10. Decisions under Risk, Uncertainty and Information

Introduction and context

Managerial economics is concerned not only with choosing among alternatives but also with choosing when the consequences of alternatives are imperfectly known. A firm may decide whether to launch a product, enter a foreign market, acquire a technology, hold inventory, insure an asset, or invest in capacity without knowing future demand, prices, regulation, exchange rates, or competitor responses. In such settings, the quality of a decision depends as much on the treatment of information and uncertainty as on the projected financial return.

Three conditions should be distinguished. Under certainty, the relevant outcomes are known. Under risk, the possible outcomes are known and probabilities can be assigned, at least approximately. Under uncertainty, the outcomes or their probabilities are incomplete, disputed, or unstable. In practice, strategic decisions often involve all three conditions simultaneously. A company may know the engineering cost of a plant, estimate probabilities for alternative demand levels, and remain deeply uncertain about a new regulation or disruptive technology.

This chapter develops an integrated approach to such decisions. Expected value and expected utility provide formal foundations; Bayesian updating shows how beliefs should change when evidence arrives; decision trees clarify sequential choices; sensitivity analysis and simulation expose the conditions under which a recommendation changes; and value-of-information analysis establishes how much research is worth purchasing. Portfolio reasoning, real-options logic, scenario planning, resilience, and behavioral analysis extend these tools to complex organizational environments.

Learning objectives

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

1. Distinguish certainty, risk, ambiguity, and deep uncertainty.

2. Apply expected monetary value and expected utility to managerial choices.

3. Construct and interpret Bayesian updates and decision trees.

4. Use sensitivity analysis, scenario analysis, and simulation to test recommendations.

5. Calculate and interpret the value of perfect and sample information.

6. Explain diversification, correlation, insurance, and portfolio reasoning.

7. Analyze flexibility and timing through real-options logic.

8. Diagnose behavioral biases and governance failures in high-stakes decisions.

9. Integrate information, resilience, ethics, and scenario planning into executive decision processes.

The architecture of uncertain decisions

A decision problem contains at least five elements: the decision maker, available actions, uncertain states of the world, consequences, and information. States are not actions; they are conditions that occur outside the manager’s control. For example, a firm may choose whether to build a large or small plant, while demand may turn out to be high, medium, or low. The payoff associated with each action-state combination is represented in a payoff matrix.

Risk is often defined as measurable uncertainty. If probabilities are available, expected-value calculations can be performed. Uncertainty in the stronger sense—sometimes called ambiguity—arises when probabilities are unknown, contested, or generated from weak evidence. Knight’s distinction remains useful because risk permits conventional probabilistic analysis, whereas uncertainty requires judgment about models, assumptions, and the credibility of evidence.

A further category is deep uncertainty. It exists when decision makers do not agree on the relevant models, the probability distributions, the values attached to outcomes, or the causal relationships linking actions to results. Climate adaptation, geopolitical disruption, and the long-run effects of an unfamiliar technology often have this character. Deep uncertainty does not make analysis impossible; it changes the aim from predicting one future to identifying strategies that remain acceptable across many plausible futures.

The distinction between objective and subjective probability is also important. Objective probability may derive from repeated frequencies or well-established statistical mechanisms. Subjective probability expresses a rationally organized belief based on evidence, expert judgment, and prior knowledge. In business, most probabilities are subjective estimates. Their usefulness depends on transparency, calibration, independence of judgments, and willingness to revise beliefs.

Expected value, expected utility, and risk preferences

The expected monetary value (EMV) of action a is:

EMV(a) = Σ p(s)X(a,s)

where p(s) is the probability of state s and X(a,s) is the monetary payoff. The manager chooses the action with the highest EMV when monetary value is an adequate representation of objectives and the decision maker is risk neutral.

Consider a product launch requiring an investment of $2 million. If demand is high, the net payoff is $5 million; if demand is low, the net payoff is negative $1 million. Suppose high demand has probability 0.6 and low demand 0.4. The EMV is 0.6($5 million) + 0.4(-$1 million) = $2.6 million. If the alternative of not launching has a payoff of zero, EMV favors launch.

Expected value is not always an adequate decision rule. A risk-averse manager may prefer a lower expected payoff with less dispersion, particularly when losses threaten liquidity, employment, solvency, or organizational legitimacy. Expected utility addresses this issue by transforming monetary outcomes through a utility function:

EU(a) = Σ p(s)u[X(a,s)]

For a risk-averse individual, the utility function is concave: each additional dollar adds less utility than the previous dollar. A common form is u(x) = ln(x), although the appropriate function depends on wealth, loss exposure, and the decision context. Risk aversion is not irrationality. It reflects diminishing marginal utility and the possibility that catastrophic losses have consequences beyond their accounting value.

The certainty equivalent is the certain amount that gives the same utility as a risky prospect. The risk premium is the difference between expected monetary value and the certainty equivalent. These concepts help managers explain why a firm may pay for insurance, accept a lower expected return for greater stability, or maintain redundant capacity.

Risk attitudes can vary by decision. A financially secure corporation may tolerate volatility in a research portfolio while adopting a conservative approach to workplace safety or debt refinancing. Thus, analysts should not classify an entire organization as simply risk seeking or risk averse without examining the relevant objective, time horizon, and constraints.

Bayesian learning and the economics of information

Bayesian reasoning provides a disciplined method for revising beliefs. Let H be a hypothesis, such as “the market is sufficiently attractive for a successful launch,” and let E be new evidence, such as favorable test-market results. Bayes’ rule is:

P(H|E) = [P(E|H)P(H)] / P(E)

The prior probability P(H) represents belief before observing the evidence. The likelihood P(E|H) assesses how probable the evidence would be if the hypothesis were true. The posterior probability P(H|E) is the revised belief.

Suppose the prior probability of strong demand is 0.4. A research report is positive in 80 percent of strong-demand cases and in 30 percent of weak-demand cases. The probability of a positive report is (0.8 × 0.4) + (0.3 × 0.6) = 0.5. The posterior probability of strong demand after a positive report is (0.8 × 0.4)/0.5 = 0.64. The report raises confidence, but does not establish certainty.

Bayesian analysis is especially valuable because it makes assumptions visible. A weak prior, correlated evidence, selective sampling, or an overestimated likelihood ratio can produce misleading posterior beliefs. Managers should ask who generated the evidence, whether the sample represents the target population, whether the information was independently obtained, and whether the measurement process itself changes behavior.

The value of information is the improvement in expected decision quality made possible by information, less the cost of obtaining it. The expected value of perfect information (EVPI) is the maximum a decision maker should pay to know the future state before choosing an action. In a payoff matrix, EVPI equals the expected payoff from selecting the best action separately in each state minus the maximum EMV available without that information. Perfect information is a benchmark rather than a realistic purchase.

The expected value of sample information (EVSI) is usually lower than EVPI because a study or forecast is imperfect. Its calculation requires the possible research results, the probability of each result, the best action after each result, and the cost of the study. Information should be purchased only when its expected incremental value exceeds its cost, including delay, confidentiality risk, and managerial attention.

Decision trees and sequential choice

A decision tree represents choices and chance events in chronological order. Decision nodes are usually shown as squares; chance nodes as circles; terminal branches contain payoffs. The tree is solved by rollback: calculate expected values at chance nodes and select the best alternative at decision nodes.

Decision trees are particularly useful when decisions unfold in stages. A firm may conduct a pilot, observe demand, and then decide whether to scale. The pilot has value not merely because it generates revenue but because it changes the subsequent decision. A tree should therefore distinguish information-generating actions from operating actions.

A well-constructed tree requires careful attention to conditional probabilities. Probabilities after a favorable pilot should not be treated as the original probabilities. Nor should the analyst combine mutually exclusive branches incorrectly. Payoffs should include investment costs, opportunity costs, taxes, financing effects, abandonment value, and relevant externalities where appropriate.

Trees can become unwieldy when they contain many stages. In those circumstances, spreadsheet models, influence diagrams, dynamic programming, or simulation may be preferable. The visual tree remains valuable as a communication device because it makes timing, reversibility, and information arrival explicit.

Sensitivity analysis, scenarios, and simulation

A decision based on a single forecast can create false precision. Sensitivity analysis varies one assumption at a time—such as price, volume, cost, or discount rate—to determine how strongly the recommendation responds. A tornado diagram ranks variables by their effect on an outcome. Break-even analysis identifies the value at which the preferred action changes.

Scenario analysis varies several connected assumptions coherently. A “rapid adoption” scenario might combine high demand, favorable regulation, and falling unit cost; a “constrained transition” scenario might combine weak demand, supply disruption, and stricter compliance requirements. Scenarios are not probability-weighted forecasts unless explicit probabilities are assigned. Their primary purpose is to test strategic robustness and stimulate discussion about leading indicators.

Monte Carlo simulation assigns probability distributions to uncertain inputs and repeatedly samples from them to generate a distribution of outcomes. Instead of asking only for expected profit, managers can examine the probability of loss, the tenth percentile of cash flow, the distribution of net present value, and the variables contributing most to variance. Correlation must be modeled carefully: exchange rates, demand, and commodity prices may move together. Treating correlated inputs as independent can materially understate risk.

Simulation does not eliminate uncertainty. It formalizes assumptions. Poor distributions, omitted risks, flawed correlations, or spurious precision produce sophisticated-looking but unreliable results. Good practice includes documenting assumptions, testing alternative distributions, validating the model against historical or operational evidence, and presenting ranges rather than a single authoritative number.

Portfolio reasoning and diversification

Portfolio reasoning extends analysis from individual projects to combinations of exposures. The expected return of a portfolio is the weighted average of component returns, but portfolio risk depends importantly on covariance. Two projects with volatile returns may form a relatively stable portfolio if their outcomes are imperfectly correlated.

Diversification reduces idiosyncratic risk, such as the failure of one supplier or one product, but cannot eliminate systematic risk affecting all holdings. In corporate settings, diversification may involve product lines, geographic markets, technologies, suppliers, or contractual arrangements. The relevant question is not whether each component appears attractive in isolation but whether it improves the risk-return characteristics of the overall portfolio.

Correlation is therefore a strategic variable. A firm dependent on several suppliers that all obtain critical inputs from the same region has less genuine diversification than its supplier count suggests. Similarly, a conglomerate may be exposed to one common interest-rate, regulatory, or technology shock despite operating in different industries.

Insurance and hedging transform exposures rather than simply reducing them. Insurance transfers specified losses to an insurer in exchange for a premium; hedging uses financial or operational instruments to offset price, currency, or interest-rate movements. Both require analysis of basis risk, counterparty risk, moral hazard, liquidity, and the possibility that protection fails precisely during systemic stress.

Real options: flexibility as an economic asset

Traditional discounted cash-flow analysis can undervalue projects when management has the ability to respond to new information. A real option is the value of managerial flexibility embedded in a physical or strategic investment. Examples include the option to delay, expand, contract, abandon, switch inputs, convert production, or stage investment.

A staged investment resembles a call option: the firm pays an initial amount to preserve the right, but not the obligation, to make a larger investment after uncertainty is partially resolved. An abandonment option limits downside exposure. Excess capacity may have option value when demand is uncertain, while modular technology may permit inexpensive expansion if adoption accelerates.

Real-options reasoning should not be used as a license to attach speculative value to every flexibility claim. The option requires uncertainty, a meaningful response opportunity, and some degree of exclusivity or control. Flexibility also has costs: modular systems may be more expensive, delay may sacrifice first-mover advantages, and maintaining multiple suppliers may reduce scale economies. The managerial task is to compare the cost of flexibility with the value of improved future choices.

Ambiguity, behavioral bias, and organizational judgment

When probabilities are ambiguous, people often display ambiguity aversion: they prefer a known-risk gamble to one with an unknown distribution, even when the expected outcomes appear similar. In strategic decisions, ambiguity can also provoke delay, political bargaining, or arbitrary reliance on a familiar forecast.

Behavioral biases distort both probability assessment and the interpretation of information. Common problems include overconfidence, anchoring on an initial estimate, confirmation bias, availability bias, optimism bias, loss aversion, escalation of commitment, and the sunk-cost fallacy. Groupthink may suppress dissent, while authority bias causes subordinates to endorse a senior executive’s forecast without adequate challenge.

Controls should be designed into the process rather than relying on individual willpower. Useful practices include independent forecasts, premortem analysis, reference-class forecasting, calibrated probability estimates, base-rate comparisons, red-team review, structured dissent, and explicit separation of advocacy from analysis. Incentives should reward accurate learning rather than merely defending prior decisions. Post-decision review must distinguish poor reasoning from an unfavorable outcome caused by an unforeseeable event; otherwise managers may learn the wrong lesson.

Resilience and scenario planning in high-stakes decisions

Efficiency-oriented decisions often minimize slack, inventory, spare capacity, and redundancy. Resilience-oriented decisions ask whether the organization can absorb shocks, continue critical functions, and recover without unacceptable loss. Resilience may be created through buffers, modularity, multiple suppliers, cross-training, cybersecurity, liquidity, recovery agreements, and decentralized authority.

Scenario planning is appropriate when prediction is unreliable but strategic choices must still be made. A rigorous process identifies driving forces, critical uncertainties, plausible alternative worlds, implications for the organization, and signposts that indicate which world may be emerging. The goal is not to select the most attractive narrative. It is to test whether a strategy remains viable and to identify low-regret actions.

High-stakes decisions should therefore be evaluated on more than expected profit. Relevant criteria may include survival probability, service continuity, distributional effects, safety, legal exposure, reversibility, stakeholder trust, and the time required to recover. Robust decision making seeks actions that perform acceptably across a wide range of futures rather than actions that maximize a fragile estimate under one preferred forecast.

An illustrative case: Northstar Energy Storage

Background

Northstar Energy Storage, a fictional mid-sized manufacturer, is considering a $40 million investment in a plant producing industrial battery modules. Demand may grow rapidly as customers decarbonize their operations, remain moderate as adoption proceeds slowly, or fall if a competing chemistry becomes commercially viable. The investment would create attractive margins under rapid growth but expose the company to substantial fixed costs under weak demand.

Situation and problem

The board must choose among three alternatives: build a full-scale plant immediately, build a smaller modular plant, or wait two years while conducting pilots and monitoring technology. The immediate full-scale option has an estimated EMV of $12 million, while the modular option has an EMV of $9 million. Waiting has an estimated EMV of $7 million if management ignores flexibility. However, the modular plant can be expanded at a moderate cost, and waiting preserves the option to enter after observing customer adoption.

The commercial team favors immediate scale and presents a favorable demand forecast. The operations team warns that the firm’s two principal suppliers are exposed to the same regional disruption. Finance emphasizes that a severe downside outcome could breach debt covenants. The board also recognizes that the available market study was commissioned by a supplier seeking a long-term contract.

Analysis

The first step is to separate model risk from parameter risk. Parameter risk concerns the probabilities and cash-flow estimates within the chosen model. Model risk concerns whether the model adequately represents technological substitution, supply concentration, and financing constraints. A sensitivity analysis shows that the full-scale project is preferred only if the probability of rapid adoption exceeds 0.55 or if construction cost remains below a specified threshold. The modular design is less profitable in the best case but produces substantially lower losses in the weak-demand case.

Northstar then uses Bayesian updating. A six-month pilot with three customers will provide evidence about reliability and willingness to pay. The pilot is not perfectly predictive, but its results have different likelihoods under rapid and weak adoption. The expected value of the pilot exceeds its cost when the analysis includes the option to cancel or expand after observing results. A Monte Carlo model further reveals that the full-scale project has a small but material probability of a cash shortfall during a weak-demand and supply-disruption combination.

The board considers a resilience-adjusted strategy: construct a modular facility, qualify a geographically separate supplier, preserve expansion rights in the site contract, and define explicit scale-up triggers based on pilot retention, order backlog, and component prices. It also establishes a downside trigger for reducing production and a quarterly review of technology developments.

Outcome

Northstar chooses the modular strategy with a staged pilot. The decision sacrifices some expected upside but purchases flexibility, reduces exposure to a correlated supply shock, and protects financial resilience. The board documents the assumptions, the evidence that would change the decision, and the executive responsible for monitoring each signpost.

Learning points

The case illustrates that the highest-EMV alternative is not automatically the best choice. Information has value when it changes action; flexibility has value when future uncertainty is material; diversification must account for correlation; and resilience may justify a lower expected return when downside consequences are severe. It also demonstrates why governance should challenge both numbers and the structure of the model that produced them.

Ethical, governance, and professional considerations

Uncertainty creates ethical obligations because probabilistic language can conceal serious consequences. Managers should disclose material assumptions, distinguish facts from judgments, report downside distributions rather than only average outcomes, and avoid presenting a model as objective when it embeds contested values. Safety, privacy, environmental harm, employee welfare, and intergenerational effects may not be adequately represented by financial payoffs.

Governance systems should establish decision rights, escalation thresholds, documentation standards, model validation, audit trails, and independent review for high-impact decisions. Regulators and professional bodies may require stress testing, capital adequacy, risk disclosures, or controls over automated decisions. A model that systematically disadvantages a stakeholder group can be ethically unacceptable even if it improves aggregate expected profit.

Artificial intelligence and advanced analytics can improve forecasting, detect patterns, and run large simulations. They can also amplify historical bias, create opaque probabilities, encourage automation bias, and fail under distribution shift. Human oversight should focus on data provenance, model drift, explainability appropriate to the decision, adversarial testing, privacy, cybersecurity, and clear accountability. An AI-generated forecast is evidence for judgment, not a substitute for responsibility.

Common limitations and managerial mistakes

Several recurring errors undermine uncertain decision analysis:

• treating an estimate as a fact or a point forecast as a prediction;
• using probabilities without explaining their source;
• ignoring base rates and relying on vivid but unrepresentative evidence;
• counting information as valuable even when it cannot change the decision;
• overlooking correlation, tail risk, or liquidity constraints;
• confusing sunk costs with relevant future costs;
• double-counting flexibility or benefits already included in cash flows;
• using scenarios as rhetorical stories without signposts or strategic implications;
• optimizing expected value while ignoring catastrophic downside;
• failing to revisit assumptions as evidence accumulates.

The remedy is not to avoid quantitative analysis. It is to combine quantitative discipline with model criticism, structured judgment, stakeholder awareness, and continuous learning.

Best-practice decision protocol

For consequential decisions, managers can use the following sequence:

1. Define the decision, objective, constraints, timing, and unacceptable outcomes.

2. Identify actions, states, causal assumptions, and information arrival points.

3. Establish base rates and develop transparent probability ranges.

4. Calculate EMV and, where relevant, expected utility and certainty equivalents.

5. Map the problem with a decision tree or influence diagram.

6. Test assumptions through sensitivity analysis, scenarios, and simulation.

7. Assess the value of additional information and the cost of delay.

8. Examine diversification, correlation, hedging, resilience, and real options.

9. Conduct a premortem, independent challenge, and ethical impact review.

10. Record triggers, ownership, monitoring indicators, and rules for updating the decision.

This protocol is not a mechanical formula. Its purpose is to make reasoning visible, comparable, and revisable.

Current developments and emerging directions

Decision analysis is increasingly connected to real-time data, digital twins, cloud simulation, machine learning, and adaptive experimentation. Organizations can update demand distributions continuously, run stress tests against synthetic disruptions, and link operational indicators to predefined decision triggers. However, greater computational power can worsen false precision if model uncertainty is neglected.

A significant emerging direction is robust and distributionally robust decision making. Rather than assuming one probability distribution, these approaches evaluate performance across a family of plausible distributions. Another is adaptive strategy: managers make an initial choice while deliberately preserving the ability to learn and revise. Climate risk, geopolitical fragmentation, cyber threats, and technology transitions are encouraging firms to combine financial valuation with resilience metrics and transition pathways.

The central development is therefore not the replacement of managerial judgment by algorithms. It is the integration of evidence, models, flexibility, organizational learning, and governance into decisions that remain defensible when the future differs from the forecast.

Chapter Summary

Decisions under risk involve known or estimated probabilities; decisions under uncertainty and ambiguity involve incomplete knowledge about outcomes, probabilities, or models. Expected monetary value is useful for risk-neutral choices, while expected utility captures risk preferences through a utility function, certainty equivalent, and risk premium. Bayesian reasoning updates prior beliefs in response to evidence, and decision trees organize sequential choices and conditional probabilities.

Sensitivity analysis identifies critical assumptions, scenario planning explores coherent alternative futures, and simulation estimates distributions rather than single outcomes. The value of information determines whether research is worth its cost. Portfolio reasoning emphasizes covariance and genuine diversification, while real-options analysis values the flexibility to delay, expand, contract, abandon, or switch. Ambiguity, behavioral bias, model risk, and organizational politics require structured challenge and governance. Finally, resilience and robust decision making encourage strategies that remain acceptable across uncertain futures, with ethical accountability and transparent assumptions.

Key Terms

Risk: A situation in which possible outcomes and sufficiently credible probabilities can be specified.

Uncertainty: A situation in which outcomes, probabilities, or causal relationships are incompletely known.

Ambiguity: Uncertainty about the appropriate probability distribution or interpretation of evidence.

Deep uncertainty: A condition in which decision makers disagree about models, probabilities, values, or causal structure.

Expected monetary value: The probability-weighted average of monetary payoffs for an action.

Expected utility: The probability-weighted average of utilities associated with possible outcomes.

Risk aversion: A preference for a certain outcome over a risky outcome with the same expected monetary value.

Certainty equivalent: The certain amount that provides utility equal to that of a risky prospect.

Risk premium: The difference between expected value and certainty equivalent.

Bayesian updating: Revision of prior beliefs using new evidence and Bayes’ rule.

Decision tree: A chronological representation of decisions, chance events, information, and payoffs.

Expected value of perfect information: The maximum expected gain from knowing the future state before acting.

Expected value of sample information: The expected gain from imperfect information obtained through research or testing.

Sensitivity analysis: Examination of how a decision changes when one or more assumptions vary.

Monte Carlo simulation: Repeated random sampling from input distributions to produce an outcome distribution.

Covariance: A measure of how two uncertain returns move together.

Real option: The value of managerial flexibility embedded in an investment or operating decision.

Resilience: The capacity to absorb disruption, maintain critical functions, and recover effectively.

Reference-class forecasting: Estimating outcomes by comparing a proposal with evidence from similar completed cases.

Model risk: The possibility that the structure or assumptions of an analytical model are inappropriate.

Review Questions

Short Answer Questions

1. Distinguish risk, uncertainty, ambiguity, and deep uncertainty.

2. What is the difference between expected monetary value and expected utility?

3. Define certainty equivalent and risk premium.

4. Why is correlation important in portfolio reasoning?

5. What is the purpose of a premortem in decision analysis?

Long Answer Questions

1. Explain how Bayesian updating can improve a managerial decision, including the role of priors, likelihoods, and posterior probabilities.

2. Describe the construction and rollback solution of a decision tree for a sequential investment decision.

3. Compare sensitivity analysis, scenario analysis, and Monte Carlo simulation, including their appropriate uses and limitations.

4. Explain how real-options logic changes the evaluation of projects involving delay, expansion, or abandonment.

5. Discuss how behavioral biases and governance mechanisms affect high-stakes decisions under ambiguity.

Analytical / Application Questions

1. A project produces $8 million with probability 0.5, $2 million with probability 0.3, and a loss of $6 million with probability 0.2. Calculate its EMV and explain why a risk-averse firm might reject it despite a positive result.

2. A pilot study is positive with probability 0.75 when demand is strong and 0.25 when demand is weak. If the prior probability of strong demand is 0.4, calculate the posterior probability of strong demand after a positive pilot result.

3. Design a decision framework for a firm facing uncertain technology substitution. Specify the role of scenarios, information value, real options, resilience measures, monitoring indicators, and governance controls.

22. Multiple Choice Questions with Answers

Question 1

A manager uses expected monetary value as the sole decision rule. Which assumption most directly supports this approach?

A. The manager is risk neutral with respect to the relevant payoffs.

B. All future states are known with certainty.

C. The manager always prefers the least variable outcome.

D. Probabilities cannot be assigned to outcomes.

Correct Answer: A. The manager is risk neutral with respect to the relevant payoffs.

Question 2

A favorable test result is more likely when a market is attractive than when it is unattractive. Bayesian updating implies that the favorable result should generally:

A. Make the attractive-market hypothesis less likely.

B. Leave the prior probability unchanged.

C. Increase the posterior probability of an attractive market.

D. Prove that the market will be attractive.

Correct Answer: C. Increase the posterior probability of an attractive market.

Question 3

Which statement best describes the expected value of perfect information?

A. The cost of commissioning a market study.

B. The maximum expected gain from knowing the state before choosing an action.

C. The expected payoff of the most profitable action in one selected state.

D. The difference between a project’s revenue and its accounting cost.

Correct Answer: B. The maximum expected gain from knowing the state before choosing an action.

Question 4

Why can a modular investment be preferable to a full-scale investment even when its immediate EMV is lower?

A. Modularity eliminates all uncertainty.

B. Modularity may preserve valuable expansion and abandonment flexibility.

C. Modularity guarantees higher demand.

D. Modularity removes the need for monitoring.

Correct Answer: B. Modularity may preserve valuable expansion and abandonment flexibility.

Question 5

In a Monte Carlo simulation, treating strongly correlated input risks as independent is likely to:

A. Make the model more realistic automatically.

B. Eliminate model risk.

C. Potentially understate joint downside exposure.

D. Guarantee a lower expected value.

Correct Answer: C. Potentially understate joint downside exposure.

Question 6

Which action is an example of genuine diversification?

A. Using five suppliers that all depend on one vulnerable region.

B. Holding several products exposed to the same single customer.

C. Combining investments whose returns are imperfectly correlated.

D. Increasing exposure to one highly profitable asset.

Correct Answer: C. Combining investments whose returns are imperfectly correlated.

Question 7

A decision maker continues funding a failing project primarily because substantial funds have already been spent. This is best described as:

A. The sunk-cost fallacy.

B. Bayesian calibration.

C. Risk pooling.

D. Reference-class forecasting.

Correct Answer: A. The sunk-cost fallacy.

Question 8

What is the principal purpose of scenario planning under deep uncertainty?

A. To identify one certain forecast.

B. To replace analysis with optimistic narratives.

C. To test strategies across plausible futures and identify signposts.

D. To eliminate the need for managerial judgment.

Correct Answer: C. To test strategies across plausible futures and identify signposts.

Question 9

Which feature most clearly creates a real option?

A. An irreversible commitment with no future choice.

B. The ability to delay or expand an investment after learning more.

C. A guaranteed return stated in a contract.

D. A historical cost that cannot be recovered.

Correct Answer: B. The ability to delay or expand an investment after learning more.

Question 10

Which governance practice most directly counters confirmation bias in a high-stakes decision?

A. Allowing only the project sponsor to review the forecast.

B. Using the initial forecast as the permanent baseline.

C. Commissioning independent challenge and structured red-team analysis.

D. Removing downside scenarios from executive reports.

Correct Answer: C. Commissioning independent challenge and structured red-team analysis.