Strategic Interaction, Information and Game Theory

Introduction

Managerial decisions are often interdependent. A firm’s pricing decision affects competitors; a supplier’s investment depends on the buyer’s future conduct; an employee’s effort depends on monitoring and rewards; and a negotiator’s offer depends on beliefs about the other party’s alternatives. In such settings, the value of an action depends not only on the decision-maker’s own choice but also on the choices, information and expectations of others.

Game theory provides a disciplined language for analysing these situations. It does not predict behaviour mechanically. Rather, it identifies strategically stable outcomes, clarifies assumptions, exposes threats and commitments, and helps managers design rules, contracts and communication strategies. Information economics complements game theory by analysing what happens when one party knows more than another.

This chapter applies strategic interaction to competitive strategy, negotiations, auctions and organisational governance. It focuses on dominant strategies, Nash equilibrium, sequential and repeated games, commitment, principal–agent relationships, moral hazard, adverse selection and signalling.

Learning Objectives

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

1. Represent a strategic situation using players, actions, information and payoffs.
2. Identify dominant strategies and calculate Nash equilibrium.
3. Analyse sequential games using backward induction and distinguish credible from non-credible threats.
4. Explain how repetition, reputation and commitment affect strategic outcomes.
5. Apply auction theory to procurement and sales decisions.
6. Diagnose moral hazard and adverse selection in employment, finance and contracting.
7. Evaluate signalling, screening and incentive-compatible mechanisms.
8. Use game-theoretic reasoning ethically in negotiations and competitive strategy.

1. The Strategic-Interaction Framework

A normal-form game specifies:

• Players: decision-makers such as firms, managers, employees or governments.
• Strategies: complete plans describing what a player will do in each relevant situation.
• Payoffs: benefits or costs associated with each combination of strategies.
• Information: what each player knows when making a decision.
• Timing: whether decisions are simultaneous or sequential.

For a two-player game, a payoff matrix records the outcome for each pair of actions. The first number in a cell is normally Player 1’s payoff and the second is Player 2’s payoff. Payoffs may represent profit, utility, market share, cost or a weighted managerial objective.

A strategy is strictly dominant if it gives a higher payoff than every alternative strategy, regardless of what the other player does. A weakly dominant strategy gives at least as high a payoff in every case and a strictly higher payoff in at least one case. Dominance is powerful because it does not depend on predicting the other player’s action.

A best response is a strategy that maximises a player’s payoff given the other player’s strategy. A Nash equilibrium is a strategy profile in which every player is choosing a best response to the strategies of the others. No player can gain by changing strategy unilaterally.

Nash equilibrium is a stability concept, not necessarily an efficient or fair outcome. A game can have several equilibria, no pure-strategy equilibrium, or an equilibrium that produces poor collective performance.

2. Dominant Strategies and Nash Equilibrium

Consider two competing firms choosing High or Low prices. Suppose their profits are represented by the following matrix:

For Firm A, Low yields 10 rather than 8 when B chooses High, and 4 rather than 2 when B chooses Low. Thus Low is strictly dominant for A. The same reasoning applies to B. The unique Nash equilibrium is (Low, Low), with payoffs (4, 4), even though both firms would prefer (High, High), which gives (8, 8). This is a version of the prisoners’ dilemma: individually rational behaviour produces an outcome that is jointly inferior.

The example illustrates the distinction between individual rationality and collective efficiency. Managers should not assume that competition will automatically produce cooperation, nor that cooperation is stable without enforcement, trust or repeated interaction.

Mixed Strategies

Some games have no equilibrium in pure strategies. In such cases, players may randomise. A mixed strategy assigns probabilities to available actions. In equilibrium, each player chooses probabilities that make the opponent indifferent among the opponent’s actions used with positive probability.

For a two-action game, if Player 1 chooses A with probability p and Player 2 chooses B with probability q, the expected payoff is:

Here, u_i(s) is Player i’s payoff in outcome s, and Pr(s) is the probability of that outcome. The formula applies when outcomes are discrete and probabilities sum to one.

Mixed strategies are relevant to product launches, sports tactics, cybersecurity and negotiations in which predictability can be exploited. Randomisation must be genuine and strategically justified; arbitrary inconsistency is not a mixed strategy equilibrium.

3. Sequential Games, Credibility and Commitment

In a sequential game, players move at different times and later players observe at least some earlier actions. These games are represented by game trees rather than only matrices. A player’s strategy must specify an action at every decision node where that player might act, including nodes that are not reached in equilibrium.

The principal solution method is backward induction: analyse the final decision first, determine the later player’s optimal response, and then use that response to analyse the earlier decision.

Suppose an incumbent firm chooses whether to accommodate or fight entry. An entrant first decides whether to enter. If entry occurs, the incumbent chooses Fight or Accommodate. Assume the entrant receives:

• Enter and face Fight: −2;
• Enter and face Accommodation: 3;
• Stay out: 0.

Suppose the incumbent receives 1 from accommodation, −1 from fighting, and 4 if entry is deterred. At the incumbent’s decision node, Accommodation is preferable to Fight because 1 > −1. Anticipating this, the entrant enters because 3 > 0. The subgame-perfect equilibrium is therefore Entry followed by Accommodation.

A threat to fight is non-credible when the player would not want to carry it out if the relevant decision node were reached. A subgame-perfect Nash equilibrium requires equilibrium behaviour in every proper subgame, eliminating such threats through backward induction.

Commitment

Commitment changes the future choice set or makes a future action costly to reverse. Examples include capacity investment, legally enforceable contracts, public price rules, escrow arrangements and delegated authority. A commitment works only when it is observable and credible. A public announcement alone may not constrain future behaviour and therefore may fail to deter an opponent.

Managers should distinguish:

Commitment can create advantage but may also reduce flexibility. Excess capacity, rigid contracts and automatic retaliation can produce inefficient escalation.

4. Repeated Games, Reputation and Cooperation

A repeated game consists of the same or related stage game played more than once. In an indefinitely repeated relationship, current actions affect future expectations and payoffs. Cooperation may be sustained even when one-shot incentives favour defection.

Let a player receive cooperative payoff C each period, deviation payoff D in the current period, and punishment payoff P thereafter. Let δ be the discount factor, where 0 ≤ δ < 1 measures the value placed on next period relative to the current period. Cooperation is sustainable under a grim-trigger strategy when:

Rearranging gives:

This condition assumes an infinite horizon, common knowledge of the stage game, observable actions and a credible punishment. It shows that cooperation is more likely when future payoffs matter greatly, deviation gains are modest, or punishment is severe but feasible.

In finite repeated games with a known final period, backward induction often undermines cooperation: if players defect in the final period, they may defect in the preceding period, and so on. However, uncertainty about the ending date, incomplete information, reputation and imperfect monitoring can change this result.

Repeated interaction supports:

• reciprocal concessions in negotiation;
• supplier development and long-term contracting;
• informal quality standards;
• reputation for reliable delivery;
• tacit coordination, subject to competition law.

It also creates risks. Punishment strategies can become excessive, errors can trigger retaliation, and tacit coordination may become illegal collusion. Managers must distinguish legitimate relationship management from agreements that unlawfully restrict competition.

5. Negotiation as Strategic Interaction

Negotiation is not simply the division of a fixed surplus. Parties may create value through trade-offs, information exchange and contingent agreements, while also competing over distribution.

A negotiator’s reservation value is the payoff from the best available alternative, often called the best alternative to a negotiated agreement. A negotiated outcome is individually rational only if each party receives at least its reservation value. The bargaining zone is the set of agreements acceptable to both sides.

Strategic variables include:

• the timing and size of offers;
• the disclosure of preferences;
• the use of deadlines;
• credible alternatives;
• issue linkage;
• contingent commitments;
• the cost of delay;
• reputation and future interaction.

Information can be private and strategically valuable. A party may conceal its reservation value, but false statements can damage trust and reputation. Ethical negotiation separates legitimate withholding from fraud, misrepresentation or exploitation of protected information.

A useful managerial practice is to map each party’s interests, outside options, possible responses and likely consequences. This converts vague bargaining into an explicit strategic model.

6. Auctions and Competitive Procurement

An auction is a mechanism that allocates an object or contract using bids. Important formats include:

• First-price sealed bid: the highest bidder wins and pays its own bid.
• Second-price sealed bid: the highest bidder wins and pays the second-highest bid.
• English auction: price rises until all but one bidder withdraw.
• Dutch auction: price falls until a bidder accepts.

In a private-value second-price auction, where each bidder knows its own valuation and values are independent, bidding one’s true valuation is a weakly dominant strategy. If valuation is v and the highest competing bid is b, bidding v wins and pays b when b < v, producing v − b ≥ 0; if b > v, losing is preferable to winning at a price above v. This result depends on private values, independent information, risk neutrality and truthful valuation of the object.

First-price auctions generally require bid shading: a bidder offers less than its valuation to retain surplus. The optimal bid depends on the number of bidders, beliefs about valuations, risk preferences and the distribution of private information. In common-value auctions, the object has approximately the same value to everyone but bidders possess different estimates. The winner’s curse occurs when winning signals that one’s estimate was overly optimistic.

For procurement, the lowest bid is not necessarily the lowest total cost. Managers should evaluate quality, delivery risk, lifecycle cost, financial capacity and contract compliance. Auction design should specify eligibility, evaluation criteria, reserve prices, disclosure rules, bid security and procedures for conflicts of interest.

7. Information Asymmetry and Principal–Agent Theory

Information is asymmetric when one party possesses relevant information that another party lacks. Principal–agent theory studies a relationship in which a principal delegates work or authority to an agent whose actions affect the principal’s outcome.

Examples include shareholders and executives, employers and employees, lenders and borrowers, governments and contractors, and buyers and suppliers. The principal typically wants the agent to take an action that is costly to the agent but valuable to the principal. Because effort or type may not be fully observable, contracts must allocate risk and create incentives.

A basic agency model can be written as:

where U_A is the agent’s utility, w is fixed compensation, b is performance-based compensation per unit of observed performance, e is effort, and c(e) is the agent’s effort cost. The principal’s expected payoff may be represented as:

where y is output and E denotes expected value. These expressions are simplified and assume risk-neutrality unless risk preferences are explicitly added.

Moral Hazard

Moral hazard arises after contracting when an agent’s action is difficult to observe and insurance or compensation weakens the consequences of that action. Examples include reduced employee effort under weak supervision, careless lending when losses are transferred to others, and excessive risk-taking when managers receive upside rewards but limited downside exposure.

Controls include:

• measurable performance standards;
• audits and monitoring;
• incentive pay balanced with risk controls;
• deferred compensation and clawbacks;
• separation of duties;
• employee ownership;
• professional norms and culture;
• termination or renewal provisions.

Incentive pay is not automatically efficient. If output is affected by random shocks, high-powered incentives may transfer excessive risk to risk-averse employees. If measurement is narrow, employees may game targets, neglect unmeasured tasks or manipulate data.

Adverse Selection

Adverse selection arises before contracting when one party has private information about type or quality. Examples include high-risk borrowers seeking insurance, low-quality suppliers competing with high-quality suppliers, and applicants whose productivity is difficult to observe.

Suppose high-quality suppliers incur cost c_H and low-quality suppliers incur c_L, with c_H > c_L, while buyers cannot directly observe quality. If both receive the same price, high-quality suppliers may exit when the price is below c_H. The market may then contain disproportionately low-quality suppliers. This is the classic “market for lemons” mechanism.

Solutions include screening, warranties, collateral, certification, references, probationary contracts, audits and self-selection menus. A screening contract offers different options so that types choose differently. For screening to work, the menu must satisfy:

• Individual rationality: each type prefers participation to its outside option.
• Incentive compatibility: each type prefers the option intended for it over the other options.

8. Signalling and Screening

Signalling occurs when the informed party takes an observable action to communicate private information. A signal is credible when it is more costly, less attractive or less easily imitated for low-quality types than for high-quality types.

Education may signal ability when high-ability individuals acquire qualifications at lower cost, although education may also build productivity. A warranty may signal product quality when poor-quality producers face higher expected repair costs. A costly bond may signal supplier reliability.

A separating equilibrium occurs when different types choose different signals and outsiders infer type from the signal. A pooling equilibrium occurs when different types choose the same signal and the signal conveys no distinction. Signals must be assessed carefully: an expensive signal may demonstrate commitment, but it may also reflect waste, privilege or regulatory barriers rather than quality.

Screening is designed by the uninformed party. Examples include diagnostic tests, insurance deductibles, probation periods, tiered pricing and requests for verifiable documentation. Effective screening uses information that is relevant to type and difficult to manipulate.

9. A Detailed Applied Case: Digital Procurement Platform

The following is an illustrative case based on common managerial situations; it is not a report of a specific company.

Background

A large manufacturer plans to award a three-year digital-maintenance contract. Two suppliers, Alpha and Beta, can bid. The manufacturer cannot directly observe long-term service quality. Alpha has a strong service record but higher operating costs. Beta has lower costs but an uncertain record. After contract award, the manufacturer will observe performance imperfectly, and switching suppliers will be costly.

Problem

A lowest-price auction may attract aggressive bids from both suppliers. However, a low-quality supplier may win, creating downtime and replacement costs. The manufacturer must choose an allocation mechanism and contract structure that address adverse selection before award and moral hazard after award.

Analysis

A simple sealed-bid auction based only on price encourages cost competition but ignores quality. Alpha may be unable to cover its higher cost at the expected price and could withdraw. This is adverse selection. After award, a fixed-price contract may encourage Beta to reduce preventive maintenance because effort is difficult to observe. This is moral hazard.

The manufacturer should also consider common-value uncertainty: both suppliers may be uncertain about the true complexity of integrating the platform. An unusually low bid may reflect overoptimism, creating a winner’s-curse risk.

Approach and Solution

The manufacturer adopts a two-stage process:

1. Prequalification using audited performance data, cybersecurity controls, financial capacity and reference checks.
2. Best-value evaluation using a transparent weighted score for price, service quality, resilience and lifecycle cost.

The contract includes:

• service-level agreements with measurable uptime and response standards;
• payment linked partly to verified performance;
• independent audits;
• data-access and transition provisions;
• a limited bonus for exceptional performance;
• penalties for avoidable failures;
• a renewal option contingent on performance;
• a contingency plan to avoid dependence on one supplier.

The manufacturer avoids basing all compensation on a single uptime measure, because outages may be caused by external events. It uses a balanced scorecard and defines force-majeure conditions in advance.

Outcome

The mechanism reduces the probability that an unsuitable supplier wins solely through a low bid. Monitoring and balanced incentives reduce avoidable underinvestment in maintenance. The manufacturer retains flexibility through transition rights, while suppliers receive clearer expectations and a fairer opportunity to demonstrate capability.

Learning Points

1. Auction design is part of strategy, not merely an administrative procedure.
2. Quality information should be gathered before contract award.
3. Performance measures must distinguish effort from uncontrollable shocks.
4. Commitment and exit rights should be designed together.
5. Transparency improves legitimacy and reduces corruption risk.

10. Limitations, Ethics and Governance

Game-theoretic models simplify reality. Payoffs may be difficult to measure; players may have bounded rationality; preferences may include fairness, identity and ethics; information may be noisy; and organisational culture may influence behaviour. Equilibrium predictions can therefore be sensitive to assumptions.

Managers should avoid treating equilibrium as a moral recommendation. A stable outcome can be exploitative, discriminatory or socially harmful. Strategic use of private information must comply with fiduciary duties, consumer-protection law, competition law, privacy requirements, procurement rules and employment regulation.

Important governance principles include:

• do not misrepresent material facts;
• disclose conflicts of interest;
• protect confidential and personal information;
• document auction and procurement criteria;
• audit algorithmic scoring for bias;
• ensure human review of high-impact decisions;
• avoid retaliation and coercive lock-in;
• design incentives that do not reward unsafe or unlawful conduct.

Technology and artificial intelligence can improve forecasting, monitoring and bid evaluation, but they also create strategic risks. If competitors use similar algorithms, behaviour may become correlated. Automated pricing can facilitate tacit coordination, while opaque models may disadvantage suppliers with limited data. AI-supported decisions should therefore include governance, explainability, data-quality controls and escalation procedures.

11. Best Practices for Managers

A disciplined strategic analysis should:

1. Identify all relevant players, including regulators, customers and affected stakeholders.
2. Separate actions, types, beliefs and observed signals.
3. Specify timing and information at each decision point.
4. Construct payoffs that include direct, indirect and reputational consequences.
5. Check for dominant strategies and best responses.
6. In sequential settings, use backward induction and test credibility.
7. Consider repeated interaction, learning and reputation.
8. Test whether contracts create adverse selection or moral hazard.
9. Use screening and signalling only when signals are relevant and ethically defensible.
10. Stress-test outcomes under alternative assumptions.
11. Protect flexibility when uncertainty is high.
12. Monitor unintended consequences and revise mechanisms over time.

Chapter Summary

Strategic interaction occurs when one decision-maker’s outcome depends on the choices of others. Game theory models such situations through players, strategies, payoffs, information and timing. Dominant strategies provide robust choices; best responses identify optimal reactions; and Nash equilibrium describes mutually consistent strategies, though not necessarily efficient outcomes.

Sequential games require backward induction and attention to credible commitments. Repeated games can support cooperation when future consequences matter, actions are observable and punishment is credible. Negotiations depend on reservation values, alternatives, information and the structure of the bargaining zone.

Auction outcomes depend on auction format, valuation structure, risk preferences and information. In principal–agent relationships, moral hazard arises from hidden action after contracting, while adverse selection arises from hidden type before contracting. Monitoring, incentive design, screening, certification and signalling can reduce these problems, but each creates costs and possible unintended effects.

Effective managers use game theory as a reasoning tool rather than as a mechanical prediction device. They combine quantitative analysis with ethical judgment, legal compliance, organisational knowledge and sensitivity to uncertainty.

Key Terms

Adverse selection: Hidden information about type or quality before a contract is made.

Best response: A strategy that maximises payoff given other players’ strategies.

Commitment: An action or arrangement that credibly constrains future choices.

Dominant strategy: A strategy that performs at least as well as alternatives regardless of others’ actions.

Game theory: The systematic study of interdependent decision-making.

Grim-trigger strategy: Cooperation continues until a deviation occurs, after which punishment is imposed permanently.

Information asymmetry: A situation in which parties possess unequal relevant information.

Moral hazard: Hidden action or reduced care after contracting because consequences are weakened.

Nash equilibrium: A strategy profile in which no player benefits from unilateral deviation.

Pooling equilibrium: An equilibrium in which different types choose the same observable action.

Principal–agent problem: Conflict arising when a principal delegates work to an agent whose actions or information are imperfectly observed.

Reservation value: Payoff available from the best alternative to an agreement.

Screening: Actions by an uninformed party to induce informed types to reveal themselves.

Separating equilibrium: An equilibrium in which different types choose different observable actions.

Signalling: An informed party’s observable action intended to communicate private information.

Subgame-perfect equilibrium: A Nash equilibrium that remains optimal in every subgame.

Winner’s curse: The risk that winning a common-value auction indicates an overly optimistic estimate.

Mixed strategy: A probability distribution over available actions.

Review Questions

Short Answer Questions

1. What distinguishes a dominant strategy from a best response?
2. Why is a Nash equilibrium not necessarily socially efficient?
3. What makes a threat credible in a sequential game?
4. Distinguish moral hazard from adverse selection.
5. What is the winner’s curse in a common-value auction?

Long Answer Questions

1. Explain how backward induction identifies a subgame-perfect equilibrium. Use an entry-deterrence example.
2. Discuss the conditions under which cooperation can be sustained in a repeated game.
3. Compare first-price and second-price auctions, including the assumptions required for truthful bidding in a second-price auction.
4. Explain how incentive contracts can reduce moral hazard while creating new risks for employees and organisations.
5. Analyse the role of signalling and screening in labour, insurance or procurement markets.

Analytical / Application Questions

1. Two firms independently choose High or Low output. Each firm earns 12 when both choose Low, 8 when one chooses High while the other chooses Low, and 4 when both choose High. The firm choosing High when the other chooses Low earns 16. Identify each firm’s dominant strategy and the Nash equilibrium. What strategic or regulatory factors might support a different outcome?
2. A buyer cannot observe supplier quality before contracting and cannot directly observe maintenance effort afterward. Design a procurement mechanism that addresses both adverse selection and moral hazard. Explain the trade-offs in your proposed screening, monitoring and incentive arrangements.
3. In a repeated supplier relationship, cooperation gives each party 10 per period, unilateral deviation gives the deviator 16 immediately, and punishment gives each party 4 per period thereafter. Using δ ≥ (D − C)/(D − P), calculate the minimum discount factor required to sustain cooperation under a grim-trigger strategy. State the assumptions behind your calculation.

Multiple Choice Questions with Answers

Question 1

Which statement best defines a Nash equilibrium?

A. An outcome that maximises total industry profit

B. A strategy profile in which no player benefits from unilateral deviation

C. A strategy that is strictly better for every player

D. An outcome chosen only when players cooperate

Correct Answer: B. A strategy profile in which no player benefits from unilateral deviation

Question 2

In a sequential game, backward induction begins by analysing:

A. The first player’s preferences

B. The average payoff across all outcomes

C. The final decision node

D. The least probable strategy

Correct Answer: C. The final decision node

Question 3

A threat is non-credible when:

A. It is publicly announced

B. It would not be optimal to carry out if the relevant situation occurred

C. It imposes a cost on the opponent

D. It is included in a written contract

Correct Answer: B. It would not be optimal to carry out if the relevant situation occurred

Question 4

Under the standard private-value assumptions of a second-price auction, a bidder’s weakly dominant strategy is to:

A. Bid zero

B. Bid above valuation

C. Bid exactly its valuation

D. Bid the average market price

Correct Answer: C. Bid exactly its valuation

Question 5

Moral hazard primarily concerns:

A. Hidden action after contracting

B. Hidden type before contracting

C. A lack of competitors

D. A government price ceiling

Correct Answer: A. Hidden action after contracting

Question 6

Adverse selection occurs primarily because:

A. All parties have identical information

B. One party has private information about type before contracting

C. Employees receive performance bonuses

D. Repeated interaction creates reputation

Correct Answer: B. One party has private information about type before contracting

Question 7

A separating equilibrium is one in which:

A. All types choose the same signal

B. No player has a best response

C. Different types choose different observable signals

D. Players randomise equally between all actions

Correct Answer: C. Different types choose different observable signals

Question 8

In the repeated-game formula δ ≥ (D − C)/(D − P), δ represents:

A. The current deviation payoff

B. The discount factor applied to future payoffs

C. The punishment payoff

D. The cost of effort

Correct Answer: B. The discount factor applied to future payoffs

Question 9

If cooperation gives 10, deviation gives 16, and punishment gives 4, the minimum discount factor under grim trigger is:

A. 0.25

B. 0.40

C. 0.60

D. 0.75

Correct Answer: C. 0.60

Question 10

Which practice most directly addresses both supplier adverse selection and post-contract moral hazard?

A. Selecting solely on the lowest initial bid

B. Using prequalification and verified performance-based monitoring

C. Removing all audit rights

D. Offering an unconditional fixed payment

Correct Answer: B. Using prequalification and verified performance-based monitoring