ESG risks in the loan book include those driven by a bank's impact on the world and the risk of financial loss due to ESG-driven defaults. For both, it is important to incorporate ESG metrics into the decisioning processes either by incorporating ESG factors into PD models or through exclusions.
Swiss banks need to incorporate ESG risk in their risk management. One reason is regulatory: Switzerland’s Climate Disclosure Ordinance makes the recommendations of the Task Force on Climate-related Financial Disclosures (TCFD) binding on larger Swiss financial institutions from 2024. The ordinance requires quantification of such risks where possible. Regulation in this area is likely to increase in line with recommendations made by Bank for International Settlement (BIS) in 2022. Beyond regulation, as ESG-related risks become more material, banks will need to accurately price that risk. Those which fail to do this will miss market opportunities and find themselves at a competitive disadvantage.
ESG Risk Taxonomy
The recent tendency has been for ESG risks to be incorporated within primary risk types, such as credit and market risk rather than treated as a standalone risk category. Of these, credit risk has received the most attention to date.
In the credit business, we see two types of ESG risk: There are those risks associated with compliance (or non-compliance) with regulation or, for example, with net-zero targets. This has to do with a bank's impact on the outside world and is referred to as inside-out or double materiality risk. For the bank, this materializes through fines, litigation or through reputational impact.
Then there is credit risk itself. This is the risk of financial loss sustained by the bank when an obligor defaults due to ESG-related factors. This type of risk can be described as outside-in.
Measuring ESG Risk in the Credit Business
As with any risk type, the foundation of credit risk management is the metric used to measure it. This metric is used to establish a risk appetite and to guide business decisions – in this case the issuance of loans – in line with that appetite.
The easiest metrics to employ here, and the bluntest instruments to use, are sectoral or regional concentrations, or concentrations mapped to ESG frameworks such as the EU taxonomy. These can be subject to limits, or in the strictest case, exclusions. Many banks currently exclude the coal industry for example. Concentration metrics are objective and easy to measure. They are of use in monitoring inside-out risk and portfolio alignment.
For the outside-in (credit risk) impact, other metrics are generally of more use. There are four main categories:
- Expected loss measures, which give an intuitive view of the risk associated with a loan;
- Regulatory capital measures, which establish the capital that a bank must hold to absorb annual losses at a 99.9% probability, according to a strictly defined methodology;
- Economic capital metrics which measure unexpected loss. These are conceptually similar to the regulatory capital view, but permit for a more bespoke modelling approach and give a portfolio view on risk;
- Scenario-based measures which are used to assess losses under a given hypothetical scenario. This is often performed for climate risk, and a separate blog entry discusses this in detail.
All the above four metric types require probabilities of default (PDs) as input. If such metrics are to be used, therefore, PDs must incorporate ESG factors.
ESG and Credit Ratings
PDs are assigned using either internal or external (agency) credit ratings. For corporate obligors, the "ESG quality" is often measured using external ESG ratings. ESG ratings should not be confused with the credit risk ratings themselves, often provided by the same agencies. Incorporation of ESG ratings into the credit risk assessment is recommended by Swiss Sustainable Finance’s report on Sustainability in Lending.
A number of studies has shown that ESG ratings are significant for PD estimation using Merton model approaches, Z-scores, historical default rates and CDS spreads to proxy PDs. Recent examples include Bannier et al. (2020) and Brogie et al. (2022). While credit ratings agencies do incorporate ESG factors in their ratings models, some research indicates that this is not always sufficient. For example, Aslan et al. (2021) shows that ESG factors have predictive power as a leading indicator for credit rating transitions.
For retail obligors, or in case the bank prefers not to rely on agencies, a scorecard approach can be adopted to produce an internal ESG rating. This uses quantitative and qualitative factors relating to the obligor, ranging from a building's energy-efficiency to a firm's anti-discrimination policy. Such an approach permits the bank to focus on the topics of most relevance for its business, and the data available. The selection of factors is a challenging exercise. Advanced methodologies make use of machine learning to incorporate media reports into the ESG rating in real time.