Forecasting, Scenarios and Early-Warning Systems

Introduction

Geopolitical risk assessment is not an exercise in predicting a single future. Political systems are open, adaptive and contested: governments change policy, firms alter behaviour, societies mobilise, and shocks propagate through financial, technological and supply-chain networks. Consequently, the central task of the geopolitical analyst is to construct disciplined judgments about plausible futures, identify the conditions that would make them more or less likely, and enable decision-makers to act before risk becomes crisis.

Forecasting, scenario planning and early-warning systems provide complementary capabilities. Forecasting estimates the likelihood or timing of specified developments. Scenario planning explores alternative combinations of uncertainty without claiming that one outcome is certain. Early-warning systems monitor indicators and signposts that may reveal whether the operating environment is moving toward a dangerous or advantageous state. Used together, these methods support strategic resilience rather than false precision.

For international businesses, the relevant questions include: Which assumptions underpin our market-entry decision? What events could invalidate them? Which signals would indicate escalating political risk? How quickly can we shift suppliers, capital or personnel? What actions are reversible now but expensive later? These questions place analysis within a decision context.

Learning Objectives

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

1. Distinguish forecasting, scenario planning, horizon scanning and early warning.
2. Conduct a structured horizon scan and identify weak signals.
3. Formulate and test key assumptions underlying geopolitical judgments.
4. Apply structured analytic techniques, trend-impact analysis and alternative-futures methods.
5. Design signposts, thresholds and escalation protocols for an early-warning system.
6. Use stress testing and scenario analysis to assess strategic exposure.
7. Evaluate the limitations, ethical risks and governance requirements of geopolitical foresight.

1. The Logic of Foresight Under Geopolitical Uncertainty

A useful distinction is between risk, uncertainty and ambiguity. Risk exists when relevant outcomes and approximate probabilities can be specified. Uncertainty exists when probabilities are contested or poorly known. Ambiguity exists when actors disagree about the meaning of events, objectives or categories themselves. Geopolitical environments commonly contain all three.

A forecast is a conditional proposition, not a prophecy. It should specify:

• the event being assessed;
• the time horizon;
• the geographic and institutional scope;
• the probability or confidence level;
• the evidence and assumptions supporting the judgment; and
• the conditions that would change the judgment.

A simple probability statement can be represented as:

where P is the assessed probability of event E, conditional on information I, assumptions A, and time horizon T. This notation does not imply that geopolitical probabilities are mechanically objective. It makes the basis of the judgment explicit and therefore open to challenge.

A decision-maker is usually concerned not only with probability but also with consequence. A simplified expected-loss expression is:

where EL is expected loss, P(E) is the probability of the event, and L(E) is the loss if it occurs. The expression assumes that loss can be represented on a meaningful common scale and that the event definition is sufficiently precise. It is not a substitute for distributional, strategic or ethical analysis. A low-probability event with catastrophic consequences may deserve attention even when its expected loss is difficult to estimate.

Forecasting and scenario planning compared

2. Horizon Scanning and Environmental Intelligence

Horizon scanning is the systematic examination of emerging trends, events, technologies, actors and weak signals that may affect an organisation. It should extend beyond conventional political reporting. A complete scan considers the interaction of geopolitical, economic, social, technological, legal, environmental and security drivers.

2.1 Scope and framing

Scanning begins with a decision question. “What could threaten our global operations over the next decade?” is too broad for routine monitoring, whereas “Which developments could disrupt our semiconductor inputs from East Asia within 24 months?” establishes a useful boundary.

The analyst should define:

• domain: supply chains, market access, sanctions, security or regulation;
• time horizon: immediate, medium-term or long-term;
• unit of analysis: country, corridor, sector, firm, alliance or network;
• impact dimensions: cost, continuity, safety, reputation, compliance and strategic position; and
• decision owner: the executive or team able to act on the result.

2.2 Sources and collection

Sources may include diplomatic statements, legislation, parliamentary records, regulatory notices, company disclosures, trade data, shipping data, satellite imagery, local-language media, academic research, expert interviews and civil-society reporting. Source diversity is important because governments, firms and communities observe different parts of reality.

Collection should distinguish facts, interpretations and assumptions. A government decree is a fact about an official act; the expectation that it will be enforced uniformly is an interpretation; the belief that enforcement will remain limited is an assumption. Mixing these categories creates overconfidence.

2.3 Weak signals

A weak signal is an early, incomplete or ambiguous indication of a potentially significant change. Examples include unusual legislative language, a new military logistics pattern, changes in procurement contracts, coordinated online narratives, elite factional rhetoric or small changes in customs enforcement.

Weak signals should not be treated as proof. Their value lies in prompting collection and analysis. A practical signal record includes:

A signal-to-noise problem arises when organisations collect more information than they can evaluate. Scanning therefore requires prioritisation, not unlimited accumulation.

3. Key-Assumption Checks and Analytic Discipline

Many strategic failures result not from a lack of information but from unexamined assumptions. A key-assumption check identifies propositions on which a judgment depends and asks whether they remain valid.

For each major assessment, analysts should list:

1. Core judgment: what is believed to be true.
2. Supporting assumptions: conditions required for the judgment to hold.
3. Evidence: observations supporting or challenging each assumption.
4. Confidence: high, medium or low, with reasons.
5. Disconfirming evidence: what would show the assumption is weakening.
6. Owner and review date: who will revisit it and when.

For example, a firm may assume that a host government will preserve investor protections because it seeks foreign capital. This assumption may fail if domestic political survival becomes more important than investment attraction, if a security crisis changes priorities, or if alternative sources of finance emerge.

Competing hypotheses

The analysis of competing hypotheses (ACH) reduces confirmation bias by comparing evidence against several explanations rather than asking only whether evidence supports a preferred view. Suppose a sudden border closure is observed. Hypotheses might include:

• H1: preparation for sustained military escalation;
• H2: temporary public-health or security precaution;
• H3: domestic political signalling;
• H4: bureaucratic failure or local initiative.

Evidence is assessed for consistency and inconsistency with each hypothesis. A diagnostic observation is more valuable when it is expected under one hypothesis but not under others. ACH does not mechanically determine the truth; it disciplines comparison and exposes residual uncertainty.

Indicators and warnings of bias

Analysts should actively check for anchoring, availability bias, mirror-imaging, groupthink, motivated reasoning and the tendency to interpret ambiguous events through a single narrative. Red-team review, independent estimates, premortems and devil’s-advocate analysis are useful safeguards.

4. Structured Analytic Techniques

Structured analytic techniques (SATs) are explicit procedures that make reasoning inspectable. They are particularly useful when evidence is incomplete and disagreement is legitimate.

4.1 Key techniques

• Decomposition breaks a complex question into smaller propositions.
• Indication and warning analysis identifies observable developments associated with a scenario.
• Analysis of competing hypotheses compares alternative explanations.
• Red teaming constructs the strongest case against a prevailing judgment.
• Premortem analysis asks why a decision assumed to have failed might have failed.
• Cross-impact analysis examines how developments reinforce or weaken one another.
• Delphi methods elicit and refine expert judgments anonymously across rounds.
• Morphological analysis combines dimensions of a problem to generate plausible configurations.
• Argument mapping displays claims, evidence, warrants and rebuttals.

No technique eliminates judgment. The purpose is to reduce hidden reasoning and make disagreement productive.

4.2 Qualitative probability language

Organisations should define probability terms. For example, “unlikely” might mean approximately 10–30 percent within a stated period, while “likely” might mean 60–80 percent. These ranges are illustrative and must be calibrated to the organisation’s practice. Without definitions, different analysts may attach radically different meanings to the same word.

Where historical data exist, calibration can be evaluated using scoring rules. The Brier score for binary forecasts is:

where N is the number of forecasts, pᵢ is the forecast probability for case i, and oᵢ is 1 if the event occurred and 0 otherwise. Lower scores indicate better probabilistic accuracy. The measure assumes comparable forecasts, well-defined outcomes and reliable records. It evaluates calibration and resolution imperfectly; it does not measure the quality of strategic advice by itself.

5. Trend-Impact Analysis

Trend-impact analysis combines a baseline trend with possible disruptive events. The analyst first identifies a trend, such as increasing regulatory scrutiny of critical minerals, then asks how discrete events might alter its trajectory.

A useful process is:

1. Establish the baseline trend and its drivers.
2. Identify events that could accelerate, interrupt or reverse it.
3. Estimate timing, direction and magnitude of impact.
4. Identify affected stakeholders and feedback effects.
5. Specify indicators that would reveal which path is emerging.
6. Translate the result into decisions and contingencies.

A trend should not be extrapolated mechanically. Political interventions can create discontinuities, and trends that appear independent may interact. For example, export controls, shipping insecurity and currency restrictions may jointly produce a larger effect than the sum of their separate impacts.

A simple impact-priority score can help organise attention:

where S is severity, E is exposure and V is vulnerability, each scored on an explicitly defined ordinal scale. This is a prioritisation heuristic, not a probability model. Multiplication is appropriate only when the organisation accepts the scale properties and understands that a zero or low score can dominate the result.

6. Scenario Planning and Alternative Futures

Scenario planning is a method for exploring several internally coherent futures. It is most valuable when uncertainty is high, the consequences of strategic choices are substantial and conventional forecasting is unreliable.

6.1 Building scenarios

A robust scenario process normally includes:

1. Focal issue: the strategic decision under examination.
2. Driving forces: political, economic, social, technological, environmental and legal factors.
3. Critical uncertainties: forces whose direction is both consequential and difficult to predict.
4. Scenario logic: combinations of uncertainties that are plausible and distinct.
5. Narratives: causal accounts explaining how the future develops.
6. Implications: effects on markets, assets, stakeholders and operations.
7. Signposts: observations that indicate movement toward a scenario.
8. Strategic responses: robust, contingent and scenario-specific actions.

Scenarios should be challenging but not fantastical. They should be causally coherent, decision-relevant and sufficiently different to expose vulnerabilities.

6.2 Two-axis and morphological approaches

The familiar two-axis matrix is useful when two uncertainties dominate, such as regional security cooperation and openness to trade. It can become misleading when it compresses a multidimensional system into four boxes. Morphological analysis is an alternative: define dimensions such as alliance structure, regulatory regime, logistics security and domestic stability, then combine values while removing logically inconsistent combinations.

6.3 Scenario quality

A scenario is not improved by dramatic prose alone. Quality tests include:

• plausibility: could the causal sequence occur?
• differentiation: does it reveal something not visible in other scenarios?
• completeness: are second-order effects considered?
• traceability: can assumptions and evidence be identified?
• decision value: would it change a choice, trigger or investment?

7. Signposts, Thresholds and Early-Warning Systems

An early-warning system converts analysis into repeated observation and action. It is a socio-technical system, not merely a dashboard. It requires data, interpretation, governance, communication and authority to respond.

7.1 Designing signposts

A signpost is an observable development associated with a scenario or risk pathway. Good signposts are:

• observable and measurable where possible;
• sufficiently specific to reduce ambiguity;
• linked to a causal mechanism;
• timely relative to the decision window;
• difficult to manipulate or supported by corroboration; and
• assigned to an owner.

Examples include a draft law entering committee, a defined change in customs clearance time, cancellation of military hotlines, unusual capital controls, or repeated attacks on a transport corridor.

7.2 Thresholds and escalation

Thresholds translate observations into categories such as normal, watch, alert and crisis. A threshold should specify the indicator, reference period, trigger condition, confidence requirement, notification path and permitted action. Multiple indicators are often preferable to one dramatic but noisy measure.

A simple alert rule can be represented as:

where zᵢ is the standardised status of indicator i, wᵢ is its assigned weight, K is the number of indicators and τ is the alert threshold. This rule assumes that indicators can be placed on a comparable scale and that additive weighting is a reasonable approximation. It should be tested against false positives and missed warnings; it should not be presented as an objective truth.

7.3 Early-warning architecture

An effective architecture includes:

1. Collection layer: internal and external data.
2. Validation layer: source reliability, duplication and anomaly checks.
3. Analytic layer: trend analysis, expert judgment and scenario comparison.
4. Decision layer: thresholds, escalation and response options.
5. Learning layer: after-action review and indicator revision.

The system must distinguish warning time from decision time. A warning is useful only if the organisation can act before the risk materialises. A ten-day warning is inadequate if relocating equipment takes six weeks.

8. Stress Testing and Strategic Resilience

Stress testing asks how a strategy, portfolio, supply chain or operating model performs under severe but plausible conditions. It should test both direct effects and cascading consequences.

A geopolitical stress test may examine:

• sudden sanctions or export controls;
• border closure or maritime disruption;
• capital repatriation restrictions;
• cyberattack on logistics or payments;
• political violence and evacuation constraints;
• expropriation or contract repudiation; and
• simultaneous shocks in correlated locations.

The analysis should identify vulnerabilities, not merely produce a dramatic scenario. Important outputs include time to impact, time to detect, time to decide, time to recover, cash-flow exposure, legal constraints and feasible mitigations.

Correlation and concentration matter. A company may believe it has diversified suppliers while all suppliers depend on the same port, currency, cloud provider or political alliance. Scenario testing should therefore map hidden common dependencies.

9. Applied Case: Protecting a Battery-Materials Supply Chain

The following is a fictional illustrative case designed for teaching purposes.

Background

Aurora Mobility, a European electric-vehicle manufacturer, sources refined battery material through suppliers in three countries. Its strategy assumes continued maritime access, stable export licensing and a cooperative relationship between the principal supplier state and the firm’s home government.

Problem

Over an 18-month horizon, analysts identify rising resource nationalism, contested elections in the supplier state and increasing naval activity near the principal shipping route. Senior management initially treats the developments as unrelated. The procurement team proposes increasing inventory, while the finance team warns that excessive stock would tie up capital.

Analysis

The risk team frames the focal question as: “Could Aurora maintain production for 90 days if access to the main supply corridor deteriorated?” A horizon scan identifies the following signals:

• draft legislation requiring domestic processing;
• delays in export-license renewals;
• public criticism of foreign-owned extraction assets;
• insurance premiums rising for the shipping corridor; and
• a new regional security agreement excluding Aurora’s home country.

A key-assumption check reveals that the company has treated “three suppliers” as diversification. In fact, all three depend on the same refining technology and two depend on the same port. The assumption of geographic diversification is therefore overstated.

The team constructs three scenarios:

Signposts are assigned to owners. Legal monitors licensing rules; logistics monitors transit time and insurance; government-affairs staff monitor political rhetoric and parliamentary action; procurement monitors supplier inventories and alternative refining capacity.

Approach and solution

Aurora adopts a staged response rather than an irreversible exit. It qualifies a fourth supplier in another region, purchases limited strategic inventory, negotiates substitution rights in contracts, and establishes a monthly executive review. A threshold is set for activating emergency shipping and inventory release when two independent indicators—such as licence delay and transit disruption—remain above defined levels for two consecutive reporting periods.

A premortem identifies likely failure modes: the alternative supplier lacks certification, inventory is concentrated in one warehouse, and legal restrictions prevent emergency re-export. These findings lead to distributed storage, earlier technical qualification and legal review of contingency contracts.

Outcome

Nine months later, the supplier state introduces a domestic-processing requirement. Because Aurora has already qualified an alternative source and negotiated a transitional licence, production falls by 8 percent for one quarter rather than stopping. The company still incurs higher costs, but its response is faster and less disruptive than those of competitors.

Learning points

The case illustrates that early warning is not simply the discovery of bad news. It is the conversion of signals into assumptions, thresholds, decisions and organisational readiness. It also demonstrates that resilience often depends on “no-regret” actions—such as qualification, legal review and information sharing—that are useful across several scenarios.

10. Governance, Ethics and the Use of AI

Forecasting systems can influence investment, employment, market access and the treatment of communities. Governance is therefore essential. Organisations should document data provenance, analytic assumptions, confidence levels, review dates and decision rights. Sensitive intelligence must be handled lawfully, with attention to privacy, human rights, sanctions compliance and export-control rules.

AI can assist with multilingual monitoring, entity resolution, anomaly detection, document classification and rapid summarisation. It can also amplify bias, reproduce unreliable sources, generate fabricated citations and conceal uncertainty behind fluent language. AI outputs should be treated as collection or analytic support, not as autonomous judgment. Human analysts must validate sources, inspect context, test alternative explanations and preserve an audit trail.

Particular care is needed when predictive systems classify political groups, communities or individuals. Apparent risk indicators may reflect surveillance bias rather than actual threat. Ethical practice requires proportionality, contestability, data minimisation and separation between strategic assessment and coercive action.

11. Limitations and Best Practices

No foresight system can remove surprise. Structural breaks, deliberate deception, cascading failures and low-frequency events may evade indicators. Early warnings can also fail because decision-makers discount inconvenient information, disagree about thresholds or lack resources to act.

Best practice includes:

• begin with a decision, not a data platform;
• separate observation, inference and assumption;
• use multiple methods and independent reviewers;
• maintain alternative hypotheses;
• specify time horizons and confidence levels;
• design indicators around causal mechanisms;
• distinguish warning time from response time;
• test for false alarms and missed events;
• update scenarios as conditions change;
• preserve dissent and red-team findings;
• link thresholds to pre-authorised actions; and
• conduct post-event reviews without hindsight bias.

The most mature organisations treat foresight as a continuous learning cycle: scan, interpret, challenge, decide, monitor, act and revise.

Chapter Summary

Forecasting, scenario planning and early-warning systems are complementary methods for managing geopolitical uncertainty. Forecasts provide conditional judgments about likely developments; horizon scanning identifies emerging issues and weak signals; scenario planning explores coherent alternative futures; stress testing examines resilience under severe conditions; and early-warning systems connect indicators to thresholds and action.

Effective practice depends on disciplined framing, explicit assumptions, competing hypotheses, structured analytic techniques and clear probability language. Signposts must be observable, causally relevant and linked to decision windows. Quantitative tools can support prioritisation and calibration, but they do not eliminate judgment or make uncertain political events mechanically predictable.

For international businesses, the value of foresight lies in improving preparedness and strategic choice. Robust systems expose hidden dependencies, identify no-regret actions, clarify escalation responsibilities and reduce the time between detection and response. Their legitimacy depends on transparent governance, ethical data use, human oversight and continual learning.

Key Terms

Horizon scanning: Systematic examination of emerging developments that may affect future decisions.

Weak signal: An early, ambiguous indication of a potentially important change.

Forecast: A conditional judgment about the likelihood or timing of an event.

Scenario planning: Exploration of several plausible, internally coherent futures.

Alternative futures: Distinct possible configurations of political, economic and social conditions.

Key assumption: A proposition that must remain valid for an assessment or strategy to hold.

Competing hypotheses: Alternative explanations evaluated against the same evidence.

Structured analytic technique: An explicit procedure designed to discipline reasoning.

Trend-impact analysis: Assessment of how discrete events may alter an established trend.

Signpost: An observable development associated with a scenario or risk pathway.

Indicator: A measure or observation used to monitor change.

Threshold: A defined condition that triggers an alert or action.

Early-warning system: An organised process linking collection, analysis, alerts and response.

Stress testing: Examination of performance under severe but plausible conditions.

Premortem: An exercise that asks why an assumed future failure occurred.

Red teaming: Deliberate construction of a strong challenge to the prevailing judgment.

Calibration: The correspondence between stated probabilities and observed frequencies.

Alert fatigue: Reduced attention caused by excessive or unreliable warnings.

No-regret action: An action beneficial or defensible across several plausible futures.

False positive: An alert indicating danger when the relevant event does not occur.

Review Questions

Short Answer Questions

1. How does scenario planning differ from forecasting?
2. What is a weak signal, and why should it not be treated as proof?
3. What information should be included in a key-assumption check?
4. Define a signpost and explain its role in an early-warning system.
5. Why must warning time be distinguished from decision time?

Long Answer Questions

1. Explain how horizon scanning can be designed for an international business facing supply-chain disruption.
2. Discuss the value and limitations of the analysis of competing hypotheses in geopolitical assessment.
3. Describe a complete process for constructing and using alternative scenarios.
4. Explain how thresholds, escalation protocols and governance determine whether an early-warning system is useful.
5. Evaluate the ethical and methodological risks of using AI in geopolitical forecasting.

Analytical / Application Questions

1. A firm’s three suppliers depend on one port and one refining technology. Design a key-assumption check and identify at least five signposts that could reveal increasing geopolitical vulnerability.
2. Construct three scenarios for a company operating in a country experiencing resource nationalism, contested elections and rising external military pressure. For each scenario, specify one robust action and three indicators.
3. Design a stress test for a 60-day closure of a critical maritime corridor. Explain the data required, the assumptions to challenge, the thresholds to use and the decisions that should be pre-authorised.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes scenario planning from forecasting?

A. Scenario planning assigns a single precise probability to one future.

B. Scenario planning explores several plausible and coherent futures.

C. Scenario planning excludes uncertainty from strategic analysis.

D. Scenario planning is limited to historical data analysis.

Correct Answer: B. Scenario planning explores several plausible and coherent futures.

Question 2

What is the primary purpose of a key-assumption check?

A. To eliminate the need for expert judgment.

B. To identify propositions on which an assessment depends and test whether they remain valid.

C. To replace all qualitative evidence with numerical data.

D. To ensure that every forecast is optimistic.

Correct Answer: B. To identify propositions on which an assessment depends and test whether they remain valid.

Question 3

Which is the best example of a signpost?

A. A general belief that regional tensions are increasing.

B. A company’s preference for diversified suppliers.

C. A new export-licensing rule entering parliamentary committee.

D. A scenario narrative with no observable indicators.

Correct Answer: C. A new export-licensing rule entering parliamentary committee.

Question 4

In the binary Brier score, what value does the outcome variable take when the forecast event occurs?

A. −1

B. 0

C. 0.5

D. 1

Correct Answer: D. 1

Question 5

Which structured analytic technique explicitly compares alternative explanations against the same evidence?

A. Analysis of competing hypotheses

B. Trend extrapolation

C. Inventory accounting

D. Stakeholder mapping only

Correct Answer: A. Analysis of competing hypotheses

Question 6

Why can a two-axis scenario matrix be problematic?

A. It always produces too many scenarios.

B. It may compress a multidimensional system into only two uncertainties.

C. It cannot contain narratives.

D. It prevents any discussion of uncertainty.

Correct Answer: B. It may compress a multidimensional system into only two uncertainties.

Question 7

Which feature most improves the usefulness of an early-warning indicator?

A. It is highly general and difficult to disconfirm.

B. It is linked to a causal mechanism and a decision window.

C. It is based on a single unverified source.

D. It triggers alerts without assigning responsibility.

Correct Answer: B. It is linked to a causal mechanism and a decision window.

Question 8

If an event has probability 0.2 and loss of 50 units, what is its simplified expected loss?

A. 0.004 units

B. 10 units

C. 50.2 units

D. 250 units

Correct Answer: B. 10 units

Question 9

What is the main purpose of a premortem?

A. To assume success and avoid discussing failure.

B. To identify how a proposed strategy might fail before implementation.

C. To calculate the exact probability of war.

D. To remove dissent from the decision process.

Correct Answer: B. To identify how a proposed strategy might fail before implementation.

Question 10

Which is an appropriate governance principle for AI-supported geopolitical analysis?

A. Accept fluent AI output without source validation.

B. Allow the system to make high-impact decisions autonomously.

C. Maintain human oversight, source checking and an audit trail.

D. Use only one data source to avoid conflicting information.

Correct Answer: C. Maintain human oversight, source checking and an audit trail.