for RRE. The impact was particularly severe for CRE, which saw a larger proportion of loans impacted and 9% of loans experiencing a change of 500bps or more.

The transition risk module involved estimating the relative impact of two different NGFS climate change mitigation pathways over a 10-year period. In the exercise, these climate transition scenarios covered impacts on the C&I and CRE portfolios. The high-transition risk scenario, Net Zero 2050, represents rapid carbon emissions reduction. In this scenario, the average PD over the 10-year horizon was around 100bps higher than the low-transition risk scenario for CRE, and around 30bps higher for C&I loans.

With the impacts differing so widely depending on the climate inputs selected, the other key factor that determined the difference in results was the difference in methodologies and analytical granularity. The FRB explicitly commented that results and approaches varied greatly across the participating banks, and methodologies are evolving. How then can one obtain relevant, clear and perceptive insights on climate in the credit portfolio in an environment of uncertainty? What are the emerging best practices?
 

Physical risk

The FRB noted that participating banks simulated different numbers of individual hazard events and provided varying levels of granularity in their physical risk analysis. The resolution used in modelling these events can significantly impact physical risk analytics. Additionally, the determination of physical impact from a climate scenario or event varies depending on the data and modelling techniques used. Some participants simulated physical risk under future climate scenarios using climate-conditioned catastrophe models, such as the Moody’s RMS models, while others used downscaled climate models. Participants who did not use catastrophe models had to develop their own vulnerability or impact modelling to determine the financial cost to counterparties of physical damage from the event.

Moody’s RMS models consider cumulative and hazard-specific impacts of events of varying severity. This is done using a large database of simulated events, including hurricanes, floods, and wildfires, conditioned to represent future climate variations. The associated damages for these possible future events are modelled at the location level, the most detailed level possible. Impact measurements for each event are calibrated based on historical insurance claims data. Moody’s use of in-house climate-conditioned catastrophe models represents industry standard practice for detailed physical risk analytics.

Participants had to make specific assumptions around the degree of insurance coverage in their credit portfolio. The FRB found this to be one of the key outstanding challenges in the exercise. The extent to which physical risk losses are covered by insurance varies according to the nature of the hazard and the property. Moody’s insurance likelihood tool models insurance coverage in mortgage portfolios and allows insurance-related assumptions to be fully integrated into the loss modelling process.

Transition risk

In the transition risk module, participating banks typically expanded the NGFS climate change scenarios to include additional macroeconomic variables. This allowed the impacts of the two selected scenarios to be resolved using more relevant economic details. For the CRE portfolio, Moody’s climate solution derives industry-standard property market indicators from climate scenarios. For C&I loans, Moody’s expertise in economic modelling helps interpret the economic impacts of climate change on specific sectors, helping to make the impacts of climate scenarios relevant to credit portfolio management .

Integrated credit modelling

The FRB noted that to estimate climate credit impacts on corporate counterparties, some participating banks used statistical methods linking obligor risk rating systems to macroeconomic variables. Other banks instead estimated the impact of climate scenarios on detailed counterparty financials. Underlying both approaches is the need of banks to not only understand the dynamics around physical and transition risk, but translate that into decision-relevant analytics that can integrate into existing risk management processes.

To comprehensively integrate into credit processes, both top-down impacts from scenarios to credit ratings and credit risk models and bottom up-counterparty financial analytics are required. Top-down impacts are important for calculating overall portfolio impacts, as in the FRB exercise, and highlighting sectors and geographies with elevated climate risk. Bottom-up counterparty financial analytics deploy climate risk impact forecasts to identify individual loans that require attention. Moody’s modelling can provide the impacts of climate scenarios on counterparty financial statement line items and balance sheets, as well as obligor credit ratings. This seamlessly overlays climate impacts, across different levels of analysis, on top of existing financial performance and credit-relevant information.
 

Learn More

The FRB pilot climate scenario analysis exercise identified a large degree of variation in modelling approaches, as well as credit impacts, in different contexts. This demonstrates the importance of implementing industry standard practices for physical risk, transition risk and integrated climate credit modelling across portfolios and climate scenarios. Moody’s risk quantification and workflow capabilities help banks analyse, act on and communicate the potential credit impact of climate by enabling identification, measurement and management of climate risk and the unlocking of opportunities in the credit portfolio. We support climate risk and opportunity analysis in all core banking asset classes, across both physical and transition risk, harnessing our expertise in credit modelling.