Climate, ESG and the Future of Bank Risk Management

Introduction

Climate change, environmental degradation, social inequality and governance failures have become material determinants of banking risk. They influence borrowers’ revenues, operating costs, asset values, insurance availability, legal liabilities, funding conditions and access to markets. Their effects may appear gradually, as in the repricing of carbon-intensive assets, or suddenly, as in a flood, cyber incident, political shock or governance scandal.

These risks are not usually separate from traditional banking risks. Rather, they are transmission channels through which credit, market, liquidity, operational, legal, reputational and strategic risks are amplified. A bank that treats climate and ESG issues merely as reporting subjects may therefore underestimate its risk profile. A sound approach embeds them in strategy, risk appetite, underwriting, valuation, stress testing, capital planning, portfolio management, disclosure and governance.

The central managerial challenge is uncertainty. Climate pathways are not forecasts in the conventional sense; they depend on policy, technology, economic behaviour and physical processes. ESG data are incomplete and sometimes incomparable. A bank must nevertheless make timely decisions using scenario analysis, conservative assumptions, expert judgement and transparent model governance.

Learning objectives

By the end of this chapter, students should be able to:

1. distinguish physical, transition, environmental, social and governance risks;
2. explain how these risks transmit into conventional banking risk categories;
3. measure exposure using indicators, scores, scenario analysis and portfolio metrics;
4. integrate ESG factors into credit assessment, collateral valuation and portfolio management;
5. design climate scenarios and interpret their limitations;
6. evaluate geopolitical risk, interconnectedness and contagion;
7. assess disclosure, regulatory and governance developments; and
8. propose a future-oriented architecture for real-time bank risk management.

1. The risk taxonomy

1.1 Physical risk

Physical risk arises from hazards and environmental changes. Acute hazards include floods, cyclones, wildfires, droughts and heatwaves. Chronic hazards include sea-level rise, changing rainfall, water stress, rising average temperature and ecosystem degradation.

Physical risk can affect a borrower through four principal mechanisms:

• Asset damage: destruction of factories, inventories, homes or infrastructure;
• Business interruption: lost production, supply delays and reduced labour productivity;
• Input and operating costs: higher water, energy, insurance or adaptation costs; and
• Collateral impairment: lower property, agricultural or infrastructure values.

Physical risk is geographically heterogeneous. A bank should therefore assess location, hazard intensity, vulnerability and adaptive capacity rather than applying a single sectoral average.

1.2 Transition risk

Transition risk results from adjustment towards a low-carbon and environmentally sustainable economy. Drivers include carbon pricing, energy regulation, litigation, technological substitution, consumer preferences and investor mandates.

A coal-fired power producer may face declining demand, higher compliance costs and stranded assets. An automobile manufacturer may face large investment requirements for electric vehicles. Conversely, transition can create opportunities for renewable energy, energy efficiency and climate-resilient infrastructure.

Transition risk is often nonlinear. A policy change, technological breakthrough or change in investor sentiment can cause rapid repricing. The same asset may therefore be safe under one pathway and impaired under another.

1.3 Environmental risk beyond climate

Environmental risk includes biodiversity loss, deforestation, soil degradation, water scarcity, pollution, hazardous waste and ecosystem-service disruption. These risks can affect production and legal permissions even when greenhouse-gas emissions are modest.

For example, a food processor dependent on a watershed may suffer material losses from water scarcity. A mining borrower may face remediation costs, licence withdrawal and litigation due to contamination. Environmental dependencies and impacts should be considered together: a firm may depend on nature while simultaneously damaging it.

1.4 Social risk

Social risk concerns impacts on employees, customers, communities and human rights. Relevant issues include labour conditions, occupational safety, access to essential services, indigenous rights, privacy, financial inclusion and product mis-selling.

Social weaknesses can produce strikes, lawsuits, regulatory sanctions, customer losses and reputational damage. Social risk is especially significant in project finance, extractive industries, agriculture, construction, healthcare and consumer lending.

1.5 Governance risk

Governance risk concerns the quality of oversight, incentives, controls and ethical conduct. It includes corruption, weak boards, inaccurate reporting, conflicts of interest, poor remuneration design, tax misconduct, cyber-control failures and failure to manage ESG matters.

Governance is an enabling condition for all other risk management. A borrower with ambitious environmental targets but unreliable data, weak controls and executive incentives based solely on short-term earnings may present greater risk than its public commitments suggest.

2. Transmission into banking risk

ESG risks become financially relevant through established risk categories.

A useful conceptual decomposition is:

This is a framework rather than an exact pricing equation. “Hazard” describes the event or change; “exposure” identifies affected assets, activities or counterparties; “vulnerability” measures sensitivity and adaptive capacity; and “financial transmission” maps the effect into cash flow, value, probability of default, loss given default or market price. The components must not be multiplied mechanically unless they have been defined on compatible, normalised scales.

For credit risk, expected loss remains conceptually relevant:

where EL is expected loss in currency units, PD is probability of default, LGD is loss given default as a proportion of exposure, and EAD is exposure at default in currency units. Climate and ESG factors may alter any of the three inputs. The relationship assumes that the inputs are consistently defined for the same horizon and that dependence and timing effects are adequately addressed. In practice, scenario analysis may produce conditional values of PD, LGD and EAD rather than a single deterministic estimate.

3. Measuring climate and ESG risk

3.1 Exposure mapping

The first task is to map counterparties, facilities, securities and collateral to relevant locations, sectors, technologies and supply chains. A bank should identify:

• borrower revenue and assets by geography;
• emissions by scope and data quality;
• energy and water intensity;
• dependence on critical ecosystems;
• insurance coverage and deductibles;
• adaptation plans and capital expenditure;
• governance, controversies and litigation; and
• maturity dates of loans and investments.

Data lineage is essential. Each metric should have an owner, source, date, estimation method, confidence level and review frequency.

3.2 Emissions metrics

Financed emissions estimate emissions associated with lending and investment. A simplified attribution approach is:

The attribution factor depends on the asset class and methodology. It must be defined consistently; for example, a lender may attribute emissions in proportion to the lender’s outstanding exposure relative to an appropriate measure of the borrower’s enterprise value or financing structure. The formula is not a universal accounting identity, and double counting can occur across institutions and asset classes.

Intensity can be expressed as:

Units might be tonnes of CO₂-equivalent per tonne of product, megawatt-hour or unit of revenue. Comparisons are meaningful only when boundaries, periods, denominators and estimation methods are comparable.

3.3 Portfolio indicators

Useful indicators include sector concentration, geography-weighted hazard exposure, financed emissions, temperature-alignment measures, energy intensity, proportion of collateral in high-risk zones, green-asset share, exposure to controversial activities and ESG-data coverage.

Portfolio concentration can be represented as:

where sᵢ is the proportion of portfolio exposure attributable to sector or counterparty i, and the sum is over all categories. The Herfindahl–Hirschman Index rises with concentration. It does not measure climate risk directly: a diversified portfolio may still be exposed to the same carbon price, water shortage or geopolitical shock.

3.4 Credit assessment integration

ESG information should enter the normal credit process rather than exist as an unconnected score. A practical sequence is:

1. identify material ESG drivers for the borrower and facility;
2. assess current exposure, controls and mitigation;
3. test cash flow under relevant scenarios;
4. evaluate collateral, insurance and covenants;
5. adjust risk grades, pricing, tenor, limits or monitoring; and
6. document judgement, evidence and uncertainty.

A score should not replace analysis. Two firms with identical emissions may have different risk because of debt maturity, pricing power, technology, location, insurance and adaptation capacity.

3.5 Scenario analysis and stress testing

Scenario analysis asks how a portfolio would behave under coherent assumptions. A scenario may include carbon prices, energy demand, interest rates, exchange rates, commodity prices, hazard intensity, GDP, unemployment and property values.

A simplified stressed loss calculation is:

The subscript s denotes the scenario. A scenario is not a probability forecast unless probabilities have been assigned and justified. Climate scenarios are particularly sensitive to time horizon, discounting, technological change, policy timing and feedback effects.

A good exercise uses several pathways, including an orderly transition, a delayed or disorderly transition and a high-physical-risk pathway. Reverse stress testing is also valuable: management asks which combination of hazards and transition shocks would breach capital, liquidity or risk appetite limits.

4. Portfolio and balance-sheet management

Climate and ESG integration affects origination, pricing and portfolio construction. Banks may use sector limits, exclusion policies, enhanced due diligence, transition-linked covenants, sustainability-linked pricing, tenor restrictions and collateral haircuts. Such tools must be evidence-based. A blanket exclusion may reduce direct exposure while increasing financing gaps or shifting risk to less regulated institutions.

Portfolio steering should distinguish between:

• green activity, which contributes to environmental objectives;
• transition activity, which may be carbon-intensive today but has a credible pathway to improvement; and
• harmful activity, where the business model is incompatible with risk appetite or legal requirements.

A transition plan should contain baselines, interim targets, capital expenditure, governance, assumptions and escalation triggers. Targets without financing and accountability are not reliable risk mitigants.

Collateral analysis requires special care. Physical hazards can reduce marketability and insurance availability. A property may retain a high current value but have a lower long-run value if buyers anticipate future flooding or regulation. Loan-to-value ratios should therefore be tested over the loan horizon, not only at origination.

5. Geopolitical risk and interconnectedness

Climate and ESG risks interact with geopolitics. Critical minerals, energy routes, food systems, sanctions, migration, water disputes and competition over technology can transmit shocks across borders. A conflict may simultaneously increase energy prices, weaken a borrower’s currency, interrupt supply chains and create liquidity stress.

Interconnectedness operates through direct exposures, common funding, shared service providers, payment systems, derivatives, correspondent banking and correlated asset holdings. Network analysis can identify central counterparties and clusters, but network centrality is not identical to loss contribution: a highly connected node may be resilient, while a less connected node may hold highly concentrated risky assets.

Banks should combine entity-level analysis with system-wide exercises. Important questions include:

• Which borrowers share the same supplier or port?
• Which portfolios depend on one energy market or technology?
• Could a cyber or geopolitical event interrupt several critical services?
• Could simultaneous margin calls create liquidity contagion?
• Does an insurer’s withdrawal amplify property or mortgage risk?

6. Governance, disclosure and regulation

Board responsibility should include approval of risk appetite, transition strategy, materiality thresholds and escalation arrangements. Senior management should assign clear ownership across risk, finance, sustainability, credit, treasury, operations and data functions. Internal audit should test controls and the reliability of disclosures.

A three-lines model is useful:

1. business and origination teams identify and manage risk;
2. independent risk and compliance functions set frameworks, challenge assumptions and monitor limits; and
3. internal audit provides independent assurance.

Disclosure frameworks increasingly emphasise governance, strategy, risk management, metrics and targets. Regulatory expectations commonly include board oversight, materiality assessment, scenario analysis, capital and liquidity implications, data governance and credible transition planning. Exact requirements differ by jurisdiction and change over time; banks must therefore maintain a regulatory inventory rather than rely on a static checklist.

Disclosure creates both discipline and risk. Unsupported claims may produce greenwashing allegations, litigation and reputational damage. A bank should disclose methodology, boundaries, estimation uncertainty, data gaps and changes in measurement. Assurance should be proportionate to materiality.

7. Real-time analytics and future risk architecture

Traditional annual reviews are insufficient for rapidly changing environmental and geopolitical conditions. A future risk architecture will combine:

• a governed ESG data platform;
• geospatial hazard layers and satellite or sensor information;
• transaction-level exposure mapping;
• event-driven alerts;
• scenario engines linked to finance and risk systems;
• model-risk controls and human review; and
• management dashboards connected to decisions.

Artificial intelligence can classify documents, detect controversies, estimate missing data and identify supply-chain relationships. However, AI does not eliminate uncertainty. Risks include biased training data, unstable relationships, false alerts, lack of explainability, privacy breaches and automated decisions unsupported by evidence. High-impact decisions should retain human accountability, audit trails, override controls and periodic validation.

Data architecture should distinguish observed, reported, estimated and modelled values. A useful data-quality record includes completeness, timeliness, consistency, provenance and confidence. The bank should avoid false precision: an estimated emissions figure with low confidence should not be presented as equivalent to an audited figure.

8. Worked application: illustrative case

Background

NorthRiver Bank finances GreenVale Foods, a large agricultural processor. GreenVale has substantial facilities in a drought-prone basin and purchases crops from farms exposed to heat stress. Its loan exposure is CU 120 million. Before climate analysis, the one-year PD is 2.0%, LGD is 40%, and EAD is CU 120 million.

Problem

The bank’s initial expected loss is:

A drought scenario is expected to reduce input supply and operating margins. Credit analysis estimates scenario values of PDₛ = 5%, LGDₛ = 50% because collateral values and recovery prospects weaken, and EADₛ = CU 125 million because the borrower draws a working-capital facility.

Analysis and approach

The bank maps the borrower’s facilities, water dependence, supplier geography, insurance and adaptation expenditure. It finds that GreenVale has partial water insurance but no tested contingency plan. Management proposes recycling investment, diversified sourcing and drought-resistant crops.

Scenario expected loss is:

The incremental scenario loss relative to the baseline is CU 2.165 million. This is not a forecast of actual loss; it is an estimate conditional on the stated assumptions.

The bank responds by requiring quarterly water-risk reporting, a minimum liquidity buffer, completion of adaptation milestones and a borrowing-base review. It also reduces concentration by setting limits on other borrowers dependent on the same basin.

Outcome and learning points

The case demonstrates that a physical hazard can affect probability of default, loss severity and exposure simultaneously. It also shows why borrower-level adaptation, collateral analysis and portfolio concentration must be considered together. The bank should revisit the assumptions, test alternative drought severities and monitor whether insurance remains available and affordable.

9. Challenges and critical evaluation

Several limitations complicate ESG risk management.

Uncertain time horizons: Bank loans may mature in three years, while climate effects unfold over decades. Long-horizon analysis can nevertheless matter for mortgages, infrastructure and strategic lending.

Data gaps: Private firms, small businesses and emerging markets may lack reliable emissions or location data. Estimation is necessary but should be transparent.

Model uncertainty: Historical relationships may fail under structural change. Climate scenarios are not interchangeable with conventional economic forecasts.

Aggregation problems: A portfolio average can conceal vulnerable borrowers, while a score can hide the mechanism of risk.

Greenwashing and transition-washing: Labels may overstate environmental performance or the credibility of transition plans.

Distributional and ethical effects: Restricting finance can harm workers, communities and developing economies. A just transition requires attention to affordability, employment and financial inclusion.

Regulatory fragmentation: Definitions, taxonomies and disclosure rules differ across jurisdictions. Banks operating internationally need controlled mapping between frameworks.

Best practice is therefore not to seek a single perfect ESG number. It is to build a decision system that combines quantitative indicators, qualitative analysis, scenario reasoning, challenge and accountability.

Chapter Summary

Climate, environmental, social and governance risks are financially material drivers of banking risk. Physical risks arise from acute hazards and chronic environmental change; transition risks arise from economic adjustment; environmental risks include pollution and ecosystem degradation; social risks concern people and communities; and governance risks concern oversight, conduct and control.

These drivers transmit into credit, market, liquidity, operational, legal, reputational and strategic risk. Measurement requires exposure mapping, reliable data, emissions and intensity metrics, concentration analysis, borrower due diligence and scenario testing. The expected-loss framework EL = PD × LGD × EAD remains useful, but ESG factors can change every component.

Effective integration embeds ESG analysis in underwriting, collateral valuation, pricing, covenants, portfolio limits, capital planning and disclosures. Geopolitical shocks and interconnected networks can amplify risks across sectors and borders. Governance, transparent methodologies and human accountability are essential, particularly when artificial intelligence and real-time data are used.

The future bank will require a dynamic risk architecture combining geospatial information, event alerts, scenario engines, strong data lineage, model-risk management and informed professional judgement. The objective is not prediction with false precision, but resilient and accountable decision-making under uncertainty.

Key Terms

Physical risk: Financial risk arising from acute or chronic physical environmental hazards.

Transition risk: Risk arising from adjustment to a low-carbon and sustainable economy.

Environmental risk: Risk associated with pollution, biodiversity, water, waste and ecosystem degradation.

Social risk: Risk affecting workers, communities, customers and human rights.

Governance risk: Risk arising from weak oversight, controls, incentives or ethical conduct.

Financed emissions: Emissions associated with a bank’s lending or investment activities.

Stranded asset: Asset that loses value because economic, technological, regulatory or environmental conditions change.

Adaptation: Actions that reduce vulnerability to actual or expected environmental impacts.

Transition plan: A time-bound strategy for moving an institution or borrower toward defined sustainability objectives.

Scenario analysis: Assessment of outcomes under a coherent set of assumptions.

Stress testing: Evaluation of resilience under adverse but plausible conditions.

Expected loss: Estimated credit loss represented conceptually by PD × LGD × EAD.

Probability of default: Likelihood that a counterparty defaults over a specified horizon.

Loss given default: Proportion of exposure expected to be lost after default.

Exposure at default: Amount exposed when default occurs.

Greenwashing: Misleading presentation of environmental performance or claims.

Materiality: Importance of an issue to financial performance, stakeholders or decisions.

Data lineage: Record of a data item’s source, transformations, ownership and history.

Interconnectedness: Degree to which institutions, markets or infrastructures depend on one another.

Nature-related risk: Risk arising from dependencies on or impacts upon ecosystems and natural resources.

Review Questions

Short Answer Questions

1. Distinguish physical risk from transition risk and provide one example of each.
2. How can ESG factors affect probability of default and loss given default?
3. Why is data lineage important in ESG risk management?
4. What is the purpose of a transition plan?
5. Explain why a portfolio can be diversified by sector yet still exposed to common climate risk.

Long Answer Questions

1. Explain how climate, environmental, social and governance risks transmit into conventional banking risk categories.
2. Discuss the advantages and limitations of using scenario analysis for climate risk management.
3. Design an ESG-integrated credit assessment process for a commercial borrower.
4. Evaluate the governance and disclosure responsibilities of a bank’s board and senior management.
5. Critically assess the potential and limitations of artificial intelligence in real-time ESG risk analytics.

Analytical / Application Questions

1. A borrower has PD = 3%, LGD = 45% and EAD = CU 80 million. Under a drought scenario, the values become 6%, 55% and CU 84 million. Calculate baseline and scenario expected loss, then interpret the difference.
2. A bank’s mortgage portfolio is concentrated in two coastal regions. Propose a framework combining hazard data, collateral valuation, insurance analysis and portfolio limits.
3. A geopolitical shock affects energy prices, shipping routes and foreign-exchange markets simultaneously. Explain how a bank should analyse direct exposures, second-round effects, liquidity contagion and management actions.

Multiple Choice Questions with Answers

Question 1

Which statement best describes transition risk?

A. Losses caused only by hurricanes and floods

B. Risk arising from economic adjustment toward a lower-carbon economy

C. Risk caused exclusively by employee misconduct

D. Risk that a payment system becomes temporarily unavailable

Correct Answer: B. Risk arising from economic adjustment toward a lower-carbon economy

Question 2

In the relationship EL = PD × LGD × EAD, what does LGD represent?

A. Probability of default

B. Liquidity gap duration

C. Loss given default

D. Loan growth denominator

Correct Answer: C. Loss given default

Question 3

A drought increases a borrower’s PD, LGD and working-capital borrowing. What does this illustrate?

A. Physical risk can affect multiple credit-risk parameters

B. Physical risk affects only market risk

C. Collateral values are independent of environmental hazards

D. Scenario analysis eliminates uncertainty

Correct Answer: A. Physical risk can affect multiple credit-risk parameters

Question 4

What is a principal limitation of an ESG score?

A. It always measures future cash flow precisely

B. It can conceal the mechanisms and uncertainty behind a rating

C. It cannot be used in any credit process

D. It automatically prevents greenwashing

Correct Answer: B. It can conceal the mechanisms and uncertainty behind a rating

Question 5

What does a higher HHI generally indicate?

A. Lower portfolio concentration

B. Higher portfolio concentration

C. Lower financed emissions

D. Higher data quality

Correct Answer: B. Higher portfolio concentration

Question 6

Which practice most directly supports credible ESG disclosure?

A. Omitting estimation uncertainty

B. Using unsupported sustainability claims

C. Documenting boundaries, methods, data gaps and assumptions

D. Replacing internal audit with marketing review

Correct Answer: C. Documenting boundaries, methods, data gaps and assumptions

Question 7

Why can a transition scenario produce rapid asset repricing?

A. Policy, technology or investor expectations can change abruptly

B. Transition risk is always gradual and predictable

C. Physical hazards cannot affect asset prices

D. Banks do not hold market-sensitive assets

Correct Answer: A. Policy, technology or investor expectations can change abruptly

Question 8

Which is an appropriate use of AI in ESG risk management?

A. Making unreviewable high-impact lending decisions

B. Detecting patterns and estimating missing data with governance controls

C. Replacing all borrower due diligence

D. Guaranteeing unbiased predictions

Correct Answer: B. Detecting patterns and estimating missing data with governance controls

Question 9

A baseline exposure has PD = 2%, LGD = 40% and EAD = CU 100 million. What is expected loss?

A. CU 0.08 million

B. CU 0.8 million

C. CU 8 million

D. CU 80 million

Correct Answer: B. CU 0.8 million

Question 10

Why should banks analyse interconnectedness in climate and geopolitical risk?

A. Common suppliers, funding and infrastructure can transmit shocks across borrowers

B. Interconnectedness applies only to equity markets

C. It guarantees that all borrowers default together

D. It makes scenario analysis unnecessary

Correct Answer: A. Common suppliers, funding and infrastructure can transmit shocks across borrowers