Climate Stress on Operational Risk

Introduction

As the Basel Committee on Banking Supervision has pointed out in their recent report Climate related financial risk – measurement methodologies, “banks and supervisors have predominantly focused on assessing credit risk, as they advance in applying methods to translate climate-related exposures into categories of financial risk”.

On the particular topic of operational risk, the Basel Committee also recommends in another document Climate related risk drivers and their transmission channels that “publicly available information regarding climate-related operational risks is scarcer than for other risk types, and therefore the whole risk category would benefit from more data and research”

As part of its regular exercise on operational risk scenario benchmarking, the ABA conducted a climate stress study of select operational risk scenarios between 2020 and 2021, the approach and results of which are particularly relevant to the Basel Committee recommendations and which we describe in detail below.

Categories of events

As we narrow down our focus to discuss the impact of climate change on banks operational risk, we consider 7 types of operational risks events: Conduct, Cyber, Disruption, Error, External Fraud, Internal Fraud and Legal.

There are potential dependencies between these operational risk categories and climate change, with the Disruption category being the most exposed, although other categories may also be affected, as shown in the table below; the Cyber, Error and Fraud operational risk categories being arguably the less impacted.

The transmission channels described in the Table 1 below are qualitative. To quantify them, it will be difficult to rely on past data, even if some basic mechanisms can be substantiated by data: for example, the relationship between climate and conflict has been quantitatively measured.

Using these transmission channels for the assessment of future operational risks, it is necessary to represent these risks by “loss generation mechanisms”, rather than data.

Operational Risk Category	Sensitivity to Physical Risk	Sensitivity to Transition Policy	Description of the possible transmission channels
Conduct		✓	New climate regulations will generate new obligations for banks and create new risks of misconduct.
Cyber		✓	Climate change may impact the geostrategic balance and increase the risk of cyber-attacks, especially by states.
Disruption	✓		Climate change increases the risk of natural disasters, disrupting not only banks but also their suppliers.
Error	✓		Increase of temperature and extreme meteorological events impacts productivity and increases the risk of error.
External Fraud	✓		Second order impact. Possible mechanisms are the increase in conflicting tensions, or the emergence of new vulnerabilities as banks adapt their business processes to the transition.
Internal Fraud	✓		Second order impact. Possible mechanisms are the increase in conflicting tensions, or the emergence of new vulnerabilities as banks adapt their business processes to the transition.
Legal	✓	✓	Climate change will expose some major companies to lawsuits and may expose their financial partners to liability. .

Structured Scenarios for Operational Risk

Structured scenarios for operational risk are specifically designed to describe loss generation mechanisms. The loss generation mechanisms are quantified using a combination of business data, external observations, expert opinion, and other data.

• The Exposure, Occurrence, Impact method (XOI), has been designed to represent and quantify loss generation mechanisms for operational risk scenarios.
• Exposure is defined by the resources exposed to the occurrence of the scenario. This can be employees for fraud, suppliers for disruption, operations for errors, products for mis-selling, etc.
• The probability of Occurrence of the scenario is assessed for each resource, and depends on the resource, on the firm controls and on external circumstances.
• The Impact of the event for one particular resource is assessed depending on the resource, on the firm controls, and on external circumstances.

In 2018, ABA launched an initiative to build and quantify structured scenarios with a bank-focused working group. This has been in particular applied to Cyber Risk in 2019. In 2020, an « ABA SSA Portal » hosting the scenarios and data has been developed and is now online.

Climate Stress on Structured Scenarios

We will now describe in detail the approach we have implemented with the Structured Scenario Work Group at ABA.

The first step was to define with the working group a set of scenarios for which there was a consensus on sensitivity to climate stress. This selection was not meant to be exhaustive but allowed us to focus on a subset of scenarios.

In the table below, we show an extract of the initial list of scenarios (35 scenarios were initially considered), and the resulting selection.

Operational Risk Category	Scenario Name	Main Sensitivity	Description of the possible transmission channel	Selected
Conduct	Corporate Client Misrepresentation	Transition	Transition may increase the probability of default of certain firms, and therefore the exposure of the bank to shareholders class actions.	✓
Conduct	Fund Improper Disclosure	Transition	Funds might be invested in non-environmentally friendly securities	✓
Conduct	Mis-selling Wholesale	Transition	Some products may depend on sectors impacted by climate change	✓
Disruption	Building Destruction	Physical	Building may be in location exposed to increased natural disasters	✓
Disruption	Datacenter Disruption	Physical	Datacenter might be in location exposed to increased natural disasters
Disruption	External Payment System Disruption	Physical	External system facility might be exposed to natural disasters
Disruption	Regional Disaster	Physical	Region might be exposed to increased probability of natural disaster	✓
Disruption	Supplier Failure	Physical Transition	Supplier might be in location exposed to increased natural disasters, or have an increased probability of default	✓
Error	Trading Algorithm Error	Transition	Market volatility may increase as the result of climate change
Error	Trading Error	Physical Transition	Probability of human error may increase as temperature increases. Market volatility may increase as the result of climate change
Internal Fraud	Unauthorized or Rogue Trading	Transition	Market volatility may increase as the result of climate change

The second step of the work was, for each scenario, to define a loss generation mechanism, and to quantify the variables of this mechanism. For the variables external to the banks, a quantification was proposed by the ABA based on research on external and robust sources.  The banks were asked to quantify their specific variables, i.e., mainly their exposure to certain risks depending on their internal controls or backup procedures.

We illustrate this process on the Regional Disaster scenario. The loss generation mechanism for this scenario is as follows and represented in a graphical model below.

A regional disaster, for instance an earthquake or a windstorm, hits a geographical area where the bank has significant assets or generates material revenue. The bank's premises - buildings, branches - and the lifelines - electricity, water, telecom, roads - in this area would be partially or fully destroyed. As a consequence, the bank would incur direct and indirect losses:

• Rebuild costs
• Relocation costs
• Loss of revenue

As shown in the table below, the quantification of the different drivers is done either from external sources, or by the individual banks.

Category	Drivers	Bank Assessment	External Assessment
Exposure	Regions	Bank’s Key Regions exposed	-
Occurrence	Regional disasters	-	Probabilities from USGS; GEM NOAA
Impact	Rebuild	Assets Value, Damage Rate	Default Damage Rate
Impact	Relocation	Relocation Cost Time to Normal	Default Time to Normal
Impact	Revenue	Dependent Revenue per Region	_

The third step of the work was to quantitatively assess the climate stress of the sensitive drivers.

For instance, the probability of regional disasters is increased, as global warming increases the frequency of hurricanes. Global warming causes a rise in sea levels and increases precipitation during hurricanes. These two factors combined increase the damage caused by hurricanes.

Based on our research, we have considered an increase of 10%-40% by the end of the 21st century for the average probability of a serious hurricanes (Category 1-2). For major hurricanes (Category 3 and more), their frequency has increased by 6% per decade over the four past decades, and an average increase of 28% is expected by 2100, but this projection varies a lot depending on the region (up to +338% for Northeast Pacific).

Results overview

Each of the participating banks has built its own assessment of each of the six operational risk scenarios included in the exercise as listed in the table below.

Operational Risk Category	Scenario Name
Conduct	Corporate Client Misrepresentation
	Fund Improper Disclosure
	Mis-selling Wholesale
Disruption	Building Destruction
	Regional Disaster
	Supplier Failure

Each assessment was then stressed according to the climate assumptions we considered.

For the sake of simplicity of this first exercise:

• The exposure to each scenario was not changed. This means that we have rather applied the anticipated climate conditions to the existing portfolio of exposures, rather than effectively trying to forecast the risk.
• We have used the most severe stress of physical risk and transition.
• Expected stress on drivers was linearized over a 30-year period, to produce an evaluation of the stress at 3 horizons: 3 years, 10 years and 30 years.

The metric stressed was the 1 in 1000 Value at Risk (VaR) of each scenario.

The results are shown below, where each line represents a bank:

• Conduct risk VaR is expected to increase by 80% in the conditions of 2050.
• Disruption risk VaR is expected to increase by 45% in the conditions of 2050.