Geopolitical Risk Modelling for Banks

Featured answer: Banks should model geopolitical risk as a cross-cutting driver transmitted through borrowers, markets, funding, payments, sanctions, technology and operations. The strongest framework combines exposure mapping, narrative scenarios, quantitative satellite models, reverse stress testing, timed capital and liquidity impacts, early-warning indicators and executable management actions.

Geopolitical risk has moved from the context section of risk reports into the core supervisory agenda. ECB Banking Supervision strengthened its focus in the 2025–27 priorities and placed resilience to geopolitical and macro-financial uncertainty at the centre of its 2026–28 priorities. In December 2025 it announced a geopolitical reverse stress test involving 110 directly supervised banks.

The implication is not that banks must forecast wars. It is that they must understand how sanctions, conflict, tariffs, trade fragmentation, sovereign stress and energy shocks transmit to their own balance sheets and operations.

Key takeaways

Treat geopolitics as a driver across established risk types, not an isolated score.
Map direct, indirect and common-factor exposure.
Use scenarios and reverse stress tests where probabilities lack credibility.
Integrate capital, liquidity and operational timing.
Pre-agree indicators, escalation and feasible management actions.

Transmission channels

Credit transmission includes borrower revenue loss, input-cost inflation, supply interruption, asset impairment and sovereign deterioration. Market transmission includes commodity, foreign-exchange, rates and spread shocks. Liquidity effects can arise through deposit behaviour, collateral calls, reduced market access and trapped currency. Operational effects include cyberattack, unavailable staff, damaged infrastructure and payment disruption. Compliance channels include sanctions screening, asset freezes and rapidly changing legal restrictions.

These channels interact. An energy shock can weaken borrowers, raise inflation, change rates, reduce collateral values and increase funding needs. A sanctions event can create operational backlog at the same time that customers demand liquidity.

Exposure mapping

Country of incorporation is inadequate. Banks should map revenue, suppliers, shipping routes, currencies, collateral, guarantors, beneficial owners, payment corridors and critical service providers. Indirect exposure may dominate direct lending. Sector concentration and common sponsors can create hidden correlation.

A network representation can connect obligors, countries, sectors and dependencies. Centrality measures help identify nodes whose disruption affects many exposures, but model output should be interpreted with data-quality limitations. The aim is prioritisation, not a single geopolitical score.

Scenario architecture

A scenario should define the event, duration, affected jurisdictions, policy response and market reaction. Risk drivers might include oil and gas prices, FX, sovereign and corporate spreads, rates, equity values, trade volumes, deposit outflow and operational downtime. Satellite models translate drivers into PD, LGD, migration, P&L, collateral, net interest income and liquidity.

For credit migration, a stressed transition matrix P* can be generated by shifting baseline transition probabilities as a function of scenario factors. One simple structure is logit(P*ᵢⱼ) = logit(Pᵢⱼ) + βᵢⱼX, with constraints so rows remain valid probabilities. The coefficients should be calibrated where evidence exists and supplemented by expert overlays with transparent rationale.

Reverse stress testing

Reverse stress testing asks which combination of shocks produces a defined adverse outcome. The ECB stated that its 2026 exercise requires banks to define a geopolitical scenario leading to at least 300 basis points of CET1 depletion. This is a supervisory exercise specification, not a universal minimum stress or capital requirement.

Internally, banks can define failure through capital, liquidity, viability, risk appetite or loss of a critical service. Optimisation or structured search then identifies combinations of credit migration, market moves, deposit outflow and operational loss capable of reaching the boundary. Management should review whether the resulting narratives are coherent and whether risk mitigants would still function.

Capital and liquidity on one timeline

Aggregate end-state loss can hide the sequence that makes a scenario dangerous. Collateral calls and outflows may arrive before asset sales, hedges or capital actions can respond. Banks should project daily liquidity in the acute phase and monthly or quarterly capital over the scenario horizon, linking common drivers and management actions.

Actions need execution constraints. Asset sales generate cash but may crystallise losses. Hedging may protect capital while increasing collateral needs. Credit tightening preserves future quality but weakens income and customer relationships. The analysis should show these trade-offs.

Hypothetical practical example

Consider a hypothetical bank with lending to exporters, commodity hedges and concentrated corporate deposits. A trade restriction reduces exporter revenue, widens sovereign and corporate spreads and weakens the local currency. Credit migration raises expected loss and RWA; hedge variation margin consumes cash; corporate deposits leave as clients fund working capital. Even if the capital ratio remains above requirement at year end, an intramonth collateral and deposit peak may breach liquidity appetite. The integrated timeline changes the management response.

Governance and early warning

Indicators should connect to exposure and action: sanctions changes, shipping disruption, energy prices, sovereign spreads, cross-border payments, deposit concentration, margin requirements and cyber alerts. Thresholds should specify escalation, not automatic conclusions. Scenario ownership must span risk, treasury, compliance, operations and the relevant businesses.

Regulatory perspective

ECB priorities and published frameworks are supervisory expectations and work programmes, not legislation creating a separate Pillar 1 geopolitical capital charge. The EBA's June 2026 Risk Assessment Report provides current sector evidence. Banks should distinguish these sources from binding CRR requirements and their own internal risk-appetite choices.

What CROs should do now

Build a cross-risk geopolitical exposure map beyond legal domicile.
Select scenarios from material transmission channels, not headlines alone.
Connect credit, market, liquidity and operational models on one timeline.
Run reverse stress tests against capital, liquidity and viability boundaries.
Challenge whether management actions remain executable under sanctions and market stress.
Define indicators, owners and escalation before the event.
Report uncertainties and data gaps explicitly.

Conclusion

Geopolitical risk becomes manageable when it is translated from narrative into exposures, transmission channels, scenarios and decisions. Precision about distant events is often illusory; precision about the bank's dependencies, thresholds and response capacity is achievable. That is the modelling standard boards and supervisors increasingly require.

References

Frequently Asked Questions

Is geopolitical risk a separate banking risk category?

Usually it is best treated as a cross-cutting driver that transmits through credit, market, liquidity, operational, compliance, business-model and capital risks.

How can banks quantify geopolitical risk?

Banks can map exposures to transmission channels, design severe but plausible and reverse-stress scenarios, translate shocks through satellite models and aggregate the timed effects on earnings, liquidity and capital.

What did the ECB focus on in its geopolitical reverse stress test?

The ECB announced in December 2025 that 110 directly supervised banks would define scenarios producing a pre-determined outcome of at least 300 basis points of CET1 depletion, with aggregate results expected in summer 2026.

Should banks assign probabilities to wars?

Not necessarily. For remote and deeply uncertain events, scenario severity, exposure sensitivity, preparedness and decision triggers are often more defensible than precise event probabilities.